Characterization of primary cilia and Hedgehog signaling during development of the human pancreas and in human pancreatic duct cancer cell lines

Authors


Abstract

Hedgehog (Hh) signaling controls pancreatic development and homeostasis; aberrant Hh signaling is associated with several pancreatic diseases. Here we investigated the link between Hh signaling and primary cilia in the human developing pancreatic ducts and in cultures of human pancreatic duct adenocarcinoma cell lines, PANC-1 and CFPAC-1. We show that the onset of Hh signaling from human embryogenesis to fetal development is associated with accumulation of Hh signaling components Smo and Gli2 in duct primary cilia and a reduction of Gli3 in the duct epithelium. Smo, Ptc, and Gli2 localized to primary cilia of PANC-1 and CFPAC-1 cells, which may maintain high levels of nonstimulated Hh pathway activity. These findings indicate that primary cilia are involved in pancreatic development and postnatal tissue homeostasis. Developmental Dynamics 237:2039–2052, 2008. © 2008 Wiley-Liss, Inc.

INTRODUCTION

Primary cilia are microtubule-based organelles that project into the extracellular environment from the mother centriole of most quiescent cells in the human body. Recent research has demonstrated that cilia function as sensory units that coordinate a variety of signal transduction pathways essential for embryonic and postnatal development as well as tissue homeostasis in the adult body (Christensen et al.,2007; Eggenschwiler and Anderson,2007; Kiprilov et al.,2008). These pathways include Hedgehog (Hh), Wnt, and platelet-derived growth factor receptor (PDGFR) signaling as well as Ca2+-signaling by means of TRP ion channels. Furthermore, primary cilia may control behavioral responses, such as occurs in the central nervous system where neuronal cilia function in a pathway that controls satiety responses (Davenport et al.,2007). Consequently, defects in ciliary assembly lead to a plethora of diseases and disorders, including kidney, liver and pancreatic cysts, blindness, obesity, diabetes, developmental disorders, and other human diseases now collectively referred to as ciliopathies (Badano et al.,2006; Satir and Christensen,2007; Yoder,2007; Fliegauf et al.,2007).

Primary cilia on the epithelium of renal tubules control tissue homeostasis probably by acting as mechanosensors that monitor the composition and flow rate of urine in the nephron (Praetorius and Spring,2003; Yoder,2007). Obstruction of this flow-sensing response in, for example, the IFT88Tg737Rpw mouse (hereafter referred to as Tg737orpk), causes disruption of tissue organization and cyst formation. The Tg737orpk mouse has a mutation in the Tg737/IFT88 gene that encodes a subunit of the IFT particle complex B required for ciliary assembly (Pazour et al.,2000; Yoder et al.,2002a; Rosenbaum and Witman,2002; Pedersen et al.,2008). Among the ciliary signaling modules involved in renal tissue homeostasis are the TRP ion channel polycystin 2 (PC-2) and polycystin-1 (PC-1), which are disrupted in human ADPKD (Yoder et al.,2002b; Pazour et al.,2002; Nauli et al.,2003). Bending of the ciliary axoneme due to fluid movement has been shown to induce a Ca2+-response (Praetorius and Spring,2001), which is dependent on PC-1 and PC-2. This calcium response is thought to be important in detecting fluid movement for maintaining tissue function, because mutations leading to its loss result in cyst development (Delmas et al.,2004; Liu et al.,2005b; Köttgen,2007; Weimbs,2007; Yoder,2007). In addition, in the absence of flow PC-1 may be proteolytically processed (Chauvet et al.,2004), leading to the nuclear translocation of the transcription factor STAT6 and co-activator P100 (Low et al.,2006). Furthermore, defects in ciliary assembly may result in up-regulation of the canonical Wnt pathway and impair the ability of noncanonical Wnt signals to suppress the canonical pathway (Gerdes et al.,2007; Corbit et al.,2008). Altered regulation of the Wnt signaling pathway in the kidney has been associated with uncontrolled cell proliferation and differentiation and was found to result in cystic kidney disease (Cano et al.,2004; Merkel et al.,2007).

In addition to kidney defects, the loss of primary cilia in the Tg737orpk mouse causes a series of abnormalities in the pancreas, including extensive cyst formation in ducts (Cano et al.,2004; Zhang et al.,2005). This may indicate a functional similarity between cilia in kidney and pancreatic duct systems. Cells of both exocrine and endocrine systems in the pancreas possess primary cilia, including islet cells and the ducts, but apparently not the acini (Kodama,1983; Ashizawa et al.,1997; Cano et al.,2004,2006; Zhang et al.,2005). Pancreas abnormalities in the Tg737orpk mouse begin with dilations of the pancreas ducts in late gestation, which after birth are accompanied by extensive formation of large, interconnected cysts as well as apoptosis and vacuolization of acini. These changes are reminiscent of chronic pancreatitis, supporting the speculation that primary cilia of ducts play an essential role in the development of the pancreas (Cano et al.,2004; Zhang et al.,2005). In the dilated ducts and cysts, PC-2 is mislocalized to intracellular compartments, the cytosolic localization of β-catenin is increased and there is an increased expression of Tcf/Lef transcription factors. This finding suggests that, as in the kidney, there is an alteration in the Wnt signaling pathway (Cano et al.,2004).

Another signaling pathway essential for pancreatic development and tissue homeostasis is the Hedgehog (Hh) pathway. Hh regulates cell proliferation and differentiation in numerous embryonic tissues and Shh is expressed in many regions of the embryo where it functions as a key organizer of tissue morphogenesis (Odent et al.,1999). One embryonic tissue where the initial absence of Shh signaling is required for development is in the pancreatic anlage, because ectopic expression of Shh leads to transformation of pancreatic mesoderm into intestinal mesenchyme (Apelqvist et al.,1997). In addition to being an important inductive signal in embryonic development, Hh signaling is required in homeostasis of mature tissues and is also implicated in many human cancers, including endodermal derived carcinomas of the esophagus, stomach, and pancreas (Beachy et al.,2004) as well as neurodegenerative disorders (Bak et al.,2003). Although the mechanism is not completely understood, it has been shown that Hh signaling is coordinated by the primary cilium (e.g., Huangfu et al.,2003; Corbit et al.,2005; Huangfu and Anderson,2005; May et al.,2005; Haycraft et al.,2005; Liu et al.,2005a; Koyama et al.,2007; Vierkotten et al.,2007; Ruiz-Perez et al.,2007; Caspary et al.,2007; Rohatgi et al.,2007; Kiprilov et al.,2008). Regulation of the Hh pathway is complex with Hh responsive Gli transcription factors having either activator or repressor functions that depend on IFT proteins. Binding of Hh to its receptor Patched (Ptc) occurs in the cilium and results in translocation of Ptc out of the cilium and into the cell body (Rohatgi et al.,2007; Kiprilov et al.,2008). In contrast, Smoothened (Smo), the transmembrane effector of the pathway whose activity is suppressed by Ptc is targeted to the cilium in the presence of Hh. This relieves the inhibitory effects of Ptc on Smo and controls the activation or deactivation of Gli transcription factors. In the absence of IFT or cilia, the repressor and activator functions of the Gli transcriptional regulators become deregulated with the resulting phenotype in a tissue being determined by whether it is the activator or repressor function of the Gli transcription factors that is most critical. For example, the processing of full-length Gli3 (Gli3FL) to repressor form (Gli3R) is affected in IFT mutants, showing the importance of functional IFT for regulating Gli transcription factors.

