Erk associates with and primes GSK-3β for its inactivation resulting in
upregulation of β-catenin
Qingqing Ding, Weiya Xia, Jaw-Ching Liu, Jer-Yen Yang, Jiahong Xia, Dung-Fang Lee, Geoffrey
Bartholomeusz, Yan Li, Yong Pan, Zheng Li, Jun Qin and Mien-Chie Hung
1Department of Molecular and Cellular Oncology, University of Texas M.D. Anderson Cancer Center,
Houston, Texas 77030
2Graduate School of Biomedical Sciences, The University of Texas, Houston, Texas 77030
3Department of Surgery, Union Hospital, Tongji Medical College, Huazhong University of Science and
Technology, Wuhan 430022, China
4Department of Hematology, Oncology and Tumor Immunology, Robert Rossle Cancer Clinic and
HELIOS Clinics, Max Delbruck Center for Molecular Medicine, Humboldt University, Charite,
Campus Buch, Berlin 13125, Germany
5Department of Biochemistry, Baylor College of Medicine, Houston, Texas 77030
6China Medical University Hospital, Taichung 404, Taiwan
ß-catenin is upregulated in many human cancers, and considered to be an oncogene.
Hepatocellular carcinoma (HCC) is one of the most prevalent human malignancies,
and individuals who are chronic hepatitis B Virus (HBV) carriers have a greater than
100-fold increased relative risk of developing HCC. Here we report a novel
mechanism by which HBV-X protein (HBX) upregulates ß-catenin. Erk, which is
activated by HBX, associates with GSK-3ß through a docking motif of GSK-3ß and
serves as a scaffold to make a bridge between GSK-3ß and p90RSK, moreover Erk
phosphorylates GSK-3ß directly, which primes GSK-3ß for its subsequent
phosphorylation at Ser 9 by p90RSK, resulting in inactivation of GSK-3ß and
upregulation of ß-catenin to facilitate cell proliferation. The novel pathway is a
general signal as it was also observed in cell lines, in which Erk-primed inactivation
of GSK-3ß was regulated by receptor tyrosine kinase HER2, growth factor IGF-1 and
TGF-ß, and further supported by immunohistochemical staining in different human
tumors including cancers of liver, breast, kidney and stomach. In conclusion, the novel
mechanism, namely Erk-primed inactivation of GSK-3ß directly links HBX or growth
factor/receptor and ß-catenin signal pathway together, and may be involved in the
development of multiple human cancers.
HBV infection is one of the major causes of HCC in humans. Epidemiological studies
showed that 80% of all HCC occurs in HBV-infected individuals. HBV genome is
composed of a partially double-stranded circular DNA and contains four overlapping
genes: S/preS, C/preC, P, and X. The X gene is the most frequently integrated viral
sequence found in HCC, which encodes a 17 kDa protein termed HBX (Arbuthnot
and Kew, 2001 P. Arbuthnot and M. Kew, Hepatitis B virus and hepatocellular
carcinoma, Int. J. Exp. Pathol. 82 (2001), pp. 77–100. Full Text via CrossRef | View
Record in Scopus | Cited By in Scopus (105)Arbuthnot and Kew, 2001 and Cromlish,
1996). HBX is a multifunction protein that can inhibit p53 function and transactivate
some transcription factors, including AP-1, NF-κB, CREB, and TBP (Benn et al.,
1996, Feitelson et al., 1993, Maguire et al., 1991, Qadri et al., 1995 and Su and
Schneider, 1996). Moreover, HBX has been implicated in the activation of many
signal transduction pathways, such as the RAS/RAF/MAPK, MEKK1/JNK,
JAK/STAT, and PI-3K/AKT pathways (Benn and Schneider, 1994, Benn et al., 1996,
Lee and Yun, 1998 and Shih et al., 2000). Malignant transformation has been
observed in certain cell lines transfected with HBx gene and in an X gene-transgenic
mouse model (Kim et al., 1991 and Terradillos et al., 1997). HBX is therefore
considered to play an important role in neoplastic transformation of hepatocytes in
HBV-infected liver; however, the molecular mechanism of HBV-induced HCC
remains poorly understood.
β-catenin is a key effector of the Wnt signaling pathway. The current model suggests
that Axin and adenomatous polyposis coli (APC) gene products serve as scaffolds to
facilitate the phosphorylation of β-catenin by GSK-3β (Nelson and Nusse, 2004 and
Peifer and Polakis, 2000), and phosphorylated β-catenin is targeted to degradation by
the ubiquitin-proteasome system (Aberle et al., 1997 and Kitagawa et al., 1999).
