Key contribution of CPEB4mediated translational control to cancer progression

Abstract

Malignant transformation, invasion and angiogenesis rely on the coordinated reprogramming of gene expression in the cells from which the tumor originated. Although deregulated gene expression has been extensively studied at genomic and epigenetic scales, the contribution of the regulation of mRNA-specific translation to this reprogramming is not well understood. Here we show that cytoplasmic polyadenylation element binding protein 4 (CPEB4), an RNA binding protein that mediates meiotic mRNA cytoplasmic polyadenylation and translation, is overexpressed in pancreatic ductal adenocarcinomas and glioblastomas, where it supports tumor growth, vascularization and invasion. We also show that, in pancreatic tumors, the pro-oncogenic functions of CPEB4 originate in the translational activation of mRNAs that are silenced in normal tissue, including the mRNA of tissue plasminogen activator, a key contributor to pancreatic ductal adenocarcinoma malignancy. Taken together, our results document a key role for post-transcriptional gene regulation in tumor development and describe a detailed mechanism for gene expression reprogramming underlying malignant tumor progression.

Full text

articles
nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012 83
Reprogramming of gene expression is a key step during cancer develop-
ment in which the aberrant control of genes primarily related to cell
proliferation, apoptosis, metabolism and invasion sets up the basis for
malignant transformation1. This phenomenon has been extensively
studied at the level of transcription, but it is becoming evident that
the post-transcriptional regulation of specific mRNA subpopulations
contributes substantially to the broad expression changes of the genes
responsible for tumoral properties of cells2–5. Coordinate regula-
tion of mRNA subpopulations is mediated by common cis-acting
elements in their 3′ untranslated regions (UTRs). This is the case for
the cytoplasmic polyadenylation element (CPE), which is present in
hundreds of mRNAs involved in cell proliferation, chromosome seg-
regation and cell differentiation6–9. CPE is bound by the CPE binding
proteins (CPEBs)10,11. CPEBs can act as either translational repressors
or activators9–13 to regulate mitotic and meiotic cell cycles and senes-
cence9,14–16, although no direct link between cancer and CPEBs has
been found. To investigate the oncogenic potential of CPEBs, we ana-
lyzed pancreatic ductal adenocarcinoma (PDA) and glioblastoma as
model systems. PDA and glioblastomas are the solid tumors with the
worst clinical prognoses, and they have been mainly studied at the
level of transcriptional regulation17,18.
We report that CPEB4 is overexpressed in PDA and that its down-
regulation in mouse xenograft models results in a marked reduction
of tumorigenic properties, primarily proliferation and vascularization,
resulting in increased survival. We also show that CPEB4 is asso-
ciated with a large number of transcripts in cancer cells, and, in
the case of tissue plasminogen activator (tPA) mRNA, the tumor-
associated overexpression of CPEB4 results in poly(A) tail elongation
and translational activation, leading to abnormal overexpression of
tPA in tumors. Our results indicate that the high amounts of CPEB4
in PDA result in a pathological reprogramming of gene expression
that drives the translational activation of mRNAs encoding factors
that sustain tumor progression. In addition, we found that CPEB4 is
also overexpressed in glioblastoma and that overexpression supports
tumoral proliferation and angiogenesis, arguing that this is a general
mechanism of gene expression regulation in cancer. These findings
may therefore hold diagnostic and therapeutic implications for PDA,
glioblastoma and, possibly, other tumor types.
RESULTS
CPEB4 is overexpressed in human PDA
To address whether CPEB-mediated mRNA regulation could have a
role in PDA, we first assessed CPEB expression in a panel of human
cell lines and tissues. Of the CPEB proteins tested, only CPEB4 was
clearly associated with the tumoral phenotype, with five out of seven
tumoral cell lines showing high amounts of CPEB4 mRNA and
1Cancer Research Programme, Hospital del Mar Research Institute (IMIM), Barcelona, Spain. 2Molecular Pathology Programme, Spanish National Cancer
Research Centre (CNIO), Madrid, Spain. 3Department of Experimental and Health Sciences, Universitat Pompeu Fabra (UPF), Barcelona, Spain. 4Center for
Genomic Regulation (CRG), Barcelona, Spain. 5Institute for Research in Biomedicine (IRB Barcelona), Barcelona, Spain. 6Department of Pathology, Hospital del
Mar, Parc de Salut Mar, Universitat Autónoma de Barcelona (UAB), Barcelona, Spain. 7Computational Genomics Group, Universitat Pompeu Fabra (UPF), Barcelona,
Spain. 8Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain. Present addresses: Richard Dimbleby Department of Cancer Research,
King’s College London, London, UK (E.O.-Z.) and Department of Anatomy and Structural Biology, Albert Einstein College of Medicine, Bronx, New York, USA (C.E.).
Correspondence should be addressed to R.M. (raul.mendez@irbbarcelona.es) or P.N. (pnavarro@imim.es).
Received 7 February; accepted 28 September; published online 4 December 2011; doi:10.1038/nm.2540
Key contribution of CPEB4-mediated translational
control to cancer progression
Elena Ortiz-Zapater1–3, David Pineda1,4, Neus Martínez-Bosch1, Gonzalo Fernández-Miranda5,
Mar Iglesias1,6, Francesc Alameda6, Mireia Moreno1, Carolina Eliscovich4, Eduardo Eyras7,8,
Francisco X Real1–3, Raúl Méndez4,5,8 & Pilar Navarro1
Malignant transformation, invasion and angiogenesis rely on the coordinated reprogramming of gene expression in the cells
from which the tumor originated. Although deregulated gene expression has been extensively studied at genomic and epigenetic
scales, the contribution of the regulation of mRNA-specific translation to this reprogramming is not well understood. Here we
show that cytoplasmic polyadenylation element binding protein 4 (CPEB4), an RNA binding protein that mediates meiotic mRNA
cytoplasmic polyadenylation and translation, is overexpressed in pancreatic ductal adenocarcinomas and glioblastomas, where
it supports tumor growth, vascularization and invasion. We also show that, in pancreatic tumors, the pro-oncogenic functions
of CPEB4 originate in the translational activation of mRNAs that are silenced in normal tissue, including the mRNA of tissue
plasminogen activator, a key contributor to pancreatic ductal adenocarcinoma malignancy. Taken together, our results document
a key role for post-transcriptional gene regulation in tumor development and describe a detailed mechanism for gene expression
reprogramming underlying malignant tumor progression.
npg © 2012 Nature America, Inc. All rights reserved.

articles
84 VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine
protein compared to a non-tumorigenic pancreatic cell line (HPDE)
(Fig. 1a,b). We also assessed CPEB4 concentrations in vivo using a set
of 190 samples from a cohort of 58 patients (Fig. 1c,d). The samples
we analyzed were from normal pancreas (n = 47), tissue with pancrea-
titis (n = 51), pancreatic intraepithelial neoplasia (PanIN) lesions (54)
and pancreatic adenocarcinomas (n = 38, with 27 well or moderately
differentiated adenocarcinomas and 11 undifferentiated adenocar-
cinomas) (Supplementary Table 1). In normal pancreas, CPEB4
expression was restricted to the islets (100%) and, occasionally, to
ductal cells (25.5%). In the inflammatory atrophic pancreas, CPEB4
was expressed in 19.6% of ducts. Notably, the expression of CPEB4
in ductal cells increased during malignant transformation, showing a
low-to-moderate frequency (34.6%) in low-grade PanINs and stronger
expression and higher frequency in high-grade PanINs (71.4%) and
in moderately differentiated or well-differentiated PDAs (88.8%).
CPEB4 expression was reduced in undifferentiated tumors (36.3%).
Taken together, these results show in vitro and in vivo correlations
between CPEB4 expression and tumoral phenotype in human PDA,
suggesting a function for this protein during tumorigenesis.