In the developing and mature pancreas, deregulated Hh signaling results in a series of diseases, including annular pancreas, diabetes mellitus, chronic pancreatitis and pancreatic cancer (Lau et al.,2006). During the early mouse embryonic stages, Hh signaling is kept at a low level in the pancreatic anlage to ensure the correct establishment of organ boundaries, as well as tissue architecture. At later developmental stages, Hh signaling is activated to promote proliferation and maturation of the tissue (Kawahira et al.,2005; Lau et al.,2006; Cano et al.,2007; van den Brink,2007). In the mature pancreas, Hh signaling maintains endocrine functions, and aberrant Hh signaling causes pancreatic ductal adenocarcinoma and chronic pancreatitis (Thayer et al.,2003; Berman et al.,2003; Kayed et al.,2003,2006; Morton et al.,2006; Cano et al.,2007).

The aim of our study was to examine the expression of primary cilia in human pancreatic ducts and to investigate the sensory competence of primary cilia with regards to the Hh signaling pathway during pancreatic development and in pancreatic adenocarcinoma cell lines, CFPAC-1 and PANC-1, which are isolated from metastatic and primary tumors, respectively (Lieber et al.,1975; Schoumacher et al.,1990). Initially, we show that localization of Hh signaling components is regulated during human pancreatic development with Gli2 and Smo accumulating in the cilium at later stages while the level of nuclear and cytosolic Gli3 expression is reduced. These changes in localization correlate with known activity of the Hh pathway during pancreas development. We demonstrate that components of the Hh pathway also localize to the cilium in CFPAC-1 and PANC-1 cells and that these cells may maintain high levels of nonstimulated Hh pathway activity. These findings indicate that Hh signaling coordinated by the primary cilium in the pancreas is essential for normal pancreatic development as well as postnatal tissue function.

RESULTS

Early Development of the Human Pancreas

To investigate the presence of primary cilia in human pancreatic progenitor cells of the initial duct epithelium we analyzed 3-μm-thick tissue sections from 7.5-week-old embryos and from 14- and 18-week-old fetuses. Hematoxylin and eosin (HE) staining of the pancreatic tissue and identification of the developing ducts at these stages are presented in Figure 1A–C. In 5- to 6-week-old embryos a dorsal and a ventral out pocketing of the foregut duodenal endoderm result in the formation of separate dorsal and ventral pancreatic primordia. During the seventh embryonic week a fusion of the two epithelial pancreatic ducts takes place followed by a rapid expansion and branching of the ductal epithelium (Fig. 1A). Pancreatic progenitor cells in this epithelium engage in two separate differentiation programs at distinct developmental stages. At the first stage from week 8 to 10, the endocrine progenitor cells appear as cell clusters budding from the central ductal epithelium. These buds form the islets of Langerhans in the central pancreas at week 12, and at midgestation, the endocrine compartment is established. At a much later stage, the exocrine progenitors are responsible for the differentiation and expansion of the exocrine acinar epithelium, which is mainly a late fetal or early perinatal event. Figure 1B shows a section of the ductal system from a 14-week-old fetus (Jackerott et al.,2006) with initial endocrine progenitors budded from the ductal epithelium, and Figure 1C shows a section from an 18-week-old fetus with endocrine progenitors budded from the ductal epithelium as well as acinar progenitors surrounding the ductal epithelium.

Figure 1.

Characterization of primary cilia in human pancreatic progenitor cells of the initial ductal epithelium. A–C: Hematoxylin and eosin (HE) staining of sections of the human developing pancreas from 7.5-week-old embryos (A), 14-week (B), and 18-week-old (C) fetuses. Lower panels show higher magnification images of the developing ducts. d, ducts; mPD, main pancreatic duct; PI, pancreatic islets budding from the duct (d); D, duodenum. D–H: Immunofluorescence microscopy (IF) analysis of primary cilia in the developing ducts at the stages presented in (A–C) by using two markers for primary cilia (arrows; anti-acetylated α-tubulin, acet. tb, red; and anti-detyrosinated α-tubulin, glu-tub, green). Nuclei were stained with DAPI (blue, 4′,6-diamidine-2-phenylidole-dihydrochloride). Shifted overlays (E) shows the combined staining with acet. tb and glu-tb, where the images for each antibody are slightly shifted from one another. Scale bars = 50 μm in A–D, 10 μm in E,G,H, 20 μm in F.

Characterization of Primary Cilia in Ducts of the Developing Human Pancreas

To identify primary cilia, we used two markers for cilia (anti-acetylated α-tubulin [acet. tb] and anti-detyrosinated α-tubulin [glu-tub]) in immunofluorescence microscopy (IF) analysis and found that epithelial cells in ducts at all stages form primary cilia that either project into the duct lumen, form spiral-like structures or align parallel to the lumen surface (Fig. 1D–H). The cilia varied in lengths from approximately 5 to 20 μm, and in some ducts long cilia seemed to form contact or bridge with cilia from neighboring epithelial cells or from cells on the opposite side of the duct (Fig. 1E,F). At the late stage of embryonic development (week 7.5), long acetylated-tubulin structures often projected from epithelial cells in a “star”-like configuration by pointing into the site of the developing duct (Fig. 1F and inset). Both short and long primary cilia were observed during all three developmental stages. Long cilia were predominantly observed in week 7.5, possibly due to the prevalence of small intercalated and intralobular type ducts that do have longer cilia (Kodama,1983). The formation and orientation of primary cilia in the human developing ducts is similar to that reported for the pancreas in the rat and mouse (e.g., Kodama,1983; Hidaka et al.,1995; Ashizawa et al.,1997; Cano et al.,2004; Zhang et al.,2005).