Mutations within the binding site of either Axin or APC, or the GSK-3β
phosphorylation site of β-catenin, result in the inability of GSK-3β to phosphorylate
β-catenin (Morin et al., 1997 and Satoh et al., 2000). Although the molecular
mechanism by which Wnt actually regulates GSK-3β is not fully elucidated yet, one
finding is that Wnt-activated Dsh causes GSK-3β dissociation from Axin (Li et al.,
1999 and Ruel et al., 1999); the other is that Wnt stimulation leads to reduction of
kinase activity of GSK-3β (Cook et al., 1996 and Ruel et al., 1999). Thus, the two
mechanisms prevent efficient β-catenin phosphorylation by GSK-3β, resulting in
stabilization of β-catenin. In addition to the Wnt-dependent stabilization of β-catenin,
it has been reported that many growth factors can stabilize β-catenin through
phosphorylation of GSK-3β to inactivate GSK-3β (Desbois-Mouthon et al., 2001,
Holnthoner et al., 2002 and Playford et al., 2000). Once β-catenin is accumulated in
the cytoplasm, it can translocate to the nucleus where it binds to Tcf/Lef and acts as its
coactivator to stimulate the transcription of target genes such as c-myc and cyclin D1
(He et al., 1998 and Tetsu and McCormick, 1999) to facilitate cell proliferation.
However, the molecular mechanism of growth factors-induced β-catenin stabilization
is not completely understood yet.
About 50%–70% of all HCC examined showed an abnormal β-catenin protein
accumulation in the cytoplasm and nucleus (de La Coste et al., 1998, Suzuki et al.,
2002 and Wong et al., 2001). However, in HCC the β-catenin mutation rate is
13%–26% (de La Coste et al., 1998 and Wong et al., 2001), the Axin mutation rate is
5%–10% (Satoh et al., 2000 and Taniguchi et al., 2002), and no mutation of APC has
been reported so far. This led us to speculate that other factors might be involved in
the accumulation of β-catenin in HCC. In this study we identify a mechanism to
stabilize β-catenin, namely Erk-primed inactivation of GSK-3β, which provides a
mechanic link for HBX- and growth factor-induced β-catenin stabilization.
HBX Activates the β-Catenin Pathway
To examine the potential relationship between HBX and β-catenin in HCC, we
analyzed the expression of HBX and β-catenin in 53 HCC samples using
immunohistochemical staining (IHC) and found that accumulation of β-catenin in
both the cytosol and nucleus significantly correlated with the expression of HBX
(Figures 1A and 1B). We then tested whether HBX may upregulate β-catenin and its
downstream target. Western blot analysis showed higher protein levels of both
wild-type β-catenin (de La Coste et al., 1998) and its downstream targets c-myc and
cyclin D1 in HBX-stable transfectants, HepG2-X and Hep3B-X, as compared to those
of parental cell lines, HepG2 and Hep3B (Figure 1C). In addition, immunofluorescent
staining indicated that there was an increase in β-catenin expression and translocation
of β-catenin to the nucleus in the HBX-stable transfectant cell lines (Figure 1D); the
rates of nuclear β-catenin are 91% versus 53% in HepG2-X and HepG2 cells, and
42% versus 6% in Hep3B-X and Hep3B cells, respectively. Consistently, the
transcriptional activity of β-catenin was also increased in HepG2-X and Hep3B-X
cells compared to their parental cells when we used a Tcf-regulated reporter to
evaluate transcriptional activity of β-catenin (Figure 1E). Moreover, flow cytometry
confirmed the more rapid progression in HBX-expressing cells whereby the
percentages of cells in S phase and G2/M phase were greater than parental cells.
However, both dominant-negative Tcf-4 (DN-Tcf-4) and siRNA-β-catenin blocked
HBX-activated transcriptional activity of β-catenin (Figure 1E) and inhibited
HBX-induced cell proliferation as indicated by significantly decreased percentage of
cells in S phase and G2/M phase (Figure 1F). These results demonstrated that HBX
can upregulate β-catenin in both hepatoma cell lines and tumor tissues, and the
HBX-activated β-catenin activity is required for the HBX-induced cell proliferation.
Figure 1.HBX Activates the β-Catenin Pathway
(A) Primary human HCC specimens were stained with antibodies specific to HBX
(AA and AC) and β-catenin (AB and AD).
