© 2014. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2014) 7, 883-894 doi:10.1242/dmm.014969
RESEARCH ARTICLE
Zebrafish reporter lines reveal in vivo signaling pathway activities
involved in pancreatic cancer
Marco Schiavone1,*, Elena Rampazzo2,*, Alessandro Casari1, Giusy Battilana2, Luca Persano3, Enrico Moro2,
Shu Liu4, Steve D. Leach4, Natascia Tiso1 and Francesco Argenton1,‡
KEY WORDS: Zebrafish, Pancreatic adenocarcinoma,
Medulloblastoma, KRAS, Reporters, TGFβ, Notch, Shh
1
Department of Biology, University of Padua, 35131 Padua, Italy. 2Department of
Molecular Medicine, University of Padua, 35131 Padua, Italy. 3Department of
Woman and Child Health, University of Padua, 35131 Padua, Italy. 4Department of
Surgery and The McKusick-Nathans Institute of Genetic Medicine Johns Hopkins
University School of Medicine, Baltimore, MD 21205, USA.
*These authors contributed equally to this work
‡
Author for correspondence (francesco.argenton@unipd.it)
This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted
use, distribution and reproduction in any medium provided that the original work is properly
attributed.
Received 22 November 2013; Accepted 23 May 2014
INTRODUCTION
Pancreatic adenocarcinoma is one of the most aggressive cancers in
industrial countries, and its incidence and mortality is still
increasing. The poor survival rate of this malignant disease indicates
that current interventions to prevent, diagnose and cure pancreatic
adenocarcinoma are far from satisfactory. Identification of the
molecular and biochemical processes that regulate the onset and
progression of pancreatic adenocarcinoma is of great relevance for
therapeutic purposes. In general, crucial signaling transduction
pathways involved in cell proliferation, stem-cell maintenance and
differentiation during embryonic development appear disrupted
during tumor formation. In pancreatic adenocarcinoma of both
acinar and ductal origins most pathways, including Sonic Hedgehog
(Shh), Wnt, Notch and transforming growth factor β (TGFβ)
signaling, have been shown to be dysregulated (Bailey and Leach,
2012). During embryonic development, Shh plays a major role in
stem-cell proliferation (Wu et al., 2010); the Wnt pathway is
involved in cell proliferation and differentiation (Yang, 2012); Notch
is responsible for stem-cell maintenance (Kwon et al., 2012); and
TGFβ controls cell and tissue homeostasis, favoring cell apoptosis
by cross-talking with other pathways such as p53 (Elston and Inman,
2012). TGFβ is also involved in tissue morphogenesis, in
cooperation with Wnt pathway, by controlling epithelial-tomesenchymal transition (EMT) and cell migration (Jing et al., 2011;
Zhou et al., 2012). During carcinogenesis, Shh and Notch pathways
seem to be involved in tumor onset, together with genomic
instability, whereas Wnt and TGFβ appear activated in cancer
progression by eliciting cell migration or neo-angiogenesis through
reciprocal cross-talk or by interactions with other pathways
(McCleary-Wheeler et al., 2012). For instance, non-canonical Wnt
has been proposed to be involved in cell proliferation and metastasis
by cross-talking with TGFβ (McCleary-Wheeler et al., 2012).
Several in vitro and in vivo studies demonstrated that, within the
tumor cell compartment, TGFβ has a dual role. By inhibiting cell
growth, it has a tumor suppressor function at early tumor stages,
whereas at later stages it mediates oncogenic effects (Heldin and
Moustakas, 2012; McCleary-Wheeler et al., 2012; Truty and Urrutia,
2007).
Animal models of human cancers provide unique insights into the
study and understanding of molecular pathways involved at both
early and late stages of malignant diseases, easing the discovery of
biomarkers and specific targets for new or more effective drug
therapies (Etchin et al., 2011; Herreros-Villanueva et al., 2012a;
Herreros-Villanueva et al., 2012b; Stoletov and Klemke, 2008). We
focused on the vertebrate teleost Danio rerio to uncover in vivo the
complex pathways regulating the biological processes underlying
pancreatic adenocarcinoma (Bailey and Leach, 2012; Goldsmith and
Jobin, 2012). Pancreatic adenocarcinoma is a solid tumor originated
by the epithelial cells of pancreatic duct and by changes in the
exocrine acinar structure as it reverts into ductal structure (Bardeesy
883
Disease Models & Mechanisms
ABSTRACT
Pancreatic adenocarcinoma, one of the worst malignancies of the
exocrine pancreas, is a solid tumor with increasing incidence and
mortality in industrialized countries. This condition is usually driven
by oncogenic KRAS point mutations and evolves into a highly
aggressive metastatic carcinoma due to secondary gene mutations
and unbalanced expression of genes involved in the specific signaling
pathways. To examine in vivo the effects of KRASG12D during
pancreatic cancer progression and time correlation with cancer
signaling pathway activities, we have generated a zebrafish model of
pancreatic adenocarcinoma in which eGFP-KRASG12D expression
was specifically driven to the pancreatic tissue by using the
GAL4/UAS conditional expression system. Outcrossing the inducible
oncogenic KRASG12D line with transgenic zebrafish reporters,
harboring specific signaling responsive elements of transcriptional
effectors, we were able to follow TGFβ, Notch, Bmp and Shh
activities during tumor development. Zebrafish transgenic lines
expressing eGFP-KRASG12D showed normal exocrine pancreas
development until 3 weeks post fertilization (wpf). From 4 to 24 wpf
we observed several degrees of acinar lesions, characterized by an
increase in mesenchymal cells and mixed acinar/ductal features,
followed by progressive bowel and liver infiltrations and, finally, highly
aggressive carcinoma. Moreover, live imaging analysis of the
exocrine pancreatic tissue revealed an increasing number of KRASpositive cells and progressive activation of TGFβ and Notch
pathways. Increase in TGFβ, following KRASG12D activation, was
confirmed in a concomitant model of medulloblastoma (MDB). Notch
and Shh signaling activities during tumor onset were different
between MDB and pancreatic adenocarcinoma, indicating a tissuespecific regulation of cell signaling pathways. Moreover, our results
show that a living model of pancreatic adenocarcinoma joined with
cell signaling reporters is a suitable tool for describing in vivo the
signaling cascades and molecular mechanisms involved in tumor
development and a potential platform to screen for novel oncostatic
drugs.
TRANSLATIONAL IMPACT
Clinical issue
With an overall 5-year survival rate of only 3-5% after diagnosis,
pancreatic adenocarcinoma is one of the most aggressive malignancies
in the industrialized world. Surgery and traditional combined therapies
are not effective enough to eradicate this deadly disease. Understanding
the cell origin and molecular mechanisms involved in its onset and
progression is a major step towards the development of novel drugs
against pancreatic adenocarcinoma, and many rodent models have been
developed during the last ten years. Zebrafish can provide a useful
complementary model to explore the characteristics of the disease in
vivo because of the organism’s amenability to live imaging and the
conserved genetic control for pancreatic development between fish and
mammals. The aim of this study was to use zebrafish to explore in vivo
the tumorigenic action of a constitutively active mutant version of the
oncogene KRAS, implicated in several types of pancreatic
adenocarcinoma.
Results
Using promoter and enhancer elements of ptf1a, a transcription factor
expressed in the exocrine pancreas and in cerebellar GABAergic
neurons, KRASG12D was specifically expressed in the pancreas and,
concomitantly, in the cerebellum using a Gal4/UAS inducible system. By
using mCherry cell signaling reporter lines, the authors observed
dysregulated activity of the Notch pathway during the early stages and
dysregulation of Smad3/TGFβ and Shh pathways during the later stages
of pancreatic adenocarcinoma. Upon comparison of pancreatic
adenocarcinoma with concomitant ptf1a-induced medulloblastoma in the
cerebellum, they conclude that TGFβ, Shh and Notch are involved in both
cancers at different stages of carcinogenesis. Furthermore, they provide
evidence that Smad3/TGFβ is controlled by KRAS in both cancers.
Interestingly, Notch and Shh signaling activities differed during tumor
onset in medulloblastoma compared with pancreatic adenocarcinoma,
indicating tissue-specific regulation of these cell signaling pathways.
Implications and future directions
This study provides insight into the signaling pathways involved in
pancreatic adenocarcinoma onset and progression. The identification of
mechanisms regulating the hallmarks of this aggressive disease and of
candidate molecules (Notch, TGFβ and Shh ligands and their effectors)
with the potential to interfere with cancer progression is an important step
towards new therapies for pancreatic adenocarcinoma. Furthermore, the
study provides a new animal model of the disease, which, coupled with
transgenic zebrafish reporter lines, provides a powerful tool for tracing
the in vivo dynamics of pancreatic tumorigenesis and for screening
candidate drugs. In the long-term, stratification of the many types of
pancreatic cancer according to genotype and developmental stage could
pave the way to the development of molecularly targeted therapies.
and DePinho, 2002; Esposito et al., 2007). Many studies have
focused on point mutations causing the constitutive activation of the
oncogene KRAS which, in turn, activates mechanisms bringing
pancreatic cancer onset (Park et al., 2008). Point mutations of the
amino acid residue at position 12 of KRAS protein (G12V and
G12D) are the most frequent changes found during the first stages
of pancreatic adenocarcinoma. Subsequent unbalanced expression
of several genes such as P53, STAT3 (Corcoran et al., 2011), SMAD4
and ARF/INK4 (Aguirre et al., 2003) brings carcinoma in situ and,
further, to metastatic carcinoma (Bardeesy et al., 2006). Several
studies on zebrafish and mice showed the involvement of Shh
signaling during early stages and its role in the stimulation of
TGFβ1 activity in duct cells during pancreatic fibrosis and at later
stages of pancreatic ductal adenocarcinoma (Jung et al., 2011).
