Tuft cells transdifferentiate to neural-like progenitor cells in the progression of pancreatic cancer

Abstract

Pancreatic ductal adenocarcinoma (PDA) is partly initiated through the transdifferentiation of acinar cells to metaplastic ducts that act as precursors of neoplasia and cancer. Tuft cells are solitary chemosensory cells not found in the normal pancreas but arise in metaplasia and neoplasia, diminishing as neoplastic lesions progress to carcinoma. Metaplastic tuft cells (mTCs) function to suppress tumor progression through communication with the tumor microenvironment, but their fate during progression is unknown. To determine the fate of mTCs during PDA progression, we have created a lineage tracing model that uses a tamoxifen-inducible tuft-cell specific Pou2f3^CreERT/+ driver to induce transgene expression, including the lineage tracer tdTomato or the oncogene Myc. mTC lineage trace models of pancreatic neoplasia and carcinoma were used to follow mTC fate. We found that mTCs, in the carcinoma model, transdifferentiate into neural-like progenitor cells (NRPs), a cell type associated with poor survival in PDA patients. Using conditional knock-out and overexpression systems, we found that Myc activity in mTCs is necessary and sufficient to induce this Tuft-to-Neuroendocrine-Transition (TNT).

Figure 1.. Lineage trace model shows tuft cell specific expression of tdTomato.

Genetic strategy to trace the fate of metaplastic tuft cells (mTCs) in pancreatic cancer progression using a dual recombinase neoplasia (KF-P2f3T) model (A) and carcinoma (KPF-P2f3T) model (B). C) Treatment regimen for neoplasia and carcinoma Models. The models have both pancreatic progression and reporter specifically in tuft cells. The KF-P2f3T model utilizes Cerulein once daily, followed by 2 weeks recovery, then once daily tamoxifen treatment for 5 days. The KPF-P2f3T model is aggressive as it contains one null p53 allele. Cerulein is not necessary and so only Tamoxifen is treated once daily for 5 days. Mice are allowed to age for different time points to determine lineage of metaplastic tuft cells in progression of pancreatic cancer. D) Dclk1 in Brown identifies tuft cells in neoplasia model (left) and carcinoma model (right). E) Specific expression of tdTomato reporter in both models. White represents Hoechst staining to mark the nucleus, tdTomato in red to show recombination of tdTomato (tdTom), Pou2f3, the master regulator of tuft cells in Green, and Acetylated alpha tubulin in Magenta to mark filamentous actin in the tufts of the mTCs. F) Dual IHC of tdTom (Teal) and CC3 (Brown) stained KF-P2f3T and KPF-P2f3T mice to determine cell death of mTCs in neoplasia and carcinoma. Black arrows indicate tdTom+ cells in lesions. Scale bars = 10 μm unless otherwise noted.

Figure 2.. Lineage tracing mTCs reveals transdifferentiation to metaplastic neuroendocrine cells (mNECs).

Dual recombinase models show fate of mTCs in progression of pancreatic cancer. A) Mice harvested 3 days post-tamoxifen treatment for labeling mTCs. Hoechst stains nucleus of cells (White), Cox1 is used to identify mTCs (Green), tdTom reporter to determine fate of mTCs (Red) and acetylated-α-tubulin (Magenta) marks filamentous actin on tuft of mTCs. B) Quantification of Percentage of mTC (White) and mTC Progeny (Pink) among ductal cells at different time points post-tamoxifen treatment in KF-P2f3T and KPF-P2f3T models. mTCs are identified by Cox1 and tdTom, mTC Progeny are only tdTom positive. (n = 3/group) p-values were calculated using a two-way ANOVA on Prism GraphPad 7. * = p < 0.05, ** = p < 0.01. C) Co-staining of tdTomato (Red) expression with remaining metaplastic ductal cell subsets. In Green, Gkn1, Lectin GS-II, and Synaptophysin (Syp) mark gastric pit-like, gastric neck-like, and neuroendocrine cells respectively in pancreatic metaplastic ducts. D) 7 weeks post-tamoxifen labeling of neoplasia and carcinoma models exposes shift of populations. Cox1 (Green) marks mTCs, tdTomato (Red) identifies reporter, Syp (White) marks neuroendocrine cells, and Blue represents DAPI Staining to identify the nucleus of a cell. E) Quantification of percentage of mTCs (Cox1+ tdTom+) in White, mTC-derived mNECs (Syp+ tdTom+) in Pink, mTC-independent mNECs (Syp+ tdTom-) in Green. (n = 3/group). p-values were calculated using a two-way ANOVA on Prism GraphPad 7. * = p < 0.05, ** = p < 0.01. Scale bars = 10 μm unless otherwise noted.

Figure 3.. Characterizing the tuft cell to neuroendocrine cell transdifferentiation (TNT).