Development of the Human Pancreas Correlates With Ciliary Localization of Hh Signaling Components in Duct Epithelial Cells

It was previously shown that the graded response to Hh-signaling controls pancreatic organogenesis in the mouse, where Hh signaling is at low levels during early embryonic stages to ensure the correct establishment of organ boundaries and tissue architecture. Hh signaling is then activated at later developmental stages to promote proliferation and maturation of the tissue (Kawahira et al.,2005; Lau et al.,2006; Cano et al.,2007; van den Brink,2007). In Hh signaling, Gli2 in vivo acts primarily as a transcriptional activator, whereas Gli3 mainly works as a transcriptional repressor to control development. To address the potential role of primary cilia in human pancreatic development, we investigated the expression and cellular distribution of Smo, Gli2, and Gli3 in developing pancreatic ducts by IF analysis. In other cell types, such as fibroblasts (Rohatgi et al.,2007) and human embryonic stem cells, hESC (Kiprilov et al.,2008), Smo becomes concentrated while Ptc levels decrease in primary cilia upon stimulation with a Hh ligand, which may result in activation or deactivation of Gli transcription factors, possibly in the cilium (Huangfu and Anderson,2006; Christensen and Ott,2007). We found that Smo (Fig. 2A) and Gli2 (Fig. 2B) are absent from primary cilia in pancreatic ducts of embryos (week 7.5), but are concentrated in the cilia of ducts of early fetuses (week 14), and even more so in 18-week-old fetuses. We then investigated the localization of Gli3 in IF analysis using two different Gli3 antibodies that were raised against amino acids 1-280 of Gli3 (Gli3-H-280) and against a peptide mapping at the N-terminus of Gli3 (Gli3-N-19). Both antibodies showed a decrease in staining intensity and cellular distribution of Gli3 during development. At week 7.5, Gli3 strongly localized to duct epithelial cells and was present in the nuclei and in an intense punctate pattern in the cytosol of duct epithelial cells. In contrast, in ducts of 14- and 18-week-old fetuses, Gli3 staining was reduced (Fig. 2C–E). As a control, a blocking peptide against anti-Gli3-N-19 decreased localization of the antibody to the duct epithelium (Fig. 2D). These results could reflect activation of the pathway because Hh is known to repress Gli3 expression (Marigo et al.,1996; Büscher and Rüther,1998; Schweitzer et al.,2000).

Figure 2.

Expression of Hedgehog signal components in human pancreatic progenitor cells of the initial ductal epithelial cells and their primary cilia from 7.5-week-old embryos and 14-week and 18-week-old fetuses. A–E: Localization of Smo (green) in 7.5-week-old embryos and 18-week-old fetuses (A), and localization of Gli2 (H-300; green, B), Gli3 (H-280, green, C), and Gli3 (N-19, green, D,E) in 7.5-week-old embryos and 14-week and 18-week-old fetuses. Primary cilia (arrows) were localized with acet. tb (red) and nuclei were stained with DAPI (blue, 4′,6-diamidine-2-phenylidole-dihydrochloride). d, ducts. Scale bars = 10 μm.

Characterization of Primary Cilia in Cultures of Human Pancreatic Duct Cell Lines

CFPAC-1 and PANC-1 are human pancreatic ductal adenocarcinoma cell lines, which form adherent epithelial monolayers in culture (Fig. 3A,B) and are often used as model cell lines for studying ion transport and signaling in human pancreas (Novak et al.,2008). PANC-1, derived from epitheliod carcinoma, forms a monolayer of heterogeneous duct cells also with some cells growing on top of each other. Actin is mainly arranged in the cortex and lamellipodia (Fig. 3B). CFPAC-1 cells, derived from a patient with cystic fibrosis containing deletion in Phe-508 in CFTR, appear homogenous and on glass surface they form a monolayer interrupted by some void spaces. In most CFPAC-1 cells actin is concentrated in well-organized stress fibers and no subcortical actin network is apparent (Fig. 3A). Also the cytosolic network of microtubules seems dispersed throughout the cells, as also observed by Hollande et al. (2005). The epithelial-like cytoskeletal arrangement of PANC-1 cells may aid the targeting and function of adenosine receptors, which regulate ion transport (Novak et al.,2008). The distribution of these receptors, carbonic anhydrase, and cytoskeleton arrangement is defective in CFPAC-1 cells (Novak et al.,2008).

Figure 3.

Characterization of primary cilia in cultures of human pancreatic adenocarcinoma cells. A,B: Adherent epithelial monolayer cultures of CFPAC-1 (A) and PANC-1 (B). Cells were either double-labeled with DAPI (blue, 4′,6-diamidine-2-phenylidole-dihydrochloride) and phalloidin (red) for nuclear and F-actin staining, respectively, or triple-labeled with DAPI (blue), phalloidin (red) and tubulin (green) (insets). C,D: Formation of primary cilia (bold arrows, acet. tb, red) in quiescent CFPAC-1 (C) and PANC-1 (D), i.e., serum-starved for 48 and 72 hr, respectively. Acet. tub (red) also stained mitotic spindles (*: metaphase cells; ¤: anaphase cells) and midbodies of cells that have recently undergone cytokinesis (open arrows). Nuclei were stained with DAPI (blue). E,F: Primary cilia (acet. tub, red, arrows) in CFPAC-1 (E) and PANC-1 (F) serum-starved for 48 and 72 hr, respectively, emerge from the centrosome marked with anti-pericentrin at the base of the cilium (Pctn, green, #, left panels) and in particular from one of the centrioles marked with anti-centrin (centrin, green, *, right panels). Nuclei were stained with DAPI (blue). G: Frequency of ciliated PANC-1 cells in the presence of 10% serum at 50 and 100% confluency and in 100% confluent cultures after 48 and 72 hr of serum starvation (0.5% serum). Scale bars = 100 μm in A,B, 10 μm. in E,F.

