Content uploaded by Mauricio J Reginato
Author content
All content in this area was uploaded by Mauricio J Reginato on Feb 04, 2014
Content may be subject to copyright.
ORIGINAL ARTICLE
Nutrient sensor O-GlcNAc transferase regulates breast cancer
tumorigenesis through targeting of the oncogenic transcription
factor FoxM1
SA Caldwell
1
,
3
, SR Jackson
1
,
3
, KS Shahriari
1
,
3
, TP Lynch
1
, G Sethi
1
, S Walker
2
,
K Vosseller
1
and MJ Reginato
1
1
Department of Biochemistry and Molecular Biology, Drexel University College of Medicine, Philadelphia, PA, USA
and
2
Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, MA, USA
Cancer cells upregulate glycolysis, increasing glucose
uptake to meet energy needs. A small fraction of a cell’s
glucose enters the hexosamine biosynthetic pathway
(HBP), which regulates levels of O-linked b-N-acetylglu-
cosamine (O-GlcNAc), a carbohydrate posttranslational
modification of diverse nuclear and cytosolic proteins. We
discovered that breast cancer cells upregulate the HBP,
including increased O-GlcNAcation and elevated expres-
sion of O-GlcNAc transferase (OGT), which is the enzyme
catalyzing the addition of O-GlcNAc to proteins. Reduc-
tion of O-GlcNAcation through RNA interference of OGT
in breast cancer cells leads to inhibition of tumor growth
both in vitro and in vivo and is associated with decreased
cell-cycle progression and increased expression of the cell-
cycle inhibitor p27
Kip1
. Elevation of p27
Kip1
was associated
with decreased expression and activity of the oncogenic
transcription factor FoxM1, a known regulator of p27
Kip1
stability through transcriptional control of Skp2. Reducing
O-GlcNAc levels in breast cancer cells decreased levels
of FoxM1 protein and caused a decrease in multiple
FoxM1-specific targets, including Skp2. Moreover, redu-
cing O-GlcNAcation decreased cancer cell invasion and
was associated with the downregulation of matrix metallo-
proteinase-2, a known FoxM1 target. Finally, pharmaco-
logical inhibition of OGT in breast cancer cells had similar
anti-growth and anti-invasion effects. These findings
identify O-GlcNAc as a novel mechanism through which
alterations in glucose metabolism regulate cancer growth
and invasion and suggest that OGT may represent novel
therapeutic targets for breast cancer.
Oncogene advance online publication, 1 March 2010;
doi:10.1038/onc.2010.41
Keywords: O-GlcNAc; OGT; FoxM1; breast cancer;
glucose metabolism; p27
Kip1
Introduction
Tumor cells have altered carbohydrate metabolism,
producing ATP primarily through glycolysis, even under
normal oxygen concentrations (Dang and Semenza,
1999). This metabolic shift in cancer cells, termed the
‘Warburg effect’, involves increased glucose uptake and
is critical in supporting cancer phenotypes (Warburg,
1956). Changes in tumor glucose uptake and meta-
bolism also alter distinct nutrient signaling pathways,
including mammalian target of rapamycin, AMP-
activated protein kinase and hexosamine biosynthetic
pathway (HBP) (Marshall, 2006). Indeed, there is
growing evidence that suggests that abnormalities within
the mammalian target of rapamycin and AMP-activated
protein kinase pathways can lead to abnormal growth
and cancer (Shaw, 2006; Guertin and Sabatini, 2007).
The majority of glucose enters glycolysis, producing
ATP, whereas approximately 2–5% of a cell’s glucose
enters the HBP (Marshall et al., 1991), resulting in the
end product uridine diphosphate (UDP)-N-acetylglucos-
amine (UDP-GlcNAc) (Hart et al., 2007). Although flux
through the HBP is likely increased in tumor cells as a
result of upregulated glucose uptake, a role for the HBP
in oncogenesis has not yet been explored.
UDP-GlcNAc is a donor substrate in the enzymatic
covalent addition of a single monosaccharide (GlcNAc)
onto serine or threonine residues. In contrast with all
other types of glycosylation, O-linked b-N-acetylglucos-
amine (O-GlcNAc) modifies a wide variety of cytosolic
and nuclear proteins. O-GlcNAc acts as novel regulatory
switch mechanism analogous to phosphorylation (Wells
et al., 2001). Cytosolic and nuclear enzymes dynamically
catalyze both the addition (O-GlcNAc transferase or
OGT) and the removal (O-GlcNAcase) of O-GlcNAc in
response to various stimuli, including tyrosine kinase
receptor activation (Vosseller et al., 2002). OGT is unique
among glycosyltransferases in its high affinity for UDP-
GlcNAc. As a consequence, OGT activity responds to
physiological changes in UDP-GlcNAc (Lubas and
Hanover, 2000), thus leading to elevated O-GlcNAc
modifications in response to increased flux through the
HBP (Buse et al., 2002). Accordingly, OGT is positioned
to function as a molecular sensor of enhanced HBP
nutrient flux, which would be expected in cancer cells.