Downregulation of CPEB4 reduces malignancy in vitro
To address the relevance of CPEB4 overexpression in tumorigenesis,
we knocked down its expression in the tumoral pancreatic cell line
RWP-1 by transfection with CPEB4 shRNA (shCPEB4). We used cells
transfected with scrambled shRNA (shCtrl) or parental cells (untrans-
fected) as controls. We monitored the efficiency of the CPEB4 down-
regulation by quantitative RT-PCR (qRT-PCR) and western blotting
(Fig. 2a,b). The proliferation rate was similar in untransfected, shCtrl
and shCPEB4 cells (Fig. 2c). In contrast, both Matrigel invasion
(Fig. 2d) and soft-agar colony formation (Fig. 2e) (used as indica-
tors of malignant transformation) were strongly reduced after CPEB4
knockdown, indicating that CPEB4 may favor tumor progression by
increasing invasiveness and anchorage-independent growth.
CPEB4 knockdown leads to reduced in vivo tumorigenicity
To determine whether CPEB4 is required for tumor progression
in vivo, we analyzed the tumorigenic properties of RWP-1 parental
cells, shCtrl cells and two different types of shCPEB4 cells (shCPEB4_2
and shCPEB4_4), which we injected subcutaneously in the flanks of
nude mice. We killed all animals 3 weeks after injection and confirmed
CPEB4 downregulation by immunohistochemistry (Supplementary
Fig. 1a,b). We found a 51% (shCPEB4_2) and a 57% (shCPEB4_4)
reduction in tumor weight and a 67% (shCPEB4_2) and an 80%
(shCPEB4_4) reduction in tumor volume after CPEB4 knockdown
(Fig. 3a). To rule out a cell-line–specific phenotype, we also down-
regulated CPEB4 expression in cells from the pancreatic cancer line
Capan-1 (Supplementary Fig. 2) and tested the effects on in vivo
tumorigenicity, which confirmed a significant decrease of tumor
weight and volume (Fig. 3b). These data indicate that CPEB4 down-
regulation in pancreatic tumoral cells results, directly or indirectly,
in slower tumor growth.
Peritoneal dissemination is one of the primary mechanisms in the
progression of pancreatic cancer19. Therefore, we analyzed tumor
formation after intraperitoneal injection of RWP-1 cells transduced
with luciferase (RWP-1 LUC cells). We killed the mice injected
with these cells when their tumors grew to over 1.5 cm in diameter
and confirmed CPEB4 downregulation by immunohistochemistr y
(Supplementary Fig. 1a,b). Luciferase activity detection showed a
clear delay of tumor formation in the mice injected with shCPEB4
cells (Fig. 3c and Supplementar y Fig. 3a). Even though they had
very different kinetics and size, our postmortem analysis showed
that both control and CPEB4 knockdown cells generated tumors
that were localized to the pancreas (with 84.2% frequency in both
groups). In the in vitro invasion samples (Fig. 2d), we consistently
found a clear trend in the control cells to develop tumors in other
organs (including the intestine, liver, kidney, stomach and ovary) at
a higher frequency than CPEB4 knockdown cells (73.6% compared
to 47.36%, respectively) (Supplementary Fig. 3b), suggesting an
impaired invasive capacity in the absence of CPEB4. Kaplan-Meier
survival curves showed that CPEB4 knockdown was associated with
a significant delay in mouse mortality (P = 0.024) (Fig. 3d). These
data show that pancreatic tumoral cells with low CPEB4 expression
have slower tumor growth and a lower capability to form tumors in
non-pancreatic tissues, resulting in a slower progression to death.
Figure 1 CPEB4 is overexpressed in PDA
cell lines and tumor tissues. (a) Expression
of CPEB4 mRNA detected by qRT-PCR in
a panel of human pancreatic cell lines.
Values, compared to hypoxanthine-guanine
phosphoribosyltransferase (HPRT) mRNA levels,
correspond to three independent experiments
Data are mean ± s.e.m. The P values
(determined by Student’s t test) are relative
to HPDE cells. *P ≤ 0.05, **P ≤ 0.01,
***P ≤ 0.001. (b) Analysis by immuno-
fluorescence of CPEB4 expression in
immortalized normal pancreatic ductal cells
(HPDE) and PDA (RWP-1 and SK-PC-1).
Scale bars, 40 µm. (c) CPEB4 expression
determined by immunohistochemistry in sections
of formalin-fixed paraffin-embedded normal
human pancreas, samples with pancreatitis,
low- and high-grade PanIN lesions and
differentiated and undifferentiated tumors.
Scale bars, 200 µm (for normal pancreas
and differentiated and undifferentiated
adenocarcinoma) or 50 µm (for pancreatitis and
low- and high-grade PanIN). (d) Quantification of CPEB4 protein expression in human pancreatic tissue samples according to histology. The stacked bars show
the percent contribution of high and low CPEB4-positive samples (Supplementary Table 1, and see the legend of Supplementary Table 1 for the classification
of high and low samples) relative to the total number of samples in each pancreatic lesion. ***P ≤ 0.001. Diff., differentiated; undiff., undifferentiated.
c
d
bHPDE RWP-1 SK-PC-1
120
100
80
60
40
Percentage of
CPEB4-positive samples
20
0
Normal
ducts
Atrophic
ducts
Low-grade
PanINs
High-grade
PanINs
Diff.
tumor
Undiff.
tumor
*** ***
High
Low
25
a
20
15
10 *
** **
*** *
*
5
Relative CPEB4
mRNA expression
0
HPDE
IMIM-PC-2
IMIM-PC-1
PANC-1
RWP-1
SK-PC-1
SK-PC-3
Capan-1
Normal pancreas
High-grade
PanIN
Pancreatitis
Differentiated
adenocarcinoma
Undifferentiated
adenocarcinoma
Low-grade PanIN
npg © 2012 Nature America, Inc. All rights reserved.

articles
nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012 85
CPEB4 knockdown reduces proliferation and vascularization
We next aimed to determine whether the differences in the growth
rates of the tumors were a result of the tumor architecture rather
than the intrinsic properties of the tumoral cells. A histopathological
characterization of tumors that developed after subcutaneous
(Fig. 4a,b) and intraperitoneal injection (Supplementary Fig. 4)
1.2
abcd
1.0
0.8
*****
***
*
*
0.6
0.4
Relative CPEB4 mRNA expression
Abs 570
Abs 590
0.2
0
2.50 0.8
0.6
0.4
0.2
0
2.00
1.50
1.00
0.50
0
1234
Time in culture (d)
5678
shCtrl 1 2 3
shCPEB4
4
CPEB490
Tubulin
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
55
e
**
**
Number of colonies
60
50
40
30
20
10
0
Untransfected
Untransfected
shCtrl
shCtrl
shCPEB4_2
shCPEB4_4
shCPEB4_2 shCPEB4_4
Figure 2 In vitro characterization of the effects of CPEB4 knockdown in RWP-1 cells.
