AZGP1 is a tumor suppressor in pancreatic cancer inducing
mesenchymal-to-epithelial transdifferentiation by inhibiting
TGF-b-mediated ERK signaling
B Kong1,3, CW Michalski1,3, X Hong1, N Valkovskaya1, S Rieder1, I Abiatari1, S Streit1,
M Erkan1, I Esposito2, H Friess1and J Kleeff1
1Department of Surgery, Technische Universita ¨t Mu ¨nchen, Munich, Germany and2Institute of Pathology, Technische Universita ¨t
Mu ¨nchen, Munich, Germany
mediated by transforming growth factor-b (TGF-b) signal-
ing leads to aggressive cancer progression. In this study, we
identified zinc-a2-glycoprotein (AZGP1, ZAG) as a tumor
suppressor in pancreatic ductal adenocarcinoma whose
expression is lost due to histone deacetylation. In vitro,
ZAG silencing strikingly increased invasiveness of pan-
creatic cancer cells accompanied by the induction of a
mesenchymal phenotype. Expression analysis of a set of
EMT markers showed an increase in the expression of
mesenchymal markers (vimentin (VIM) and integrin-a5)
and a concomitant reduction in the expression of epithelial
markers (cadherin 1 (CDH1), desmoplakin and keratin-19).
Blockade of endogenous TGF-b signaling inhibited these
morphological changes and the downregulation of CDH1,
as elicited by ZAG silencing. In a ZAG-negative cell line,
human recombinant ZAG (rZAG) specifically inhibited
exogenous TGF-b-mediated tumor cell invasion and VIM
expression. Furthermore, rZAG blocked TGF-b-mediated
ERK2 phosphorylation. PCR array analysis revealed that
ZAG-induced epithelial transdifferentiation was accompa-
nied by a series of concerted cellular events including a shift
in the energy metabolism and prosurvival signals. Thus,
epigenetically regulated ZAG is a novel tumor suppressor
essential for maintaining an epithelial phenotype.
Oncogene (2010) 29, 5146–5158; doi:10.1038/onc.2010.258;
published online 28 June 2010
Keywords: pancreatic cancer; AZGP1; EMT; MET;
transdifferentiation; transforming growth factor-b
Zinc-a2-glycoprotein (AZGP1, ZAG) is a secreted
41-kDa protein whose function has not yet been fully
elucidated. ZAG has initially been identified and
purified in human serum already in 1961 (Burgi and
Schmid, 1961). Using immunohistochemical studies,
it has been found that ZAG is expressed mainly in
epithelial cells of the breast, the prostate, the liver and
various other gastrointestinal organs (Tada et al., 1991).
In line with its production by secretory epithelial cells,
ZAG is found in a number of body fluids (Frenette
et al., 1987; Ohkubo et al., 1990). Recently, it has been
shown that ZAG might be involved in carcinogenesis
and tumor differentiation. Thus, in prostate and breast
cancer, expression of ZAG has been linked to poorer
tumor differentiation (Diez-Itza et al., 1993; Hale et al.,
2001) though the underlying mechanisms remain elusive.
ZAG has also been associated with cancer cachexia due
to its high level of amino-acid sequence homology with
tumor-derived lipid-mobilizing factor (Russell et al.,
2004) and because in a mouse model of ZAG-producing
tumors, ZAG stimulated lipolysis in adipocytes leading
to cachexia (Bing et al., 2004).
Epithelial-to-mesenchymal transdifferentiation (EMT)
leading to aggressive cancer progression is considered to
be a crucial event in the carcinogenesis of pancreatic
ductal adenocarcinoma (PDAC). EMT is characterized
by the loss of epithelial characteristics and by the
acquisition of a mesenchymal phenotype (Brabletz et al.,
2005). During carcinogenesis and development of
a more mesenchymal phenotype, less motile (epithelial)
cells restart developmental programs (that is, transform-
ing growth factor-b (TGF-b), notch, Wnt and hedgehog
pathways) to gain migratory and invasive properties
(Ellenrieder et al., 2001; Brabletz et al., 2005; Dembinski
and Krauss, 2009; Wang et al., 2009). This involves the
loss of cell–cell junctions and a reorganization of the
actin cytoskeleton, as well as the acquisition of
resistance to apoptosis, and thus of resistance to
chemotherapy (Robson et al., 2006; Wang et al.,
2009). The complexity of the underlying signaling
processes and the adoption of a completely different
phenotype imply that EMT is not a simple change in
migratory/invasive capabilities but rather a switch into a
different cellular program. Importantly, these ‘features’
seem to necessitate that the cell forfeits proliferative
capacities to survive under anchorage-independent
conditions, as recently shown in K-ras/extracellular
signal-regulated kinase (ERK)-activated breast cancer
Received 14 January 2010; revised 11 May 2010; accepted 28 May 2010;
published online 28 June 2010
Correspondence: Dr J Kleeff, Department of Surgery, Technische
Universita ¨ t Mu ¨ nchen, Ismaningerstrasse 22, Munich 81675, Germany.
3These authors contributed equally to this work.
Oncogene (2010) 29, 5146–5158
& 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10
cells (Evdokimova et al., 2009). Similarly in pancreas
cancer, a subpopulation of slowly dividing cancer cells
show an EMT-resembling morphology (Dembinski and
Krauss, 2009). On the molecular level, EMT is defined
by downregulation of epithelial differentiation markers
(for example, cadherin 1 (CDH1)) and by the transcrip-
tional induction of mesenchymal markers, such as
vimentin (VIM) and CDH2. In spite of its high
activation threshold, EMT can be induced by members
of the TGF-b family including TGF-b1 and BMP4
(Ellenrieder et al., 2001; Hamada et al., 2007). Though
TGF-b is a tumor suppressor in the early stages of
PDAC, it contributes to tumor progression by promot-
ing EMT and invasion in the late cancer stages
(Ellenrieder et al., 2001; Hamada et al., 2007). Con-
stitutively active Ras signaling—as a consequence of K-
ras mutations that are present in 70–90% of PDAC—
has been shown to be indispensable for TGF-b-mediated
induction of EMT (Longnecker and Terhune, 1998;
Ellenrieder et al., 2001; Roberts and Wakefield, 2003;
Horiguchi et al., 2009). These data suggest that TGF-b
cooperates with Ras/ERK signaling in advanced stages
of PDAC promoting invasion and metastasis through
EMT. Therefore, EMT not only entails the orchestra-
tion of promigratory/invasive events but also reorgani-
zation of proliferation to allow for survival in different
microenvironments. As we show ZAG to be widely
lost in pancreatic cancer, we hypothesize that it might
have tumor-suppressor functions and that it might be
involved in EMT in PDAC.
Expression of ZAG is decreased in primary
and metastatic PDAC
Quantitative real-time-PCR (QRT–PCR) of bulk nor-
mal pancreatic tissues (n¼19) and PDAC (n¼57)
tissues revealed a significant downregulation of ZAG
mRNA expression in PDAC (?6.5?, Po0.0001;
Figure 1a). Though ZAG staining was predominantly
detected in acinar and peripheral cells of islets in the
normal pancreas (Figure 1b), its expression in normal
ducts was generally absent (Supplementary Figure 1).
In contrast, ZAG staining in cancer cells was lost in 74%
of the PDAC cases (42/57) and only 21% (12/57) and
5% (3/57) of the samples showed moderate or strong
ZAG staining intensities in cancer cells (Figure 1c, right
panel). Consistent with our results, published serial
analysis of gene expression data of isolated primary
PDAC cells demonstrated that 50% (12/24) of cancers
showed low expression of ZAG (less than 100 tags),
whereas 37.5% (9/24) and 12.5% (3/24) of the cells
expressed moderate (100–1000 tags) or high (more than
1000 tags) ZAG levels (Jones et al., 2008; Figure 1c, left
1164368/DC1). Moreover, 39 metastatic PDAC tissues
consisting of liver metastasis (23), lymph node metas-
tasis (6), peritoneal metastasis (6), omentum metastasis
(3) and muscle metastasis (1) were stained for ZAG.