(B) Summary of the expression of β-catenin and HBX in human HCC specimens.
(C) Western blot analysis showed that wt β-catenin and c-myc and cyclin D1, two
downstream targets of β-catenin, are abundant in the HBX-expressing HepG2 and
Hep3B cells. The arrow marks the position of wt β-catenin, and a known mutant
β-catenin (de La Coste et al., 1998) migrating below the wt β-catenin was also
detected in HepG2 and HepG2-X cells.
(D) An immunofluorescence assay was performed on different cell lines as indicated
to measure the expression and cellular localization of β-catenin (green). The nucleus
was stained with DAPI (red).
(E) A reporter assay was performed to measure the transcriptional activity of β-catenin.
Cells were transfected with Tcf-regulated reporter gene Top or mutant reporter Fop
plus vector control, dominant-negative Tcf4 or siRNA-β-catenin as indicated for 24 hr,
and then the ratio of Top and Fop was used to measure the transcriptional activity of
β-catenin. The error bars represent standard deviation.
(F) DN-Tcf4 and siRNA-β-catenin inhibit HBX-induced cell proliferation. The
proliferative activity of hepatoma cells was measured by flow cytometry analysis.
Cells were transfected with vector control, dominant-negative Tcf4, and
siRNA-β-catenin plus GFP (ratio 5:1), respectively, as indicated for 48 hr, and then
FACS was performed as described in the Experimental Procedures.
HBX Stabilizes β-Catenin through Inactivation of GSK-3β
We next investigated the mechanism in which HBX increases β-catenin expression.
There was no difference in the mRNA level of β-catenin in HepG2, HepG2-X, and
Hep3B, Hep3B-X, as shown by Northern blot assay of total RNA (Figure 2A).
However, proteasome inhibitors Lactacystin and MG132 significantly increased the
β-catenin protein level in the parental cells (without HBX), but not in the HBX
transfectants (Figure 2B), suggesting that HBX may inhibit proteasomal degradation
of β-catenin, resulting in upregulation of β-catenin. It is known that phosphorylation
of β-catenin by GSK-3β causes its degradation by the ubiquitin-proteasome system;
therefore, we examined whether the stabilization of β-catenin in HBX transfectants
may result from inhibition of GSK-3β activity to phosphorylate β-catenin. To this end,
we compared kinase activity of GSK-3β between HBX transfectants and their parental
cells. Although CK1 is known to be required as a priming kinase for GSK-3β
phosphorylation of β-catenin in vivo (Amit et al., 2002 and Liu et al., 2002), in vitro
β-catenin can be directly phosphorylated at S33, S37, and T41 residues by GSK-3β
without priming by CK1 (Liu et al., 2002) (Figure S1). Thus, we used GST-tagged
β-catenin protein as a substrate to measure the kinase activity of GSK-3β. The kinase
assay indicated that the ability of GSK-3β to phosphorylate β-catenin was
significantly lower in HepG2-X and Hep3B-X cells as compared to HepG2 and
Hep3B cells (Figure 2C). Consistent to this notion, we found that phosphorylation of
GSK-3β at the Ser9 residue, which is known to inactivate GSK-3β kinase activity
(Cohen and Frame, 2001), in HepG2-X and Hep3B-X cells was much stronger than
that detected in HepG2 and Hep3B cells (Figure 2D). Furthermore, transfection of
constitutively active GSK-3β (S9AGSK-3β) into HepG2-X and Hep3B-X cells
resulted in a significant suppression of transcriptional activity of β-catenin (Figure 2E)
as compared to wild-type GSK-3β, indicating that constitutively active GSK-3β can
reverse HBX-upregulated β-catenin activity. Taken together, these data suggest that
HBX may upregulate β-catenin through inactivation of GSK-3β.
Figure 2.HBX Stabilizates β-Catenin through Inactivation of GSK-3β
(A) A Northern blot of total RNA showed comparable mRNA levels of β-catenin in
HepG2, HepG2-X and Hep3B, Hep3B-X.
(B) Cells as indicated were treated with proteasome inhibitor Lactacystin (10 μM) or
MG132 (20 μM). Cell lysates were blotted with anti-β-catenin antibody.
(C) Kinase activity of GSK-3β. GSK-3β was immunoprecipitated and incubated with
GST-β-catenin as described in the Experimental Procedures. Reaction products were
analyzed by Western blotting.