(a) CPEB4 mRNA expression in parental RWP-1 cells and in cells transfected with shCtrl
or four different CPEB4-targeting shRNAs (shCPEB4_1 through shCPEB4_4). Values
correspond to three independent experiments. Data are mean ± s.e.m. The P values
(determined by Student’s t test) are relative to shCtrl cells. *P ≤ 0.05. shCPEB4_2 and
shCPEB4_4 showed the greatest reduction of CPEB4 concentrations and were selected for
further experiments. (b) Downregulation of CPEB4 protein concentrations in control cells
and in cells transfected with shCPEB4, as analyzed by western blot. Tubulin concentrations
are shown as the loading control. One representative experiment (n = 3) is shown. (c) Proliferation
rate of parental RWP-1 cells or cells harboring shCtrl or shCPEB4 analyzed using a 3-(4.5-
dimethylthiazol-2-yl)-2.5-diphenyltetrazolium bromide (MTT) assay (three independent
experiments). Data are mean ± s.e.m. Abs 570, absorbance at 570 nm. (d) Matrigel invasion
of parental, shCtrl and shCPEB4 RWP-1 cells. Invasion (after 72 h) was measured by absorbance at 590 nm (Abs 590) (three independent
experiments). Data are mean ± s.e.m. The P values (determined by Student’s t test) are relative to shCtrl cells. ***P ≤ 0.001. (e) Soft agar colony
formation of parental RWP-1 cells and cells expressing shCtrl or shCPEB4. The graph shows the number of colonies (mean ± s.e.m.) after 4 weeks
of three independent experiments. The P values (determined by Student’s t test) are relative to shCtrl cells. **P ≤ 0.01. Shown at the right are
representative images of the colonies formed by each type of cell.
Untransfected
a b
1.0 0.5
0.4
0.3
0.2
0.1
0
1.2
0.9
0.6
0.3
0
0.8
0.6 **
** **
**
***
0.4
Weight (g)
Weight (g)
Volume (cm3)
0.2
0
Untransfected
shCtrl shCPEB4_2 shCPEB4_4
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
1.2
0.9
0.6
0.3
0
**
**
Volume (cm3)
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
c d
*
Untransfected shCtrl
shCPEB4_2 shCPEB4_4
1.0
0.8
0.6
0.4
0.2
Control
shCPEB4
0
0 10 20 30 40
Time (d)
50 60
Cumulative survival
Figure 3 Effects of CPEB4 knockdown in vivo on tumorigenicity in nude mice.
(a) Tumorigenicity of untransfected, shCtrl, shCPEB4_2 and shCPEB4_4 RWP-1
cells (1 × 106) after subcutaneous injection in the flanks of nude mice (n = 10).
One representative mouse and the tumors it formed are shown. The quantification of
tumor weight and volume is shown on the right. Box and whisker plots show the mean,
quartiles and tenth and ninetieth percentiles of the data. The P values (determined
by Student’s t test) are relative to tumors formed by shCtrl cells. *P ≤ 0.05, **P ≤ 0.01,
***P ≤ 0.001. Scale bars, 1 cm. (b) Quantification of the tumor weight and volume
of untransfected, shCtrl, shCPEB4_2 and shCPEB4_4 Capan-1 cells (1 × 106) after
subcutaneous injection in the flanks of nude mice (n = 8). Box and whisker plots
show the mean, quartiles and tenth and ninetieth percentiles of the data. The P values
(determined by Student’s t test) are relative to tumors formed by shCtrl cells. *P ≤ 0.05,
**P ≤ 0.01. (c) RWP-1 LUC cells were injected intraperitoneally, and bioluminescent
activity was measured weekly. Two representative mice from each group (n = 10) after
3 weeks of injection are shown. Scale bars, 1 cm. (d) Kaplan-Meier survival curves in tumor-bearing nude mice injected with control or knockdown
CPEB4 (shCPEB4) RWP-1 LUC cells. Control denotes the group of mice injected with untransfected or shCtrl RWP-1 LUC cells (n = 19), and shCPEB4
denotes the group of mice injected with shCPEB4_2 and shCPEB4_4 RWP-1 LUC cells (n = 19). The P values (determined by log-rank test) are relative
to mice from the control group. *P ≤ 0.05.
npg © 2012 Nature America, Inc. All rights reserved.

articles
86 VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine
showed that tumors from shCPEB4 cells had a reduced prolifera-
tion rate (determined by Ki67 labeling) and decreased microvessel
density (by von Willebrand factor (vWF) labeling) compared to
parental and shCtrl RWP-1 cells. Notably, tumors from shCPEB4
cells showed less ischemic necrotic areas and increased stroma, as
detected by α smooth muscle actin (α-SMA) staining (Fig. 4c). The
fact that we found a similar increase in stroma formation in tumors
derived from subcutaneous and intraperitoneal injection indicates
that this increase is not a consequence of tumor size (as, in the mice
injected intraperitoneally, we collected all tumors when they reached
~1.5 cm3) or the time required to develop the tumor (as, in the mice
injected subcutaneously, we killed them all at the same time) but,
rather, was an effect directly related to low CPEB4 expression that
resulted in an increased amount of tumor-associated fibrosis. We
also observed reduced cell proliferation, tumoral angiogenesis and
increased stroma formation in tumors from Capan-1 shCPEB4 cells
(Supplementary Fig. 5).
These results indicate that, although CPEB4 is not required in a
cell-autonomous manner for cellular proliferation, reduced amounts
of CPEB4 affect both the proliferation rate of tumors and their archi-
tecture by reducing vascularization and increasing stroma formation,
with the consequence of reduced tumor growth and invasion during
in vivo pancreatic tumorigenesis.
CPEB4 associates with a large number of CPE-containing mRNAs
We next aimed to determine whether the overexpression of CPEB4
in PDA resulted in the abnormal translational activation of mRNAs
encoding ‘pro-tumoral’ factors. To obtain a genome-wide picture
of potential CPEB4 targets in PDA, we performed RNA immuno-
precipitation (RIP) of CPEB4 from RWP-1 cell extracts and iden-
tified the associated mRNAs by Illumina Solexa ultrasequencing
(Supplementar y Fig. 6). We identified 842 mRNAs significantly
(P < 0.05) associated with CPEB4 when compared with the control
RNAs immunoprecipitated with IgG (Supplementary Tables 2 and 3).
Untransfected
a
b
c
vWF
100 8
6
4
2
0
*** *** *
*
**
** **
80
60
40
20
Percentage of
Ki67-positive cells
Percentage of area with
positive vWF staining
0
Subcutaneous Intraperitoneal
Ki67
**
*
***
100 µm100 µm100 µm
100 µm100 µm
*
*
*
*
*
**
**
*
*
shCtrl shCPEB4_2 shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Subcutaneous Intraperitoneal
Subcutaneous Intraperitoneal
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
H&E
α-SMAα-SMA H&E
shCtrl shCPEB4_2 shCPEB4_4
Figure 4 Effects of CPEB4 knockdown on proliferation, vessel density and stroma in xenograft tumors. (a) Immunostaining of Ki67 and vWF in tumors
formed by untransfected, shCtrl or shCPEB4 (shCPEB4_2 and shCPEB4_4) RWP-1 cells injected subcutaneously in nude mice. Scale bars, 100 µm.
(b) Quantification of Ki67 and vWF immunohistochemistry staining shown by the percentage of positively stained cells compared to total number of
cells per field. The P values (determined by Student’s t test) relative to shCtrl are shown. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. Data are mean ± s.e.m.