Among these samples we detected, apart from two (5%)
ZAG-positive liver metastases (Figure 1d, right lower
panel), only ZAG-negative (Figure 1d) metastatic cancer
cells. In addition, ZAG staining was performed on a
tissue Pancreatic Intraepithelial Neoplasia (PanIN)
array containing 44 PanIN lesions of different grades.
It is of note that scattered ZAG-expressing cells were
observed in 48% (21/44) of the lesions, though the
appearance of such scattered ZAG-positive cells was not
correlated to the grade of the lesions (Figure 1e).
Comparison of the staining intensities in PanIN lesions,
primary and metastatic PDAC revealed striking differ-
ences in expression levels (Figure 1f, 48 vs 26 vs 5%,
Po0.05). As ZAG expression was mainly observed in
acinar and endocrine cells in the normal pancreas, we set
out to investigate its expression in pancreatic endocrine
tumors (PETs) and in acinar cell carcinomas (ACCs).
Although ZAG expression was detected in 50% (7/14)
of the PET samples and two out of three PET cell lines
(Supplementary Figures 2A and B), only 18% (2/11) of
the ACC samples were positive for ZAG (Supplemen-
tary Figure 2C). Altogether, the rate of ZAG-positive
cancer cells in PET was generally higher than that in
PDAC and ACC (Figure 1g, 50 vs 26 and 18%).
Few cancer cell lines express and secrete ZAG
ZAG mRNA expression and ZAG protein (cell lysates
and supernatants) were only found in two (Aspc-1 and
Capan-1) out of seven tested cell lines (Aspc-1, Capan-1,
Colo-357, Mia-PaCa-2, Panc-1, T3M4 and Su86.86;
Figures 2a and b). As there are three N-linked
glycosylation sites on human ZAG (as assessed by
X-ray crystallography) (Ohkubo et al., 1990; Rolli et al.,
2007), we set out to determine whether glycosylation
contributes to cellular transport of ZAG. In line with
this assumption, inhibition of N-linked glycosylation by
tunicamycin significantly decreased ZAG secretion in
Aspc-1 and Capan-1 cells (Figure 2c). Though there
were no differences between ZAG expression and
secretion in cancer cell lines, we hypothesized that
secretion might be altered in vivo (potentially due to
deglycosylation). Therefore, we measured ZAG levels in
the sera of PDAC patients (n¼16) and healthy
volunteers (n¼14) revealing no significant differences
between the groups (Figure 2d).
Histone deacetylase inhibition reconstitutes ZAG
expression and induces MET in pancreatic cancer cells
To assess the mechanism underlying downregulation/
loss of ZAG in most of the tested PDAC cells, we
performed demethylation and deacetylation-inhibition
assays. Although the histone deacetylase (HDAC)
inhibitor Trichostatin A (TSA) reconstituted ZAG
expression in the two ZAG-negative cell lines T3M4
and Panc-1 on both the mRNA and protein levels
(Figures 3a and b), demethylation using 5-aza-20-
deoxycytidine (5-aza) had no such effect (data not
shown). Similar results were also obtained from two
additional cell lines (Colo-357 and Su86.86), though to a
lesser extent than in Su86.86 cells (Supplementary
AZGP1 in pancreatic cancer
B Kong et al
(n¼19) and PDAC (n¼57) tissues was analyzed by QRT–PCR as described in the Materials and methods section. Data are presented
as relative expression (normalized to the median expression of ZAG in PDAC tissues). (b) Normal pancreatic acinar cells and islet
cells are immunopositive for ZAG (magnifications, ?200, ?630 and ?400). (c) Analysis of published serial analysis of gene
expression data (Jones et al., 2008): 50% of primary PDAC cells express low levels of ZAG, whereas 37.5 and 12.5% show moderate
and high ZAG expression levels, respectively (left panel, http://www.sciencemag.org/cgi/content/full/sci;1164368/DC1). In the majority
(74%) of the analyzed primary PDAC tissue sections, cancer cells are ZAG-negative, whereas 21 and 5% exhibit weak and strong
staining of ZAG, respectively (right panel, ?200). (d) In 95% of the metastatic PDAC tissue sections (liver (?200), omentum (?200),
peritoneum (?200), muscle (?100) and lymph node (?200)), cancer cells are ZAG-negative (right panel). Only 5% (two liver
metastases) of metastatic tissues are weakly ZAG-positive (right lower panel, ?100). (e) ZAG-expressing cells (middle panel, ?200,
right panel, ?630) were observed in 48% of PanIN lesions; 52% of the lesions were ZAG-negative (left panel, ?200).
(f) Immunohistochemically, the rate of ZAG-positive cells in PanIN lesions, primary and metastatic PDAC, gradually decreases
from 48 to 26 and 5%. (g) The percentage of ZAG-positive samples in PET is generally higher than in PDAC or ACC (50 vs 26
Expression of ZAG is decreased in primary and metastatic PDAC. (a) Expression of ZAG mRNA in normal pancreas
AZGP1 in pancreatic cancer
B Kong et al
Figures 3A and B). We corroborated these findings by
treating Panc-1 cells with TSA, which induced a time-
(Figure 3c). Chromatin immunoprecipitation revealed
a significantly increased binding of acetylated histone
H3 to the ZAG promoter (Figure 3d), suggesting an
open chromatin conformation within that region. As it
has been shown recently that inhibition of HDAC in
Panc-1 cells abolished TGF-b1-mediated EMT (Fritsche
et al., 2009; von Burstin et al., 2009), we set out to
determine whether inhibition of HDACs in Panc-1 cells
would influence their steady-state differentiation status.
To this end, we performed QRT–PCR to compare
expression of a panel of EMT markers (CDH1, keratin-
19 (KRT19), dentin sialoprotein (DSP), VIM and
integrin-a 5) between dimethyl sulfoxide (DMSO)- and
TSA-treated cells. This analysis revealed that TSA
treatment induced a more epithelial-like phenotype in
Panc-1 cells characterized by increased expression of
epithelial markers (KRT19 and CDH1) and a concomi-
tant downregulation of a mesenchymal marker (VIM,
Figure 3e). The changes in CDH1 and VIM expression
were also confirmed on the protein level (Figure 3f).
Consistently, TSA treatment significantly decreased
cancer cell invasion (Figure 3g, upper panel), whereas
(Figure 3g, lower panel).