(D) Cell lysates as indicated were blotted with anti-p-GSK-3β (Ser9) and
(E) Cells were transfected with Top or Fop plus wild-type or constitutively active
GSK-3β as indicated, and then a Tcf-regulated reporter assay was performed to detect
the transcriptional activity of β-catenin. The error bars represent standard deviation.
The Erk/p90RSK Pathway Is Required for HBX-Mediated Inactivation of
We next investigated which signal pathway is involved in HBX-mediated
phosphorylation of GSK-3β at Ser9. It is known that three kinases, p90RSK, a
downstream kinase of Erk, and AKT and PKC, that can be activated by HBX
(Murakami, 2001) can phosphorylate GSK-3β at the Ser9 residue (Eldar-Finkelman,
2002). To delineate which kinase is responsible for phosphorylation of GSK-3β, we
exploited specific inhibitors to block these kinases: Calphostin C to block PKC,
LY294002 and Wortmannin to block the PI-3K/PKB pathway, and PD98059 to block
the Erk/p90RSK pathway. A reporter assay showed that only PD98059 inhibited
HBX-upregulated β-catenin transcriptional activity in a dose-dependent manner
(Figure 3A), while other inhibitors did not (data not shown). Furthermore, the
dominant-negative Erk also inhibited HBX-upregulated β-catenin transcriptional
activity (Figure 3B). Consistently,
phosphorylation of GSK-3β at Ser9 and the protein level of β-catenin in HepG2-X
and Hep3B-X cells (Figure 3C). Taken together, the results suggest that the
HBX-activated Erk pathway plays a major role in the HBX-induced β-catenin
upregulation. To determine whether p90RSK is required for the phosphorylation of
GSK-3β at Ser9 by the HBX-activated Erk pathway, small interfering RNA
(siRNA-RSK) was used to block p90RSK expression. We found that treatment of
HBX transfectants with siRNA-RSK resulted in a reduced level of β-catenin and
GSK-3β phosphorylation at Ser9 when compared with nonspecific siRNA treatment
(Figure 3D), and siRNA-RSK also suppressed the HBX-upregulated transcriptional
activity of β-catenin (Figure 3E). These results indicated that the Erk/p90RSK
pathway is required for phosphorylation of GSK-3β at the Ser9 residue and
upregulation of β-catenin by HBX.
PD98059 significantly inhibited the
Figure 3.HBX Inactivates GSK-3β through the Erk/p90RSK Pathway
(A) Cells as indicated were transfected with Top or Fop and treated with increasing
dose of MEK1/Erk inhibitor PD98059 (10, 20, and 50 μM), and then a Tcf-regulated
reporter assay was performed to detect the transcriptional activity of β-catenin. The
error bars represent standard deviation.
(B) Cells were transfected with Top or Fop plus DN-Erk1 or DN-Erk2 as indicated,
and then a Tcf-regulated reporter assay was performed to detect the transcriptional
activity of β-catenin. The error bars represent standard deviation.
(C) Cell lysates, treated or untreated with PD98059, were blotted with antibodies as
indicated. The arrow marks the position of wt β-catenin. The p-Erk1 was not activated
in HepG2-X cells, and accordingly, DN-Erk1 did not inhibit β-catenin activity in this
(D) Cells as marked were transfected with siRNA-RSK or nonspecific siRNA as a
negative control, and then cell lysates were blotted with antibodies as indicated.
(E) Cells were transfected with Top or Fop plus siRNA-RSK or nonspecific siRNA as
a negative control, and then a Tcf-regulated reporter assay was performed to detect the
transcriptional activity of β-catenin. The error bars represent standard deviation.