(c) H&E and α-SMA staining of tissue sections of the different tumors after subcutaneous and intraperitoneal injection of untransfected, shCtrl or
shCPEB4 RWP-1 cells. Tumors from shCPEB4 cells showed less ischemic necrotic areas (asterisks) and more stromal component between tumoral cells
(arrowheads and α-SMA staining). Scale bars, 200 µm.
b
0.5 1.0 120
100
80
60
40
20
0
0.8
0.6
0.4
0.2
0
0.4
0.3
0.2
0.1
Percent of inputPercent of input
Percent of input
Percent of input
Percent of input
Percent of input
Percent of input
Percent of input
Percent of input
0
4.0
3.0
2.0
1.0
0
4.0
3.0
2.0
1.0
0
4.0
3.0
2.0
1.0
0
4.0 8.0
6.0
4.0
2.0
0
3.0
2.0
1.0
0
0.5
0.4
0.3
0.2
0.1
0
MT2A MT1E tPA OAZ2 RPS27
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
Percent of input
0.5
0.4
0.3
0.2
0.1
0
TMOD3PORCNORC4LH2AFVIDH1
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
lgG
CPEB4
0.9
a
0.8
0.7
Proportion above
enrichment score threshold
0.6
0.5
–4 –3 –2 –1 0
Enrichment score
1 2 3
0.9
0.8
0.7
Proportion above
enrichment score threshold
0.6
0.5
–4 –3 –2 –1 0
Enrichment score
1 2 3 4
Figure 5 Illumina Solexa sequencing analysis of mRNAs bound to CPEB4
in RWP-1 cells. (a) Representation of the proportion of mRNAs with
activation CPEs having an enrichment score greater than or equal to each
cutoff value on the x-axis. At right is a representation of the proportion
of mRNAs with repression CPEs having an enrichment score greater
than or equal to each cutoff value on the x-axis. The proportions were
calculated over the total number of mRNAs for which we could find a CPE
of either type. (b) Validation by qRT-PCR of mRNAs bound to CPEB4. The
fold enrichment of target sequences in the CPEB4 immunoprecipitates
compared to IgG and to the input fraction is shown. Data are mean ± s.e.m.
npg © 2012 Nature America, Inc. All rights reserved.

articles
nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012 87
The 3′ UTRs of the mRNAs coimmunoprecipitated with CPEB4
mRNAs were enriched for activation arrangements of CPEs that are
able to recruit the polyadenylation complex6. In contrast, repression
CPE arrangements, which can assemble the translational repression
complex but not recruit the cytoplasmic polyadenylation machinery6,
were under-represented (Fig. 5a). A gene ontology analysis of the
CPEB4-associated mRNAs (Supplementary Fig. 7) showed a signifi-
cant (P ≤ 0.05) enrichment in cellular functions relevant to tumor-
igenesis (cell cycle, transcriptional regulation, apoptosis and DNA
damage); the most significant enrichment was for mRNAs encod-
ing general translation factors and ribosomal proteins, which are
differentially regulated and are key mediators of cellular transfor-
mation5,20. We further validated ten of the mRNAs associated with
CPEB4 with a validation rate of 90% (Fig. 5b and Supplementary
Table 4). Notably, the only mRNA identified in the original RIP
ultrasequencing that was not validated in the RIP qRT-PCR analysis
(RPS27) was also the only mRNA that did not contain CPEs. These
results indicate that, when overexpressed in PDA, CPEB4 is associ-
ated with a large number of mRNAs that are potential targets for
tumor-specific translational regulation.
tPA expression in pancreas is regulated by CPEB4
To test whether the association of these transcripts with CPEB4
resulted in their differential regulation in PDA, we next analyzed
in detail the relevance of CPEB4 in the translation of tPA mRNA,
one of the transcripts that was more enriched in the CPEB4
coimmunoprecipitation (Supplementary Table 2) and that is known
to be overexpressed in pancreatic tumors21–27. In addition, tPA mRNA
has been previously found to be regulated by CPEB1-mediated cyto-
plasmic polyadenylation in mouse oocytes and neurons28,29. When we
analyzed tPA mRNA expression in PDA compared to normal acinar
or ductal cells, we saw a clear discrepancy between the amounts
of protein and mRNA. Thus, tPA mRNA levels were high both in
normal ductal cells, in which tPA protein is absent, and in PDAs,
which express high amounts of tPA protein (Fig. 6a), suggesting that
tPA mRNA is stored silently in normal pancreatic ducts and is trans-
lationally activated in PDA.
To assess whether tPA mRNA was translationally regulated by cyto-
plasmic polyadenylation, we measured the length of the tPA mRNA
poly(A) tails in normal and tumoral samples by anchored RT-PCR6.
In normal pancreas, tPA mRNA had a short poly(A) tail, whereas
a b c
ef
d
3′ UTR of tPA
ARE
PUM
Cyclin B1
CPE
tPA
CPE
CPE CPE HEX
CPE
CPE CPE
CPE HEX
ARE HEX
020 40 60 80 100 120 140
Translation stimulation + progesterone
rey or Renilla Luc
(percentage C: B1)
HEX
PUM
ARE
ARE
ARE
AREARE
ARE CPE CPE HEX
HEX
HEX
CPE
70
Relative tPA mRNA expression
60
50
40
30
20
10
Acinar Ductal
Normal
PDA
Tumor
Endocrine
64
shCtrl
shCPEB4_2
shCPEB4_4
tPA (supernatants)
tPA (total extracts)
Tubulin
64
55
Relative tPA mRNA
expression
1.2
1.0
0.8
0.6
shCtrl 2 4
shCPEB4
0.4
0.2
0
shCtrl
SubcutaneousIntraperitoneal
shCPEB4_2 shCPEB4_4
shCtrl 2 4
*** ***
shCPEB4
Percent of tPA
staining
120
100
80
60
40
20
0shCtrl 2 4
***
***
shCPEB4
Percent of tPA
staining
120
100
80
60
40
20
0
A90
A20
1 2
Normal Tumor
1 2
A90
A20
Wild type
RWP-1
SK-PC-1
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
Untransfected
shCtrl
shCPEB4_2
shCPEB4_4
tPA mRNA
tPA mRNA
polyadenylation
A90
A20
A0
Figure 6 tPA expression is regulated at the
translational level by CPEB4 in normal and
tumoral pancreatic tissues. (a) Relative tPA
mRNA levels, compared to HPRT, detected
by qRT-PCR in normal human acinar and
ductal pancreatic cells, PDA and endocrine
tumors. Data are mean ± s.e.m. (b) Analysis
of the polyadenylation of tPA transcripts from mouse normal pancreas and ductal tumor
from Ela-myc mice (top) or normal human pancreas and RWP-1 and SK-PC-1 cells (bottom).
At the top, the complementary DNA was detected by labeling with [32P] γ-dATP followed by autoradiography. 1 indicates the internal control of tPA
mRNA expression, and 2 indicates polyadenylation. A20 and A90 indicate the respective length of poly(A) tail. (c) A schematic representation of the
3′ UTR of human tPA mRNA (top) and luciferase Renilla activity (bottom) of synthetic chimeric mRNAs containing the firefly luciferase coding sequence
fused upstream from the indicated tPA 3′ UTRs injected into oocytes. The intact 3′ UTR of cyclin B1 and mutants of the three CPEs therein were used
as controls. Translation of the nonrepressed control mRNA was adjusted to 100%. (d) Polyadenylation analysis of tPA transcripts from untransfected,
shCtrl or shCPEB4 RWP-1 cells. (e) tPA RNA and protein expression after CPEB4 knockdown in RWP-1 cells. HPRT (for RNA) or tubulin (protein) was
used for normalization. Data are mean ± s.e.m. (f) tPA immunohistochemistry in tumors from nude mice after subcutaneous or intraperitoneal injection
of RWP-1 untransfected, shCtrl or shCPEB4 cells. Scale bars, 100 µm. At the bottom is quantification of tPA knockdown in tumors, calculated by the
percentage of tPA staining intensity and compared to shCtrl tumors. The P values (determined by Student’s t test) relative to shCtrl are shown.
***P ≤ 0.001. Data are mean ± s.e.m. One representative experiment (n = 3) is shown (a–d).
npg © 2012 Nature America, Inc. All rights reserved.

articles
88 VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine
in ductal tumors and in PDA cell lines (RWP-1 and SK-PC-1), the
poly(A) tail was elongated (Fig. 6b), suggesting that the translation-
ally silent tPA mRNA in normal pancreas becomes polyadenylated in
pancreatic cancer. To further characterize these changes in poly(A)
tail length, we sought to identify the regulatory cis-acting elements in
the tPA mRNA 3′ UTR microinjecting reporter transcripts of Xenopus
laevis oocytes.