To study the time course of primary cilia formation the cells were analyzed by IF microscopy both in cultures with 10% serum (at 50% and 100% confluency) and following serum starvation of confluent cultures in medium with 0.5% serum for periods of 48 and 72 hr (Fig. 3G). At 50% confluency and in the presence of 10% serum PANC-1 cells are most often in interphase growth and mitosis, such that very few cells are ciliated. At 100% confluency, approximately 4% of the cells had formed single primary cilia, presumably as a consequence of cellular contact inhibition of growth. After 48 and 72 hr of serum starvation, the frequency of ciliated cells increased to approximately 12 and 46%, respectively, and cilia had lengths of approximately 10–20 μm (Fig. 3D, left panel, and 3G). Similar results on appearance of cilia were obtained with cultures of CFPAC-1 cells; these cells, however, started to loose contact with the polylysine-coated glass coverslips at 72 hr of serum starvation. The relatively low percentage of ciliated cells in pancreatic cell lines is in sharp contrast to that of other cell types such as cultured fibroblast, in which more than 95% of the cells are ciliated after 24 hr of serum starvation (Schneider et al.,2005). The low frequency of ciliated cells in the adenocarcinoma cell lines could be related to their cancerous origin because cells that have an increased rate of growth more rarely enter growth arrest, which is required for assembly and maintenance of the primary cilium (Satir and Christensen,2007). Indeed, IF analysis revealed many mitotic and dividing cells in the CFPAC-1 and PANC-1 cultures as evidenced by the presence of mitotic spindles and midbodies (Fig. 3C,D, right panels). As in other cell types, the primary cilia emanate from the centrosomes (labeled by anti-Pctn, Fig. 3E,F left panels) and, in particular, from one of the two centrioles (labeled by anti-centrin, Fig. 3E,F right panels), presumably the mother centriole, which also functions as the basal body (Satir and Christensen,2007). PANC-1 cells grew longer cilia (10–20 μm) compared with CFPAC-1 cells (5–15 μm), where in some cases the short cilia projected vertically into the medium, visualized as “dots” in the IF analysis (Fig. 3C, middle panel).

Localization of Ptc and Smo to Primary Cilia of Human Pancreatic Duct Cell Lines

To evaluate whether Hh-signaling and primary cilia might be linked to development of cancer in adult pancreas, we next investigated whether components of the Hh signaling system are present in CFPAC-1 and PANC-1 primary cilia. Initially, we demonstrate by IF analysis that as in other cell types, Ptc and Smo localized to primary cilia of both cell lines (Fig. 4A,F). As a control, we exogenously expressed Smo from a YFP-Smo construct and show that it localized to the cilia of CFPAC-1 cells (Fig. 4G). The Ptc antibody recognized Ptc as a single protein band in Western blot analysis (Fig. 4B, upper panel), and this band was removed in the presence of anti-Ptc blocking peptide (Fig. 4C). Expression of Ptc at the mRNA level was previously shown to be up-regulated as a consequence of the autonomous operation of an active Hh signaling process in pancreatic cancer cells in which Ptc mRNA in PANC-1 cells is up-regulated at a higher level than that of CFPAC-1 cells (Berman et al.,2003; Thayer et al.,2003). Consistent with these earlier findings, we show that the protein level of Ptc in PANC-1 was higher than that in CFPAC-1 cells (Fig. 4B, upper and lower panels). As further controls for antibody specificity, we show that Ptc localized to primary cilia of wild-type mouse embryonic fibroblasts (wt MEFs) but not to mutant Ptc−/− MEFs (Fig. 4D) as previously demonstrated by Rohatgi et al. (2007). Then we stimulated wt MEFs with the Smo agonist, SAG, that activates Hh-signaling (Chen et al.,2002) to analyze changes in protein levels of Ptc, which is up-regulated in response to activation of the Hh pathway. As shown in Figure 4E, the Ptc is present at a low level in wt MEFs in the absence of SAG, whereas it increased after Hh pathway activation. This increase was dependent on the presence of the primary cilium, because the level of Ptc was kept at a low level in the presence of SAG in MEFs derived from the Tg737orpk mouse (Fig. 4E). These results support the idea that the cilium is required for activation of the Hh pathway, and that aberrant Hh signaling in pancreatic cancer may be associated with increased Hh signaling by means of the primary cilium.

Figure 4.

Localization of Patched (Ptc) and Smoothened (Smo) in cultures of quiescent human pancreatic adenocarcinoma cells. Primary cilia are marked with arrows and were localized with acet. tb (red). A: Localization of Ptc (green) in PANC-1 cells (upper panels) and CFPAC-1 cells (lower panels). B: The upper panel shows sodium dodecyl sulfate-polyacrylamide gel electrophoresis, SDS-PAGE, and Western blotting analysis of anti-Ptc cross-reactivity to proteins in CFPAC-1 and PANC-1 cells. The lower panel shows protein relative levels of Ptc in CFPAC-1 and PANC-1 cells. Values are given relative to the level of Ptc in CFPAC-1 cells (n = 3). C: Western blotting analysis of anti-Ptc cross-reactivity in the absence and in the presence of blocking peptide in PANC-1 cells. D: Localization of Ptc (green) to the primary cilium in wild-type mouse embryonic fibroblasts (wt MEFs, upper panels) and in mutant Ptc−/− MEFs (lower panels). E: Protein levels of Ptc before and after stimulation with SAG for 24 hr in quiescent wt and Tg737orpk MEFs. F: Localization of Smo (green) in PANC-1 cells (upper panels) and CFPAC-1 cells (lower panels). G: Localization of YFP-Smo (green) in CFPAC-1 cells. Localization of Smo following 24 hr SAG stimulation in CFPAC-1 cells. Scale bars = 10 μm.

Another observation was that Ptc localized only weakly to the primary cilium (Fig. 4A), while Smo localized very strongly and in a punctate pattern in the cilium of the pancreatic duct cancer cells (Fig. 4F). This is in sharp contrast to that observed in cultures of normal cells such as MEFs and hESCs, where Ptc is highly concentrated in the cilium and Smo is mostly absent from the cilium in nonstimulated cells (Rohatgi et al.,2007; Kiprilov et al.,2008). Stimulation of the pancreatic duct cells with SAG did not further increase ciliary localization of Smo in CFPAC-1 cells (Fig. 4H), supporting the conclusion that the Hh pathway is maintained at a high and autonomous level in these cells.