Received 11 September 2009; revised 2 November 2009; accepted 25
January 2010
Correspondence: Dr MJ Reginato, Department of Biochemistry and
Molecular Biology, Drexel University College of Medicine, 245 North
15th Street, Philadelphia, PA, USA.
E-mail: Mauricio.Reginato@drexelmed.edu
3
These authors contributed equally to this work.
Oncogene (2010), 1–12
&
2010 Macmillan Publishers Limited
All rights reserved 0950-9232/10 $32.00
www.nature.com/onc
O-GlcNAc is known to influence protein–protein interac-
tions (Roos et al., 1997); therefore, modulations of
O-GlcNAc may alter the formation of specific protein
complexes involved in oncogenic signaling. Modulation
of O-GlcNAc levels is linked to growth/survival pheno-
types such as cell-cycle progression and altered mitotic
phosphorylation patterns (Boehmelt et al., 2000; Zhu
et al.,2001;O’Donnellet al., 2004; Slawson et al., 2005),
showing that a proper balance between O-GlcNAcation
and phosphorylation is required for normal cell growth.
Recently, it was shown that p53-deficient mouse embryonic
fibroblasts, which increase glycolysis, display increased O-
GlcNAcation on a number of proteins (Kawauchi et al.,
2009). Thus, abnormal levels of O-GlcNAc in cancer cells
may contribute to deregulated posttranslational control of
protein function linked to oncogenic phenotypes.
A number of transcription factors are known to be
modified by O-GlcNAc, suggesting that this glucose-
sensing mechanism can directly link nutrient status to gene
expression (Comer and Hart, 1999). Elevated expression
or activity of FoxM1 is associated with the development
and progression of numerous cancers (Wonsey and
Follettie, 2005; Myatt and Lam, 2007). FoxM1 serves as
a key regulator of cell proliferation during organ
development by controlling transcription of genes critical
for G1/S and G2/M progression (Myatt and Lam, 2007),
including Skp2 during G1/S and Nek2, Survivin and
PLK1 during G2/M. Recently, FoxM1 overexpression
was found to correlate with ErbB2 (HER2) status in breast
cancers (Bektas et al., 2008). FoxM1 has also been shown
to regulate cellular invasion through the transcriptional
regulation of matrix metalloproteinases (MMPs) (Wang
et al., 2007). Thus, targeting FoxM1 or its regulators has
been proposed as a viable therapeutic strategy for treating
cancer (Myatt and Lam, 2008).
In this study, we provide the first evidence that OGT and
O-GlcNAc levels are elevated in breast cancer cells, and that
reducing abnormally high O-GlcNAcation inhibits cancer
cell growth in vitro and in vivo, and also reduces breast
cancer cell invasion. Decreasing O-GlcNAc levels through
knockdown of OGT in cancer cells promotes elevation of
the cell-cycle regulator p27
Kip1
and reduces expression of
FoxM1, in addition to a number of FoxM1 targets. Indeed,
regulation of FoxM1 may provide a mechanism through
which decreased levels of O-GlcNAc inhibit breast cancer
phenotypes, as we also found that inhibition of invasion by
targeting OGT was associated with reduction in the FoxM1
transcriptional target MMP-2. Our data suggest that tumor
progression is associated with elevated O-GlcNAcation,
which deregulates critical factors in oncogenic growth and
invasion. In addition, we show that pharmacological
inhibition of OGT may be a valuable strategy for normal-
izing oncogenic phenotypes in breast cancer transformation.
Results
Breast cancer cell lines upregulate O-GlcNAc and OGT
levels
To determine whether levels of O-GlcNAc-modified
proteins are altered in cancer cells, we compared normal
mammary epithelial cells with established breast
cancer cells or oncogene-overexpressing cells. We found
that MCF-10A cells overexpressing the activated
form of ErbB2 (NeuT) and the breast cancer cell lines
SKBR3 and MDA-MB-453 contain elevated levels of
O-GlcNAc-modified proteins compared with normal
human immortalized mammary epithelial MCF-10A
cells (Figure 1a). We then examined whether the increase
in O-GlcNAc-modified proteins in breast cancer cell
lines was related to altered expression of OGT, the
enzyme responsible for catalyzing O-GlcNAc addition
to proteins. We found that OGT is overexpressed in five
different breast cancer cell lines when compared with
normal MCF-10A and MCF-12A cells (Figure 1b). The
increase in OGT protein levels may be due to an increase
in RNA levels, as we found that ErbB2-overexpressing
cells and MDA-MB-231 cells contain elevated OGT
RNA levels compared with normal MCF-10A cells
(Figure 1c). Furthermore, we searched the Oncomine
database and found OGT levels elevated in invasive
ductal carcinoma compared with normal breast tissue
(Supplementary Figure 1). We thus show, for the first
time, that breast cancer cells have elevated levels of
O-GlcNAc and OGT.