However, how the unbalance of Shh, TGFβ, Notch and Bmp
signaling pathways reflects in pancreatic adenocarcinoma
progression remains to be unveiled. To trace in vivo the activity of
Shh, TGFβ, Notch and Bmp, we used zebrafish transgenic lines
884
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
expressing the fluorescent reporter mCherry under control of
specific responsive elements recognized by downstream regulators
of each pathway.
To obtain a zebrafish model for pancreatic carcinogenesis, we took
inspiration from a zebrafish model established by Park et al. (Park et
al., 2008) based on the constitutive activation of KRASG12V oncogene
in the pancreas under control of ptf1a promoter elements. In this work,
a new zebrafish model of pancreatic adenocarcinoma was reproduced
by using the conditional Gal4/UAS expression system (Liu and
Leach, 2011), driving the expression of mutated KRASG12D in the
exocrine pancreas of zebrafish derived from outcrosses with TGFβ,
Notch, Bmp and Shh reporter lines expressing the fluorescent reporter
mCherry. The selected ptf1a promoter also drives conditional
KRASG12D expression in cerebellum; thus, we were able to assess
KRAS activity during the early stages of cerebellar development by
obtaining a concomitant model of putative pediatric medulloblastoma
(MDB) (Gilbertson et al., 2006). By comparing the expression of
mCherry reporters we were able to reveal in vivo signaling pathways
elicited after oncogenic KRAS constitutive activation in both pancreas
and cerebellum, showing in vivo how TGFβ, Notch and Shh are
involved in pancreatic adenocarcinoma and during MDB
carcinogenesis.
RESULTS
Expression of eGFP-KRASG12D and Kaplan-Meier analysis in
a pancreatic adenocarcinoma and medulloblastoma
zebrafish model based on a Gal4/UAS expression system
To generate a zebrafish model of pancreatic adenocarcinoma, eggs
derived from Tg(ptf1a:Gal4) outcrosses were injected with
Tol2(UAS:eGFP-KRASG12D) plasmid (Liu and Leach, 2011; Park et
al., 2008); we call these samples “Tg(ptf1a:Gal4)/UAS:eGFPKRASG12D injected”. As a control, we used outcrosses of the stable
transgenic line Tg(ptf1a:eGFP) expressing cytoplasmic enhanced
green fluorescent protein (eGFP) in cells from both pancreas and
cerebellum. In particular, we observed tissue-specific expression of
eGFP both in pancreas and cerebellum in all 75 Tg(ptf1a:eGFP)
collected controls, faithfully recapitulating the endogenous pattern
of ptf1a expression (Fig. 1A) (Lin et al., 2004; Sellick et al., 2004;
Zecchin et al., 2004). In order to obtain a significant number of
cancer lesions, we collected 120 Tg(ptf1a:Gal4)/UAS:eGFPKRASG12D injected animals. All the selected Tg(ptf1a:Gal4)/
UAS:eGFP-KRASG12D injected larvae expressed eGFP-KRASG12D
in cerebellum but only 50% of them were also expressing eGFPKRASG12D in the pancreas (Fig. 1B). This might reflect the fact that,
as already described, expression of ptf1a in cerebellum is stronger
than in pancreas (Zecchin et al., 2004).
Our second aim was the phenotypic characterization of collected
fish: 120 Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected individuals
and 75 Tg(ptf1a:eGFP) controls. We focused our attention on fish
survival and appearance of severe or lethal signs of disease. The
observation time ranged from 1 to 100 weeks post fertilization (wpf).
As reported in Fig. 1C, we found one peak of death at 1-2 wpf and
two peaks of sickness at 4-6 wpf and 15-32 wpf for
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D. Thus, about 92% of
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected fish showed mortality
and severe disease signs between 2 and 32 wpf, whereas the
remaining 8% survived. Tg(ptf1a:eGFP) controls showed two peaks
of mortality: the first at 1-2 wpf and the second at 72-100 wpf,
showing a normal survival pattern. The phenotypic differences
between Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected and
Tg(ptf1a:eGFP) were statistically significant according to uncoupled
two-tailed Student’s t-tests performed on both groups. In order to
Disease Models & Mechanisms
RESEARCH ARTICLE
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
understand the reason for three different peaks of lethality for
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected fish, we made a more
detailed observation. In particular, we observed that 60 out of 120
(50%) samples died after 2 wpf at larval stage; 35 out of 120 (30%)
got sick at juvenile stage, showing strong motility disruption and
swimming defects before dying; 15 out of 120 (12%) were collected
at early adulthood, between 3 and 6 months post fertilization (mpf),
as soon as they showed big protruding masses in the belly, later
demonstrated to be pancreatic cancer (as shown in Fig. 2); and 10 out
of 120 (8%), expressing eGFP-KRASG12D only in cerebellum, reached
adulthood without any alteration. Histological analysis of cerebellum
from 35 out of 120 fish demonstrated that severe motility defects at
3-6 wpf, also related to a strong decrease in survival, were linked to
a cerebellar dysplasia or to exacerbating medulloblastoma, resembling
the pediatric form of human cerebellar cancer (shown in
supplementary material Fig. S1).
Histological features of pancreatic tumors
We followed pancreatic tissue growth by observing the progressive
increase in eGFP fluorescence under control of ptf1a promoter
elements. To characterize eGFP-positive pancreas extracted from
Tg(ptf1a:eGFP) and Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected
samples, we performed histological analysis with hematoxylin-eosin
(H&E) assay. As previously reported (Wan et al., 2006), we found
that normal exocrine pancreas from all Tg(ptf1a:eGFP) fish is
mainly characterized by arborized clusters of pancreatic acini
surrounded by adipose tissue and flanked by bowel loops and liver
(Fig. 2A). We collected 35 out of 120 Tg(ptf1a:Gal4)/UAS:eGFPKRASG12D injected samples between 2 and 5 wpf because of strong
motility defects. All 35 fish expressed eGFP-KRASG12D in the
cerebellum whereas only 16 out of 35 showed small focal eGFPKRASG12D-positive lesions in the exocrine pancreas, which was
enlarged compared with normal tissue. We performed histological
analysis on the 16 small abdominal eGFP-KRASG12D-positive focal
lesions to discriminate pancreatic lesions induced by KRASG12D
during the first stages of carcinogenesis. Acinar hyperplasia, an early
lesion observed during the development of pancreatic tumors
showing an acinar phenotype, was seen in 6 out of 16 samples.
Acinar hyperplasia is characterized by quite-organized, although
abnormally abundant, acinar tissue that almost totally replaced the
duct ephitelium (Fig. 2B left panel, 2C). The other ten animals also
showed stromal enrichment progressively destroying the compact
structure of exocrine tissue, becoming fibrous such as in acute
pancreatitis (Fig. 2B right panel, 2B′). All 15 Tg(ptf1a:Gal4)/
UAS:eGFP-KRASG12D injected fish collected between 14 and 24 wpf
showed widespread and protruding abdominal eGFP-KRASG12Dpositive masses in the gut region (supplementary material Fig.
S2A,B). Histological analysis showed defined features of
malignancy, such as the invasion of normal pancreas and
surrounding organs. These tumors displayed a wide degree of
heterogeneity with respect to histological patterns of differentiation,
including pancreatic adenocarcinoma with acinar phenotype,
pancreatic adenocarcinoma with mixed acinar and ductal features as
already described (supplementary material Fig. S2A′,B′) (Park et al.,
2008). In particular, we found that all 15 analyzed tumors showed a
predominant acinar and mucinous phenotype and non-ductal
differentiation. Out of 15 tumors, 7 displayed dramatic enrichment
of stroma that infiltrated duct lumen, formation of strong reactive
ductal-like structures and increased mucinous features, such as the
presence of goblet cells similar to those observed in human and
mouse pancreatic mucinous adenocarcinoma of acinar origin
(Fig. 2D-F). These mucinous features were also demonstrated by
staining pancreatic slices with Alcian Blue (Fig. 2H,I). The other
eight samples revealed pancreatic adenocarcinoma of predominant
acinar phenotype with mixed acinar/ductal features, mainly
characterized by disorganized proliferation of cells with
885
Disease Models & Mechanisms
Fig. 1. Tissue specificity expression of eGFP and eGFP-KRASG12D under control of ptf1a promoter and Kaplan–Meier analysis. (A,B) Lateral views,
anterior to the right showing that EGFP (A) and eGFP-KRASG12D (B) are specifically expressed in cerebellum (red arrows) and pancreas (white arrows) of
larvae observed at 5 dpf. (C) Kaplan–Meier analysis of dead or severely sick animals revealed lethality for about 92% of Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D
as measured at 24 wpf. Differences between Tg(ptf1a:eGFP) and Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D were statistically significant with P<0.05. Number of
sample for each group are indicated.