ZAG inhibits invasion of pancreatic cancer cells
by blocking TGF-b-mediated EMT
Though ZAG RNAi in Aspc-1 cells significantly
decreased its expression (Figure 4a, left panel) compared
with controls, no significant effect on cell growth was
observed (Figure 4a, middle panel). Accordingly, treat-
ment of Aspc-1 cells with human recombinant ZAG
(rZAG) did not alter proliferation (Supplementary
Figure 4A). However, RNAi-mediated ZAG silencing
strikingly increased invasiveness of Aspc-1 cells, which
was partially reversible by treatment with rZAG
(though to a lesser extent when using lower concentra-
tions of rZAG; Figure 4a, right panel). As this increase
in invasiveness was accompanied by significant pheno-
typic changes of cellular morphology (Figure 4b, left
panel; more epithelial structure of control cells vs a
fibroblast-like phenotype of many siRNA-treated cells
forming scattered cell clusters; confirmed by actin
staining, Figure 4b), we hypothesized that ZAG RNAi
might induce a mesenchymal program. Therefore, we
performed QRT–PCR to the compare expression
patterns of a panel of EMT markers (CDH1, KRT19,
DSP, VIM, ITGA5, ITGB1, ITGAV and ITGA3)
between control and ZAG-silenced Aspc-1 cells. This
assay revealed that the epithelial markers CDH1, DSP
and KRT19 were significantly downregulated whereas
the mesenchymal markers VIM and ITGA5 were
upregulated (Figure 4b, middle panel). The changes in
CDH1 and VIM expression were confirmed at the
protein level (Figure 4b, right panel). In addition, we
validated the results of the invasion assays as well as the
changes in morphology and EMT markers using two
ZAG siRNA molecules (Supplementary Figures 4B and
C). Because TGF-b signaling is considered to be one of
the major driving forces of EMT in PDAC, we
speculated that the inhibitory effect of ZAG on EMT
Out of seven tested cell lines (Aspc-1 (A), Capan-1 (Ca), Colo357 (Co), Mia-PaCa2 (Mi), Panc-1 (P), T3M4 (T), Su86.86 (S)), only Aspc-1
and Capan-1 expressed ZAG. (b) Correspondingly, ZAG protein is found in both supernatants and cell lysates of Aspc-1 and Capan-1 cells
by immunoblot analysis; loading control: GAPDH. (c) Tunicamycin treatment shifts the molecular weight of ZAG from 43kDa
(glycosylated) to 35kDa (cell lysates) and significantly decreases ZAG levels in the supernatants of Aspc-1 and Capan-1 cells. One
representative immunoblot out of two independent experiments is shown. (d) ZAG levels in the sera of PDAC patients (n¼16) and healthy
volunteers (n¼14) were determined by enzyme-linked immunosorbent assay. No significant (NS) differences were found between the groups.
Few cancer cell lines express and secrete ZAG. (a) mRNA expression analysis (QRT–PCR) of ZAG in pancreatic cancer cell lines.
AZGP1 in pancreatic cancer
B Kong et al
might be attributed to a (direct/indirect) blockade of
endogenous TGF-b signaling. Furthermore, it has
recently been shown that in Smad4-deficient pancreatic
cancer cell lines, autocrine activation of the TGF-b
receptor system occurs (Subramanian et al., 2004). As
Aspc-1 cells are Smad4 negative due to a point mutation
(Wan et al., 2005), we selectively inhibited type I TGF-b
receptor/ALK5 (SB525354) and both type I and type II
TGF-b receptors (LY364847) after having silenced
ZAG. Though ALK5/type I TGF-b receptor blockade
by SB525354 had no effect, the combined inhibition of
type I and type II TGF-b receptors by LY364847
reversed the morphological changes elicited by ZAG
silencing (Figure 4c, left panel). Furthermore, LY364847
treatment alone induced upregulation of CDH1 con-
firming an important role of endogenous TGF-b in
reconstitutes ZAG mRNA expression in ZAG-negative cell lines (Panc-1 and T3M4). QRT–PCR assays were repeated three times and
data are shown as relative expression (normalized to Aspc-1 levels). (b) Correspondingly, ZAG protein is found in both supernatants
and cell lysates of Panc-1 and T3M4 cells after TSA treatment (immunoblot analysis). One representative immunoblot out of the three
independent experiments is shown. (c) TSA treatment induces a time-dependent histone H3 acetylation in Panc-1 cells. One
representative immunoblot out of the two independent experiments is shown. (d) Binding of acetylated histone H3 to the ZAG
promoter (Panc-1 cells) is significantly increased following a 2h TSA treatment (white bars, compared with control, black bars).
(e) TSA treatment in Panc-1 cells induces an MET program as documented by the increased expression of epithelial markers with a
concomitant decreased expression of mesenchymal markers (QRT–PCR; CDH1: E-cadherin, KRT19: keratin 19, VIM: vimentin,
ITGA5: integrin-a5, DSP: desmoplakin, *Po0.05; relative expression fold of control). (f) Changes in CDH1 and VIM levels are
confirmed by immunoblot analysis. (g) In a Matrigel invasion assay, TSA treatment significantly decreases Panc-1 cell invasiveness
(upper panel), whereas no effect on cell proliferation (lower panel) is seen. Values shown are the mean±s.e.m., fold of control (DMSO)
obtained from three independent experiments, *Po0.05.
Histone deacetylase inhibition reconstitutes ZAG expression and induces MET in pancreatic cancer cells. (a) TSA treatment
AZGP1 in pancreatic cancer
B Kong et al
(1# and 2#) of specific siRNA or negative control siRNA (control) at 48h (left panel). Proliferation of Aspc-1 cells as assessed by
3-(4,5-dimethylthiazole-2-yl) 2,5-diphenyltetrazolium bromide assays following ZAG siRNA: ZAG siRNA does not change the
proliferation rates (middle panels). Data from three independent experiments are expressed as mean±s.e.m. In a Matrigel invasion
assay, ZAG silencing strikingly increases Aspc-1 cancer cell invasiveness, which is partially reversible by pre-incubation with rZAG
(right panel). Values shown are the mean±s.e.m. fold of negative control (control) obtained from three independent experiments,
*Po0.05. (b) ZAG RNAi in Aspc-1 induces fibroblast-like changes in cellular morphology (left panel, phase contrast and actin
labeling). The morphological changes were accompanied by the decreased expression of epithelial markers with a concomitant
increased expression of mesenchymal markers (QRT–PCR, middle panel; CDH1: E-cadherin, KRT19: keratin 19, DSP: desmoplakin,
VIM: vimentin, ITGA5: integrin-a5, ITGB1: integrin-b 1, ITGAV: integrin-aV, ITGA3: integrin-a3, *Po0.05; relative expression fold
of control). Changes in CDH1 and VIM levels are confirmed by immunoblot analysis (right panel). (c) Blockade of endogenous TGF-b
signaling by LY 364947 reverses the morphological changes elicited by ZAG silencing in Aspc-1 cells (left panel, phase contrast).
Treatment with LY 364947 and PD 98059 increases expression of CDH1 and reverts downregulation of CDH1 induced by ZAG RNAi
(QRT–PCR, middle panel). No effects on VIM expression are seen (right panel, relative expression fold of control in DMSO treated
group), SB 525354 has no effect on either the expression of CDH1 or VIM. (d) In vitro cell invasion assay: rZAG reverts TGF-b-
induced invasion of Panc-1 cancer cells (left panel). Expression of CDH1 and VIM after 48h of incubation with TGF-b1 (with or
without rZAG); loading control: GAPDH. One representative immunoblot out of three independent experiments is shown.
ZAG inhibits invasion of pancreatic cancer cells by blocking TGF-b-mediated EMT. (a) ZAG silencing with two sets
AZGP1 in pancreatic cancer
B Kong et al
EMT/MET; although LY 364847 had no influence on
basal or induced expression of VIM by ZAG silencing
(Figure 4c, right panel), it totally abolished the down-
regulation of CDH1 after ZAG RNAi (Figure 4c,
middle panel). In addition, inhibition of ERK signaling
using an inhibitor of mitogen-activated protein kinase
kinase (MEK, PD-98059) induced a similar effect as
LY364847 (Figure 4c, middle and right panel). These
data suggest that both TGF-b and ERK signaling are
involved in EMT mediated by the loss of ZAG. To
corroborate these findings, we used the highly invasive
pancreatic cancer cell line Panc-1, which also possesses
an intact TGF-b signaling pathway but does not express
ZAG, for further analyses. Consistent with the recently
reported data (Ellenrieder et al., 2001), invasion assays
revealed that exogenous TGF-b1 increased invasiveness
of Panc-1 cells. Though rZAG alone had no effect on
invasiveness, it strongly attenuated TGF-b1-mediated
tumor cell invasion and VIM upregulation (Figure 4d). It
is interesting that in comparison with the TGF-b receptor
blocking experiments, ZAG did not change TGF-b-
induced CDH1 suppression (Figure 4d, right panel).
