Cell of origin affects tumour development and phenotype in pancreatic ductal adenocarcinoma

Abstract

Objective: Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive tumour thought to arise from ductal cells via pancreatic intraepithelial neoplasia (PanIN) precursor lesions. Modelling of different genetic events in mice suggests both ductal and acinar cells can give rise to PDAC. However, the impact of cellular context alone on tumour development and phenotype is unknown.

Design: We examined the contribution of cellular origin to PDAC development by inducing PDAC-associated mutations, Kras^G12D expression and Trp53 loss, specifically in ductal cells (Sox9CreER;Kras^LSL-G12D;Trp53^flox/flox ('Duct:KP^cKO ')) or acinar cells (Ptf1a^CreER;Kras^LSL-G12D;Trp53^flox/flox ('Acinar:KP^cKO ')) in mice. We then performed a thorough analysis of the resulting histopathological changes.

Results: Both mouse models developed PDAC, but Duct:KP^cKO mice developed PDAC earlier than Acinar:KP^cKO mice. Tumour development was more rapid and associated with high-grade murine PanIN (mPanIN) lesions in Duct:KP^cKO mice. In contrast, Acinar:KP^cKO mice exhibited widespread metaplasia and low-grade as well as high-grade mPanINs with delayed progression to PDAC. Acinar-cell-derived tumours also had a higher prevalence of mucinous glandular features reminiscent of early mPanIN lesions.

Conclusion: These findings indicate that ductal cells are primed to form carcinoma in situ that become invasive PDAC in the presence of oncogenic Kras and Trp53 deletion, while acinar cells with the same mutations appear to require a prolonged period of transition or reprogramming to initiate PDAC. Our findings illustrate that PDAC can develop in multiple ways and the cellular context in which mutations are acquired has significant impact on precursor lesion initiation, disease progression and tumour phenotype.

Keywords: lineage tracing; pancreatic cancer; tumor development; tumor heterogeneity.

Figure 1.. Kras^G12D-expressing ductal cells form tumors earlier than acinar cells in the absence of Trp53.

(A) Schematic of the alleles in the Sox9CreER;Kras^LSL-G12D;Trp53^flox/flox;R26R^YFP (Duct:KP^cKO) and Ptf1a^CreER;Kras^LSL-G12D;Trp53^flox/flox;R26R^YFP (Acinar:KP^cKO) mouse models used in this study. Tamoxifen (TM) injection induces Cre-mediated DNA recombination and results in expression of oncogenic Kras^G12D from the Kras^LSL-G12D (“K”) allele and the YFP lineage label from the R26R^YFP allele. In addition, TM injection induces the deletion of exons 2–10 from the Trp53^flox (“P^cKO”) allele and loss of p53 expression. (B) Duct:KP^cKO (n=19) and Acinar:KP^cKO (n=9) mice were injected three times on alternating days with TM beginning at three-four weeks of age. The mice were monitored until they reached their humane endpoint to determine survival duration. (C) The median disease specific survival of Duct:KP^cKO and Acinar:KP^cKO mice (82 vs. 128 days, p<0.0001). Mice euthanized due to non-pancreatic reasons were censored (hash marks). (D) Representative gross anatomical photographs of the mouse abdomen from Sox9CreER;Trp53^flox/flox;R26R^YFP (Control), Duct:KP^cKO, and Acinar:KP^cKO mice. White or black dashed lines outline either normal parenchyma or tumors, respectively. p.i., post tamoxifen injection. Scale bars: 5 mm.

Figure 2.. Duct:KP^cKO and Acinar:KP^cKO mice develop PDAC.

Representative whole section (A) and high-magnification (B) images of Sox9CreER;Trp53^flox/flox;R26R^YFP (Control), Duct:KP^cKO, and Acinar:KP^cKO pancreata stained with hematoxylin and eosin (H&E). Tumors are outlined with dashed lines in (A). (B) H&E staining shows that PDAC in Duct:KP^cKO and Acinar:KP^cKO mice are predominantly moderately-to-poorly differentiated. (C) Immunohistochemistry for ductal cell marker, Cytokeratin 19, and the YFP lineage marker demonstrates that duct-like tumors in Duct:KP^cKO and Acinar:KP^cKO mice arise from ductal and acinar cells, respectively. (D) Low- (top row) and high-(bottom row) magnification images of H&E staining of the different histological tumor phenotypes observed in Duct:KP^cKO and Acinar:KP^cKO mice. The small-gland phenotype was observed in both mouse models. The mucinous-gland and large-gland phenotypes were found more often in Acinar:KP^cKO mice and rarely, if ever, in Duct:KP^cKO mice. Scale bars: 5 mm (A), 100 μm (B), 200 μm (D, top), and 50 μm (D, bottom and C).