The 3′ UTR of tPA mRNA contains two putative CPEs in proxi-
mity to the hexanucleotide polyadenylation signal (HEX)6,10 and two
potential AU-rich elements (AREs). AREs are well known regula-
tors of mRNA deadenylation and stability7 (Fig. 6c). To determine
whether these elements were functional, we fused the tPA 3′ UTR
or variants in which these cis-acting elements had been inactivated
downstream of the firefly luciferase open reading frame, and we
then injected them into stage IV oocytes together with Renilla firefly
mRNA. As a control, we used the 3′ UTR of cyclin B1 and a variant in
which the three CPEs were mutated6. Stimulation with progesterone,
to activate CPEBs, resulted in a marked increase in luciferase activ-
ity when the full-length sequence of the 3′ UTR of tPA mRNA was
present, but this stimulation was largely prevented when the CPE or
HEX sequences were deleted or mutated. In contrast, we observed
an increase in luciferase activity when the AREs were deleted from
chimeric mRNA, which is consistent with their role in deadenylation
(Fig. 6c). These results show that the 3′ UTR of human tPA contains
functional CPEs and AREs, which may account for its translational
control and the changes in poly(A) tail length observed in cell cul-
tures and tissues.
To further test whether CPEB4 regulates tPA synthesis, we per-
formed a poly(A) test assay of tPA mRNA from RWP-1 cells trans-
fected with shCPEB4. Downregulation of CPEB4 resulted in tPA
mRNA with a shorter poly(A) tail (Fig. 6d) and, accordingly, a
decrease of tPA protein expression without affecting tPA mRNA levels
(Fig. 6e). We confirmed these results in vivo, where subcutaneous and
intraperitoneal xenografts from shCPEB4 cells also had reduced tPA
expression (Fig. 6f). Together, these data show that tPA expression is
differentially regulated by CPEB4-mediated translational control in
tumoral cells compared to normal pancreas, leading to the accumula-
tion of tPA that is associated with PDA progression.
To determine the relative contribution of tPA translational regula-
tion to the tumoral phenotype observed after CPEB4 depletion, we
performed a rescue experiment using ectopic expression of tPA in
xenografted RWP-1 cells depleted of CPEB4, which resulted in an
increased number of tumors (Supplementary Fig. 8). This observa-
tion suggests that, although CPEB4 binds to hundreds of mRNAs
(Supplementary Table 2), tPA mRNA translation is a key con-
tributor to the decreased in vivo tumorigenicity that occurs after
CPEB4 downregulation.
CPEB4 promotes tumor growth and vascularization in gliomas
The crucial biological role of CPEB4 in PDA raised the question
of whether it also has a similar role in the development of other
tumor types. To explore this possibility, we analyzed the expres-
sion of CPEB4 in human glioblastomas. CPEB4 was absent from
normal astrocytes but was ver y abundant in high-grade gliomas
(Supplementary Fig. 9a). In addition, CPEB4 downregulation in cells
from the T98G line resulted in reduced tumor size (Supplementar y
Figs. 9b and 10), cellular proliferation (Supplementary Fig. 9c,d;
measurement of Ki67) and microvessel density (Supplementary
Fig. 9c,d; measurement of vWF). Thus, the contribution of CPEB4 to
tumor malignancy seems to extend to other tumor types, suggesting
that translational control by CPEB4 may be a general mechanism to
control gene expression during carcinogenesis.
DISCUSSION
PDA is a complex genetic disease resulting from the unbalanced
expression of many tumor-promoting or antitumoral factors imping-
ing on essential signal transduction pathways, for example, the K-Ras,
Hedgehog, TGF-β and Wnt-Notch pathways17. Although PDA develop-
ment has been associated with a number of genetic and epigenetic
alterations, it is now clear that this is only the first layer of genetic
reprogramming associated with tumor progression and that post-
transcriptional regulation may also participate in the cellular changes
that result in malignant transformation2–4,30–37. The altered expression
and activity of general translation factors has previously been observed
in tumors4,5. However, a direct functional link between the differen-
tial expression of mRNA-specific translational regulators and tumor
development has not been established. Here we show that CPEB4 is
specifically upregulated in high-grade PanIN lesions and in well or
moderately differentiated PDA. In tumoral cells, CPEB4 associates with
and translationally activates hundreds of mRNAs, engaging an addi-
tional level of gene regulation (Supplementar y Table 2). To test this
concept, we studied the expression of tPA, one of the transcripts that
is enriched in the CPEB4 immunoprecipitate. tPA is a member of the
plasminogen system, a family of proteins that are essential components
of tumor progression in several types of tumors, including PDA38–43.
tPA protein is absent in normal pancreas and is overexpressed in 85%
of PDA, and tPA protein has a role in tumor proliferation, migration,
invasion and angiogenesis21–27. However, tPA mRNA is similarly
expressed in both normal and tumoral pancreatic samples. Therefore,
translational activation of tPA mRNA as a result of the tumor-specific
overexpression of CPEB4 is consistent with its expression pattern and
its function in tumors. Indeed, we found that tPA mRNA is translation-
ally activated in tumors through the recruitment of CPEB4 to CPEs in
its 3′ UTR. tPA mRNA is also translationally regulated by CPEB1 in
mouse oocytes28,44,45 and neurons29. However, in tumors, pathological
activation of tPA mRNA is mediated by CPEB4.
In addition to tPA transcript, CPEB4-associated mRNAs are
significantly (P ≤ 0.05) enriched in a number of cellular functions
that are relevant to tumorigenesis, including RAS-related mole-
cules, cell signaling components (Smad family member 3 (Smad3),
phosphoinositide-3-kinase (PI3K), calcium/calmodulin-dependent
protein kinase II (CamKII) and G-protein–coupled receptor), chro-
matin remodeling proteins (histone deacetylases and MYST histone
acetyltransferase), cyclins, apoptosis-related molecules (caspase 8,
apoptosis-related cysteine peptidase (CASP8) and B-cell CLL/lymphoma 2
(BCL2) binding component 3), stress and inflammation factors
(interleukin-32, hypoxia inducible domain family, member 1A (HIG1),
interferon receptor 2 and heat shock 70-kDa protein 4 (Hsp70)), meta-
bolic enzymes (isocitrate dehydrogenase 1 (IDH1)) and genes related
to cell migration and metastasis (encoding matrix metallopeptidase 7
(MMP-7), tPA, β-catenin and Twist) (Supplementary Table 2).
However, the most significant enrichment was for mRNAs encoding
translation factors (ribosomal proteins, eukaryotic translation initiation
factor 3, subunit E (eIF3e) and eIF3h), a cellular function that has been
previously associated with malignant transformation4,46–48. mRNA
from a number of pro-fibrogenic factors (MMP-7, SMAD3 and serpin
peptidase inhibitor, clade B (ovalbumin), member 1 (SerpinB1)) was
also associated with CPEB4. CPEB4-associated transcripts are enriched
in CPEs in proximity to the HEX, which is the hallmark of CPE-
mediated cytoplasmic polyadenylation and translational activation6.
npg © 2012 Nature America, Inc. All rights reserved.

articles
nature medicine VOLUME 18 | NUMBER 1 | JANUARY 2012 89
CPEB4-mediated reprogramming of gene expression in PDA has
a marked effect on at least three of the processes considered to be
hallmarks of cancer1. First, CPEB4 correlates with increased tumoral
growth, and its downregulation results in a substantial reduction
of cellular proliferation in the tumoral context but not in a cell-
autonomous manner. Second, CPEB4 knockdown results in a reduc-
tion of tumor angiogenesis, suggesting that this effect may contribute
to the changes in cell proliferation observed in vivo. Even though
reduced vascularization could explain the reduced tumoral size and
provide a clear link with the translational control of tPA, we cannot
rule out additional functions for CPEB4 in a tumoral-niche–specific
manner, for example, by regulating cell-cell attachment, tumor-stromal
interactions or other events. Third, cells overexpressing CPEB4 show
an advantage for tissue colonization and invasion, although further
studies will be required to better understand the role of CPEB4 in
this process. This work shows that overexpression of CPEB4 in PDA
activates the translation of a wide range of transcripts, subverting its
normal function in meiosis, and confers relevant advantages to cancer
cells for tumor growth and progression. Notably, ectopic expression
of genes normally related to germline traits was recently found to be
associated with tumor growth in Drosophila49.