ZAG induces mesenchymal-to-epithelial
transdifferentiation by modulating TGF-b-mediated
Both Panc-1 and Aspc-1 cells carry a constitutively
active K-Ras mutation. As it has been shown that
constitutive activation of Ras is required for TGF-b-
induced EMT in Panc-1 cells (Ellenrieder et al., 2001;
Horiguchi et al., 2009), we analyzed whether ZAG was
involved in the interaction between Ras/ERK and TGF-
b signaling. To rule out the possibility that ZAG might
have a direct effect on the transduction of the TGF-b
signal, we examined the phosphorylation of Smad2
following treatment with exogenous TGF-b1 (with or
without incubation with
Although TGF-b1 time-dependently induced phospho-
rylation of Smad2, neither rZAG alone nor rZAG in
combination with TGF-b1 had any additional effect
(Figure 5a). As exogenous TGF-b1 causes sustained
phosphorylation of ERK2, which is essential for TGF-
b1-mediated EMT (Ellenrieder et al., 2001), we eval-
uated ERK2 phosphorylation following treatment with
TGF-b1 for 24h (±rZAG). These assays revealed that
the TGF-b1-mediated increase in ERK2 phosphoryla-
tion wasblocked bypreincubation
(Figure 5b). Because TGF-b1-mediated inhibition of
cell growth is partially transduced by transcriptional
induction of the cell-cycle inhibitor p21(WAF1/Cip1;
p21), regulated by Smad4 (Finkel, 1996), we used p21 as
an internal control for the activation of the Smad4-
dependent TGF-b pathway. These experiments revealed
that in Panc-1 cells with intact Smad4, TGF-b1
treatment induced expression of p21 that remained
unchanged by treatment with rZAG (Figure 5b),
suggesting that rather the cross talk between TGF-b
and Ras/ERK signaling than the Smad4-dependent
pathway is blocked by rZAG. Accordingly, in Aspc-1
(Smad4-negative) cells, ZAG silencing, as well as TGF-
b1 treatment, caused sustained phosphorylation of
ERK2, presumably due to (facilitation of) activation
of endogenous TGF-b signaling (Figure 5c). Interes-
tingly, ZAG RNAi alone or in combination with TGF-b1
induced p21 expression (Figure 5c).
ZAG at the crossroads of cellular differentiation
and energy control
To further characterize signal transduction pathways
modulated by ZAG, we used an RT–PCR-based
of Smad2 following TGF-b treatment with or without rZAG. Loading control: GAPDH. (b) Increased phospho-ERK2 following
incubation with TGF-b1 is reverted by pre-treatment with rZAG; p21 levels are unaltered (Panc-1 cells). One representative
immunoblot out of three independent experiments is shown. (c) ZAG silencing increases phosphorylation of ERK2 and p21 levels,
which is further enhanced following incubation with TGF-b1. One representative immunoblot out of three independent experiments
ZAG exerts MET effects by modulating TGF-b-mediated ERK signaling. (a) Time-dependent, increasing phosphorylation
AZGP1 in pancreatic cancer
B Kong et al
pathway array to compare expression changes of a large
number of genes after ZAG silencing in Aspc-1 cells
(Figure 6a). In line with our findings above, BMP4 and
CDKN1B were increased, which reflected the activation
of endogenous TGF-b signaling (Lecanda et al., 2009).
In particular, exogenous human recombinant BMP4
(rBMP4) treatment induced an EMT program in Panc-1
and Su86.86 cells (Supplementary Figures 5A and B),
which is consistent with published results (Ellenrieder
et al., 2001; Hamada et al., 2007). Furthermore,
BCL2L1, CCND1 and MYC, which are survival path-
way-related genes, were upregulated underscoring the
importance of prosurvival signals during EMT (Biliran
et al., 2005; Freemantle et al., 2007; Barbie et al., 2009).
EGR1, which represents a mitogenic pathway, was
downregulated, consistent with the observed increased
expression of the cell-cycle inhibitor p21 (Lim et al.,
2008). Surprisingly, HK2 and CEBPB, which are
insulin-responsive factors, were highly induced after
ZAG RNAi. We thus speculate that the increased levels
of HK2, which catalyzes one of the first steps in glucose
metabolism, reflect an increased demand for glucose
following ZAG RNAi. Increased expression of CEBPB and HK2 (insulin-responsive factors) with the downregulation of GYS1
indicates a shift towards a high-glucose metabolic rate. Increased expression of CDKN1B and BMP4 reflects activation of endogenous
TGF-b signaling. In addition, changes in BCL2L1, CCND1, MYC and EGR1 suggest an increased prosurvival signal. (b) QRT–PCR
was performed to confirm expression changes of HK2, GYS1 and BMP4 after ZAG RNAi in Aspc-1 cells; relative expression fold of
control (mean±s.e.m.) obtained from three independent experiments. (c) The changes in HK2 are confirmed by immunoblot analysis.
(d) HK2 silencing with specific siRNA or negative control siRNA (control) at 72h in Panc-1 and Su86.86 cells (lower panel). In a
Matrigel invasion assay, HK2 silencing strikingly decreases Panc-1 and Su86.86 cancer cell invasiveness, whereas no influence on cell
proliferation is observed (upper panel). Values shown are the mean±s.e.m., fold of negative control (control) obtained from three
independent experiments, *Po0.05. (e) Schematic interference model: ZAG modulates TGF-b and Ras/ERK signaling.
ZAG at the crossroads of cellular differentiation and energy control. (a) Results of the real-time PCR pathway finder array
AZGP1 in pancreatic cancer
B Kong et al
(and thus an increased energy need) when cancer cells
undergo EMT. Correspondingly, GYS1—an enzyme
responsible for adding glucose monomers to glycogen—
was downregulated. These results were confirmed by
independent QRT–PCR assays (Figure 6b), and in-
creased expression of HK2 was confirmed on the protein
level (Figure 6c). To interrogate whether HK2 is
functionally relevant for EMT and cell invasion, we
chose the highly invasive pancreatic cancer cell lines
Panc-1 and Su86.86 for further analysis. As expected,
HK2 RNAi suppressed HK2 expression in Panc-1
and Su86.86 cells (Figure 6d, lower panel). Though
downregulation of HK2 did not significantly affect cell
(Figure 6d, upper panel; confirmed using another HK2
siRNA; data not shown). However, no changes in the
expression of EMT markers after silencing HK2
were observed in Panc-1 and Su86.86 cells (data not
shown). These results suggest that a certain level of HK2
or glucose metabolism is essential for maintaining the
basal invasiveness of cancer cells rather than their
differentiation status. Therefore, we speculate that these
molecules cooperatively promote EMT and cell invasion
following a loss of ZAG. To conclude, loss of ZAG in
Aspc-1 seems not only to induce EMT but to also
reprogram the cells within a series of concerted cellular
events consisting of increased energy consumption,
a prosurvival signal and a lower proliferation rate
In K-ras-transformed pancreatic cancer cells, epithelial
plasticity and EMT are increasingly seen as epigenetic
phenomena (Singh et al., 2009), which is supported by
recent findings in other malignancies, such as breast
cancer in which a signal from the tumor microenviron-
ment (for example, TGF-b) can induce phenotypic and
gene expression changes associated with de novo
epigenetic events important for EMT (Dumont et al.,
2008). In general, methylation (regulated by DNA
methyltransferases and histone deacetylation (regulated
by HDACs) are two main mechanisms through which
epigenetic regulation is conferred (Jaenisch and Bird,
2003; Glozak and Seto, 2007). By removing acetyl
groups from histones, it is found that HDACs create a
non-accessible chromatin conformation that reduces the
transcription of genes (that is, those which are
implicated in the control of cell growth differentiation
and apoptosis) (Glozak and Seto, 2007). In PDAC,
HDAC inhibitors have been shown to block cell
proliferation to promote differentiation and to induce
apoptosis (Fritsche et al., 2009; Von Burstin et al.,
2009); we showed that TSA blocks cell invasion and
promotes mesenchymal-to-epithelial transdifferentiation
(MET) in human PDAC cells, therefore, these might
constitute promising anticancer agents—given that
gene-specific control would be possible. The link
between the epigenetic HDAC machinery, EMT and
metastasis has also recently been shown using highly
metastatic (mesenchymal phenotype) cancer cells de-
rived from a genetically engineered mouse pancreas
cancer model; in these cells, silencing of CDH1 was
mediated by a transcriptional repressor complex con-
taining snail, HDAC1 and HDAC2 (von Burstin et al.,
2009). In our study, we now show that the expression of
ZAG—an EMT related gene—is also lost due to histone
transfection against HDAC1-7 failed to significantly
reconstitute ZAG expression in Panc-1 cells (data not
shown), we hypothesize that a so far unknown co-
repressor or repressor complex exists. Using a set of
in vitro experiments, we showed that ZAG specifically
blocks signaling between TGF-b and Ras/ERK. Over-
expression of TGF-b ligands and their receptors as well
as activating mutations of K-Ras occurs at a very high
frequency in PDAC and interactions of these two major
pathways have been shown to be crucial in pancreatic
carcinogenesis (Lemoine et al., 1992; Friess et al., 1993).