IF staining of Cox1 (Red) and Syp (Green) in neoplasia and carcinoma models determines that mNECs in the KPF-P2f3T model relies on the mTC population. A) Cox1 (Red) and Syp (Green) markers are stained in Pou2f3 double knock-in models of pancreatic neoplasia and carcinoma: (Left) KF; Pou2f3^{CreERT/CreERT} and (Right) KPF; Pou2f3^{CreERT/CreERT}, correspondingly to investigate the role of tuft cells in mNEC genesis. Green Arrows denote mNECs, Red Asterisks denote stromal Cox1⁺ staining identifying Cox1⁺ Immune cells. B) Pou2f3, the master regulator of tuft cells (Cyan) is stained with Cox1, marker of tuft cells (Red), and Syp, neuroendocrine marker (Green) to determine if mNECs also express Pou2f3. C) Dual IHC of Syp (Teal) and Ki67 (Brown) of neoplasia and carcinoma models of pancreas to determine proliferation of mNECs. D) Co-IF of mTC marker, Cox1 (Green), lineage tracer, tdTom (Red), mNEC marker, Syp (White), incorporated EdU (Magenta) to mark proliferating cells, and nuclear marker, Hoechst (Blue) in KPF-P2f3T mice treated with EdU over the course of 3 days post-tamoxifen treatment. E) Yellow arrows mark rare transitionary cells that are marked with both Syp (Green) and Cox1 (Red). Green arrow identifies cells that are only expressing Syp, mNECs. Red arrow marks cells only expressing Cox1, mTCs. Scale bars = 10 μm unless otherwise noted.

Figure 4.. Identifying distinctions between mNECs in the progression of PDA.

Neoplasia and carcinoma PDA models were stained with markers of different islet compartments to determine differences between mNECs between the KF-P2f3T and KPF-P2f3T models. B) Islets and lesions from KF-P2f3T and KPF-P2f3T are shown to determine co-localization of Syp (Red) and markers of the different compartments of the Islet (Green). PanCK (Cyan) was used to distinguish islet cells from mNECs, and Hoechst was used to mark the nucleus of each cell. Glucagon (Gcg), Insulin (Ins), Somatostatin (Sst), Ghrelin (Ghrl), and Pancreatic Polypeptide (Pp) were stained to identify the α cells, β cells, δ cells, ε cells and γ cells respectively. Scale bars = 10 μm unless otherwise noted.

Figure 5.. mTC-derived mNECs are identified in PDA Carcinoma.

IHC and IF are used to confirm the identity of mTC Progeny in dedifferentiated carcinoma of KPF-P2f3T model. A) IHC of dedifferentiated carcinoma stained with Dclk1 (Left) and tdTomato (Right) in Brown. B) Dedifferentiated carcinoma of KPF-P2f3T model is stained with Hoechst (Cyan), Syp (White), and tdTomato (Red) to determine identity of mTC progeny. Scale bars = 50 μm

Figure 6.. Identification of mTC-derived Nrxn3+ NRPs in Carcinoma

Co-staining of Syp (Red), Nrxn3 (Green), PanCK (Magenta), and Hoechst (White) in KF-P2f3T neoplasia model and KPF-P2f3T carcinoma model at A) 3 days post-tamoxifen, B) 7 weeks post-tamoxifen, and C) KPF-P2f3T dedifferentiated carcinoma. Scale bars = 10 μm unless otherwise noted.

Figure 7.. cMyc Mediated development of mNECs.

Overexpression and knockdown of Myc in neoplasia and carcinoma models show dependence of Myc expression on transition of mTCs to NRPs in PDA progression. A) KF-MOE 7 weeks post-tamoxifen treatment are stained with tdTom to identify cells recombined by tamoxifen treatment and cells overexpressing Myc (Red), Syp to mark neuroendocrine cells (Green), Acetylated-α-tubulin to mark filamentous actin of cells (Magenta), and Hoechst to identify nucleus of all cells (White) to detect mTC-derived mNECs. B) KF-MOE 7 weeks post-tamoxifen treatment tissue is stained for co-IF to determine expression of Syp (Green), tdTom (Red), Nrxn3 (Magenta), and Hoechst (White). C) Quantification of mTCs (Cox1+ tdTom+) in White, mTC derived mNECs (Syp+ tdTom+) in Pink and mTC independent mNECs (Syp+ tdTom-) in Green for KF-P2f3T and KF-MOE models. Mice were sac’d after 3 days and 7 weeks post tamoxifen treatment to label tuft cells with tdTomato and overexpress cMyc specifically in tuft cells. (n = 3/group) p-values were calculated using a two-way ANOVA on Prism GraphPad 7. * = p < 0.05. D) Quantification of mTCs (White), mTC Derived mNECs (Pink), and mTC Independent, (Green) in KPF-P2f3T and KPF-P2M^Fl/+ models. Mice were sac’d at 7 weeks post tamoxifen treatment to label tuft cells with tdTom and flox cMyc expression in tuft cells specifically. (n = 3/group) p-values were calculated using a two-way ANOVA on Prism GraphPad 7. * = p < 0.05, **= p<0.005. E) Co-staining of Cox1 (Green), tdTom (Red), PanCK (Magenta), and Hoechst (White) in KPF-P2M^Fl/+ at 7 weeks post-tamoxifen treatment to identify co-localization of mTC marker, Cox1, with tdTom (Top). Co-staining of combined antibodies for Ghrl, Gcg, Sst, Ppy, and Ins to mark cells expressing “Hormones” to identify EECs (Green), tdTom used to identify cells recombined and Myc knockout cells (Red), PanCK to mark pancreatic epithelium/lesions (Magenta), and Hoechst to mark nucleus of cells (White) in KPF-P2M^Fl/+ at 7 weeks post-tamoxifen treatment to identify EECs and tdTom+ cells as separate cell populations when Myc is conditionally knocked out in carcinoma model. Scale bars = 10 μm unless otherwise noted.