Figure 1 Breast cancer cells contain elevated O-GlcNAcation and
OGT levels. (a) Protein lysates from indicated cell lines were
collected for immunoblot analysis and probed with indicated
antibodies. (b) Normal mammary epithelial cells MCF-10A and
MCF-12A and breast cancer cells MCF-10A-ErbB2, BT20, MDA-
MB-231, MDA-MB-453 and MDA-MB-468 were lysed and
subjected to immunoblot analysis with indicated antibodies. (c)
Total RNA was harvested from MCF-10A, MCF-10A-ErbB2 and
MDA-MB-231 cells, and levels of OGT mRNA were quantified by
QRT–PCR and normalized to cyclophilin A. Normalized OGT
mRNA levels are presented relative to MCF-10A. Values represent
mean and s.e. of at least three independent experiments,
* represents Student’s t-test, P-valueo0.05.
OGT required for breast tumorigenesis
SA Caldwell et al
2
Oncogene
OGT is required for malignant growth of transformed
breast cancer cells in vitro
To test whether reducing high levels of O-GlcNAcation
alters breast cancer phenotypes, we targeted OGT
through RNA interference (RNAi) in MCF-10A-ErbB2
cells. The efficiency of two different OGT shRNA lenti-
viral constructs was confirmed by western blotting. We
detected at least a 50% knockdown of OGT protein
compared with cells infected with control (scrambled)
shRNA sequence (Figure 2a). We then tested whether
the reduction of OGT led to a global decrease in
O-GlcNAcation. Cells were treated with or without
the specific O-GlcNAcase inhibitor 9D (Macauley et al.,
2005) to block enzymatic removal of O-GlcNAc. MCF-
10A-ErbB2 cells infected with control shRNA showed a
significant increase in O-GlcNAcation in the presence
of 9D (Figure 2b). However, cells infected with RNAi
targeting OGT had significantly decreased basal
O-GlcNAcation, and completely blocked the 9D-in-
duced elevation of O-GlcNAcation (Figure 2b); decrease
in OGT levels led to a significant reduction in O-
GlcNAc-modified proteins. These cells were then placed
in three-dimensional (3D) culture assays or soft agar
assays to determine the effect of reducing OGT levels on
cancer cell growth. Under 3D conditions, reduction of
OGT by RNAi in MCF-10A-ErbB2 cells caused a
dramatic inhibition of oncogenic phenotypes, including
decreased cell growth and an eight-fold decrease in cell
number at day 12 (Figure 2c) compared with control
RNAi cells. In addition, reduction of OGT levels in
MCF-10A-ErbB2 cells showed a similar decrease in
colony formation in soft agar assays (Supplementary
Figure 2). To test whether the reduction of abnormally
high levels of O-GlcNAcation could alter breast cancer
phenotypes independent of ErbB2, we knocked down
OGT levels in the highly transformed breast cancer cell
line MDA-MB-231, which does not overexpress ErbB2.
MDA-MB-231 cells stably infected with RNAi against
OGT showed a three-fold decrease in soft agar colony
formation (Figure 2d) and resulted in significant
inhibition of growth under 3D conditions compared
with control-infected cells (data not shown). Knock-
down of OGT in parental MCF-10A cells did not
significantly block growth or ability to form acinar
structures in 3D culture (Supplementary Figure 3D),
suggesting that reducing OGT levels in nontransformed
cells is not cytotoxic. Consistent with the idea that
elevated OGT contributes to tumor cell growth, over-
expression of OGT in MCF-10A-ErbB2 cells increased
the number of soft agar colonies (data not shown).
Thus, we showed that OGT and abnormally elevated
levels of O-GlcNAc are required for and may contribute
to transformed growth of breast cancer cells in vitro.
OGT is required for tumorigenic growth of human breast
cancer cells in vivo
We next examined a role for OGT in promoting
oncogenic phenotypes in vivo. To test this, we performed
orthotopic xenografts of MDA-MB-231 cells stably
expressing either OGT shRNA or control shRNA. OGT
knockdown and decreased basal O-GlcNAcation were
verified by western blot analysis at the time of injection
(Figure 3a). Control and OGT knockdown cells
were then injected directly into contralateral mammary
fat pads of immunocompromised Nu/Nu mice to avoid
inter-animal variations. A four-fold decrease in tumor
volume was observed in mice injected with OGT
knockdown cells compared with control cells at the
end of the 8-week experiment (Figure 3b). At necropsy,
of mice injected with cells expressing scrambled shRNA,
84% developed visible tumors that could be excised;
only 41% of mice injected with cells containing OGT-1
shRNA (Figure 3c) and 40% of mice injected with
cells containing OGT-2 developed visible tumors (data
not shown). Tumor mass measurements from OGT
knockdown cells showed a similar four-fold reduction
compared with tumors from control cells (Supple-
mentary Figure 4A). Importantly, tumors that even-
tually grew from OGT knockdown cells restored
OGT expression (Figure 3d) and had a similar Ki-67
expression (Supplementary Figure 4B), suggesting a
strong selective pressure against tumor cells deficient in
OGT. These data indicate the importance of OGT in
tumor cell growth in vivo.