CPEB4 overexpression and its contribution to tumor growth
and angiogenesis are not restricted to PDA: they also occur in glio-
blastomas, suggesting that CPEB4-mediated regulation of gene
expression might be a more general mechanism in cancer. Indeed,
gene profiling in other types of tumors has shown changes in mRNAs
encoding different members of the CPEB family of proteins50–55. In
addition, CPEB4 protein is overexpressed in a large variety of tumors
(17 out of a total of 20 tumor types listed at http://www.proteinatlas.
org/ENSG00000113742), suggesting that this overexpression might
be a general mechanism in tumoral development and a possible new
therapeutic target.
METHODS
Methods and any associated references are available in the online
version of the paper at http://www.nature.com/naturemedicine/.
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTS
The authors acknowledge F. Gebauer, S. Aznar-Benitah, A. García de
Herreros and J. Valcárcel for critical comments on the manuscript and for other
valuable contributions. We also thank S. Hahn (Department of Molecular
GI-Oncology, University of Bochum, Germany) and M. Buchholz (Department
of Gastroenterology, Endocrinolog y and Metabolism, Philipps-University of
Marburg, Marburg, Germany) for providing normal pancreas RNA,
O. Casanovas (Translational Research Laboratory, Catalan Institute of Oncology
(ICO)–Bellvitge Biomedical Research Institute (IDIBELL), L’Hospitalet
de Llobregat, Spain) for his help with the T98G nude mice injections,
J.R. González-Vallinas (Computational Genomics Group, Universitat Pompeu
Fabra, Parc de Recerca Biomèdica de Barcelona (PRBB)) for assistance with
Gene Ontology tools and E. Castillo from the Ultrasequencing Unit (Centre for
Genomic Regulation, PRBB), S. Mojal from the Statistics Unit (IMIM, PRBB)
and T. Lobato from the Histopathological Unit (IMIM, PRBB) for technical
assistance. This work was funded by the research grants Instituto de Salud
Carlos–Fondos Europeos de Desarrollo Regional (FEDER) (PI080421) from the
Ministerio de Ciencia e Innovación (MICINN) and grants from Fundació La
MaratóTV3 (051110), the American Institute for Cancer Research (AICR)
(11-0086) and Generalitat de Catalunya (2009SGR1409) to P.N.; grants
BFU2008-02373 and Consolider RNAREG CSD2009-00080 from the MICINN
and grants from Fundació La MaratóTV3 (051110), AICR (11-0086) and
Generalitat de Catalunya (2009SGR1436) to R.M.; grant SAF2007-60860
and Consolider ONCOBIO from the MICINN and a grant from the VI EU
Framework Programme MolDiag-PaCa project to F.X.R.; grants Consolider
RNAREG CSD2009-00080 and BIO2008-01091 from the MICINN to E.E.;
and grants from the Instituto de Salud Carlos III FEDER (RD09/0076/00036)
and the Xarxa de Bancs de tumors sponsored by the Pla Director d’Oncologia de
Catalunya to the MARBiobanc. P.N. is supported by the Instituto de Salud Carlos
III and the Departament de Sanitat de la Generalitat de Catalunya. D.P. holds
a Juan de la Cierva grant from the MICINN. N.M.-B. holds a grant from the
Fundación Ramón Areces. C.E. was supported by a fellowship from the DURSI
(Generalitat de Catalunya) and Fons Social Europeu (ESF).
AUTHOR CONTRIBUTIONS
E.O.-Z. performed the experiments shown in Figures 1–6 and in Supplementary
Figures 1, 3, 4 and 6–10, prepared the figures for the manuscript (except Fig. 5b)
and contributed to the experimental design of the study and the preparation of
the manuscript. D.P. generated the majority of the plasmids used in the study,
designed the primers for the RIP validation experiments, helped with the frog
surgery, injection of the oocytes, reporter analyses and in vivo experiments and
prepared Figures 5b, 6d and 6f and Supplementary Figure 2. N.M.-B. established
the RWP-1 LUC cell line, helped with the in vivo experiments, performed
immunohistochemistry analyses of the Capan-1 xenografts and performed all the
statistical analyses. D.P. and N.M.-B. prepared Figure 3 and Supplementary
Figure 5 and made useful contributions to the experimental design and the
interpretation of the data. G.F.-M. generated Capan-1 shCtrl and shCPEB4 cells
and contributed to the interpretation of the data. M.I. supplied all the human
pancreatic samples used in the study and helped with the histopathological
analyses of all the immunohistochemistries performed in samples from both
mice and humans. F.A. supplied human glioblastoma samples and helped with
the histopathological and immunohistochemistries analyses in tumors from both
mice and humans. M.M. helped with the in vivo experiments. C.E. generated the
data shown at the bottom of Figure 6b. E.E. performed the analysis of the data
generated in the Illumina Solexa sequencing, performed the motif analysis of the
3′ UTR sequences shown in Figure 5a, generated the data shown in
Supplementary Figures 6 and 7 and contributed to the preparation of the
manuscript. F.X.R. contributed to the study design, data analysis and manuscript
preparation. P.N. and R.M. directed the study and wrote the manuscript, which all
authors approved.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/naturemedicine/.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. Hanahan, D. & Weinberg, R.A. The hallmarks of cancer. Cell 100, 57–70
(2000).
2. Mayr, C., Hemann, M.T. & Bartel, D.P. Disrupting the pairing between let-7 and
Hmga2 enhances oncogenic transformation. Science 315, 1576–1579 (2007).
3. Lee, Y.S. & Dutta, A. The tumor suppressor microRNA let-7 represses the HMGA2
oncogene. Genes Dev. 21, 1025–1030 (2007).
4. Sonenberg, N. & Hinnebusch, A.G. Regulation of translation initiation in eukaryotes:
mechanisms and biological targets. Cell 136, 731–745 (2009).
5. Silvera, D., Formenti, S.C. & Schneider, R.J. Translational control in cancer.
Nat. Rev. Cancer 10, 254–266 (2010).
6. Piqué, M., Lopez, J.M., Foissac, S., Guigo, R. & Mendez, R. A combinatorial code
for CPE-mediated translational control. Cell 132, 434–448 (2008).
7. Belloc, E. & Mendez, R. A deadenylation negative feedback mechanism governs
meiotic metaphase arrest. Nature 452, 1017–1021 (2008).
8. Eliscovich, C., Peset, I., Vernos, I. & Mendez, R. Spindle-localized CPE-mediated
translation controls meiotic chromosome segregation. Nat. Cell Biol. 10, 858–865
(2008).
9. Novoa, I., Gallego, J., Ferreira, P.G. & Mendez, R. Mitotic cell-cycle progression is
regulated by CPEB1- and CPEB4-dependent translational control. Nat. Cell Biol.
12, 447–456 (2010).
10. Mendez, R. & Richter, J.D. Translational control by CPEB: a means to the end.
Nat. Rev. Mol. Cell Biol. 2, 521–529 (2001).
11. Groppo, R. & Richter, J.D. Translational control from head to tail. Curr. Opin. Cell
Biol. 21, 444–451 (2009).
12. Igea, A. & Mendez, R. Meiosis requires a translational positive loop where CPEB1
ensues its replacement by CPEB4. EMBO J. 29, 2182–2193 (2010).