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
Fig. 2. Different pancreas transformations induced by KRASG12D. (A) Control exocrine pancreas characterized by organized acinar clusters surrounded by
fat cells and ducts. (B, left; C). Pre-tumoral acinar hyperplasia induced by KRASG12D seen in 6 out of 16 analyzed samples at 1 and 2 mpf. Acinar tissue
extension over the entire exocrine pancreas replaced the ductal ephitelium. (B, right) Pre-tumoral fibrotic pancreas tissue with stromal enrichment, reactive
ducts and a few acinar cells (arrowheads) interspersed inside disorganized exocrine tissue was seen in 10 out of 16 collected samples. (D) End-stage
mucinous pancreatic adenocarcinoma of acinar origin with goblet cells (arrowheads) interspersed into exocrine tissue. (E) End-stage mucinous pancreatic
adenocarcinoma with reactive ducts (arrowheads), increasing stroma and goblet cells. (F) End-stage mucinous adenocarcinoma with reactive ducts losing their
morphology (arrowheads), stromal enrichment and disorganization of acinar tissue. (G) End-stage pancreatic adenocarcinoma with mixed acinar/ductal
features showing an expansion of acinar tissue (black asterisk), its disorganization and duct hyperactivity (white asterisk). (H) Control exocrine pancreas Alcian
Blue-negative samples showing organized acinar and duct structure. (I) End-stage mucinous Alcian Blue-positive pancreatic adenocarcinoma with goblet cells
and small duct-like structures interspersed in a disorganized acinar tissue. Tumors shown in D, E, F and I were seen in 7 out of 15 analyzed samples between
3 and 5 mpf. Tumor shown in G was seen in 8 out of 15 analyzed samples between 3 and 5 mpf. (B′,D′,E′) More detailed magnifications of B (right), D and E.
Arrowheads indicate acinar cells in B′, goblet cells in D′ and duct-like structures in E′. Age of analyzed samples is indicated. Scale bars: 50 μm.
recognizable acinar morphology and an increasing number of
ductal-like structures inside the exocrine pancreas (Fig. 2G).
To explore the reason for strong motility defects and decrease in
survival of juveniles between 3 and 6 wpf, we performed H&E
staining on 35 out of 120 fishes expressing eGFP-KRASG12D in
886
cerebellum. As expected by eGFP-KRASG12D tissue-specific
expression driven by ptf1a, we found the hallmark features of an
undifferentiated medulloblastoma in both the external granular layer
and ventricular zone in 5 out of 35 samples analyzed at 1 mpf
(Aldinger and Elsen, 2008; Pascual et al., 2007). In contrast to
Disease Models & Mechanisms
RESEARCH ARTICLE
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
control cerebellum of Tg(ptf1a:eGFP) fish (supplementary material
Fig. S1A,B), the tumor appears as sheet-like areas of small, round,
blue cells with scant cytoplasm and dense hyperchromatic nuclei
(supplementary material Fig. S1C,D), resembling classic human
cerebellar cancer (supplementary material Fig. S1E,F), which
originates from granule cell progenitors located in the external
granular layer (EGL) of the cerebellum (Marino, 2005). The other
30 collected samples, regularly expressing eGFP-KRASG12D in
cerebellum, showed a normal histology with a few interspersed cells
containing dense hyperchromatic nuclei, as seen in cerebellum
dysplasia (data not shown).
Cell hyperproliferation inside tumor lesions, low degree of apoptosis
and changes in the tumor microenvironment (such as inflammation
burst, neoangiogenesis, increase in cell stemness and EMT) are the
main hallmarks of pancreatic cancer progression (Bardeesy and
DePinho, 2002; Hanahan and Weinberg, 2011). In order to analyze
some of these, we performed histology and immunohistochemistry
(IHC) assays on 7 out of 15 collected pancreatic tumors.
To understand the progressive enlargement of eGFP-KRASG12Dpositive masses, we assessed cell proliferation by using an antibody
that specifically labels the PCNA (proliferating cell nuclear antigen)
protein, a specific marker for cells in S phase of the cell cycle.
Significantly higher expression levels of PCNA were seen in
eGFP-KRASG12D-positive masses, confirming a robust cell
hyperproliferation during tumor development (Fig. 3A,B, Fig. 4;
supplementary material Fig. S3B′).
In order to specifically label mesenchymal and epithelial cells, we
analyzed the expression of E-Cadherin, N-Cadherin and Vimentin.
The E-Cadherin to N-Cadherin switch is linked to the increase in
mesenchymal features and is also involved in the mechanism of
EMT (Nakajima et al., 2004; Rhim et al., 2012). It has been well
described in pancreatic tumor progression and also during
transdifferentiation processes such as the acinar to ductal metaplasia
(Wu et al., 2012). E-Cadherin is a transmembrane protein involved
in cell-cell adhesion. Whereas normally produced in acinar tissue of
Tg(ptf1a:eGFP), we observed a slight decrease in E-Cadherin
expression in acinar tissue of Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D
injected fish (Fig. 3D). As shown in Fig. 4, reduction in the
expression of E-Cadherin was not statistically significant in the
pancreatic adenocarcinoma, possibly due to the normal levels of
expression in the duct and endothelial cells, which hindered the
differences between the exocrine pancreas and tumor mass (Fig. 3D;
supplementary material Fig. S3D′).
N-Cadherin is a transmembrane protein mainly expressed during
embryonic development and commonly found in cancer cells. It
supports a mechanism of transendothelial migration leading to
metastasis (Goonesinghe et al., 2012; Lirdprapamongkol et al.,
2012; Scanlon et al., 2013). IHC analysis showed a statistically
significant increase in N-Cadherin expression in eGFP-KRASG12Dpositive clusters, whereas its expression was totally absent in the
pancreas of Tg(ptf1a:eGFP) controls (Fig. 3E,F, Fig. 4;
supplementary material Fig. S3F′).
Analysis of mesenchymal markers was completed by observing
the expression of Vimentin, the major cytoskeletal component of
mesenchymal cells, previously used as the main sign of cells
undergoing EMT (Savagner, 2010). We observed a Vimentin
increase during pancreatic tumor progression in all 7 samples
examined that were collected between 14 and 24 wpf
(Fig. 3G,H′,H′′). Not all Vimentin-positive cells are also eGFP-
Fig. 3. Cell hyperproliferation, EMT and apoptosis induction during
pancreatic cancer. Markers of EMT (E-Cadherin, N-Cadherin and Vimentin),
cell proliferation (PCNA) and apoptosis were assessed by
immunohistochemistry on pancreatic tissue of both control Tg(ptf1a:eGFP)
and tumor-prone Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D lines. (A,B) The cell
proliferation marker PCNA (white arrowheads) is increased in tumor-prone
fish (B) compared with controls (A). Cropped image evidenced by white
square is reported in supplementary material Fig. S3B′. (C,D) No differences
in E-Cadherin (white arrowheads) expression was shown between pancreas
of tumor-prone fish (D) compared with controls (C). Cropped image
evidenced by white square is reported in supplementary material Fig. S3D′.
(E,F) N-Cadherin (white arrowheads) is highly expressed in
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D fish (F) whereas it is almost absent in
controls (E). Cropped image evidenced by white square is reported in
supplementary material Fig. S3F′. (G-H′′) A slight increase in Vimentin
expression (white arrowheads) was seen in both tumor mass and stroma of
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D fish (H′,H′′) whereas it resulted low
expressed in controls (G). Apoptosis level was higher in tumor-prone fish (J)
than in controls (I) as indicated by white arrowheads in J. Analyses were
performed on 7 tumor and 7 control samples. All samples were analyzed at
3 mpf. Scale bars: 100 μm.
887
Disease Models & Mechanisms
Activation of EMT, proliferation and apoptosis during
pancreatic cancer progression
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
KRASG12D-positive, indicating that the increase in mesenchymal
cells is detected not only in tumor mass but also in the tumor stroma
(Fig. 3H′). Thus, in agreement with previous data, we observed an
increased expression of Vimentin in mesenchymal cells arising
during cancer progression (Rhim et al., 2012; Satelli and Li, 2011).
Except for the slight E-Cadherin decrease, all the results obtained
comparing EMT and proliferation markers in eGFP-KRASG12Dpositive pancreatic tumors were statistically significant compared
with control pancreas (Fig. 4).
Finally, to assess apoptosis we used the TUNEL (terminal
deoxynucleotidyl transferase dUTP nick end labeling) assay.
Compared with control Tg(ptf1a:eGFP), we observed a significant
increase in apoptosis in pancreatic adenocarcinoma of
Tg(ptf1a:Gal4;UAS:eGFP-KRASG12D) injected animals (Fig. 3I,J;
Fig. 4).