Inhibition of OGT decreases cell-cycle progression and
induces p27
Kip1
expression through regulation of FoxM1
in breast cancer cells
To investigate further the growth-inhibitory effect of
OGT knockdown in breast cancer cells, we conducted a
cell-cycle analysis by propidium iodide staining and flow
cytometry. Reduction of OGT in MCF-10A-ErbB2 cells
caused a significant accumulation of cells in the G1
phase within 48 h compared with control shRNA-
infected cells: 72% G1 content in OGT shRNA-infected
cells relative to 47% in control shRNA cells (Figure 4a).
With OGT shRNA, we also observed a significant
decrease in the S- and G2/M-phase population com-
pared with control. In addition, we found a two-fold
decrease in Ki-67 staining in MCF-10A-ErbB2 and
MDA-MB-231 cells expressing OGT shRNA (Supple-
mentary Figure 5A). We did not detect an increase in
the sub-G1 population of cells nor did we detect a
significant change in DNA fragmentation at this time
point (data not shown), suggesting that targeting OGT
had minimal effects on apoptosis. Reducing OGT in
normal MCF-10A cells caused a slight increase in G1
population, but neither this (Supplementary Figure 3B)
nor changes in Ki-67 staining (Supplementary Figure
3C) were statistically significant.
Increase in the population of cells in G1 suggests that
cell-cycle regulators may be altered by reducing OGT
expression. Knockdown of OGT results in a significant
induction of p27
Kip1
levels and reduction of prolifera-
ting cell nuclear antigen in MCF-10A-ErbB2 cells
(Figure 4b), as well as in MDA-MB-231 cells (Supple-
mentary Figure 5B), consistent with cell-cycle arrest at
G1. The regulation of p27
Kip1
is highly complex; it is well
established that oncogenic signaling, including receptor
tyrosine kinase, c-Src and mitogen-activated protein
OGT required for breast tumorigenesis
SA Caldwell et al
3
Oncogene
kinase activation in cancer cells is associated with
increased p27
Kip1
proteolysis (Chu et al., 2008). Yet,
knockdown of OGT in MCF-10A-ErbB2 cells did not
reduce activity of ErbB2, c-Src, Erk (extracellular
signal-regulated kinase) (Supplementary Figure 6) or
Akt (Figure 3b) as measured with respective phospho-
specific antibodies. As p27
Kip1
mRNA levels were not
decreased in cells depleted of OGT (data not shown),
we considered alternative pathways regulating p27
Kip1
degradation.
Degradation of p27
Kip1
is primarily regulated by the
SCF
SKP2
E3 ubiquitin ligase complex (Chu et al., 2008).
This complex includes the F-Box protein Skp2 that
targets cyclin-dependent kinase inhibitors for degrada-
tion during the G1/S transition. We found that OGT
knockdown in MCF-10A-ErbB2 cells (Figure 4b)
and MDA-MB-231 cells (Supplementary Figure 5B)
decreases Skp2 expression. One level of Skp2 regu-
lation is through transcriptional activation by FoxM1
(Wang et al., 2005). We found that in MCF-10A-ErbB2
Figure 2 Knockdown of OGT reduces O-GlcNAcation and inhibits growth of MCF-10-ErbB2 and MDA-MB-231 cells in vitro.
(a) MCF-10A-ErbB2 cells were infected with control, OGT-1 or OGT-2 shRNA pLKO.1 lentivirus, and protein lysates were collected
48 h after infection for immunoblot analysis with indicated antibodies. (b) MCF-10A-ErbB2 cells infected as described in panel a for
48 h and were treated for 24 h with the indicated concentrations of 9D before lysis and immunoblot analysis with the indicated
antibodies. (c) MCF-10A-ErbB2 cells infected, as described in panel a, and placed in a 3D morphogenesis assay. Cells were imaged and
counted at indicated time points. (d) MDA-MB-231 cells were infected as above and lysed 48 h after infection and subjected to
immunoblot analysis with indicated antibodies (left) or, placed in soft agar assays (right). Colonies were stained 14 days later and
quantified. Insert: image showing representative number and size of colonies. Values represent mean and s.e. of at least three
independent experiments, * represents Student’s t-test, P-valueo0.05.
OGT required for breast tumorigenesis
SA Caldwell et al
4
Oncogene
(Figure 4b) and MDA-MB-231 cells (Supplementary
Figure 5B), reducing OGT expression leads to signifi-
cant decreases in FoxM1 protein levels. FoxM1 can
regulate progression from the G1 to S phase, and is also
known to be a key regulator during G2/M transition.