Localization of Gli Transcription Factors to Primary Cilia of Human Pancreatic Duct Cell Lines

Gli2 and Gli3 are the primary transcription factors that are being regulated in Hh signaling. The full-length transcription factors mainly function as transcriptional activators, but in the absence of the Hh signaling they may undergo proteolytic processing and function as transcriptional repressors (Wang et al.,2000; Litingtung et al.,2002; Pan et al.,2006). To investigate the expression and localization of Gli2 in growth-arrested PANC-1 and CF-PAC-1 cells we used three different antibodies raised against amino acids 841-1140 mapping near the C-terminus of Gli2 (Gli2-H-300), a peptide mapping near the N-terminus of Gli2 (Gli2-N-20), and a peptide mapping within an internal region of Gli2 (Gli2-G-20). All three antibodies localized uniquely to the primary cilium and most often as an enrichment at the tip of the cilium (Fig. 5A,B, left panels), similar to what has been observed in primary cilia of the murine limb bud and hESC (Haycraft et al.,2005; Kiprilov et al.,2008). Ciliary localization was abolished in the presence of blocking peptides to the two peptide antibodies (Fig. 5B, right panels). Also, localization of Gli2 to the ciliary tip was confirmed by transfection and overexpression of a green fluorescent protein (GFP) -Gli2 construct in NIH3T3 fibroblasts (Fig. 5C). To confirm antibody specificity, we performed Western blotting analysis with the peptide antibodies in the absence and in the presence of blocking peptides. As shown in Fig. 5D, Gli2-N-20 specifically recognized both full-length (ca. 170 kDa) and processed (ca. 70 kDa) forms of Gli2 that function as activator (Gli2(FL)) and repressor (Gli2(R)) forms in Hh signaling, respectively (Pan et al.,2006). In contrast, Gli2-G-20 recognized only Gli2(FL) (Fig. 5E). This finding suggests that the activator forms of Gli2 are present in the primary cilium and that ciliary Gli2 may be part of the Hh signaling machinery that is up-regulated in pancreatic cancer cells.

Figure 5.

Expression and localization of Gli transcription factors in quiescent human pancreatic adenocarcinoma cells and NIH3T3 fibroblasts. Primary cilia are marked with arrows and were localized with acet. tb (red). A: Localization of Gli2 (H-300; green) to primary cilia of CFPAC-1 cells. The inset in the lower panel shows a shifted overlay of the cilium (red) and Gli3 (H-300, green). B: Localization of Gli2 (N-20) and Gli2 (G-20, both in green) in the absence and in the presence of their corresponding blocking peptides in PANC-1 and CFPAC-1 cells. C: Localization of green fluorescent protein (GFP) -Gli2 (green) in NIH3T3 cells. D,E: SDS-PAGE and Western blotting analysis of anti-Gli2 (N-20) (D) and anti-Gli2 (G-20) (E) specificity to proteins in CFPAC-1 and PANC-1 cells in the absence and in the presence of their corresponding blocking peptides. Anti-Gli2 (G-20) only recognized a band at ca. 170 kDa. Abbreviations: Gli2(FL): Full length (activator) form of Gli2; Gli2(R): Processed (repressor) form of Gli2. F: SDS-PAGE and Western blotting analysis of anti-Gli3 (N-19) and anti-Gli3 (H-280) specificity to proteins in CFPAC-1 cells. Gli3(FL): Full-length (mild activator) form of Gli3; Gli3(R): Processed (repressor) form of Gli3. Anti-Gli (N-19) recognized both Gli3 forms, while Gli3 (H-280) predominantly recognized the repressor form of Gli3. G: Western blotting analysis of anti-Gli3 (N-19) specificity in the absence and in the presence of its blocking peptide in CFPANC-1 cells. H,I: Changes in localization of Gli3 (H-280) in nonstimulated (left panels) vs. SAG-stimulated (right panels) in CFPAC-1 (H) and NIH3T3 (I) cells. In nonstimulated NIH3T3 cells, Gli3 localized similar to that of GFP-Gli3(R), but after SAG stimulation the nuclear and cytosolic levels of Gli3 were markedly decreased. In CF-PAC-1 cells the nuclear and cytosolic levels of Gli3 were low in both stimulated and nonstimulated cells. J: Expression of two different GFP-Gli3 (green) constructs in quiescent NIH3T3 cells, where the repressor form of Gli3 (GFP-Gli3(R)) strongly localizes to the nucleus and to cytosolic puncta (left panels) and the full-length form of Gli3 (GFP-Gli3(FL)) localizes to the primary cilium (right panels). Insets: Shifted overlays with high resolution of GFP in the cilium. Nuclei were stained with DAPI (blue, 4′,6-diamidine-2-phenylidole-dihydrochloride). Scale bars = 5 μm in A, 10 μm in J.

We then analyzed lysates from CFPAC-1 cells by Western blotting analysis using the Gli3 antibodies, Gli3-N-19 and Gli3-H-280. Gli3-N-19 recognized both full-length (Gli3(FL) at ca. 180 kD) and processed forms (Gli3(R) at ca. 80 kDa) of Gli3 (Fig. 5F,G). In contrast, Gli3-H-280 predominantly recognized the processed version of Gli3 (Fig. 5F), supporting the conclusion that this antibody mainly identifies the repressor form of Gli3 under the experimental conditions carried out in this work. These data further support the assumption that the observed decrease in staining intensity and cellular distribution of Gli3 in the epithelium of the pancreatic ducts during fetal development (Fig. 2) is due to the increased activation of the Hh signaling that favors the processing of transcriptional activators of the Hh pathway. Based on these observations, it was therefore interesting to investigate the expression and localization of Gli3(R) in pancreatic cancer cells upon stimulation of the Hh pathway using Gli3-H-280 in IF analysis. As seen in Figure 5H, Gli3-H-280 stained CFPAC-1 cells very weakly in both the absence and in the presence of SAG. In contrast, Gli3-H-280 strongly stained the nucleus as well as small cytoplasmic puncta in NIH3T3 cells and this staining was largely absent in the presence of SAG (Fig. 5I). Furthermore, we used transfected NIH3T3 cells with GFP constructs of either the GFP-Gli3(R) and Gli3(FL) to evaluate the cellular localization of activator and repressor forms of Gli3. Similar to that seen in nonstimulated NIH3T3 cells with Gli3-H-280 (Fig. 5I) and previously in mouse limb bud mesenchyme cells, we found that GFP-Gli3(R) localized predominantly to the nucleus and to small cytosolic puncta, but not to the primary cilium (Fig. 5J, left panels). In contrast, GFP-Gli3(FL) seemed to localize to the primary cilium with minimal localization in the nucleus and without cytosolic puncta (Fig. 5J, right panels). These results are in line with our conclusion that Gli3-H-280 recognizes the repressor form of Gli3, which is concentrated in the nucleus in the absence of Hh signaling, and that Gli3-repressor forms are expressed at a low level in CFPAC-1 cells.