Erk Docks to GSK-3β at DEF Domain and Phosphorylates GSK-3β at the 43T
p90RSK has been known to phosphorylate GSK-3β at Ser9 for quite some time
(Stambolic and Woodgett, 1994); however, there is no evidence indicating direct
association between these two molecules (Frodin and Gammeltoft, 1999). As an Erk
downstream target, p90RSK is also known to interact directly with Erk (Frodin and
Gammeltoft, 1999); however, it is not known whether there is direct interaction
between Erk and GSK-3β. Interestingly, we identified a site (291FKFP) in the C
terminus of GSK-3β that is similar with the consensus Erk-docking motif, FXFP,
known as the DEF domain (Jacobs et al., 1999 and Lee et al., 2004). These motifs are
highly conserved among different species (Figure 4A), suggesting that Erk may
interact with GSK-3β. Taken together, this led us to hypothesize that Erk may serve as
a scaffold to hold GSK-3β and p90RSK in a complex to facilitate GSK-3β
phosphorylation by p90RSK. In support of this notion, a coimmunoprecipitation
experiment demonstrated that Erk was associated with GSK-3β and p90RSK at the
same time in all four cells (Figure 4B). Furthermore, the result of double
immunoprecipitation clearly shows that the three proteins, Erk, GSK-3β, and p90RSK,
are in the same complex (Figure S2). To validate the Erk docking site on GSK-3β, we
mutated the potential docking site 291FKFP on a GSK-3β kinase-dead mutant
(GST-GSK-3β KD) by substituting the phenylalanine residues Phe 291 and 293 with
alanine (FKF to AKA) to produce GST-GSK-3β KD291/3A. A GST-GSK-3β
pull-down assay demonstrated that GST-GSK-3β has an association with Erk and
p90RSK (lane 1, Figure 4C), but disrupting the DEF domain of GSK-3β significantly
reduces the association of GSK-3β with Erk and p90RSK (lane 2, Figure 4C). The
results support the notion that Erk docks to GSK-3β at the DEF domain and bridges
GSK-3β and p90RSK together, which may facilitate GSK-3β phosphorylation by
p90RSK. As Erk interacts with GSK-3β, we then ask whether Erk may phosphorylate
GSK-3β. To this end, we identified five putative Erk-phosphorylation motifs, 43TP,
275TP, 309TP, 324TP, and 330TP in GSK-3β. These motifs are highly conserved
among different species. Mass spectrometry was used to identify phosphorylation
sites of GSK-3β by Erk, and 43T was the only one found to be phosphorylated by Erk
(Figure 4D). To further confirm phosphorylation of 43T, we constructed mutants of
GSK-3β at 43T and generated antibody against the phosphorylation site of GSK-3β,
Thr43, to examine whether Erk could phosphorylate GSK-3β in vivo. Mutation of
GSK-3β 43T significantly inhibited the phosphorylation of GSK-3β by Erk (lane 3,
Figures 4E and 4F) even though it still interacted with Erk (lane 3, Figure 4C).
Moreover, the phosphorylation level of GSK-3β at Thr43 is significantly higher in the
HepG2-X and IGF-1-treated HepG2 cells in which Erk is also highly phosphorylated
compared to HepG2 cells (lanes 4 and 2 versus lane 3, Figure 4G), while Erk/MAPK
inhibitor PD98059 can dramatically inhibit GSK-3β phosphorylation at Thr43 in the
HepG2-X and IGF-1-treated HepG2 cells (lanes 5 and 1 versus lanes 4 and 2, Figure
4G). Furthermore, consistent with our hypothesis, mutating the Erk-docking motif
(DEF domain) of GSK-3β, which impaired the association of GSK-3β with Erk (lane
2, Figure 4C), also significantly inhibited the phosphorylation of GSK-3β by Erk
(lane 2, Figure 4E). These data demonstrate that Erk docks to GSK-3β at the DEF
domain to bridge GSK-3β and p90RSK together and phosphorylates GSK-3β at the
Figure 4.Erk Docks to GSK-3β at the DEF Domain and Phosphorylates GSK-3β at
the 43T Residue
(A) Comparison of the amino acid sequences of the Erk-docking site (DEF domain) in
GSK-3β with other known Erk substrates.
(B) Endogenous Erk was immunoprecipitated from cell lysates as indicated, and
immunocomplexes were blotted with antibodies specific to GSK-3β, Erk, and
(C) GSK-3β interacts with Erk directly and with p90RSK indirectly though the DEF
domain. Purified GST-GSK-3β proteins as indicated were mixed with Hep3B-X cell
lysate and then pulled down with glutathione-sepharose 4B beads. Samples were
subjected to immunoblot with anti-Erk, anti-p90RSK, and anti-GST (for detection of
GST- GSK-3β) antibodies.
(D) Mass spectral analysis of GSK-3β phosphorylation by Erk. GSK-3β was
phosphorylated at the 43T residue in vitro by Erk2.
(E) Mutant GST-GSK-3β protein as indicated was incubated with active Erk2, and the
kinase assay was performed as described in the Experimental Procedures.