Indeed, we find that FoxM1-specific targets involved
in G2/M phase, including Survivin, Nek2 and PLK1,
are also decreased in OGT knockdown cells, both in
MCF-10A-ErbB2 (Figure 4c) and MDA-MB-231 cells
(Supplementary Figure 5B).
To begin addressing the mechanism of how OGT
regulates FoxM1 levels, we examined effects of OGT
knockdown in MDA-MB-231 cells stably expressing
exogenous FoxM1. Reducing OGT levels caused
downregulation of stably overexpressed wild-type
FoxM1 protein (Figure 4d), suggesting that OGT and
Figure 3 OGT is required for tumorigenic growth of human breast cancer cells in vivo.(a) MDA-MB-231 cells expressing control,
OGT-1 or OGT-2 shRNA were lysed and analyzed by immunoblot analysis with indicated antibodies before injection into mice.
(b) Mean tumor volume (mm
3
) of MDA-MB-231 cells expressing control (n¼37), OGT-1 (n¼17), or OGT-2 (n¼20) shRNA injected
into mammary fat pad of immunocompromised mice at indicated week. Values represent mean and s.e. of independent experiments,
* represents Student’s t-test, P-valueo0.05. (c) Top, representative mammary fat pad tumor in mice transplanted with MDA-MB-231
control shRNA and OGT-1 shRNA cells 8 weeks after injection, and, bottom, resected tumors 8 weeks after injection of MDA-MB-231
cells expressing control or OGT-1 shRNA. (d) Tumors resected from mice were lysed and analyzed by immunoblot analysis with
indicated antibodies.
OGT required for breast tumorigenesis
SA Caldwell et al
5
Oncogene
O-GlcNAcation may regulate FoxM1 posttranscription-
ally. Recent studies have identified the N terminus of
FoxM1 as being a substrate for ubiquitin-mediated
degradation, contributing to the normal changes in
FoxM1 levels across the cell cycle (Laoukili et al., 2008)
(Park et al., 2008). The N terminus of FoxM1 contains
destruction box (D box) and KEN-box sequences,
short degradation motifs recognized by the anaphase-
promoting complex E3 ubiquitin ligase (Park et al.,
2008) (Laoukili et al., 2008). FoxM1 regulation by
O-GlcNAcation required the N terminus of FoxM1, as
protein levels of a deletion mutant missing the first 209
amino acids of FoxM1 (DN-DKEN-FoxM1) were no
longer decreased by reducing OGT expression as
compared with wild-type FoxM1 (Figure 4d). To test
whether overexpression of wild-type or mutant FoxM1
can rescue cell growth defect caused by downregulating
OGT, we placed these cells in 3D culture. Cells
overexpressing either the wild type or mutant of FoxM1
were able to partially overcome the inhibitory effect of
OGT silencing on cell growth in 3D culture (Figure 4e).
In addition, knockdown of FoxM1 with RNAi in MCF-
10A-ErbB2 cells or MDA-MB-231 cells caused in-
creased expression of p27
Kip1
, inhibition of growth in
3D culture and soft agar results similar to that observed
in OGT knockdown cells (data not shown). The
reduction of FoxM1 protein is not a part of a global
alteration in protein degradation, as we did not detect
changes in levels of other Fox transcription family
members, including FOXO3a (Supplementary Figure 7).
Moreover, we found that reduction of OGT levels led to
no significant change in the expression of a number of
Figure 4 Knockdown of OGT inhibits cell-cycle progression, elevates p27
Kip1
expression and reduces FoxM1 expression in breast
cancer cells. (a) Cell-cycle analysis of MCF-10A-ErbB2 cells 48 h after lentiviral infection with control, OGT-1 or OGT-2 shRNA.
Cells were collected, stained with propidium iodide and analyzed by flow cytometry. Cell-cycle distribution was determined using
Guava Cytosoft Software (Millipore, Billerica, MA, USA). (b,c) Cell lysates were collected from MCF-10A-ErbB2 cells 48 h after
lentiviral infection with control, OGT-1, or OGT-2 shRNA. Lysates were analyzed by immunoblot analysis with indicated antibodies.
(d) MDA-MB-231 cells were infected with retroviruses encoding control vector (pBabe), wild-type FoxM1 (pBabe-FoxM1-WT) or
mutant FoxM1 (pBabe-FoxM1-DN/DKen). After stable selection, cells were infected with lentivirus containing control or OGT-1
shRNA for 48 h, lysed and analyzed by immunoblot analysis with indicated antibodies. (e) MDA-MB-231 cells stably expressing
control vector, wild-type FoxM1 or mutant FoxM1 were infected with OGT-1 shRNA lentivirus. Cells were then placed in 3D
morphogenesis assay, imaged at day 5 and counted at day 8. Values represent mean and s.e. of at least three independent experiments,
* represents Student’s t-test, P-valueo0.05.