In addition, the Ras/ERK cascade has been shown to be
activated by members of the TGF-b family contributing
to TGF-b-mediated effects, such as EMT (Ellenrieder
et al., 2001; Horiguchi et al., 2009). The TGF-b signal
transducers Smad2 and Smad3 have also been shown to
be phosphorylated by ERK (de Caestecker et al., 1998;
Kretzschmar et al., 1999). Recent studies revealed that
ERK2 phosphorylation through TGF-b, which occurs
in a Ras-dependent way, is essential for TGF-b-
mediated EMT (Ellenrieder et al., 2001; Horiguchi
et al., 2009). At the same time, it is important to
consider that activating K-Ras mutations in the
pancreatic cancer cells do not result in constitutively
increased ERK signaling (Giehl et al., 2000). Interes-
tingly, in a mouse colorectal cancer model, the K-Ras
tumor phenotype was associated with an attenuated
signaling through the mitogen-activated protein kinase
pathway (Haigis et al., 2008). Thus, it is still not clear
whether, due to relatively ‘low’ levels of basal Ras/ERK
activity, the threshold of activation of this pathway (that
is, by TGF-b) is also relatively low. In our study, we
showed that TGF-b1-mediated ERK2 phosphorylation
was blocked by ZAG and that this effect was associated
with EMT and an increased invasive potential of two
pancreatic cancer cell lines, which carry the constitu-
tively active Ras mutation. ZAG seems to function like
a specific barrier that interrupts TGF-b/ERK signaling.
We furtheridentified (http://scansite.mit.edu)
growth factor receptor-binding protein, 2-Src homology
2 (Grb2–SH2, TLKDIVEYYNDSNGS) domain, within
the ZAG sequence. The Grb2–SH2 domain has been
described as docking site for activated receptors and,
therefore, is important in the oncogenic Ras signal
transduction pathway (Lung and Tsai, 2003; Benfield
et al., 2007). We thus speculate that ZAG might exert its
effect on TGF-b and Ras/ERK signaling through the
Grb2–SH2 domain. However, a synthesized small
peptide including this region failed to reconstitute the
function of the full-length protein in our experimental
setup (data not shown), adding another layer of
complexity to the underlying mechanism. Nevertheless,
AZGP1 in pancreatic cancer
B Kong et al
the exact molecular mechanisms deserve (and require)
Though obesity and insulin resistance are potential
risk factors for developing pancreatic cancer (Michaud
et al., 2001; Calle et al., 2003), the underlying molecular
mechanisms remain elusive. A recent study revealed that
high-fat-diet-induced obesity in mice accelerated the
development of PanIN in the context of embryonic K-
Ras activation through changes in energy metabolism.
These data suggested changes in energy metabolism
rather than insulin resistance as key elements in early
carcinogenesis of PDAC (Khasawneh et al., 2009).
Recent evidence has also established a link between
low serum levels of ZAG and obesity in humans;
however, whether ZAG serum levels are also related to
insulin resistance is still under debate (Marrades et al.,
2008). Nonetheless, the antidiabetic properties of human
ZAG have recently been shown in type 2 diabetes in
the ob/ob mouse model (Russell and Tisdale, 2010).
Furthermore, ZAG-deficient animals fed by a standard
or a lipid-rich diet were heavier than wild-type mice
(Rolli et al., 2007). Therefore, we hypothesized that
ZAG might be involved in early carcinogenetic events in
the pancreas through a modulation of energy meta-
bolism pathways. In line with this assumption, silencing
of ZAG in PDAC cells lines upregulated the two insulin
responsive-factors HK2 and CEBPB, whereas GYS1
levels were decreased. These data suggest that the cancer
cells are ‘pushed’ toward a high-glucose metabolic state
that seems to be important for cell invasion. Corre-
spondingly, silencing of HK2 in highly invasive cancer
cells strikingly attenuated their invasiveness, which
further underscores an essential role of energy meta-
bolism in cellular invasion events. Moreover, such a
shift in energy metabolism occurs distinctively within
EMT. Though our study is limited by in vitro tissue
culture conditions with supraphysiological levels of
glucose, oxygen and growth factors, it seems possible
that ZAG might serve as a critical mediator adjusting
energy metabolism according to the differentiation
status of cancer cells. Because cancer cells consume
glucose at an accelerated rate with reduced oxidative
phosphorylation even in the presence of oxygen (known
as the ‘Warburg effect’), such a capacity confers a
distinct competitive advantage compared with normal
epithelial cells (Mathupala et al., 2006; Christofk et al.,
2008). As a consequence, cancer cells that have under-
gone EMT—inducing a high-glucose metabolic rate—
may have gained further malignant ‘advantages’ com-
pared with the more epithelial, earlier cancer cell stages.
Such advantages may help these cells to migrate or
survive under even more unfavorable conditions, that is,
in ‘new’ microenvironments as present after having
metastasized to distant organs.
In conclusion, our data provide evidence that ZAG
affects the differentiation status of pancreatic cancer by
blocking the cross talk between TGF-b and Ras/ERK
signaling. EMT elicited by loss of ZAG is accompanied
by a series of concerted cellular events including a shift
in energy metabolism, prosurvival signals and reduced
cell proliferation. We speculate that targeting this cross
talk by restoring the function of ZAG might constitute
a new rational approach to reduce/reverse EMT in
Materials and methods
Patients and tissue sampling
Tissue sampling and processing was performed as previously
described (Zhang et al., 2007). Histological examination was
carried out by an experienced pathologist (IE). The use of
tissue for this study was approved by the local ethics
committees and written informed consent was obtained from
the patients before the operation.