(F) Mutant GST-GSK-3β protein as indicated was incubated with active Erk2, and
then samples were subjected to immunoblot with anti-p-GSK-3β (Thr43) and
anti-GST (for detection of GST-GSK-3β) antibodies.
(G) Cell lysates of HepG2 and HepG2-X, treated with IGF-1 (40 ng/ml) or PD98058
(20 μM), were blotted with antibodies as indicated.
Erk Docking to GSK-3β and Phosphorylating GSK-3β Are Required for
HBX-Mediated Inactivation of GSK-3β and Upregulation of β-Catenin
We next asked whether Erk docking to GSK-3β and phosphorylating GSK-3β have
effects on the HBX-mediated inactivation of GSK-3β and upregulation of β-catenin.
Mutation at either the docking motif (GSK-3β291/3A) or the phosphorylation site
(GSK-3β43A), which prevents 43T phosphorylation, reduced HBX-mediated Ser9
phosphorylation of GSK-3β (Figure 5A), indicating that GSK-3β docking to Erk and
phosphorylation at the 43T residue by Erk are required for HBX-induced inactivation
of GSK-3β. Moreover, mutation of the 43T residue to aspartate (GSK-3β43D) retains
Ser9 phosphorylation of GSK-3β (Figure 5A), suggesting that 43T phosphorylation
serves as a priming site for Ser9 phosphorylation of GSK-3β. The two mutants,
GSK-3β291/3A and GSK-3β43A, can cause significant ubiquitination of β-catenin as
compared to wild-type GSK-3β and primed GSK-3β(GSK-3β43D) (lanes 3 and 5
versus lanes 1 and 7, Figure 5B), which is independent of Erk inhibitor PD98059
(lanes 4 and 6 versus lanes 3 and 5, Figure 5B); however, ubiquitination of β-catenin
caused by wild-type GSK-3β is increased by treatment with Erk inhibitor in the
Hep3B-X cells (lane 2 versus lane 1, Figure 5B). The results suggest that GSK-3β
docking to Erk and phosphorylation by Erk are required for inhibition of β-catenin
ubiquitination and therefore likely for β-catenin stabilization. Consistent with this
notion, the two mutants, GSK-3β291/3A and GSK-3β43A, decreased the
HBX-upregulated protein level of β-catenin and its downstream target c-myc (Figure
5C) and also suppressed HBX-upregulated β-catenin transcriptional activity showed
by a Tcf-regulated reporter assay as compared to wild-type GSK-3β and primed
GSK-3β(GSK-3β43D) (Figure 5D). Taken together, these results indicate that both
GSK-3β291/3A and GSK-3β43A mutants function as constitutively active GSK-3β to
downregulate β-catenin, and suggest that GSK-3β association with Erk and
phosphorylation at the 43T residue by Erk are required for the subsequent
phosphorylation at Ser9 to inactivate GSK-3β and upregulate β-catenin.
Figure 5.Erk Docking to GSK-3β and Phosphorylation of GSK-3β Are Required for
HBX-Mediated Inactivation of GSK-3β and Upregulation of β-Catenin
(A) GSK-3β mutants as indicated were transfected to Hep3B-X cells, and then
GSK-3β was immunoprecipitated and the complexes were subjected to immunoblot
with specific p-Ser9 and p-Thr43 GSK-3β antibodies.
(B) Ubiquitination analysis. Hep3B-X cells were transfected with HA-ubiquitin,
β-catenin, and GSK-3β-wt, 43A, 43D, or 291/3A, then treated with PD98059 and
MG132 as indicated. β-catenin was immunoprecipitated and blotted with
(C) Hep3B-X cells were transfected with GSK-3β-wt, 43A, 43D, or 291/3A, then
harvested for immunoblot as indicated.
(D) Cells were transfected with Top or Fop plus GSK-3β-wt, 43A, 43D, or 291/3A as
indicated, and then a Tcf-regulated reporter assay was performed to detect the
transcriptional activity of β-catenin. The error bars represent standard deviation.
HER2 and IGF-1 in Breast Cancer, Like HBX in Hepatoma, Can Upregulate
β-Catenin through Inactivation of GSK-3β by the Erk Pathway
To investigate whether a similar relationship between Erk and GSK-3β observed in
HCC can also be found in other cancer types, we searched for cell lines that express a
high level of both p-Erk and β-catenin. Out of five breast cancer cell lines we
screened, MDA-MB-435 and MCF-7/H18 (HER2-transfected MCF-7) were identified
to have relatively high levels of p-Erk, β-catenin, and β-catenin transcription activity
(Figure 6A). Immunofluorescence staining also indicated that expression of β-catenin
was increased and translocated to the nucleus in the MCF-7/H18 cell lines as
compared to the parental MCF-7 (Figure 6B). Treatment with Erk inhibitor PD98059
(Figure 6C), or transfection by GSK-3β mutants 291/3A or 43A (Figure 6D), resulted
in inhibition of transcriptional activity of β-catenin in these breast cancer cells.