OGT required for breast tumorigenesis
SA Caldwell et al
6
Oncogene
transcription factors implicated in breast cancer, includ-
ing p53, c-Myc and nuclear factor-kB (Supplementary
Figure 7). As other Fox transcription family members
have been shown to be directly modified by O-GlcNAc,
we tested whether FoxM1 is modified by O-GlcNAc.
FoxM1 immunoprecipitated from MDA-MB-231 cells
overexpressing wild-type FoxM1 did not show any O-
GlcNAc modifications, whereas endogenous Sp1, a
transcription factor known to be modified by O-GlcNAc
(Han and Kudlow, 1997), was highly modified under
similar conditions (data not shown). Thus, our data
show that breast cancer cell growth inhibition by
targeting OGT is associated with increased cell-cycle
arrest at G1, elevated expression of p27
Kip1
and specific
posttranscriptional downregulation of the oncogenic
transcription factor FoxM1 and its targets. However,
FoxM1 is not directly O-GlcNAcated, suggesting that
OGT may be regulating FoxM1 indirectly.
OGT regulates breast cancer cell invasion
We observed that breast cancer cells with OGT knock-
down produced fewer invasive protrusions when cul-
tured under 3D conditions (Figure 5a), suggesting that
reduction of elevated O-GlcNAcation may inhibit
cellular invasion. To test this directly, we placed
MCF-10A-ErbB2 cells targeted with OGT or control
shRNA in transwell invasion assays. Knockdown of
OGT led to a three-fold decrease in invasion compared
with controls (Figure 5b). Breast cancer invasion and
metastasis is associated with elevated levels of MMP-2
(Duffy et al., 2000). As FoxM1 regulates expression of
MMP-2 in pancreatic cancer cells (Wang et al., 2007),
we examined levels of MMP-2 in OGT knockdown cells.
Indeed, we found a two-fold decrease in the expression
of MMP-2 at both mRNA (Figure 5c) and protein levels
(Figure 5d) in MCF-10A-ErbB2 cells when OGT is
knocked down. Knockdown of FoxM1 in MCF-10A-
ErbB2 cells also leads to decreased MMP-2 levels
and invasion (data not shown). Thus, our data suggest
that OGT regulates cancer cell invasion by modu-
lating MMP-2 expression, possibly by the regulation
of FoxM1.
OGT inhibitor blocks breast cancer growth and invasion
We have recently identified novel inhibitors of OGT
catalytic activity (Gross et al., 2005). OGT inhibitor
(OGTi) treatment of MCF-10A-ErbB2 cells reduced
O-GlcNAc levels (Figure 6a) and dramatically decreased
growth in soft agar (Figure 6b) and 3D culture assays
(Figure 6c). Pharmacological inhibition of OGT led to
decreased FoxM1 expression, which correlated with
elevation of p27
Kip1
levels (Figure 6a). Similar to OGT
knockdown, a decrease in invasive protrusions from
cells treated with OGTi in 3D culture was observed, and
a six-fold decrease in cell invasion of MCF-10A-ErbB2
cells was observed in response to treatment with OGTi
using transwell invasion assays (Figure 6d). Similar
inhibitory effects on FoxM1 levels, cell growth and
invasion were observed in MDA-MB-231 cells treated
with OGTi (Supplementary Figure 8). Treatment of
parental MCF-10A acinar structures with OGTi at day
14 for 48 h did not cause cytotoxic effects or disruption
of acinar architecture (data not shown).
Thus, we show in this study for the first time that
OGT and O-GlcNAcation is elevated in cancer cells,
and that normalization of these levels by two indepen-
dent methods (RNAi knockdown of OGT and pharma-
cological inhibition) reduces tumor growth and
invasion. Elevated O-GlcNAcation and OGT levels
appear to contribute to cancer cell growth and invasion,
at least in part by regulating the stability of the
oncogenic transcription factor FoxM1 and its down-
stream targets.
Discussion
Glucose flux through the HBP, leading to modifications
of nuclear and cytoplasmic proteins by O-GlcNAc, is
emerging as a key regulator for many biological
processes and disease states. OGT regulation of the
insulin pathway has been implicated in insulin resistance
associated with type II diabetes (Vosseller et al., 2002)
(Yang et al., 2008), and O-GlcNAc alterations has also
been associated with neurodegenerative diseases, includ-
ing Alzheimer’s disease (Liu et al., 2004). In this study,
we show for the first time that OGT and O-GlcNAc
modifications are elevated in cancer cells, and that
normalization of these levels reduces tumor growth and
invasion. Elevated O-GlcNAc and OGT may contribute
to cancer cell growth and invasion, in part by regulating
the oncogenic transcription factor FoxM1.