Seven pancreatic cancer cell lines—ASPC1, Capan1, Colo357,
MiaPaca2, SU86.86, Panc1 and T3M4—were cultured was
previously described (Zhang et al., 2007). PET cell lines were
cultured in Dulbecco’s modified Eagle’s medium (Bon-1 and
QGP-1) or RPMI 1640 (CM) cell culture media supplemented
with 10% fetal bovine serum, 100U/ml penicillin and 100mg/
ml streptomycin at 371C, 5% CO2. The inhibitor of
methylation, 5-aza (5mM; Sigma-Aldrich, St Louis, MO,
USA), and the inhibitor of HDAC, TSA (0.1 and 1mM;
Sigma-Aldrich), were added to proliferating or confluent cells
(Panc-1, T3M4, Colo-357 and Su86.86) for 72 and 24h. Cells
were then lysed and processed for total RNA extraction or
whole cellular extract preparation as described thereafter. For
functional assays, cells were treated with TSA for 72h, and
were then subjected to invasion and proliferation assays in the
absence of TSA. For the human recombinant BMP4 (rBMP4)
assays, Panc-1 and Su86.86 cells were treated with rBMP4
Nordenstadt, Wiesbaden, Germany) in serum-free medium
for 6 consecutive days before the extraction of protein.
The mRNA extraction, cDNA synthesis and QRT–PCR were
performed as described previously (specimens used: normal
n¼19; PDAC n¼57) (Michalski et al., 2007a). Data are
presented as relative expression fold. b-Actin and hypo-
xanthine phosphoribosyltransferase 1 were used as house-
keeping genes. Sequences of primers are provided in Supple-
mentary data section.
Immunohistochemistry was performed using the Dako Envi-
sion System (Dako Cytomation GmbH, Hamburg, Germany)
as previously described (Michalski et al., 2007b). In brief,
tissue sections or a PanIN array (Esposito et al., 2007) were
incubated with rabbit anti-human ZAG polyclonal antibodies
(1:500, catalog no. RD181093100; BioVendor, Heidelberg,
Germany) at 41C overnight followed by incubation with
a horseradish-peroxidase-linked goat anti-rabbit antibody
(Dako Cytomation GmbH), followed by a reaction with
diaminobenzidine and counterstaining with Mayer’s hemato-
xylin. To confirm antibody specificity, we incubated tissue
sections in the absence of the primary antibodies with negative
control rabbit or mouse IgG. Under these conditions, no
immunostaining was detected.
Immunofluorescence was performed as previously described
(Michalski et al., 2008). In brief, cytoskeleton actin was labeled
AZGP1 in pancreatic cancer
B Kong et al
with Alexa Fluor 488 phalloidin (Molecular Probes, Eugene,
with DAPI and antifading medium (Gel/mount; Abcam,
Serum collection and enzyme-linked immunosorbent assay
The serum collection and enzyme-linked immunosorbent assay
were performed as previously described (Zhang et al., 2007).
The study was approved by the ethics committee of the
University of Heidelberg and written informed consent was
obtained from all individuals from whom serum samples were
collected. In brief, rabbit anti-ZAG antibody (0.05mg/ml,
catalog no. sc-11358; Santa Cruz Biotechnology, Santa-Cruz,
CA, USA) was used as the capture antibody and mouse anti-
ZAG antibody (1mg/ml, catalog no. sc-13585; Santa Cruz
Biotechnology) was used as the detection antibody.
Cultured pancreatic cancer cells were lysed in ice-cold RIPA
buffer (catalog no. 9806; Cell Signaling, Danvers, MA, USA)
containing one tablet EDTA-free protease inhibitor cocktail
(Roche Diagnostics, Mannheim, Germany) for 10min. For the
detection of phospho-p44/42 and phospho-Smad2, we used a
cell lysis buffer (catalog no. 9803; Cell Signaling). The western
blot analysis was performed as previously described (Ketterer
et al., 2009). The following antibodies were used: mouse anti-
ZAG (catalog no. sc-13585, 1:500; Santa Cruz Biotechnology),
rabbit anti-phospho-p44/42 mitogen-activated protein kinase
(Erk1/2) (Thr202/Tyr204; 1:1000, catalog no. 9101; Cell
Signaling), rabbit anti-p21Waf1/Cip1 (12D1; 1:1000, catalog
no. 2947; Cell Signaling), rabbit anti-phospho-Smad2 (Ser465/
467;138D4; 1:1000, catalog no. 3108; Cell Signaling), rabbit
anti-CDH1 (24E10; 1:5000, catalog no. 3195; Cell Signaling),
mouse anti-VIM (SKU catalog no. 18-0052, 1:15000; Invitro-
gen, Carlsland, CA, USA), rabbit anti-HK2 (C64G5, 1:2000,
catalog no. 2867; Cell signaling) or rabbit anti-GAPDH
(1:5000; Santa Cruz Biotechnology) overnight at 41C.
Chromatin immunoprecipitation assay
the Magna ChIP A Chromatin Immunoprecipitation Kit
(catalog no. 17-610; Millipore, Temecula, CA, USA) according
to the manufacturer’s instructions. Equal amounts of cross-
linked chromatin from 1?106Panc-1 cells treated with TSA
(1mM) or DMSO for 2h were precipitated with either 10ml
acetylated-H3 antibody or 10ml normal rabbit IgG overnight.
The total chromatin (1%) in each reaction was saved as an
‘input’ for later quantification. After purification, the enriched
DNA was quantified by RT–PCR using primers specific for
the ZAG promoter. The amount of immunoprecipitated
DNA in each sample is represented as signal relative to input
chromatin. Each yielded value was then normalized to control
(DMSO treated). The assays were repeated three times.
Sequences of primers are provided in Supplementary data
Synthetic siRNA oligonucleotides for ZAG and negative
control siRNA were purchased from Ambion (#s1848_1#
and s1849_2#; Applied Biosystems, Darmstadt, Germany) and
were prepared and stored according to the manufacturer’s
instructions. siRNA transfections were carried out according
to the manufacturer’s instructions. siPORT NeoFX Transfec-
tion Agent (Ambion Applied Biosystem, Austin, TX, USA)
transfection reagent was used. The final concentration of both
the control and specific oligonucleotides was 40nM. The
efficacy of the siRNA transfection was ascertained by
immunoblot analysis after 48h of transfection. Synthetic
siRNA oligonucleotides for HK2 were obtained from Qiagen
(catalog nos SI03021935_1# and SI00287329_2#, Qiagen,
Hilden, Germany) and transfected with HiPerFect transfection
reagent according to the manufacturer’s instructions. The
efficacy of HK2 downregulation was assessed after 72h of
transfection. The sequences of siRNA oligonucleotides are
provided in Supplementary data section.
Cell growth was determined using the 3-(4,5-dimethylthiazole-
2-yl)2,5-diphenyltetrazolium bromide (5mg/ml in phosphate-
buffered saline; Sigma-Aldrich) colorimetric growth assay as
previously described (Zhang et al., 2007). All assays were
performed in triplicates and were repeated three times.
To assess cell invasion in vitro, we used 24-well Matrigel
invasion chambers with 8-mm pore sizes (BD Biosciences, San
Jose, CA, USA) and reconstituted them with 600ml serum-free
Dulbecco’s modified Eagle’s medium in both the top and the
bottom chambers for 2–4h. Cells were trypsinized and were
seeded into the top chamber at a density of 5?104cells (Aspc-1)
or 1.25?104cells (Panc-1 and Su86.86) per well in 500ml
Dulbecco’s modified Eagle’s medium containing 0.5% fetal
calf serum. The outer chambers contained 0.7ml of medium
(10% fetal calf serum). According to each experimental setup,
TGF-b (CF111; Millipore) or rZAG was added to the top
chambers at a dose of 10ng/ml, 1 or 0.01mg/ml. After
incubation at 371C for 24h, cells remaining attached to the
upper surface of the membrane were carefully removed with
cotton swabs, whereas cells that reached the underside of the
chamber were stained with hematoxylin and eosin and were
counted. All experiments were repeated three times.