Moreover, in breast cancer cell line MDA-MB-231 and hepatoma cell line Hep3B,
growth factor IGF-1 can lead to β-catenin upregulation and phosphorylation of Erk
and GSK-3β(Ser9). Once treated with Erk inhibitor PD98059 or siRNA-Erk,
phosphorylated Erk was reduced in the IGF-1-treated cells, and concurrently,
downregulation of β-catenin and inhibition of GSK-3β(Ser9) phosphorylation were
observed (Figure 6E), suggesting that IGF-1-induced β-catenin upregulation is
mediated by the Erk/GSK-3β pathway. To further examine whether this relationship
identified in the cell lines can also be observed in human tumor tissues, we compared
the levels of phosphorylated GSK-3β and β-catenin in six phosphorylated
Erk-negative and six phosphorylated Erk-positive human breast tumors. To prevent
interference by Akt, which is also able to phosphorylate GSK-3β at Ser9, the above 12
tumor tissues were preselected to be negative for phosphorylated Akt by IHC (Hu et
al., 2004). Strongly supporting the notion, GSK-3β was highly phosphorylated at Ser9
in six of six phosphorylated Erk-positive breast tumors, and β-catenin accumulated in
both the cytoplasm and nucleus in these tumor samples, while p-GSK-3β(Ser9) and
nuclear β-catenin were not detected in all six p-Erk negative specimens (Figure 6F).
In addition, we also examined a number of human primary tumor specimens by IHC
using a tumor tissue array. Among 12 individual tumor specimens, the same
functional relationship between p-Erk, p-GSK-3β, and accumulation of β-catenin was
found in two tumor entities, stomach carcinoma and renal cell carcinoma (Figure S3),
suggesting that the Erk/GSK-3β/β-catenin relationship may be a general phenomenon
frequently activated in multiple human tumors.
Figure 6.HER2 and IGF-1 Upregulate β-Catenin through Inactivation of GSK-3β by
the Erk Pathway in Breast Cancer
(A) Lysates of breast cancer cell lines were blotted with antibody as indicated, and a
Tcf-regulated reporter assay was performed as in Figure 1E.
(B) An immunofluorescence assay was performed on cell lines as indicated to
measure the expression and cellular localization of β-catenin.
(C) Cells as marked were transfected with Top or Fop, then treated with PD98059. A
Tcf-regulated reporter assay was performed as in Figure 1E. The error bars represent
(D) Cells were transfected with Top or Fop plus GSK-3β mutants as indicated, and
then a Tcf-regulated reporter assay was performed as in Figure 1B. The error bars
represent standard deviation.
(E) MDA-MB-231 and Hep3B cells were pretreated with PD98059 (20 μM) or
siRNA-Erk1+2 before stimulation with IGF-1 (40 ng/ml). The expression of β-catenin
and the status of GSK-3β and Erk were examined by Western blotting.
(F) Tissue sections from p-Akt-negative breast cancer (FA–FH) were stained with
specific antibody against p-Erk (FB and FF), p-Ser9 of GSK-3β (FC and FG), and
β-catenin (FD and FH).
HCC is one of the most prevalent human malignancies and a leading cause of death in
many countries, mainly in Asia and Africa, and it is clearly linked to viral infection.
Individuals who are chronic HBV carriers have a greater than 100-fold increased
relative risk of developing HCC (Arbuthnot and Kew, 2001). Effects of HBX on
transcriptional activation, signal transduction, and DNA repair implicate it in the
etiology of HBV-associated HCC. It is well known that oncogenic virus often
provides some clues about how cancer might develop. For instance, when mutation in
APC, Axin, and β-catenin cannot elucidate β-catenin stabilization, the
latency-associated nuclear antigen protein of Kaposi sarcoma-associated herpesvirus
relocalizes GSK-3β, which leads to cytoplasmic depletion of GSK-3β and
accumulation of β-catenin (Fujimuro et al., 2003), and the large T antigen of the
human polyomavirus JC virus can activate β-catenin through directly binding it (Gan
and Khalili, 2004). Mutations in APC, Axin, and β-catenin represent only 50% of the
upregulation of β-catenin in hepatoma, raising an interesting possibility for a
relationship between HBX and β-catenin. In this study, we found that accumulation of
β-catenin in human HCC samples was significantly correlated with HBX expression,
and importantly, we identified a mechanism to stabilize β-catenin involving GSK-3β
phosphorylation and inactivation by the HBX-activated Erk/RSK pathway, which will
help us further understand the roles of HBX in the HBV-related HCC.