13. Huang, Y.S., Kan, M.C., Lin, C.L. & Richter, J.D. CPEB3 and CPEB4 in neurons:
analysis of RNA-binding specificity and translational control of AMPA receptor GluR2
mRNA. EMBO J. 25, 4865–4876 (2006).
14. Groisman, I. et al. Control of cellular senescence by CPEB. Genes Dev. 20,
2701–2712 (2006).
15. Richter, J.D. CPEB: a life in translation. Trends Biochem. Sci. 32, 279–285
(2007).
16. Burns, D.M. & Richter, J.D. CPEB regulation of human cellular senescence, energy
metabolism, and p53 mRNA translation. Genes Dev. 22, 3449–3460 (2008).
npg © 2012 Nature America, Inc. All rights reserved.

articles
90 VOLUME 18 | NUMBER 1 | JANUARY 2012 nature medicine
17. Jones, S. et al. Core signaling pathways in human pancreatic cancers revealed by
global genomic analyses. Science 321, 1801–1806 (2008).
18. Parsons, D.W. et al. An integrated genomic analysis of human glioblastoma
multiforme. Science 321, 1807–1812 (2008).
19. Iacobuzio-Donahue, C.A. et al. DPC4 gene status of the primary carcinoma correlates
with patterns of failure in patients with pancreatic cancer. J. Clin. Oncol. 27,
1806–1813 (2009).
20. van Riggelen, J., Yetil, A. & Felsher, D.W. MYC as a regulator of ribosome biogenesis
and protein synthesis. Nat. Rev. Cancer 10, 301–309 (2010).
21. Paciucci, R. et al. Isolation of tissue-type plasminogen activator, cathepsin H, and
non- specific cross-reacting antigen from SK-PC-1 pancreas cancer cells using
subtractive hybridization. FEBS Lett. 385, 72–76 (1996).
22. Paciucci, R., Tora, M., Diaz, V.M. & Real, F.X. The plasminogen activator system
in pancreas cancer: role of t-PA in the invasive potential in vitro. Oncogene 16,
625–633 (1998).
23. Díaz, V.M., Planaguma, J., Thomson, T.M., Reventos, J. & Paciucci, R. Tissue
plasminogen activator is required for the growth, invasion, and angiogenesis of
pancreatic tumor cells. Gastroenterology 122, 806–819 (2002).
24. Ryu, B. et al. Relationships and differentially expressed genes among pancreatic
cancers examined by large-scale serial analysis of gene expression. Cancer Res.
62, 819–826 (2002).
25. Iacobuzio-Donahue, C.A. et al. Highly expressed genes in pancreatic ductal
adenocarcinomas: a comprehensive characterization and comparison of the
transcription profiles obtained from three major technologies. Cancer Res. 63,
8614–8622 (2003).
26. Aguilar, S. et al. Tissue plasminogen activator in murine exocrine pancreas cancer:
selective expression in ductal tumors and contribution to cancer progression.
Am. J. Pathol. 165, 1129–1139 (2004).
27. Missiaglia, E. et al. Analysis of gene expression in cancer cell lines identifies
candidate markers for pancreatic tumorigenesis and metastasis. Int. J. Cancer 112,
100–112 (2004).
28. Huarte, J., Belin, D., Vassalli, A., Strickland, S. & Vassalli, J.D. Meiotic maturation
of mouse oocytes triggers the translation and polyadenylation of dormant tissue-type
plasminogen activator mRNA. Genes Dev. 1, 1201–1211 (1987).
29. Shin, C.Y., Kundel, M. & Wells, D.G. Rapid, activity-induced increase in tissue
plasminogen activator is mediated by metabotropic glutamate receptor-dependent
mRNA translation. J. Neurosci. 24, 9425–9433 (2004).
30. Gardina, P.J. et al. Alternative splicing and differential gene expression in colon
cancer detected by a whole genome exon array. BMC Genomics 7, 325
(2006).
31. Karni, R. et al. The gene encoding the splicing factor SF2/ASF is a proto-oncogene.
Nat. Struct. Mol. Biol. 14, 185–193 (2007).
32. Srebrow, A. & Kornblihtt, A.R. The connection between splicing and cancer. J. Cell
Sci. 119, 2635–2641 (2006).
33. Karni, R., Hippo, Y., Lowe, S.W. & Krainer, A.R. The splicing-factor oncoprotein
SF2/ASF activates mTORC1. Proc. Natl. Acad. Sci. USA 105, 15323–15327
(2008).
34. Cooper, C. et al. Increasing the relative expression of endogenous non-coding steroid
receptor RNA activator (SRA) in human breast cancer cells using modified
oligonucleotides. Nucleic Acids Res. 37, 4518–4531 (2009).
35. Pio, R. & Montuenga, L.M. Alternative splicing in lung cancer. J. Thorac. Oncol.
4, 674–678 (2009).
36. Mayr, C. & Bartel, D.P. Widespread shortening of 3′ UTRs by alternative cleavage
and polyadenylation activates oncogenes in cancer cells. Cell 138, 673–684
(2009).
37. Ji, Z. & Tian, B. Reprogramming of 3′ untranslated regions of mRNAs by alternative
polyadenylation in generation of pluripotent stem cells from different cell types.
PLoS ONE 4, e8419 (2009).
38. Blasi, F. & Sidenius, N. The urokinase receptor: focused cell surface proteolysis,
cell adhesion and signaling. FEBS Lett. 584, 1923–1930 (2010).
39. Ossowski, L. Plasminogen activator dependent pathways in the dissemination of
human tumor cells in the chick embryo. Cell 52, 321–328 (1988).
40. Aguirre-Ghiso, J.A., Liu, D., Mignatti, A., Kovalski, K. & Ossowski, L. Urokinase
receptor and fibronectin regulate the ERK(MAPK) to p38(MAPK) activity ratios that
determine carcinoma cell proliferation or dormancy in vivo. Mol. Biol. Cell 12,
863–879 (2001).
41. Smith, H.W. & Marshall, C.J. Regulation of cell signalling by uPAR. Nat. Rev. Mol.
Cell Biol. 11, 23–36 (2010).
42. Ulisse, S., Baldini, E., Sorrenti, S. & D’Armiento, M. The urokinase plasminogen
activator system: a target for anti-cancer therapy. Curr. Cancer Drug Targets 9,
32–71 (2009).
43. McMahon, B. & Kwaan, H.C. The plasminogen activator system and cancer.
Pathophysiol. Haemost. Thromb. 36, 184–194 (2008).
44. Huarte, J., Belin, D. & Vassalli, J.D. Plasminogen activator in mouse and rat oocytes:
induction during meiotic maturation. Cell 43, 551–558 (1985).
45. Huarte, J. et al. Transient translational silencing by reversible mRNA deadenylation.
Cell 69, 1021–1030 (1992).
46. Gandin, V. et al. Eukaryotic initiation factor 6 is rate-limiting in translation, growth
and transformation. Nature 455, 684–688 (2008).
47. Ruggero, D. et al. The translation factor eIF-4E promotes tumor formation and
cooperates with c-Myc in lymphomagenesis. Nat. Med. 10, 484–486 (2004).
48. Livingstone, M., Atas, E., Meller, A. & Sonenberg, N. Mechanisms governing the
control of mRNA translation. Phys. Biol. 7, 021001 (2010).
49. Janic, A., Mendizabal, L., Llamazares, S., Rossell, D. & Gonzalez, C. Ectopic
expression of germline genes drives malignant brain tumor growth in Drosophila.
Science 330, 1824–1827 (2010).
50. Gaspar, C. et al. Cross-species comparison of human and mouse intestinal polyps
reveals conserved mechanisms in adenomatous polyposis coli (APC)-driven
tumorigenesis. Am. J. Pathol. 172, 1363–1380 (2008).