In vivo reporter analysis of the role of Shh, Notch and TGFβ
in pancreatic tumors
Models of pancreatic adenocarcinoma have been useful in identifying
genetic, molecular and biochemical processes regulating tumor
progression (Salnikov et al., 2012). A wide variety of alterations in
signaling pathways play pivotal roles in the pathogenesis of
pancreatic tumors, affecting acinar or epithelial compartments
together with the surrounding stromal microenvironment
888
(McCleary-Wheeler et al., 2012). KRAS gene mutations are
considered the main cause of pancreatic adenocarcinoma onset
together with hyperactivation of the Shh signaling pathway (Jung
et al., 2011; Park et al., 2008). Other molecular pathways
involved during embryonic development, such as Notch and TGFβ,
have been shown to be unbalanced in pancreatic adenocarcinoma
but, for most, their specific role in tumor development remains to
be investigated. We attempted to answer these questions by
using specific transgenic lines Tg(2xID1BRE:nlsmCherry)ia17;
Tg(12xSBE:nlsmCherry)ia15; Tg(EPV.Tp1-Mmu.Hbb:nlsmCherry)ia7;
Tg(12xGli-Hsv.Ul23:nlsmCherry)ia10, reporting Bmp, TGFβ, Notch
and Shh signaling pathways, respectively. All the lines were
previously characterized for a correct reporter expression by using
specific signaling inhibitors: LDN193189 for Bmp; SB431542 for
TGFβ; DAPT for Notch; and cyclopamine for Shh (supplementary
material Fig. S4). These lines were outcrossed with Tg(ptf1a:Gal4)
and injected with Tol2(UAS:eGFP-KRASG12D) as described in
supplementary material Fig. S5. In order to identify which of these
pathways are involved in pancreatic cancer onset, we performed
confocal microscope analysis at 3, 5, 7, 30 and 60 days post
fertilization (dpf), by using 10 eGFP-positive larvae from
Tg(ptf1a:eGFP) and Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D injected
lines for each time point. Significance of data from each time point
was confirmed by using the ANOVA test.
Disease Models & Mechanisms
Fig. 4. Significant changes in levels of
mesechymal, apoptosis and
proliferation markers. Quantification of
IHC results by counting double-positive
cells expressing eGFP-KRASG12D and
PCNA, E-Cadherin, TUNEL, N-Cadherin or
Vimentin in both Tg(ptf1a:eGFP) and
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D.
Results are reported as percentage of
double eGFP-KRASG12D/PCNA, eGFPKRASG12D/E-Cadherin, eGFP-KRASG12D /
TUNEL, eGFP-KRASG12D/N-Cadherin and
eGFP-KRASG12D/Vimentin-positive cells
normalized to DAPI counts. Results from 7
Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D
samples were all statistically significant,
compared with 7 control Tg(ptf1a:eGFP)
samples, except for E-Cadherin values.
Error bars indicate s.e.m. for all analyzed
samples; *P<0.05, **P<0.01 according to
uncoupled Student’s t-test.
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
In agreement with previous evidence showing that deregulation
of Wnt signaling occurs later in pancreatic tumor development and
differentiation, linked to neo-vessels formation and tumor metastasis
(Lowy et al., 2003; Morris et al., 2010), we did not find any Wnt/catenin reporter activity during pancreatic tumor onset (not shown).
We analyzed the Bmp signaling pathway, which is rarely involved
in pancreatic tumor development, in an indirect manner, mediated
by the PI3K/Akt pathway rather than Ras/MAPK (Chen et al., 2011;
Handra-Luca et al., 2012; Virtanen et al., 2011). At 30 and 60 dpf,
we did not observe any changes in Bmp signaling response in
eGFP-KRASG12D fish compared with Tg(ptf1a:eGFP) controls
(Fig. 5H). This result is in agreement with other studies, showing
that Bmp is more needed in pancreatic development and
differentiation during embryonic and larval stages than in tumor
onset (Tiso et al., 2002; Xu et al., 2011).
We then verified whether Notch signaling, reported to be
upregulated during both chronic inflammation and early stages of
tumorigenesis, could play a main role (Su et al., 2006). We observed
a strong increase in Notch reporter activity in acinar cells of the
exocrine pancreas from Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D
injected animals (Fig. 5C,D) compared with Tg(ptf1a:eGFP)
controls in which the reporter expression was totally ductal
(Manfroid et al., 2012). By quantitative analysis, we documented a
significant increase in Notch reporter activity at 5 dpf, which
consistently remained upregulated at 60 dpf (Fig. 5F). Thus, our
observation suggested an elevated Notch signaling activity during
onset and progression of pancreatic cancer.
We next investigated the role of the Smad3/TGFβ signaling
pathway by analyzing a specific mCherry reporter driven by
Smad3 responsive elements; Smad3 is reported to be activated in
pancreatic adenocarcinoma as a response to the antiproliferating
function of TGFβ (Ungefroren et al., 2011). We observed a strong
increase in TGFβ signaling responsive cells during pancreatic
cancer onset between 30 and 60 dpf (Fig. 5A,B). In particular,
in vivo analysis of the Smad3/TGFβ pathway in
Tg(12xSBE:nlsmCherry)ia15 fish at 2 mpf showed a strong increase
889
Disease Models & Mechanisms
Fig. 5. TGFβ, Notch, Shh and Bmp
signaling activities during pancreatic
cancer onset. TGFβ, Notch, Shh and Bmp
signaling activities were evaluated by
observing mCherry expression using confocal
microscope analysis. (A-D) Pictures of full Zstacks from 30 and 60 dpf samples. Controls
obtained by outcrossing the Tg(ptf1a:eGFP)
zebrafish line with TGFβ (A) and Notch (C)
mCherry reporter lines. TGFβ (B) and Notch
(D) signaling activity in tumor lines. (EH) Quantitative analysis of eGFP-KRASG12Dpositive cells expressing TGFβ (E), Notch (F),
Shh (G) and Bmp (H) mCherry reporters by
VOLOCITY 6.0 software. For each time point,
10 eGFP-positive samples were analyzed.
Results for TGFβ (E) and Notch (F) signaling
were statistically significant. Error bars
indicate s.e.m. for all analyzed samples;
**P<0.01, ***P<0.005 according to ANOVA
test.
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
in mCherry expression in three out of five fish. In these fish the
reporter activity was clearly detectable in virtually all eGFPKRASG12D-positive cells, indicating a possible KRASG12Ddependent activity for Smad2/3/4 (supplementary material
Movie 1). Quantitative analysis of mCherry expression showed
that the slight increase in TGFβ response seen in
Tg(12xSBE:nlsmCherry)ia15 fish was not significant at 30 dpf,
whereas it started to be significant at 2 mpf (Fig. 5E). These data
demonstrated that TGFβ is particularly involved during early
stages of tumor development (Heldin and Moustakas, 2012; Truty
and Urrutia, 2007). We also studied the expression of TGFβ
at later stages of pancreatic tumor by performing the analyses
on fish at 3-6 mpf. Our results confirmed the activation of
the Smad3/TGFβ pathway, showing a strong increase in
12xSBE:mCherry expression rising at 16 wpf both in tumor mass
(eGFP-KRASG12D-positive cells) and stromal microenvironment
(eGFP-KRASG12D-negative cells) (supplementary material Fig.
S6A,B; Movie 2).
Finally, during the early stages of carcinogenesis, the Shh reporter
(i.e. the mCherry reporter linked to Gli1 responsive element)
showed no Shh signaling until 30 dpf with a very slight increase at
2 mpf in tumor stroma (Fig. 5G).
The involvement of the Smad3/TGFβ signaling pathway during
pancreatic tumor progression was confirmed by real-time PCR,
showing a significant increase (P<0.005) in mCherry mRNA levels
in Tg(ptf1a:Gal4)/UAS:eGFP-KRASG12D compared with the control
(Fig. 6A). During later stages of pancreatic tumor, we also found a
strong expression of specific genes related to stemness (e.g. cdh2
(N-Cadherin) and nestin) (Fig. 6B), confirming the activation of
EMT. Besides a significant increase in p53 onco-suppressor and
TGFβR1 mRNA inside the eGFP-KRASG12D-positive tumor mass
(Fig. 6C), other pathways were directly or indirectly involved in
pancreatic tumor progression: Notch and Wnt pathway activities
increased at later stages of pancreatic tumor, as demonstrated by a
890
significant increase in notch1a, her9, myc and cyclinD1 mRNA
levels (Fig. 6D,E).
In order to verify whether the unbalanced expression of Shh, Notch
and TGFβ signaling pathways was similar in different tumors induced
by KRASG12D, we compared our reporter analyses on pancreatic
adenocarcinoma with MDB data. In particular, to verify the reporter
activities during MDB onset and progression, we analyzed eGFPKRASG12D and mCherry expression by confocal microscopy at 3, 5,
7 and 30 dpf (supplementary material Fig. S7). We observed the
inhibition of canonical Notch signaling and the increase in Smad3bmediated TGFβ signaling at tumor onset (supplementary material Fig.
S7). This is in agreement with results previously described in human
and mouse models of MDB and mouse models of MDB (Aref et al.,
2013; Guessous et al., 2008; Rodini et al., 2010). Smad3/TGFβ
activity continued to be sustained till 30 dpf with a very slow
decrease. Although we detected a general inhibition of canonical
Notch signaling pathway during MDB development, basal Notch
activity in KRASG12D-positive cells was observed at 30 dpf (MDB
later stages), suggesting that Notch might be involved in MDB
progression (supplementary material Fig. S8). Sustained Smad3/TGFβ
and Shh activities in KRASG12D-positive cells were observed,
underscoring the importance of these pathways in both onset and
progression of MDB (supplementary material Fig. S6C,D; Movie 3).