Upregulation of β-catenin occurred in a variety of cancers, such as colorectal, breast,
and ovarian cancers (Lin et al., 2000, Morin et al., 1997 and Rask et al., 2003), and its
ability to induce neoplastic transformation of normal cells proves that β-catenin is a
bona fide oncogene (Kolligs et al., 1999 and Orford et al., 1999). Inhibition of
GSK-3β is essential for the stabilization of β-catenin. GSK-3β inactivation can be
achieved by two primary mechanisms. One mechanism is Wnt-dependent inhibition
(Cook et al., 1996, Itoh et al., 2000, Li et al., 1999 and Ruel et al., 1999); the other
mechanism is phosphorylation of GSK-3β at the Ser9 residue (Cohen and Frame,
2001). GSK-3β inhibition can be caused by a number of growth factors including
IGF-1, EGF, PDGF, FGF-2, HGF, TGF-β, TNF-α, etc. (Cheon et al., 2004,
Desbois-Mouthon et al., 2001, Holnthoner et al., 2002 and Papkoff and Aikawa, 1998),
and one virus protein, the latent membrane protein 2A of Epstein-Barr virus
(Morrison et al., 2003), has been reported to increase β-catenin stabilization through
Ser9 phosphorylation of GSK-3β. However, in certain cell types inhibition of GSK-3β
via Ser9 phosphorylation is not sufficient to induce β-catenin accumulation (Ding et
al., 2000 and Yuan et al., 1999), suggesting that the consequences of changed GSK-3β
activity on β-catenin stabilization are cell type dependent. For instance, reduced
GSK-3β activity caused by insulin stimulation leads to β-catenin accumulation in
HepG2 cells and cardiomyocytes (Desbois-Mouthon et al., 2001 and Haq et al., 2003),
but not in human embryonic kidney 293 cells (Ding et al., 2000). In the current study,
we observed that in hepatoma cell lines HepG2 and Hep3B, inactivation of GSK-3β
via Ser9 phosphorylation is involved in HBX-mediated β-catenin stabilization.
Another important finding is identification of a mechanism to elucidate accurately
how the Erk/p90RSK pathway phosphorylates and inactivates GSK-3β. Like HBX in
hepatoma, HER2 and IGF-1 in breast cancer cells can also upregulate β-catenin
through inactivation of GSK-3β via Ser9 phosphorylation by the Erk/RSK pathway. In
addition, we noticed that in WI-38 fibroblast cells, TGF-β-mediated upregulation of
β-catenin also required Erk phosphorylation and Ser9 phosphorylation of GSK-3β
(Figure S4). Thus, the functional Erk/GSK-3β/β-catenin relationship may be a general
phenomenon in multiple biological systems.
Our results, together with previous findings, provide a plausible mechanism (Figure 7)
for how MAPK/Erk stabilizes β-catenin to facilitate cell proliferation. Growth
factor/receptor or HBX activate the Erk pathway. On one hand, Erk phosphorylates
and activates p90RSK, and on the other hand, Erk serves as an adaptor and priming
kinase through binding to the DEF domain of GSK-3β and phosphorylating GSK-3β
at the 43T residue, to facilitate phosphorylation of GSK-3β by p90RSK on Ser9,
which leads to the inactivation of GSK-3β. Thus, β-catenin is stabilized in the
cytoplasm and translocated to the nucleus, and then it interacts with Tcf/Lef and
upregulates its downstream targets such as cyclin D1 and c-myc, which will facilitate
cell proliferation. In conclusion, the mechanism, namely Erk-primed inactivation of
GSK-3β, directly links HBX or growth factor/receptor and the β-catenin signal
pathway together and may be involved in the development of multiple human cancers.
Figure 7.Model for Erk Association with and Priming of GSK-3β for Its Inactivation
Resulting in Upregulation of β-Catenin