51. Sääf, A.M. et al. Parallels between global transcriptional programs of polarizing
Caco-2 intestinal epithelial cells in vitro and gene expression programs in normal
colon and colon cancer. Mol. Biol. Cell 18, 4245–4260 (2007).
52. Read, T.A. et al. Identification of CD15 as a marker for tumor-propagating cells in
a mouse model of medulloblastoma. Cancer Cell 15, 135–147 (2009).
53. Chen, H.Y. et al. A five-gene signature and clinical outcome in non-small-cell lung
cancer. N. Engl. J. Med. 356, 11–20 (2007).
54. Israel, A., Sharan, R., Ruppin, E. & Galun, E. Increased microRNA activity in human
cancers. PLoS ONE 4, e6045 (2009).
55. Hansen, C.N. et al. Expression of CPEB, GAPDH and U6snRNA in cervical and
ovarian tissue during cancer development. APMIS 117, 53–59 (2009).
npg © 2012 Nature America, Inc. All rights reserved.

nature medicine
doi:10.1038/nm.2540
ONLINE METHODS
Cell culture, CPEB4 knockdown and tPA overexpression. Human pancreatic
cells have been previously described56 . T98G cells, derived from a human
glioblastoma tumor, were purchased from America Type Culture Collection.
All cells were cultured in DMEM supplemented with 10% FBS. All experi-
ments using human cell lines were approved by the Ethical Committee for
Clinical Research of IMIM–Hospital del Mar. shRNA against CPEB4 and the
corresponding scramble control (shCtrl) were obtained from Sigma (siMis-
sion) and transfected into RWP-1, RWP-1 LUC, Capan-1 or T98G cells with
Lipofectamine and Plus reagent (Invitrogen). Selection was performed for
3 d with puromycin (3 µg ml−1 for RWP-1 and RWP-1 LUC cells, 2 µg ml−1
for Capan-1 cells and 1 µg ml−1 for T98G cells). CPEB4 concentrations were
analyzed by western blot and qRT-PCR. For the in vivo rescue experiment,
RWP-1 shCtrl and RWP-1 shCPEB4_2 cells were transfected with an empty
vector, pcDNA3 or a pcDNA3 vector expressing the coding sequence of human
tPA26. Selection was performed for 5 d with neomycin (700 µg ml−1 ).
Generation of RWP-1 LUC cells. For the generation of RWP-1 LUC cells, the
pLHCX plasmid (a gift from C. Fillat, Center for Genomic Regulation, Barcelona,
Spain) was used to produce lentiviral particles in the Phoenix Ampho HEK293T
cell line. After transduction, stable clones expressing luciferase were selected
with hygromycin (0.3 mg ml−1); luciferase activity was measured using the Dual-
Luciferase Reporter Assay (Promega). A clone with a medium rate of luciferase
activity was selected.
MTT cell proliferation assay. Cell growth was determined by a MTT colori-
metric growth assay. Briefly, 1 × 103 cells per well were plated in 96-well plates
in complete medium and cultured to confluence. Each day, cell growth was
determined by adding MTT solution (1 mg ml−1) in DMEM without FBS for 4 h
at 37 °C. The dye was then extracted with a solution of 0.1 N HCl and 10% Triton
X-100 in isopropanol, and absorbance was measured with a 570-nm filter.
Cell invasion assays. For the invasion assays, Transwells (Costar) coated with
Matrigel (BD Bioscience) were used as previously described56. Briefly, 30,000
RWP-1 cells were seeded in DMEM without FBS and with 1% BSA and were
allowed to invade for 72 h. Cells that passed to the lower compartment of the
filter were fixed, and absorbance (at 590 nm) was quantified after crystal violet
staining for 5 min at 25 °C and 10% acetic acid extraction. The absorbance
measurements listed in Figure 2 are equivalent to the following number of
cells: 15,000 (untransfected), 16,600 (shCtrl), 7,230 (shCPEB4_2) and 8,840
(shCPEB4_4).
Soft-agar growth assay. Anchorage-independent growth was evaluated as
described57. Briefly, 1 × 104 cells were plated in complete DMEM containing
0.3% soft agar in 6-cm plates over a layer of solidified DMEM containing 0.7%
soft agar. Medium was added twice a week to maintain humidity. After 5 weeks,
colonies were stained with MTT (0.5 mg ml−1) for 3 h at 37 °C and counted.
In vivo tumorigenicity assay. RWP-1 (1 × 106), Capan-1 (1 × 106) or T98G
cells (6 × 106) were injected subcutaneously into the two posterior flanks of
male BALB/c nude mice (Charles River). Mice were followed weekly, and
when tumors in any of the groups (mice injected with untransfected, shCtrl,
shCPEB4_2 or shCPEB4_4 cells) reached around 1 cm3, determined by palpa-
tion (at 3 weeks), all mice were killed, and the tumors were weighed, measured
and processed. For intraperitoneal injections, 1 × 106 RWP-1 LUC cells were
injected in the peritoneal cavity, and tumor growth was assessed weekly by bio-
luminescence until mice were killed (when the bioluminescence signal indicated
a tumor size of ≥1.5 cm3 or when mice presented with symptoms of suffering
or health deterioration). For bioluminescence detection and quantification,
mice were anesthetized, and the substrate D-firefly-Luciferin (Xenogen) was
administered intraperitoneally (32 mg per kg of body weight). Luciferase activity
was visualized and quantified using an in vivo bioluminescent system (IVIS 50
Xenogen) and a Living Image 2.20.1 software overlay on Igor Pro4.06A software
(WaveMetrics) as previously described58. Luciferase activity was quantified from
non-saturated images by measuring the total amount of emitted light recorded
by the charge-coupled device camera. A total of ten (for RWP-1 cells and two
flanks of subcutaneous injection), eight (for Capan-1 cells and two flanks of sub-
cutaneous injection), four (for T98G cells and two flanks of subcutaneous injec-
tion) and ten (for RWP-1 LUC cells and one intraperitoneal injection) mice were
used for each condition (untransfected, shCtrl, shCEPB4_2 and shCPEB4_4).
For the analysis of tumor incidence in the pancreas and other organs (invasion)
and of survival, mice were clustered into two groups: a control group (mice
injected with untransfected and shCtrl RWP-1 LUC cells) and a shCPEB4 group
(mice injected with shCPEB4_2 or shCPEB4_4 RWP-1 LUC cells). All animal
procedures met the guidelines of European Community Directive and were
approved by the PRBB ethical committee.
Statistical analyses. All results were evaluated using the SPSS statistical soft-
ware package. Different statistical tests were used according to the type of data
analyzed (Student’s t test,
χ
2, Fisher’s or log-rank tests), as is indicated in figure
legends. P ≤ 0.05 was considered statistically significant.
Additional methods. Descriptions of the tissue samples, immunohistochemistry
and histopathological analyses, plasmid construction, oligonucleotides, immuno-
fluorescence microscopy, RNA extraction, qRT-PCR, translation of luciferase
reporters in X. laevis oocytes, poly(A) tail testing, western blot, zymography,
RNA immunoprecipitation, Illumina Solexa sequencing (RIP-Seq) and ultra-
sequencing analyses can be found in the Supplementary Methods.
56. Roda, O. et al. Galectin-1 is a novel functional receptor for tissue plasminogen
activator in pancreatic cancer. Gastroenterology 136, 1379–1390 (2009).
57. Rizzino, A. Behavior of transforming growth factors in serum-free media:
an improved assay for transforming growth factors. In Vitro 20, 815–822 (1984).
58. Huch, M. et al. Urokinase-type plasminogen activator receptor transcriptionally
controlled adenoviruses eradicate pancreatic tumors and liver metastasis in mouse
models. Neoplasia 11, 518–528 (2009).
npg © 2012 Nature America, Inc. All rights reserved.