DISCUSSION

Recent research has shown that primary cilia coordinate signal pathways, which are essential in mammalian development, tissue homeostasis and behavioral responses. Accordingly, defective primary cilia assembly and maintenance lead to a plethora of diseases and disorders, now collectively referred to as ciliopathies (Badano et al.,2006; Fliegauf et al.,2007; Satir and Christensen,2007; Yoder,2007). Previous reports have shown that loss of primary cilia in the Tg737orpk mouse causes a series of developmental defects in exocrine and endocrine tissues of the pancreas, including extensive cyst formation in ducts (Cano et al.,2004; Zhang et al.,2005). However, the function of primary cilia in pancreatic organogenesis and adult tissue homeostasis is largely unknown. In this report, we demonstrate that essential components of the Hedgehog (Hh) signaling pathway localize to primary cilia of both human pancreatic progenitor cells of the initial ductal epithelium and in cultures of human pancreatic adenocarcinoma duct cell lines, and that expression of ciliary Hh components may be linked to pancreatic organogenesis and pancreatic cancer.

Hh Signaling and Primary Cilia in the Developing Human Pancreas

Previous studies have shown that the graded response to activation of the Hh signaling pathway is critical in pancreatic organogenesis in the mouse, where Hh signaling is activated only at late developmental stages (Kawahira et al.,2005; Lau et al.,2006; Cano et al.,2007; van den Brink,2007). It is likely that organogenesis of the human pancreas also depends on the graded response to Hh signaling, because the transcriptional network that controls pancreatic development is highly conserved in mammals (Wilson et al.,2003), although the time point of onset of development and maturation of the endocrine system differs in humans and rodents (Ostrer et al.,2006). Recently, the primary cilium was shown to act as a sophisticated switch by which cells turn Hh signaling on and off by the regulated movement of Smo into the cilium and Ptc out of the cilium upon stimulation of the Hh pathway in fibroblasts (Rohatgi et al.,2007) and hESC (Kirprilov et al.,2008), where Smo probably functions to activate Gli transcription factors (Christensen and Ott,2007). To analyze the significance of the primary cilium in Hh signaling during organogenesis of the human pancreas, we studied sections of the developing human pancreas at different developmental stages. These stages included embryonic stage 7.5 weeks, which is before development of the endocrine system, and stages 14 and 18 weeks, where endocrine progenitors bud from the ductal epithelium to form the endocrine compartment (Fig. 1). Based on the recent data from Rohatgi et al. (2007), our findings that Smo and Gli2 are absent from pancreatic primary cilia at embryonic stage 7.5, but highly concentrated in cilia in 14- and 18-week-old fetuses (Fig. 2) suggest that the Hh pathway has become activated during developmental stages. This further supports the conclusion that development of the human pancreas is regulated by a graded Hh signaling response, and that this could be coordinated by the primary cilium. Indeed, the loss of Gli3 staining in the ductal epithelium after week 7.5, that is, before formation of the endocrine system, is consistent with the interpretation that Gli3 staining at week 7.5 may represent the up-regulated level of the repressor form of Gli3 (Fig. 5F), consistent with low level of Hh signaling at this developmental stage. Therefore, disruption of pancreatic development in mice with defects in primary ciliary assembly (Cano et al.,2004,2006; Zhang et al.,2005), may partly be due to loss of coordinated Hh signaling during genesis of the pancreas.

Another interesting observation is the presence of long cilia-like structures that seem to bridge with corresponding structures emerging from adjacent or opposite cells of the lumen surface (Fig. 1). This is particularly prominent in the developing ducts at week 7.5. Bridging of primary cilia was also observed in collecting ducts of the mouse developing kidney (Liu et al.,2005b), but the function of physical contact between individual primary cilia is unknown. Although highly speculative at this point, the intertwining of cilia could be part of the sensory signaling machinery that controls the initial development of ducts and perhaps endocrine progenitors. In the developing kidney, the formation of tubular systems is controlled by inversin, which acts as a molecular switch between the canonical and nonconical Wnt pathways (Simons et al.,2005), and inversin potentially signals from the primary cilium to suppress β-catenin expression in the canonical pathway and to up-regulate the noncanonical pathway to organize planar cell polarity and correct tubular formation (Christensen et al.,2007). Because loss of the primary cilium in the developing mouse pancreas is associated with increased β-catenin expression (Cano et al.,2004; Zhang et al.,2005), it is possible that cilia in the developing pancreas regulate both Wnt and Hh signaling, although it is presently unknown whether Hh or Wnt signaling is affected by physical contact between primary cilia in the ductal epithelium. Further experiments are required to explore this.

Hh Signaling and Primary Cilia in Human Pancreatic Duct Cell Lines

Hh is implicated as an important mediator of human pancreatic carcinoma, as evidenced by, for example, the increased expression of positive Hh regulators and decreased expression of negative Hh regulators in CFPAC-1 or PANC-1 cells (Thayer et al.,2003; Berman et al.,2003). Here, we show that essential components of the Hh pathway, including Smo, Ptc, and Gli2, are present in primary cilia of human pancreatic ductal adenocarcinoma cell lines, CFPAC-1 and PANC-1, consistent with the idea that the primary cilium continues to coordinate Hh signaling in cells derived from the mature pancreas. We also suggest that aberrant Hh signaling in these cancer cell lines may be associated with the autonomous activation of the signaling pathway in the cilium as judged by high levels of Smo and low levels of Ptc in the cilium, which is accompanied by the formation of Gli2(FL) forms in the cilium and low levels of Gli3(R) in the nucleus.