We also detected a rise in mesenchymal marker expression, as
reported by significant increases in nestin and notch1b signaling
together with a decrease in gfap, olig4, olig2 and ngn1. Significant
increase in stat1b and TGFβ signaling and a strong reduction in p53
expression (supplementary material Fig. S9) were compatible with
hyperproliferation of KRASG12D-positive cells responsible for MDB
progression (Rodini et al., 2010).
DISCUSSION
In the first zebrafish model of pancreatic cancer, a mutant KRAS
oncogene alone was able to induce pancreatic adenocarcinoma of
Disease Models & Mechanisms
Fig. 6. Expression of genes and
activation of pathways involved
in pancreatic adenocarcinoma
progression (real-time PCR data).
Real-time PCR assay was used to
confirm IHC data. (A,B) Increasing
mRNA levels of TGFβ:mCherry
reporter (A), cdh2 and nestin
(B) were significant in
Tg(ptf1a:Gal4;UAS:eGFP-KRASG12D)
compared with the control
Tg(ptf1a:eGFP. (C-E) Additional
real-time PCR experiments revealed
significantly unbalanced expression
of several genes including tp53,
TGFbR1b (C), notch1a, cyclinD1,
her9 (D) and MYCn (E). Results are
reported as mRNA level normalized
to -actin. Error bars indicate s.e.m.
for all analyzed data. Both Student’s
t-tests and ANOVA tests were
performed to statistically analyze
data; *P<0.05, **P<0.01, ***P<0.005
comparing data between
Tg(ptf1a:Gal4;UAS:eGFP-KRASG12D)
and Tg(ptf1a:eGFP).
predominantly acinar or mucinous phenotype (Park et al., 2008). To
develop this cancer model, a transgenic BAC system expressing
eGFP linked to oncogenic KRASG12V in zebrafish pancreas under
control of ptf1a promoter elements was generated. In this work, we
were able to reproduce in zebrafish several KRASG12D-dependent
pancreatic cancers by using a conditional Gal4/UAS expression
system. This strategy allowed us to follow pancreatic carcinogenesis
both in space and time. Furthermore, we used the KRASG12D rather
than KRASG12V mutated oncogene because of its major expression
frequency in human cancers. KRASG12D is also known to give rise to
the most aggressive types of pancreatic adenocarcinomas and has a
major role in cancer onset and maintenance (Rachagani et al., 2011).
In agreement with the previously generated fish model, we
reproduced pancreatic cancers displaying some similarities and
some differences to both human and mouse models. Similarly to
human and mouse, zebrafish pancreatic tumors showed a highly
aggressive behavior and propensity for metastatic spread, as
demonstrated by an increased expression of mesenchymal markers
and infiltration of adjacent organs (Fig. 2 and Fig. 3), which
ultimately led to the death of carrier animals. Unlike most human
pancreatic cancers, however, we frequently observed features of
non-ductal differentiation. In fact, both histological and
immunohistochemical analyses confirmed a mixed acinar/ductal
carcinoma of predominant acinar or mucinous phenotypes,
becoming ductal at later stages of tumor development (Fig. 2)
(Maitra and Leach, 2012; Park et al., 2008; Stanger et al., 2005).
Moreover, Park et al. found that pancreatic progenitor cells
expressing oncogenic KRAS under control of ptf1a promoter
undergo normal specification and migration, but fail to differentiate
(Park et al., 2008). This block in differentiation results in abnormal
persistence of an undifferentiated progenitor pool. We confirmed
this result by observing the upregulation of N-Cadherin and
Vimentin mesenchymal markers (Figs 3, 4; supplementary material
Fig. S3). Further, we found the upregulation of PCNA and p53
(Figs 3, 4, 6), demonstrating the cell hyperproliferation at the tumor
site, a feature already reported in mouse and human tumors (Conradt
et al., 2012; Ghosh and Leach, 2011; Gironella et al., 2007).
Moreover, we tried to unveil the most relevant molecular
pathways involved in KRAS-mediated pancreatic tumor
development by taking advantage of transgenic zebrafish reporter
lines expressing responsive elements known to be pathway specific
(supplementary material Fig. S4) (Moro et al., 2013). According to
what has already been shown ex vivo on human pancreatic cancer
(Bailey and Leach, 2012; De La O et al., 2008; Hu et al., 2012), we
confirmed in vivo the upregulation of the Notch signaling pathway
during tumor onset (Fig. 5), followed by the upregulation of the
TGFβ/Smad3 pathway during tumor progression (Fig. 5;
supplementary material Movies 1, 2). Nevertheless, the Notch
pathway also remained active at later stages, possibly through the
upregulation of Notch1a, as demonstrated by real-time PCR (Fig. 6)
and confirming previous observations in human and mouse
(Mysliwiec and Boucher, 2009; Wang et al., 2006b). Another
evidence of TGFβ and Notch signaling upregulation at later stages
of pancreatic adenocarcinoma was the concomitant upregulation of
the gene encoding the TGFβ- and Notch-related target CyclinD1
(Fig. 6), involved both in neo-vessel formation and cancer
metastases during the later stages of tumor development (Kornmann
et al., 1999; Wang et al., 2006a; Wang et al., 2006b).
All described data demonstrate in vivo the leading role of the
TGFβ pathway during mid and late stages of tumor progression. In
particular, TGFβ started to increase at 1 mpf, reaching maximum
expression at 2 mpf when mCherry reporter activity for TGFβ was
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
detectable in most of the eGFP-KRASG12D-positive cells (Fig. 5;
supplementary material Movie 1). At later stages, as confirmed also
by real-time PCR data (Fig. 6), levels of TGFβ signaling remained
high but its upregulated activity was particularly observed in stromal
cells surrounding the tumor (supplementary material Movie 2; Fig.
S6). These two different spatiotemporal patterns of TGFβ expression
confirm the idea that TGFβ has a double role: at early stages of
carcinogenesis, the activity of TGFβ in most KRASG12D-positive
cells is likely to be post-mitotic, possibly acting as tumor suppressor;
by contrast, at later stages, TGFβ expression in stromal cells
surrounding the tumor could indicate a proinflammatory and
oncogenic effect (McCleary-Wheeler et al., 2012). Furthermore, we
observed high activity of the Notch pathway during pancreatic
tumor onset, whereas Shh activity remained unaffected until 2 mpf
when it started to be induced in a few eGFP-KRASG12D-positive
cells. These results demonstrate the progressive increase in TGFβ
and Shh activities following KRASG12D constitutive activation
during mid and advanced stages of pancreatic cancer progression,
respectively.
Although there is little evidence demonstrating the involvement
of mutated KRAS in MDB development (Gilbertson et al., 2006), we
were able to reproduce a cancer model with features of pediatric
MDB due to the specific expression of KRASG12D in ptf1aexpressing tissues (supplementary material Figs S1, S5-S9). Fish
harboring medulloblastoma started to show strong motility
disruption at 15 dpf and died at 45 dpf. To confirm the importance
of TGFβ, Shh and Notch signaling pathways during carcinogenesis,
we evaluated their activity during MDB onset and progression in
parallel with pancreatic adenocarcinoma. As in pancreatic
adenocarcinoma, we found a progressively increasing number of
eGFP-KRASG12D-positive cells activating TGFβ during MDB
development. This result suggests that TGFβ is possibly regulated
by the same mechanism, involving constitutive activation of KRAS
oncogene during carcinogenesis of both pancreatic adenocarcinoma
and MDB. Conversely, Notch and Shh signaling activities were
observed to be different in the two cancers. In particular, Notch
signaling was inhibited at early stages of MDB development,
whereas in pancreatic adenocarcinoma, the co-activation of Notch
and KRASG12D in acinar cells might be involved in acinar to ductal
metaplasia during the early stages of carcinogenesis (De La O et al.,
2008). We observed high Shh activity at MDB onset (supplementary
material Movie 3; Fig. S6), confirming what is already known for
human Shh-dependent MDBs (Bhatia et al., 2012). Notably, no
Shh activity was seen during pancreatic adenocarcinoma onset. In
other words, by comparing pancreatic adenocarcinoma with
medulloblastoma, we were able to postulate that Smad3/TGFβ and
Notch activities might be linked to the constitutive activation of
KRASG12D, the activity of Smad3/TGFβ and Notch reporters being
present in KRASG12D-expressing cells in both types of cancer. By
contrast, Shh activity was upregulated in KRASG12D-positive
cerebellum but not in the pancreas where it was confined to the
stroma of the pancreatic tumor at later stage of carcinogenesis, in
agreement with previous observations (Park et al., 2008).
These results suggest that zebrafish cancer models, coupled with
transgenic zebrafish reporter lines, are powerful tools for the
analysis in vivo of the initiating events of pancreatic tumorigenesis
and the sequences of hallmark progression during cancer
development. Moreover, Notch, TGFβ and Shh ligands and their
effectors could be interesting targets for high-throughput screening
of drugs to be used in efficacious combined or sequential therapy of
several malignant diseases such as MDB and pancreatic
adenocarcinoma.