Most cancers exhibit altered metabolism, including
increased aerobic glycolysis and a dependence on glyco-
lytic pathways for ATP generation. To serve the less-
efficient energy-producing glycolytic route, tumor cells
increase glucose uptake. Increased glucose consumption
may lead to increased shunting to the HBP. Consistent
with this idea, a recent study has shown that elevated
glycolysis associated with loss of p53 in mouse embryonic
fibroblasts leads to increased O-GlcNAc modifications
(Kawauchi et al., 2009). Our results show that breast
cancer cells known to have increased glycolysis, such as
MDA-MB-231 (Gatenby and Gillies, 2004; Gallagher
et al., 2007), contain elevated O-GlcNAc modifications
and increased OGT levels. However, it is not clear whether
elevation of glycolysis in cancer cells directly leads to
increased flux through the HBP and consequential increase
in O-GlcNAc modifications, or whether transformation
by oncogenes or loss of tumor suppressors may regulate
OGT expression or activity. Nonetheless, our data indicate
that elevated O-GlcNAcation links cancer cell alterations
of metabolic pathways to transformed cell growth and
invasion signals.
The induction of p27
Kip1
by knockdown of OGT is
significant, as many breast cancer therapies directly
upregulate p27
Kip1
protein. Furthermore, the magnitude
of breast cancer cell growth inhibition by therapies
including the anti-ErbB2 antibody Herceptin closely
parallels the level of p27
Kip1
induced (Yakes et al., 2002).
OGT required for breast tumorigenesis
SA Caldwell et al
7
Oncogene
However, we found little change in signaling pathways
associated with ErbB2 activation or breast cancer in
general in OGT knockdown cells compared with
controls, suggesting that decreased O-GlcNAcation
alters p27
Kip1
stability through mechanisms independent
of inhibiting ErbB2 signaling. Nevertheless, it is almost
certain that additional functionally significant effects of
OGT knockdown/inhibition on cellular signaling are
occurring, and it will be important to elucidate these.
FoxM1 is a well-characterized regulator of p27
Kip1
and
cell growth and transcriptionally regulates Skp2, the
specific recognition factor for p27
Kip1
ubiquitination. In
glioma cells, RNAi knockdown of FoxM1 led to
increased p27
Kip1
levels associated with a decrease in
Skp2 protein (Liu et al., 2006). In pancreatic cells, RNAi
against FoxM1 led to a decrease in metastasis and
angiogenesis, correlating with a reduction of MMP-2
and vascular endothelial growth factor expression
(Wang et al., 2007). Consistent with these data, OGT
knockdown in breast cancer cells led to a reduction in
invasion and downregulation of the FoxM1 target
MMP-2. As FoxM1 is highly expressed in proliferating
tumor cells and contributes to metastasis, it is currently
being considered as a viable therapeutic target for a
Figure 5 OGT knockdown blocks invasion and reduces MMP-2 expression in breast cancer cells. (a) MCF-10A-ErbB2 cells
expressing control, OGT-1 or OGT-2 shRNA were placed in 3D culture. At day 8, cells were fixed and stained for confocal microscopy
with indicated antibodies. (b) MCF-10A-ErbB2 cells infected with control, OGT-1 or OGT-2 shRNA were placed in transwell invasion
chambers for 24 h. Invading cells were DAPI stained and counted. (c) Total RNA from MCF-10A-ErbB2 cells infected with control,
OGT-1 or OGT-2 shRNA were collected and assayed for OGT and MMP-2 expression using QRT–PCR, normalized to Cyclophilin
A. Data expressed as normalized expression relative to control shRNA. (d) Cell lysates from MCF-10A-ErbB2 cells expressing control,
OGT-1 or OGT-2 shRNA were collected and analyzed by immunoblotting with indicated antibodies. Values represent mean and s.e. of
at least three independent experiments, * represents Student’s t-test, P-valueo0.05.
OGT required for breast tumorigenesis
SA Caldwell et al
8
Oncogene
number of cancers (Gartel, 2008). Our results show that
targeting OGT with first-generation OGTis may be a
novel way to modulate FoxM1 expression in breast and
perhaps other cancers.
FoxM1 expression increases during the G1 to S phase
after cyclin E/cyclin-dependent kinase 2-mediated
(Major et al., 2004) and Ras/Mek/Erk kinase-mediated
phosphorylation (Ma et al., 2005). However, it is
unlikely that the increase in G1-phase cells and down-
regulation of FoxM1 expression in response to OGT
knockdown is due to loss of Mek/Erk signaling, as no
decrease in Erk activation is observed in OGT knock-
down cells. Recent studies have identified FoxM1 as
being a substrate for ubiquitin-mediated degradation,
contributing to the normal changes in FoxM1 levels
across the cell cycle (Laoukili et al., 2008) (Park et al.,
2008). O-GlcNAcation of p53 has been shown to protect
against ubiquitin-mediated degradation (Yang et al.,
2006) and increase the half-life of steroid nuclear
receptors (Cheng and Hart, 2001). However, although
the Fox family member FoxO1 has recently been shown
to be modified by O-GlcNAc (Housley et al., 2008), we
were unable to detect O-GlcNAc modifications on
FoxM1. One possibility is that hyper-O-GlcNAcation
of regulators of FoxM1 in breast cancer cells blocks
FoxM1 degradation, increasing FoxM1 protein levels
and contributing to transformation. In this model,
as O-GlcNAc levels are decreased, FoxM1 becomes
susceptible to proteosomal degradation, accounting for
its decreased expression level and inhibition of trans-
formed phenotypes. Our data show that the N terminus
of FoxM1 is required for regulation by OGT. Recent
studies have shown that the N terminus of FoxM1
contains both D-box D- and KEN-box sequences that
are required for proteolytic targeting by anaphase-
promoting complex-Cdh1 adaptor (Park et al., 2008)
(Laoukili et al., 2008), thus suggesting that changes
in O-GlcNAcation may indirectly regulate FoxM1
degradation.