Figure 3.. Tumor burden in Duct:KP^cKO and Acinar:KP^cKO mice at humane endpoint is similar, but reaches the peak amount earlier in Duct:KP^cKO mice.

Quantification of the pancreatic area displaced by tumor area in individual Duct:KP^cKO (n=8) and Acinar:KP^cKO (n=8) mice at their humane endpoint (66.0% ± 8.7% vs. 60.0% ± 9.7%, p=0.4) plotted against time post tamoxifen injection (p.i.). All values shown as mean ± SEM.

Figure 4.. Tumors arise earlier in Duct:KP^cKO compared to Acinar:KP^cKO mice.

(A) Schematic describing the experimental design. Duct:KP^cKO and Acinar:KP^cKO mice (n=4) were injected with tamoxifen at 3–4 weeks of age and euthanized at 2, 4, 6 and 8 weeks post injection (p.i.) or 4, 6, 8, 12, and 16 weeks p.i. for Duct:KP^cKO or Acinar:KP^cKO mice, respectively. Quantification of the number of tumors present (B) and the cross-sectional diameter of each tumor (C) in these Duct:KP^cKO and Acinar:KP^cKO mice revealed that tumors initiated earlier from ductal compared to acinar cells. Trend lines in the graph indicate significant correlations between time and tumor number (B) or size (C). All values shown as mean ± SEM. mm, milimeters. **, p<0.01 and ***, p<0.001.

Figure 5.. Kras^G12D expression and loss of p53 induces a spectrum of mPanIN lesions from acinar cells, but predominately high-grade mPanIN from ductal cells.

(A) Representative images of hematoxylin and eosin stained normal duct (arrowhead), mPanIN1, 2 or 3 lesions (arrows) found in Acinar:KP^cKO and Duct:KP^cKO mice. No mPanIN1 lesions were observed in Duct:KP^cKO mice. (B) Quantification of the average number of mPanIN lesions of each grade present per section per mouse at the indicated time points post tamoxifen injection (p.i.) in Duct:KP^cKO and Acinar:KP^cKO mice. The number of mPanIN1 in the Acinar:KP^cKO line at 16 weeks post injection was set to 1 and the other circles represent the fraction of mPanIN present per time point or grade in Duct:KP^cKO and Acinar:KP^cKO mice relative to that sample. Immunohistochemistry for proliferation marker Ki67 (C) and quantification of the Ki67⁺ cells per mPanIN3 (D) in Duct:KP^cKO and Acinar:KP^cKO mice. Scale bar: 50 μm (A), 100 μm (C).

Figure 6.. Glandular areas containing gastric mucin expression are more prevalent in tumors from Acinar:KP^cKO compared to Duct:KP^cKO mice.

Representative images of tumor areas from Duct:KP^cKO and Acinar:KP^cKO mice stained with hematoxylin and eosin (H&E)(A-B, left panels) or Alcian blue (AB) (A, right panels) or Mucin 5AC (Muc5AC or M5AC) (B, right panels). Light Alcian blue staining in the stroma or Muc5AC staining in blood vessels or blood cells was classified as negative. The Muc5AC staining is likely an artifact of the anti-mouse secondary antibody. Quantification of the percent of AB (C) or Muc5AC (D) positive tumor area in Duct:KP^cKO and Acinar:KP^cKO mice (AB: 20.7% vs 2%, respectively, p<0.001 (***) and Muc5AC: 5.3% vs. 1.4%, respectively, p<0.05 (*)). (E) Representative images of tumor areas from Duct:KP^cKO and Acinar:KP^cKO mice stained with Cytokeratin 20 (CK20). Quantification of the percent of CK20 positive tumor area in Duct:KP^cKO and Acinar:KP^cKO mice (p<0.0001 (****)). All values shown as mean ± SEM. Scale bar: 100 μm (A-B).