891
Disease Models & Mechanisms
RESEARCH ARTICLE
RESEARCH ARTICLE
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
MATERIALS AND METHODS
Quantitative RT-PCR and statistical analysis
Generation of transgenic reporter zebrafish lines for Shh, Bmp,
TGFβ and Notch pathways
Total mRNA was isolated from the pancreas and cerebellum of zebrafish
embryos using Trizol (Invitrogen, Carlsbad, CA) and 0.5 μg of total RNA
reverse-transcribed using SuperScript RNaseH-Reverse Transcriptase
(Invitrogen, Carlsbad, CA). Quantitative RT-PCR reactions were run in
triplicate using Brilliant® SYBR® Green QPCR Core Reagent Kit
(Stratagene, La Jolla, CA). Fluorescent emission was recorded in real time
(Sequence Detection System 7900HT, Applied Biosystems, Carlsbad, CA).
Gene expression analysis was completed using the comparative Ct method
of relative quantification (Spatuzza et al., 2008). PCR amplification
conditions consisted of 40 cycles with primer annealing at 60°C. Sequences
of specific primers used in this work are listed in supplementary material
Table S1. Primers were designed using the software Primer 3
(http://bioinfo.ut.ee/primer3-0.4.0/input.htm). PCR amplicons were
previously evaluated on agarose gel and, during SYBR green analyses,
primer dissociation curves were checked in each run to ensure primer
specificity on human and zebrafish mRNA. Relative RNA quantities were
normalized to β-actin.
Statistical analyses to compare results for Tg(ptf1a:Gal4)/UAS:eGFPKRASG12D injected samples and Tg(ptf1a:eGFP) controls were performed
using uncoupled Student’s two-tailed t-test and Microsoft Excel 2011 or
Prism GraphPad software package. The ANOVA test was performed to
analyze statistical differences in pathway expression at different time points
of cancer development.
Generation of pancreatic adenocarcinoma zebrafish model and
analysis of molecular pathways involved in tumor onset
We injected Tol2(UAS:eGFP-KRASG12D) plasmid into one-cell stage
embryos derived from incrosses or outcrosses of Tg(ptf1a:Gal4)jh16
zebrafish line with signaling mCherry reporter lines. Zebrafish
reporter lines used in this work were Tg(2xID1BRE:nlsmCherry)ia17,
Tg(12xSBE:nlsmCherry)ia15, Tg(EPV.Tp1-Mmu.Hbb:nlsmCherry)ia7 and
Tg(12xGli-HSV.Ul23:nlsmCherry)ia10 (supplementary material Fig. S5).
Approximately 150 embryos were raised, all expressing eGFP according
to the expected ptf1a pattern. Transcutaneous eGFP expression was
evaluated at 1-week intervals until development of a tumor mass, when
fishes were euthanized for further histological and histochemical
evaluation. Larvae at 3, 5, 7 dpf and 1 mpf were photographed live using
a NIKON C2 H600L confocal microscope with 20× and 40× water
dipping objectives. Lasers used to excite fluorophores were 488 nm for
eGFP and 561 nm for mCherry. The number of single eGFP-KRASG12Dpositive cells also expressing the mCherry reporter was calculated using
VOLOCITY 6.0 software (PerkinElmer, Waltham, MA) on eGFP-positive
confocal acquired images. Embryos and larvae were anesthetized using
Tricaine and mounted in 0.8% low melting agarose on a glass lid before
photographing. The project, with protocol number 18746, was examined
and approved by the Ethical Committee of the University of Padua.
Histology and immunohistochemistry of tumors
Pancreas and cerebellum of eGFP-KRASG12D-positive fish and
Tg(ptf1a:eGFP) controls were dissected and fixed in PBS containing 4%
paraformaldehyde overnight at +4°C. Collected tissues were paraffinembedded, cut as 1 μm slices and examined histologically using the standard
H&E and Alcian Blue methods to analyze cell morphology and tissue
structure. Tissue slices were immunostained with DAPI, to label cell nuclei,
and antibodies anti-E-Cadherin (ab53033, Abcam, Cambridge, UK), anti-NCadherin (ab12221, Abcam, Cambridge, UK), anti-Vimentin (M7020, Dako,
Glostrup, Denmark), and anti-PCNA (M0879, Dako, Glostrup, Denmark),
according to standard procedures. A TUNEL assay protocol (Invitrogen,
Carlsbad, CA) was used to detect apoptosis.
892
This article is part of a Special Issue, Spotlight on Zebrafish: Translational Impact.
See all the articles in the issue at http://dmm.biologists.org/content/7/7.toc.
Acknowledgements
We acknowledge Prof. Giuseppe Basso, Dott. Luigi Pivotti and Dott. Martina
Milanetto for their kind help in this work.
Competing interests
The authors declare no competing financial interests.
Author contributions
M.S., E.R. and F.A. conceived and designed experiments; M.S., E.R., G.B. and
L.P. performed experiments; M.S., E.R. and F.A. analyzed the data; E.M., A.C.,
S.L., S.D.L. and N.T. contributed reagents/materials/analysis tools; and M.S., E.M.,
N.T. and F.A. wrote the paper.
Funding
The work is supported by the European Union [grant number ZF-HEALTH CT2010-242048], the Cariparo Project ‘An in vivo reporter platform for cancer studies
and drugs screening’, the AIRC Project IG 10274 and the Ministry of Health [grant
number RF-2010-2309484].
Supplementary material
Supplementary material available online at
http://dmm.biologists.org/lookup/suppl/doi:10.1242/dmm.014969/-/DC1
References
Aguirre, A. J., Bardeesy, N., Sinha, M., Lopez, L., Tuveson, D. A., Horner, J.,
Redston, M. S. and DePinho, R. A. (2003). Activated Kras and Ink4a/Arf deficiency
cooperate to produce metastatic pancreatic ductal adenocarcinoma. Genes Dev. 17,
3112-3126.
Aldinger, K. A. and Elsen, G. E. (2008). Ptf1a is a molecular determinant for both
glutamatergic and GABAergic neurons in the hindbrain. J. Neurosci. 28, 338-339.
Aref, D., Moffatt, C. J., Agnihotri, S., Ramaswamy, V., Dubuc, A. M., Northcott, P.
A., Taylor, M. D., Perry, A., Olson, J. M., Eberhart, C. G. et al. (2013). Canonical
TGF-β pathway activity is a predictor of SHH-driven medulloblastoma survival and
delineates putative precursors in cerebellar development. Brain Pathol. 23, 178-191.
Bailey, J. M. and Leach, S. D. (2012). Signaling pathways mediating epithelialmesenchymal crosstalk in pancreatic cancer: Hedgehog, Notch and TGFβ. In
Pancreatic Cancer and Tumor Microenvironment (ed. P. J. Grippo and H. G.
Munshi), Chapter 7. Trivandrum, India: Transworld Research Network.
Bardeesy, N. and DePinho, R. A. (2002). Pancreatic cancer biology and genetics.
Nat. Rev. Cancer 2, 897-909.
Bardeesy, N., Cheng, K. H., Berger, J. H., Chu, G. C., Pahler, J., Olson, P., Hezel,
A. F., Horner, J., Lauwers, G. Y., Hanahan, D. et al. (2006). Smad4 is dispensable
for normal pancreas development yet critical in progression and tumor biology of
pancreas cancer. Genes Dev. 20, 3130-3146.
Bhatia, B., Potts, C. R., Guldal, C., Choi, S., Korshunov, A., Pfister, S., Kenney, A.
M. and Nahlé, Z. A. (2012). Hedgehog-mediated regulation of PPARγ controls
Disease Models & Mechanisms
The generation of all reporter vectors was performed by using the Multisite
Gateway-based construction kit (Tol2 kit), provided by Chi-Bin Chien of the
University of Utah (USA) (Kawakami et al., 2004; Kwan et al., 2007). All
sequences for the responsive elements were PCR amplified using oligos
carrying the HindIII and BamHI restriction sites and cloned into the
p5EMCS vector. For generation of a transgene containing a TGFβresponsive element, a (CAGA)12 box (also called 12xSBE for Smad3
binding element) was selectively amplified with adaptor oligos (Dennler et
al., 1998). To generate a Notch reporter transgenic line, a Notch-responsive
box (donated by Nathan Lawson, University of Massachusetts Medical
School, Worcester, MA) containing 12× multimerized oligonucleotide
sequences, corresponding to 12 copies of the Epstein Barr Virus terminal
protein 1 (TP1) promoter (12 Rbp-Jκ binding sites), has been integrated into
the zebrafish germline (Parsons et al., 2009). To obtain the vector carrying
Bmp-responsive elements, a transcription cassette (−1046;-863) with the
minimal promoter region and the TATA box of the human ID1 gene was
isolated from the plasmid pID183 (López-Rovira et al., 2002). The Shhresponsive construct is described elsewhere (Corallo et al., 2013). Each
p5EMCS containing the responsive elements was recombined with middle
pME-MCS vectors carrying the fluorescent protein coding sequences
(GFP/mCherry). The final vectors containing the reporter were sequenced
on both strands to verify the correct orientation of the responsive elements,
minimal promoter, fluorescent protein coding sequence and SV40 poly-A.
A pCS2FA-transposase construct necessary to generate transposase mRNA
was included in the Tol2 Kit. To finally obtain the transgenic reporters lines,
eggs derived from wild-type zebrafish were co-injected with transposase
mRNA and reporter plasmids. Larvae were analyzed with fluorescence
microscopy at different developmental stages. Stable lines were isolated
from F1 carriers.
metabolic patterns in neural precursors and shh-driven medulloblastoma. Acta
Neuropathol. 123, 587-600.