Detection of EMT markers, phospho-ERK and p21
To analyze the effect of TGF-b and rZAG on EMT, we grew
Panc-1 cells to 70% confluence in Dulbecco’s modified Eagle’s
medium containing 10% fetal calf serum. Cells were washed
twice in serum-free medium and were starved for 24h in
serum-free medium, and were finally treated for 48h with
TGF-b (10ng/ml) or carrier protein. rZAG was added at a
concentration of 1mg/ml, 90min before the application of
TGF-b. For Aspc-1, cells were (siRNA) transfected as
described. At 24h after transfection, the medium was removed
and the cells were washed twice in serum-free medium and
were starved for 12h in serum-free medium. The cells were
then treated for 36h with TGF-b (10ng/ml) or carrier protein.
SB525354 (200nM), LY-364947 (25mM) and PD98059 (50mM,
all from Tocris Bioscience, Bristol, UK) were used to treat
Aspc-1 cells for 24h in serum-free medium before RNA
extraction. For the detection of phospho-ERK and p21,
a similar experimental setup was carried out with the exception
that TGF-b treatment was restricted to 24h. All experiments
were repeated three times.
RT–PCR pathway finder array
All the reagents and materials for the PCR array (PAHS-014)
were purchased from SABiosciences (Frederick, MD, USA).
The assay was performed according to the manufacturer’s
instruction. Datawere analyzedusing theweb-based
AZGP1 in pancreatic cancer
B Kong et al
software from SABiosciences (http://www.sabiosciences.com/
For statistical analyses, the GraphPad Prism 5 Software
(GraphPad, San Diego, CA, USA) was used. The w2-test was
used to compare the numbers of ZAG-positive samples
between PanIN lesions, primary and metastatic PDAC, as
well as the frequency of ZAG-positive samples in PETs,
primary PDAC and ACCs. Unless otherwise stated, an
unpaired t-test was used for group-wise comparisons. The
level of statistical significance was set at Po0.05. Results are
expressed as mean±standard error of the mean (s.e.m.) unless
Conflict of interest
The authors declare no conflict of interest.
We thank Felicitas Altmayr, Tanja Rossmann-Bloeck, Manja
Thorwirth and Carmen Marthen for excellent technical support.
This study was in part supported by the European Union (within
the framework of the ‘MolDiagPaca’ project; to JK, CWM and
HF) and by the commission for clinical research of the TU
Munich (KKF). BK received a fellowship from the Scholarship
Council of the Ministry of Education of China.
Barbie DA, Tamayo P, Boehm JS, Kim SY, Moody SE, Dunn IF et al.
(2009). Systematic RNA interference reveals that oncogenic
KRAS-driven cancers require TBK1. Nature 462: 108–112.
Benfield AP, Whiddon BB, Clements JH, Martin SF. (2007).
Structural and energetic aspects of Grb2-SH2 domain-swapping.
Arch Biochem Biophys 462: 47–53.
Biliran Jr H, Wang Y, Banerjee S, Xu H, Heng H, Thakur A et al.
(2005). Overexpression of cyclin D1 promotes tumor cell growth
and confers resistance to cisplatin-mediated apoptosis in an elastase-
myc transgene-expressing pancreatic tumor cell line. Clin Cancer
Res 11: 6075–6086.
Bing C, Bao Y, Jenkins J, Sanders P, Manieri M, Cinti S et al. (2004).
Zinc-alpha2-glycoprotein, a lipid mobilizing factor, is expressed in
adipocytes and is up-regulated in mice with cancer cachexia. Proc
Natl Acad Sci USA 101: 2500–2505.
Brabletz T, Jung A, Spaderna S, Hlubek F, Kirchner T. (2005).
Opinion: migrating cancer stem cells—an integrated concept of
malignant tumour progression. Nat Rev Cancer 5: 744–749.
Burgi W, Schmid K. (1961). Preparation and properties of Zn-alpha 2-
glycoprotein of normal human plasma. J Biol Chem 236:
Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. (2003).
Overweight, obesity, and mortality from cancer in a prospectively
studied cohort of U.S. adults. N Engl J Med 348: 1625–1638.
Christofk HR, Vander Heiden MG, Harris MH, Ramanathan A,
Gerszten RE, Wei R et al. (2008). The M2 splice isoform of
pyruvate kinase is important for cancer metabolism and tumour
growth. Nature 452: 230–233.
De Caestecker MP, Parks WT, Frank CJ, Castagnino P, Bottaro DP,
Roberts AB et al. (1998). Smad2 transduces common signals
from receptor serine-threonine and tyrosine kinases. Genes Dev 12:
Dembinski JL, Krauss S. (2009). Characterization and functional
analysis of a slow cycling stem cell-like subpopulation in pancreas
adenocarcinoma. Clin Exp Metastasis 26: 611–623.
Diez-Itza I, Sanchez LM, Allende MT, Vizoso F, Ruibal A, Lopez-
Otin C. (1993). Zn-alpha 2-glycoprotein levels in breast cancer
cytosols and correlation with clinical, histological and biochemical
parameters. Eur J Cancer 29A: 1256–1260.
Dumont N, Wilson MB, Crawford YG, Reynolds PA, Sigaroudinia
M, Tlsty TD. (2008). Sustained induction of epithelial to
mesenchymal transition activates DNA methylation of genes
silenced in basal-like breast cancers. Proc Natl Acad Sci USA 105:
Ellenrieder V, Hendler SF, Boeck W, Seufferlein T, Menke A,
Ruhland C et al. (2001). Transforming growth factor beta1
treatment leads to an epithelial-mesenchymal transdifferentiation
of pancreatic cancer cells requiring extracellular signal-regulated
kinase 2 activation. Cancer Res 61: 4222–4228.
Esposito I, Kleeff J, Abiatari I, Shi X, Giese N, Bergmann F et al.
(2007). Overexpression of cellular inhibitor of apoptosis protein 2 is
an early event in the progression of pancreatic cancer. J Clin Pathol
Evdokimova V, Tognon C, Ng T, Sorensen PH. (2009). Reduced
proliferation and enhanced migration: two sides of the same coin?
Molecular mechanisms of metastatic progression by YB-1. Cell
Cycle 8: 2901–2906.
Finkel E. (1996). High hopes for p21 in cancer treatment. Lancet
Freemantle SJ, Liu X, Feng Q, Galimberti F, Blumen S, Sekula D
et al. (2007). Cyclin degradation for cancer therapy and chemopre-
vention. J Cell Biochem 102: 869–877.
Frenette G, Dube JY, Lazure C, Paradis G, Chretien M, Tremblay
RR. (1987). The major 40-kDa glycoprotein in human prostatic
fluid is identical to Zn-alpha 2-glycoprotein. Prostate 11: 257–270.
Friess H, Yamanaka Y, Buchler M, Ebert M, Beger HG, Gold LI et al.
(1993). Enhanced expression of transforming growth factor beta
isoforms in pancreatic cancer correlates with decreased survival.
Gastroenterology 105: 1846–1856.
Fritsche P, Seidler B, Schuler S, Schnieke A, Gottlicher M, Schmid
RM et al. (2009). HDAC2 mediates therapeutic resistance of
pancreatic cancer cells via the BH3-only protein NOXA. Gut 58:
Giehl K, Skripczynski B, Mansard A, Menke A, Gierschik P. (2000).
Growth factor-dependent activation of the Ras–Raf–MEK–MAPK
pathway in the human pancreatic carcinoma cell line PANC-1
carrying activated K-ras: implications for cell proliferation and cell
migration. Oncogene 19: 2930–2942.
Glozak MA, Seto E. (2007). Histone deacetylases and cancer.
Oncogene 26: 5420–5432.
Haigis KM, Kendall KR, Wang Y, Cheung A, Haigis MC, Glickman
JN et al. (2008). Differential effects of oncogenic K-Ras and N-Ras
on proliferation, differentiation and tumor progression in the colon.
Nat Genet 40: 600–608.
Hale LP, Price DT, Sanchez LM, Demark-Wahnefried W, Madden JF.