In support of the hypothesis that the primary cilium may comprise the site for aberrant Hh signaling in human pancreatic cancer tumorigenesis, we show that in contrast to normal cells, which require Hh pathway stimulation in the cultures for Smo to enter the cilium and activate Hh target genes (Rohatgi et al.,2007; Kiprilov et al.,2008), Smo is strongly localized to the primary cilium in both CFPAC-1 and PANC-1 cells, even in the absence of external stimulation of the Hh pathway by SAG. Further and in contrast to primary cultures of mouse embryonic fibroblasts, Ptc is expressed at a higher level than that of MEFs, which require the primary cilium and stimulation of the Hh pathway to increase their expression of Ptc. The activator form of Gli2, Gli2(FL), localizes to the tip of primary cilium, which may comprise the site for activation of Gli transcription factors in Hh signaling. We then examined the expression and localization of Gli3 before and after SAG stimulation by comparing the localization of endogenous Gli3 by IF assessed using Gli3-H-280, which recognizes the processed form of Gli3 with localization of GFP-Gli3(FL) and GFP-Gli3(R) expressed in transfect growth-arrested NIH3T3 fibroblasts. As shown in Figure 5J, only GFP-Gli3(R) localized to the nucleus and to cytoplasmic puncta. These data are consistent with the conclusion that the repressor form is absent from the primary cilium and is repressing Hh target genes in the absence of Hh signaling. We then performed IF analysis in NIH3T3 cells in the absence and in the presence of SAG and compared the staining with Gli3-H-280 to localization of the protein expressed from the GFP constructs. As shown in Figure 5I (left panels) anti-Gli3 stains nonstimulated cells similar to that of the GFP-Gli3(R) construct. Upon activation of the Hh pathway nuclear Gli3 staining is strongly reduced, supporting the conclusion that the antibody in IF analysis is detecting the repressor form of Gli3. Finally, we examined the localization of Gli3 in CFPAC-1 cells, which showed very weak cytosolic and nuclear staining of Gli3 in both stimulated and nonstimulated cells. This result would be expected if Hh signaling activity is elevated in these cells because Hh signaling represses Gli3 expression (Marigo et al.,1996; Büscher and Rüther,1998; Schweitzer et al.,2000), and is further evidenced by the accumulation of Smo in the cilium of these cells.

CONCLUSIONS

Our data indicate a functional role of the primary cilium in coordinating Hh signaling during development of the human pancreas and potentially in tumorigenesis of the adult human pancreas. During development the primary cilium may sense and relay the graded response to activation of the Hh signaling pathway that controls human pancreatic organogenesis at the onset of fetal development. In the adult, the autonomous activation of the Hh pathway in pancreatic tumorigenesis may be linked to aberrant Hh signaling in the primary cilium of the pancreatic cells. Further analysis will be required to understand the potential significance of Gli processing in the cilium and how the switch between activator and repressor forms of Gli2 and Gli3 is regulated by the cilium during development and tumorigenesis of the pancreas and potentially of other human tissues and organs. Furthermore, it will be important to identify the molecular mechanisms that coordinate translocation of Smo and Ptc in and out of the cilium, which ultimately may impinge on the activation and deactivation of the Hh pathway, which is of relevance in human health and development.

EXPERIMENTAL PROCEDURES

Cell Cultures

The human pancreatic exocrine duct cell lines PANC-1 (ATCC, #CRL-1469) and CFPAC-1 (ATCC, #CRL-1918; passages 8–36) were from the American Type Culture Collection and were cultured in Dulbecco′s modified Eagles medium with glutamax (DMEM; Invitrogen) and Iscove's modified Dulbecco's medium (IMDM; Invitrogen), respectively. Both growth media were supplemented with 10% heat inactivated fetal calf serum (FCS; Invitrogen) and 1% penicillin-streptomycin (Penicilin G sodium: 10,000 U/ml and streptomycin G sodium: 10,000 μg/ml) (P/S; Invitrogen) and kept in a humidified air chamber at 37°C and 5% CO2. Passing of cells was performed by trypsination. NIH3T3 Swiss mouse embryonic fibroblasts and wt, Ptc−/− and Tg737orpk mouse embryonic fibroblasts (MEFs) were cultured as described previously for MEFs (Schneider et al.,2005).

Tissues

Human embryos and fetuses were obtained from extra uterine pregnancies, Caesarean sections, or prostaglandin induced legal abortions donated to the Developmental Biology Unit, ICMM, at the Panum Institute, University of Copenhagen, Denmark. The embryos and fetuses ranged from 8 to 180 mm crown-rump length (CRL), corresponding to 6th–19th ovulation weeks. Informed consent was obtained according to the guidelines of the Helsinki Declaration II. Additional samples from legal first trimester abortions from the Laboratory of Reproductive Biology, Rigshospitalet, and Frederiksberg Hospital (both University of Copenhagen) were also included in this study. Informed consent was obtained according to the Helsinki declaration II and approved by the ethical committee of Copenhagen and Frederiksberg Communities (KF 258206). Calculation of embryonic and fetal age was based on information about the last menstrual period, and measurements of CRL and foot lengths.

Antibodies, Blocking Peptides, and Staining Reagents

Primary antibodies (dilutions in parenthesis for IF analysis): mouse monoclonal antibodies from SigmaAldrich: anti-acetylated α-tubulin (acet. tb, cat. no. T6793; 1:5,000), anti-β-actin (cat. no. A5441; 1:5,000), and α-tubulin (cat. no. T5168; 1:2,000). Rabbit polyclonal anti-detyrosinated α-tubulin (glu-tub; ab48389; 1:500) was from Abcam. Polyclonal antibodies from Santa Cruz Biotechnology: goat anti-centrin (cat. no. sc-8719; 1:500), goat anti-pericentrin (Pctn; cat.no. sc-28145; 1:500), rabbit anti-Gli2 transcription factor [Gli2(H-300), cat. no. sc-28674; Gli2(N-20), cat. no. sc-20290; Gli2(G-20), cat no. sc-20291; all diluted 1:200], rabbit anti-Gli3 transcription factor [Gli3(H-280), cat. no. sc-20688; Gli3(N-19), cat. no. sc-6155; all diluted at 1:100], goat anti-Patched (Ptc; cat. no. sc-6149; 1:200). Two different rabbit anti-Smoothened were used: (1) Smo (cat. no. LS-A2668; 1:100) from MBL, and (2) Smo (cat. no. 38686) from Abcam. Secondary antibodies for immunofluorescence microscopy analysis (all from Molecular Probes and diluted at 1:600): Alexa Fluor 568-conjugated goat anti-mouse IgG (cat. no. A11019), Alexa Fluor 568-conjugated rabbit anti-mouse IgG (cat. no. A11061), Alexa Fluor 488-conjugated goat anti-rabbit IgG (cat. no. A11008), and Alexa Fluor 488-conjugated donkey anti-goat IgG (cat. no. A11055). Secondary antibodies for Western blot analysis (all from Jackson ImmunoResearch and diluted 1:5,000): alkaline phosphatase conjugated goat anti-rabbit, rabbit anti-goat, and goat anti-mouse. Nuclei were stained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI, 1:600) from Molecular Probes, and F-actin was stained with rhodamine phalloidin (1:100; Molecular Probes, Invitrogen). Blocking peptides sc-6155P, sc20290P, sc-20291P, and sc- sc-6149P were from Santa Cruz Biotechnology.