Chen, X., Liao, J., Lu, Y., Duan, X. and Sun, W. (2011). Activation of the PI3K/Akt
pathway mediates bone morphogenetic protein 2-induced invasion of pancreatic
cancer cells Panc-1. Pathol. Oncol. Res. 17, 257-261.
Conradt, L., Henrich, A., Wirth, M., Reichert, M., Lesina, M., Algul, H., Schmid, R.
M., Kramer, O. H., Saur, D. and Schneider, G. (2012). Mdm2 inhibitors synergize
with topoisomerase II inhibitors to induce p53-independent pancreatic cancer cell
death. Int. J. Cancer 132, 2248-2257.
Corallo, D., Schiavinato, A., Trapani, V., Moro, E., Argenton, F. and Bonaldo, P.
(2013). Emilin3 is required for notochord sheath integrity and interacts with Scube2
to regulate notochord-derived Hedgehog signals. Development 140, 4594-4601.
Corcoran, R. B., Contino, G., Deshpande, V., Tzatsos, A., Conrad, C., Benes, C.
H., Levy, D. E., Settleman, J., Engelman, J. A. and Bardeesy, N. (2011). STAT3
plays a critical role in KRAS-induced pancreatic tumorigenesis. Cancer Res. 71,
5020-5029.
De La O, J. P., Emerson, L. L., Goodman, J. L., Froebe, S. C., Illum, B. E., Curtis,
A. B. and Murtaugh, L. C. (2008). Notch and Kras reprogram pancreatic acinar
cells to ductal intraepithelial neoplasia. Proc. Natl. Acad. Sci. USA 105, 1890718912.
Dennler, S., Itoh, S., Vivien, D., ten Dijke, P., Huet, S. and Gauthier, J. M. (1998).
Direct binding of Smad3 and Smad4 to critical TGF beta-inducible elements in the
promoter of human plasminogen activator inhibitor-type 1 gene. EMBO J. 17, 30913100.
Elston, R. and Inman, G. J. (2012). Crosstalk between p53 and TGF-β Signalling. J.
Signal Transduct. 2012, 294097.
Esposito, I., Seiler, C., Bergmann, F., Kleeff, J., Friess, H. and Schirmacher, P.
(2007). Hypothetical progression model of pancreatic cancer with origin in the
centroacinar-acinar compartment. Pancreas 35, 212-217.
Etchin, J., Kanki, J. P. and Look, A. T. (2011). Zebrafish as a model for the study of
human cancer. Methods Cell Biol. 105, 309-337.
Ghosh, B. and Leach, S. D. (2011). p53: Guardian of pancreatic epithelial identity. Cell
Cycle 10, 1717.
Gilbertson, R. J., Langdon, J. A., Hollander, A., Hernan, R., Hogg, T. L., Gajjar, A.,
Fuller, C. and Clifford, S. C. (2006). Mutational analysis of PDGFR-RAS/
MAPK pathway activation in childhood medulloblastoma. Eur. J. Cancer 42, 646649.
Gironella, M., Seux, M., Xie, M. J., Cano, C., Tomasini, R., Gommeaux, J., Garcia,
S., Nowak, J., Yeung, M. L., Jeang, K. T. et al. (2007). Tumor protein 53-induced
nuclear protein 1 expression is repressed by miR-155, and its restoration inhibits
pancreatic tumor development. Proc. Natl. Acad. Sci. USA 104, 16170-16175.
Goldsmith, J. R. and Jobin, C. (2012). Think small: zebrafish as a model system of
human pathology. J. Biomed. Biotechnol. 2012, 817341.
Goonesinghe, A., Luan, X. M., Hurlstone, A. and Garrod, D. (2012). Desmosomal
cadherins in zebrafish epiboly and gastrulation. BMC Dev. Biol. 12, 1.
Guessous, F., Li, Y. and Abounader, R. (2008). Signaling pathways in
medulloblastoma. J. Cell. Physiol. 217, 577-583.
Hanahan, D. and Weinberg, R. A. (2011). Hallmarks of cancer: the next generation.
Cell 144, 646-674.
Handra-Luca, A., Hammel, P., Sauvanet, A., Ruszniewski, P. and Couvelard, A.
(2012). Tumoral epithelial and stromal expression of SMAD proteins in pancreatic
ductal adenocarcinomas. J. Hepatobiliary Pancreat. Sci. 20, 294-302.
Heldin, C. H. and Moustakas, A. (2012). Role of Smads in TGFβ signaling. Cell
Tissue Res. 347, 21-36.
Herreros-Villanueva, M., Hijona, E., Cosme, A. and Bujanda, L. (2012a). Mouse
models of pancreatic cancer. World J. Gastroenterol. 18, 1286-1294.
Herreros-Villanueva, M., Zubia-Olascoaga, A. and Bujanda, L. (2012b). c-Met in
pancreatic cancer stem cells: therapeutic implications. World J. Gastroenterol. 18,
5321-5323.
Hu, H., Zhou, L., Awadallah, A. and Xin, W. (2012). Significance of Notch1-signaling
pathway in human pancreatic development and carcinogenesis. Appl.
Immunohistochem. Mol. Morphol. 21, 242-247.
Jing, Y., Han, Z., Zhang, S., Liu, Y. and Wei, L. (2011). Epithelial-mesenchymal
transition in tumor microenvironment. Cell Biosci 1, 29.
Jung, I. H., Jung, D. E., Park, Y. N., Song, S. Y. and Park, S. W. (2011). Aberrant
Hedgehog ligands induce progressive pancreatic fibrosis by paracrine activation of
myofibroblasts and ductular cells in transgenic zebrafish. PLoS ONE 6, e27941.
Kawakami, K., Imanaka, K., Itoh, M. and Taira, M. (2004). Excision of the Tol2
transposable element of the medaka fish Oryzias latipes in Xenopus laevis and
Xenopus tropicalis. Gene 338, 93-98.
Kornmann, M., Tangvoranuntakul, P. and Korc, M. (1999). TGF-beta-1 up-regulates
cyclin D1 expression in COLO-357 cells, whereas suppression of cyclin D1 levels is
associated with down-regulation of the type I TGF-beta receptor. Int. J. Cancer 83,
247-254.
Kwan, K. M., Fujimoto, E., Grabher, C., Mangum, B. D., Hardy, M. E., Campbell, D.
S., Parant, J. M., Yost, H. J., Kanki, J. P. and Chien, C. B. (2007). The Tol2kit: a
multisite gateway-based construction kit for Tol2 transposon transgenesis constructs.
Dev. Dyn. 236, 3088-3099.
Kwon, S. M., Alev, C., Lee, S. H. and Asahara, T. (2012). The molecular basis of
Notch signaling: a brief overview. Adv. Exp. Med. Biol. 727, 1-14.
Lin, J. W., Biankin, A. V., Horb, M. E., Ghosh, B., Prasad, N. B., Yee, N. S., Pack, M.
A. and Leach, S. D. (2004). Differential requirement for ptf1a in endocrine and
exocrine lineages of developing zebrafish pancreas. Dev. Biol. 270, 474-486.
Lirdprapamongkol, K., Chiablaem, K., Sila-Asna, M., Surarit, R., Bunyaratvej, A.
and Svasti, J. (2012). Exploring stemness gene expression and vasculogenic
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
mimicry capacity in well- and poorly-differentiated hepatocellular carcinoma cell
lines. Biochem. Biophys. Res. Commun. 422, 429-435.
Liu, S. and Leach, S. D. (2011). Screening pancreatic oncogenes in zebrafish using
the Gal4/UAS system. Methods Cell Biol. 105, 367-381.
López-Rovira, T., Chalaux, E., Massagué, J., Rosa, J. L. and Ventura, F. (2002).
Direct binding of Smad1 and Smad4 to two distinct motifs mediates bone
morphogenetic protein-specific transcriptional activation of Id1 gene. J. Biol. Chem.
277, 3176-3185.
Lowy, A. M., Fenoglio-Preiser, C., Kim, O. J., Kordich, J., Gomez, A., Knight, J.,
James, L. and Groden, J. (2003). Dysregulation of beta-catenin expression
correlates with tumor differentiation in pancreatic duct adenocarcinoma. Ann. Surg.
Oncol. 10, 284-290.
Maitra, A. and Leach, S. D. (2012). Disputed paternity: the uncertain ancestry of
pancreatic ductal neoplasia. Cancer Cell 22, 701-703.
Manfroid, I., Ghaye, A., Naye, F., Detry, N., Palm, S., Pan, L., Ma, T. P., Huang, W.,
Rovira, M., Martial, J. A. et al. (2012). Zebrafish sox9b is crucial for
hepatopancreatic duct development and pancreatic endocrine cell regeneration.
Dev. Biol. 366, 268-278.
Marino, S. (2005). Medulloblastoma: developmental mechanisms out of control.
Trends Mol. Med. 11, 17-22.
McCleary-Wheeler, A. L., McWilliams, R. and Fernandez-Zapico, M. E. (2012).
Aberrant signaling pathways in pancreatic cancer: a two compartment view. Mol.
Carcinog. 51, 25-39.