(2001). Zinc alpha-2-glycoprotein is expressed by malignant
prostatic epithelium and may serve as a potential serum marker
for prostate cancer. Clin Cancer Res 7: 846–853.
Hamada S, Satoh K, Hirota M, Kimura K, Kanno A, Masamune A
et al. (2007). Bone morphogenetic protein 4 induces epithelial-
mesenchymal transition through MSX2 induction on pancreatic
cancer cell line. J Cell Physiol 213: 768–774.
Horiguchi K, Shirakihara T, Nakano A, Imamura T, Miyazono K,
Saitoh M. (2009). Role of Ras signaling in the induction of snail by
transforming growth factor-beta. J Biol Chem 284: 245–253.
Jaenisch R, Bird A. (2003). Epigenetic regulation of gene expression:
how the genome integrates intrinsic and environmental signals. Nat
Genet 33(Suppl): 245–254.
AZGP1 in pancreatic cancer
B Kong et al
Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, Angenendt P et al. Download full-text
(2008). Core signaling pathways in human pancreatic cancers
revealed by global genomic analyses. Science 321: 1801–1806.
Ketterer K, Kong B, Frank D, Giese NA, Bauer A, Hoheisel J et al.
(2009). Neuromedin U is overexpressed in pancreatic cancer and
increases invasiveness via the hepatocyte growth factor c-Met
pathway. Cancer Lett 277: 72–81.
Khasawneh J, Schulz MD, Walch A, Rozman J, Hrabe de Angelis M,
Klingenspor M et al. (2009). Inflammation and mitochondrial fatty
acid beta-oxidation link obesity to early tumor promotion. Proc
Natl Acad Sci USA 106: 3354–3359.
Kretzschmar M, Doody J, Timokhina I, Massague J. (1999). A
mechanism of repression of TGFbeta/Smad signaling by oncogenic
Ras. Genes Dev 13: 804–816.
Lecanda J, Ganapathy V, D’Aquino-Ardalan C, Evans B, Cadacio C,
Ayala A et al. (2009). TGFbeta prevents proteasomal degradation
of the cyclin-dependent kinase inhibitor p27kip1 for cell cycle arrest.
Cell Cycle 8: 742–756.
Lemoine NR, Jain S, Hughes CM, Staddon SL, Maillet B, Hall PA
et al. (1992). Ki-ras oncogene activation in preinvasive pancreatic
cancer. Gastroenterology 102: 230–236.
Lim JH, Jung CR, Lee CH, Im DS. (2008). Egr-1 and serum response
factor are involved in growth factors- and serum-mediated
induction of E2-EPF UCP expression that regulates the VHL-HIF
pathway. J Cell Biochem 105: 1117–1127.
Longnecker DS, Terhune PG. (1998). What is the true rate of K-ras
mutation in carcinoma of the pancreas? Pancreas 17: 323–324.
Lung FD, Tsai JY. (2003). Grb2 SH2 domain-binding peptide analogs
as potential anticancer agents. Biopolymers 71: 132–140.
Marrades MP, Martinez JA, Moreno-Aliaga MJ. (2008). ZAG, a lipid
mobilizing adipokine, is downregulated in human obesity. J Physiol
Biochem 64: 61–66.
Mathupala SP, Ko YH, Pedersen PL. (2006). Hexokinase II:
cancer’s double-edged sword acting as both facilitator and gate-
keeper of malignancy when bound to mitochondria. Oncogene 25:
Michalski CW, Laukert T, Sauliunaite D, Pacher P, Bergmann F,
Agarwal N et al. (2007a). Cannabinoids ameliorate pain and reduce
disease pathology in cerulein-induced acute pancreatitis. Gastro-
enterology 132: 1968–1978.
Michalski CW, Maier M, Erkan M, Sauliunaite D, Bergmann F,
PacherP etal. (2008).Cannabinoids
inflammation and fibrosis in pancreatic stellate cells. PLoS One 3:
Michalski CW, Shi X, Reiser C, Fachinger P, Zimmermann A, Buchler
MW et al. (2007b). Neurokinin-2 receptor levels correlate with
intensity, frequency, and duration of pain in chronic pancreatitis.
Ann Surg 246: 786–793.
reduce markers of
Michaud DS, Giovannucci E, Willett WC, Colditz GA, Stampfer MJ,
Fuchs CS. (2001). Physical activity, obesity, height, and the risk of
pancreatic cancer. JAMA 286: 921–929.
Ohkubo I, Niwa M, Takashima A, Nishikimi N, Gasa S, Sasaki M.
(1990). Human seminal plasma Zn-alpha 2-glycoprotein: its
purification and properties as compared with human plasma
Zn-alpha 2-glycoprotein. Biochim Biophys Acta 1034: 152–156.
Roberts AB, Wakefield LM. (2003). The two faces of transforming
growth factor beta in carcinogenesis. Proc Natl Acad Sci USA 100:
Robson EJ, Khaled WT, Abell K, Watson CJ. (2006). Epithelial-to-
mesenchymal transition confers resistance to apoptosis in three
murine mammary epithelial cell lines. Differentiation 74: 254–264.
Rolli V, Radosavljevic M, Astier V, Macquin C, Castan-Laurell I,
Visentin V et al. (2007). Lipolysis is altered in MHC class I zinc-
alpha(2)-glycoprotein deficient mice. FEBS Lett 581: 394–400.
Russell ST, Tisdale MJ. (2010). Antidiabetic properties of zinc-alpha2-
glycoprotein in ob/ob mice. Endocrinology 151: 948–957.
Russell ST, Zimmerman TP, Domin BA, Tisdale MJ. (2004).
Induction of lipolysis in vitro and loss of body fat in vivo by zinc-
alpha2-glycoprotein. Biochim Biophys Acta 1636: 59–68.
Singh A, Greninger P, Rhodes D, Koopman L, Violette S, Bardeesy N
et al. (2009). A gene expression signature associated with ‘K-Ras
addiction’ reveals regulators of EMT and tumor cell survival.
Cancer Cell 15: 489–500.
Subramanian G, Schwarz RE, Higgins L, McEnroe G, Chakravarty S,
Dugar S et al. (2004). Targeting endogenous transforming growth
factor beta receptor signaling in SMAD4-deficient human pancrea-
tic carcinoma cells inhibits their invasive phenotype1. Cancer Res
Tada T, Ohkubo I, Niwa M, Sasaki M, Tateyama H, Eimoto T.
(1991). Immunohistochemical localization of Zn-alpha 2-glycopro-
tein in normal human tissues. J Histochem Cytochem 39: 1221–1226.
Von Burstin J, Eser S, Paul MC, Seidler B, Brandl M, Messer M et al.
(2009). E-cadherin regulates metastasis of pancreatic cancer in vivo
and is suppressed by a SNAIL/HDAC1/HDAC2 repressor com-
plex. Gastroenterology 137: 361–371.
Wan M, Huang J, Jhala NC, Tytler EM, Yang L, Vickers SM et al.
(2005). SCF(beta-TrCP1) controls Smad4 protein stability in
pancreatic cancer cells. Am J Pathol 166: 1379–1392.
Wang Z, Li Y, Kong D, Banerjee S, Ahmad A, Azmi AS et al. (2009).
Acquisition of epithelial-mesenchymal transition phenotype of
gemcitabine-resistant pancreatic cancer cells is linked with activa-
tion of the notch signaling pathway. Cancer Res 69: 2400–2407.
Zhang W, Erkan M, Abiatari I, Giese NA, Felix K, Kayed H et al.
(2007). Expression of extracellular matrix metalloproteinase inducer
(EMMPRIN/CD147) in pancreatic neoplasm and pancreatic stellate
cells. Cancer Biol Ther 6: 218–227.
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)
AZGP1 in pancreatic cancer
B Kong et al