Immunohistochemistry of the Developing Human Pancreas

Tissue specimens were dissected into appropriate tissue blocks and fixed for 12–24 hr at 4°C in one of the following fixatives: 10% buffered formalin, 4% Formol-Calcium, Lillie's AAF, or Bouin's fixatives. The specimens were dehydrated with graded alcohols, cleared in xylene, and embedded in paraffin wax (Merck, melting point 52°C). Serial sections, 3–5 μm thick, were cut in transverse, sagittal, or horizontal planes and placed on silanized slides. Representative sections of each series were stained with hematoxylin and eosin, or with toluidine blue. For immunohistochemistry (IHC) sections were dewaxed, rehydrated and washed in phosphate buffered saline (PBS; 137 mM NaCl, 2.6 mM KCl, 6.5 mM Na2HPO4, and 1.5 mM KH2PO4) as previously described (Teilmann and Christensen,2005), followed by rinsing with blocking buffer (5% bovine serum albumin in PBS) for 15 min before incubation with primary antibodies for at least 1.5 hr at room temperature or overnight at 4°C. The sections were then washed three times in blocking buffer, incubated for 45 min in dark with fluorochrome-conjugated secondary antibodies and DAPI in blocking buffer. After washing, sections were mounted in PBS with 70% glycerol and 2% N-propylgallate and sealed with nail polish. Samples were observed on an Eclipse E600 microscopes (Nikon, Tokyo, Japan) with EPI-FL3 filters and MagnaFire cooled CCD camera (Optronics, Goleta, CA), and digital images were processed using Adobe Photoshop 6.0.

Immunocytochemistry

Cells were grown on 25-mm-diameter sterile polylysine-coated glass coverslips for 3–5 days to ∼80% confluency, followed by serum starvation in DMEM/IMDM with 0.5% (PANC-1 and CFPAC-1) and 0% (MEF) FCS and 1% P/S for 24, 48, and 72 hr, to induce growth arrest and formation of primary cilia. The cells were briefly washed in PBS, fixed for 15 min in 4% paraformaldehyde followed by wash in blocking buffer and then permeabilized with 1% Triton X-100 in PBS for 10 min. The coverslips were rinsed with blocking buffer for 30 min before incubation with antibodies as described above for IHC, and mounted on microscope slides in anti-fade mounting solution and sealed with nail polish. Samples were observed on an Eclipse E600 microscopes (Nikon, Tokyo, Japan) with EPI-FL3 filters and MagnaFire cooled CCD camera (Optronics, Goleta, CA), and digital images were processed using Adobe Photoshop 6.0.

Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis and Western Blot Analysis

Cell lysates were prepared by using 1% sodium dodecyl sulfate (SDS) lysis buffer. Lysates were sonicated and centrifuged to separate off cell debris. Protein concentrations were measured using a DC Protein Assay from Bio-Rad. Proteins were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) as described by Christensen et al. (2001). Briefly, proteins were separated under reducing conditions with NuPAGE 10% Bis-Tris precast gels (Invitrogen), followed by electrophoretic transfer to nitrocellulose membranes (Invitrogen). Before incubation with antibodies for 2 hr/over night the membranes were blocked with 5% milk/TBST (10 mM Tris/HCl [pH 7.5], 120 mM NaCl, 0.1% Tween 20) and 0.5% Na-azide. Antibodies were diluted in 5% milk/TBST as indicated: anti-Ptc (1:200), anti-Gli2 (N-20) and (G-20) (1:200), anti-Gli3 (H-280) and (N-19) (1:100), anti-β-actin (1:5,000), and anti-α-tubulin (1:2,000). Membranes were washed several times in TBST followed by incubation with alkaline phosphatase-conjugated secondary antibodies. Blots were developed with BCIP/TNBT from KPL. The developed blots were scanned and band intensity was estimated from arbitrary densitometric values obtained using UN-SCAN-IT software. The blots were further processed for publication in Adobe Photoshop version 6.0.

SAG Stimulation

Cultures of PANC-1, CF-PAC1 cells and NIH3T3 fibroblasts (80% confluency) were serum starved for 48–72 hr and incubated in the presence and in the absence of 1 μM SAG (Alexis Biochemicals, San Diego, CA) for 0 and 24 hr, followed by IF analysis with anti-Smo and anti-Gli3 and by WB analysis with anti-Ptc and anti-Gli2. Primary cilia were visualized with anti-acet. tb and nuclei with DAPI. All images were taken with equivalent time exposures.

Transfection of Cells With Fluorescent-Tagged Plasmids

CFPAC-1 cells were cultured on coverslips placed in six-well test plates (NUNC A/S, Denmark) and grown to 70% confluency. The medium was changed to DMEM/IMDM for 1 hr, followed by incubation with 1 μg/ml plasmid (YFP-Smo, provided by P. Beachy) and 5 μl of lipofectamine 2000 (Invitrogen) per well. After 4 hr of transfection the medium was changed to DMEM/IMDM with 10% FCS, and after the cells had reached 90% confluency they were serum-starved for 48 hr to induce growth arrest. Cells were then fixed and permeabilized and subjected to immunocytochemistry analysis as described above. NIH3T3 cells were cultured and transfected under the same conditions as above, except that serum-free DMEM was used during the 4-hr transfection period. The following plasmids were used: full length Gli2::GFP, GFP-Gli2(FL), full length Gli3::GFP, GFP-Gli3(FL), Gli3repressor::GFP, GFP-Gli(R) (Buttitta et al.,2003). All the vectors were constructed in the pShuttle backbone (Clontech).

Acknowledgements

We thank Drs. Melissa Lutterodt and Anne Grete Byskov (Laboratory of Reproductive Biology, Rigshospitalet, Denmark) for expert help with collecting of human embryonic tissues. We also thank Dr. Kathryn Anderson (Memorial Sloan-Kettering Cancer Center, NY, USA) for the gift of the Ptc−/− MEFs, and Simon Krabbe and Mette R. Hansen (Department of Biology, University of Copenhagen, Denmark) for help with cultures of human pancreatic ductal adenocarcinoma cell lines. This work was supported by the Lundbeck Foundation and the Danish Science Research Council (S.T.C.), the Novo Nordisk/Novozymes Foundation (S.K.N.), the Danish Science Research Council (I.N.), funds from the Department of Biology, University of Copenhagen (C.A.C.), and by a National Institute of Health Grant (B.K.Y.).

Ancillary