Moro, E., Vettori, A., Porazzi, P., Schiavone, M., Rampazzo, E., Casari, A., Ek, O.,
Facchinello, N., Astone, M., Zancan, I. et al. (2013). Generation and application of
signaling pathway reporter lines in zebrafish. Mol. Genet. Genomics 288, 231-242.
Morris, J. P., 4th, Cano, D. A., Sekine, S., Wang, S. C. and Hebrok, M. (2010). Betacatenin blocks Kras-dependent reprogramming of acini into pancreatic cancer
precursor lesions in mice. J. Clin. Invest. 120, 508-520.
Mysliwiec, P. and Boucher, M. J. (2009). Targeting Notch signaling in pancreatic
cancer patients—rationale for new therapy. Adv. Med. Sci. 54, 136-142.
Nakajima, S., Doi, R., Toyoda, E., Tsuji, S., Wada, M., Koizumi, M., Tulachan, S. S.,
Ito, D., Kami, K., Mori, T. et al. (2004). N-cadherin expression and epithelialmesenchymal transition in pancreatic carcinoma. Clin. Cancer Res. 10, 4125-4133.
Park, S. W., Davison, J. M., Rhee, J., Hruban, R. H., Maitra, A. and Leach, S. D.
(2008). Oncogenic KRAS induces progenitor cell expansion and malignant
transformation in zebrafish exocrine pancreas. Gastroenterology 134, 2080-2090.
Parsons, M. J., Pisharath, H., Yusuff, S., Moore, J. C., Siekmann, A. F., Lawson, N.
and Leach, S. D. (2009). Notch-responsive cells initiate the secondary transition in
larval zebrafish pancreas. Mech. Dev. 126, 898-912.
Pascual, M., Abasolo, I., Mingorance-Le Meur, A., Martínez, A., Del Rio, J. A.,
Wright, C. V., Real, F. X. and Soriano, E. (2007). Cerebellar GABAergic progenitors
adopt an external granule cell-like phenotype in the absence of Ptf1a transcription
factor expression. Proc. Natl. Acad. Sci. USA 104, 5193-5198.
Rachagani, S., Senapati, S., Chakraborty, S., Ponnusamy, M. P., Kumar, S., Smith,
L. M., Jain, M. and Batra, S. K. (2011). Activated KrasG¹²D is associated with
invasion and metastasis of pancreatic cancer cells through inhibition of E-cadherin.
Br. J. Cancer 104, 1038-1048.
Rhim, A. D., Mirek, E. T., Aiello, N. M., Maitra, A., Bailey, J. M., McAllister, F.,
Reichert, M., Beatty, G. L., Rustgi, A. K., Vonderheide, R. H. et al. (2012). EMT
and dissemination precede pancreatic tumor formation. Cell 148, 349-361.
Rodini, C. O., Suzuki, D. E., Nakahata, A. M., Pereira, M. C., Janjoppi, L.,
Toledo, S. R. and Okamoto, O. K. (2010). Aberrant signaling pathways in
medulloblastomas: a stem cell connection. Arq. Neuropsiquiatr. 68, 947-952.
Salnikov, A. V., Liu, L., Platen, M., Gladkich, J., Salnikova, O., Ryschich, E.,
Mattern, J., Moldenhauer, G., Werner, J., Schemmer, P. et al. (2012). Hypoxia
induces EMT in low and highly aggressive pancreatic tumor cells but only cells with
cancer stem cell characteristics acquire pronounced migratory potential. PLoS ONE
7, e46391.
Satelli, A. and Li, S. (2011). Vimentin in cancer and its potential as a molecular target
for cancer therapy. Cell. Mol. Life Sci. 68, 3033-3046.
Savagner, P. (2010). The epithelial-mesenchymal transition (EMT) phenomenon. Ann.
Oncol. 21 Suppl. 7, vii89-vii92.
Scanlon, C. S., Van Tubergen, E. A., Inglehart, R. C. and D’Silva, N. J. (2013).
Biomarkers of epithelial-mesenchymal transition in squamous cell carcinoma. J.
Dent. Res. 92, 114-121.
Sellick, G. S., Barker, K. T., Stolte-Dijkstra, I., Fleischmann, C., Coleman, R. J.,
Garrett, C., Gloyn, A. L., Edghill, E. L., Hattersley, A. T., Wellauer, P. K. et al.
(2004). Mutations in PTF1A cause pancreatic and cerebellar agenesis. Nat. Genet.
36, 1301-1305.
Spatuzza, C., Schiavone, M., Di Salle, E., Janda, E., Sardiello, M., Fiume, G.,
Fierro, O., Simonetta, M., Argiriou, N., Faraonio, R. et al. (2008). Physical and
functional characterization of the genetic locus of IBtk, an inhibitor of Bruton’s
tyrosine kinase: evidence for three protein isoforms of IBtk. Nucleic Acids Res. 36,
4402-4416.
Stanger, B. Z., Stiles, B., Lauwers, G. Y., Bardeesy, N., Mendoza, M., Wang, Y.,
Greenwood, A., Cheng, K. H., McLaughlin, M., Brown, D. et al. (2005). Pten
constrains centroacinar cell expansion and malignant transformation in the
pancreas. Cancer Cell 8, 185-195.
Stoletov, K. and Klemke, R. (2008). Catch of the day: zebrafish as a human cancer
model. Oncogene 27, 4509-4520.
Su, Y., Büchler, P., Gazdhar, A., Giese, N., Reber, H. A., Hines, O. J., Giese, T.,
Büchler, M. W. and Friess, H. (2006). Pancreatic regeneration in chronic
pancreatitis requires activation of the notch signaling pathway. J. Gastrointest. Surg.
10, 1230-1242.
893
Disease Models & Mechanisms
RESEARCH ARTICLE
RESEARCH ARTICLE
Wang, Z., Zhang, Y., Li, Y., Banerjee, S., Liao, J. and Sarkar, F. H. (2006b). Downregulation of Notch-1 contributes to cell growth inhibition and apoptosis in pancreatic
cancer cells. Mol. Cancer Ther. 5, 483-493.
Wu, S. M., Choo, A. B., Yap, M. G. and Chan, K. K. (2010). Role of Sonic hedgehog
signaling and the expression of its components in human embryonic stem cells.
Stem Cell Res. 4, 38-49.
Wu, Q., Miele, L., Sarkar, F. H. and Wang, Z. (2012). The role of EMT in pancreatic
cancer progression. Pancreat. Disord. Ther. 2, 1000e121.
Xu, X., Browning, V. L. and Odorico, J. S. (2011). Activin, BMP and FGF pathways
cooperate to promote endoderm and pancreatic lineage cell differentiation from
human embryonic stem cells. Mech. Dev. 128, 412-427.
Yang, Y. (2012). Wnt signaling in development and disease. Cell Biosci 2, 14.
Zecchin, E., Mavropoulos, A., Devos, N., Filippi, A., Tiso, N., Meyer, D., Peers, B.,
Bortolussi, M. and Argenton, F. (2004). Evolutionary conserved role of ptf1a in the
specification of exocrine pancreatic fates. Dev. Biol. 268, 174-184.
Zhou, B., Liu, Y., Kahn, M., Ann, D. K., Han, A., Wang, H., Nguyen, C., Flodby, P.,
Zhong, Q., Krishnaveni, M. S. et al. (2012). Interactions between β-catenin and
transforming growth factor-β signaling pathways mediate epithelial-mesenchymal
transition and are dependent on the transcriptional co-activator cAMP-response
element-binding protein (CREB)-binding protein (CBP). J. Biol. Chem. 287, 7026-7038.
Disease Models & Mechanisms
Tiso, N., Filippi, A., Pauls, S., Bortolussi, M. and Argenton, F. (2002). BMP
signalling regulates anteroposterior endoderm patterning in zebrafish. Mech. Dev.
118, 29-37.
Truty, M. J. and Urrutia, R. (2007). Basics of TGF-beta and pancreatic cancer.
Pancreatology 7, 423-435.
Ungefroren, H., Groth, S., Sebens, S., Lehnert, H., Gieseler, F. and Fändrich, F.
(2011). Differential roles of Smad2 and Smad3 in the regulation of TGF-β1-mediated
growth inhibition and cell migration in pancreatic ductal adenocarcinoma cells:
control by Rac1. Mol. Cancer 10, 67.
Virtanen, S., Alarmo, E. L., Sandström, S., Ampuja, M. and Kallioniemi, A. (2011).
Bone morphogenetic protein -4 and -5 in pancreatic cancer—novel bidirectional
players. Exp. Cell Res. 317, 2136-2146.
Wan, H., Korzh, S., Li, Z., Mudumana, S. P., Korzh, V., Jiang, Y. J., Lin, S. and
Gong, Z. (2006). Analyses of pancreas development by generation of gfp transgenic
zebrafish using an exocrine pancreas-specific elastaseA gene promoter. Exp. Cell
Res. 312, 1526-1539.
Wang, Z., Banerjee, S., Li, Y., Rahman, K. M., Zhang, Y. and Sarkar, F. H. (2006a).
Down-regulation of notch-1 inhibits invasion by inactivation of nuclear factor-kappaB,
vascular endothelial growth factor, and matrix metalloproteinase-9 in pancreatic
cancer cells. Cancer Res. 66, 2778-2784.
Disease Models & Mechanisms (2014) doi:10.1242/dmm.014969
894