In summary, this study is the first to link OGT and
O-GlcNAcation to cancer cell growth and invasion and
identifies novel regulation of the oncogenic transcrip-
tion factor FoxM1 by altered O-GlcNAcation. Thus,
the nutrient-sensing roles of OGT may link abnormal
Figure 6 Pharmacological OGT inhibition reduces O-GlcNAcation and blocks growth and invasion of MCF-10A-ErbB2 cells.
(a) MCF-10A-ErbB2 cells were treated with control (DMSO) or 500 mMOGTi for 48 h. Cells were lysed and proteins analyzed by
immunobloting with indicated antibodies. (b) MCF-10A-ErbB2 cells were placed in soft agar assay and treated with control or 500 mM
OGTi for 14 days. Cell were stained, colonies were counted and imaged (inset). (c) MCF-10A-ErbB2 cells were placed in 3D culture
and treated with OGTi (500 mM) or control. At day 8, phase images of the acini were acquired. (d) MCF-10A-ErbB2 cells were placed in
transwell invasion slides in the presence of control or 500 mMOGTi. Values represent mean and s.e. of at least three independent
experiments, * represents Student’s t-test, P-valueo0.05.
OGT required for breast tumorigenesis
SA Caldwell et al
9
Oncogene
metabolic states in cancer cells to deregulation of
critical growth and transformation factors such as
FoxM1, and pharmacological modulation of enzymes
regulating O-GlcNAcation may be a novel therapeutic
strategy in cancer.
Materials and methods
Materials
The O-GlcNAcase inhibitor, 9D, was provided by David
Vocadlo (Simon Fraser University, Burnaby, BC, Canada).
Growth-factor-reduced Matrigel was purchased from BD
Biosciences (Franklin Lakes, NJ, USA). Antibodies such as
anti-actin, anti-FoxM1 and anti-proliferating cell nuclear
antigen were obtained from Santa Cruz Biotechnology (Santa
Cruz, CA, USA); anti-phospho-Akt (Ser473), anti-phospho-Akt
(T308), anti-AKT and anti-MMP-2 were purchased from Cell
Signaling (Danvers, MA, USA); anti-OGT was obtained from
Sigma (St Louis, MO, USA); anti-p27
Kip1
, anti-Nek2, anti-PLK1
and anti-integrin-a5 were purchased from BD-Biosciences; anti-
Skp2 was from Zymed (Carlsbad, CA, USA); anti-integrin-a6
was obtained from Chemicon (Billerica, MA, USA). Anti-O-
GlcNAc antibody (CTD110.6) (Comer et al., 2001) and OGTi
(Gross et al., 2005) have been described previously.
Cell culture and viral infections
MCF-10A, MCF-12A SKBR-3, MDA-MB-231, MDA-MB-
453, MDA-MB-468, MCF-7 and BT-20 cells were acquired
from ATCC (American Type Culture Collection, Manassas,
VA, USA) and cultured following the instructions of ATCC.
MCF-10A cells were grown as described previously (Reginato
et al., 2003). Constitutively active ErbB2 mutant (pBabe-
NeuT) was kindly provided by Danielle Carroll (Harvard
Medical School, Boston, MA, USA). pBabe-Flag-FoxM1 and
pBabe-Flag-DN-DKEN-FoxM1 (was created by cloning Flag-
FoxM1 cDNA insert from pcDNA3-Flag-FoxM1 and
pcDNA3-Flag- DN-DKEN-FoxM1 plasmids (Laoukili et al.,
2008) (kindly provided by Rene H Medema, University
Medical Center, Utrecht, The Netherlands) into the BamHI
and EcoRI sites of pBabe-puro. MDA-MB-231 cells over-
expressing FoxM1 were generated using vesicular stomatitis
virus G protein-pseudotyped retroviruses and were infected
and selected as described previously (Reginato et al., 2003).
Immunoblotting
Cell lysates from 1–5 10
6
cells were prepared in RIPA lysis
buffer (150 mMNaCl, 1% NP40, 0.5% DOC, 50 mMTris-HCl
at pH 8, 0.1% SDS, 10% glycerol, 5 mMEDTA, 20 mMNaF
and 1 mMNa
3
VO
4
) supplemented with 1 mg/ml each of
pepstatin, leupeptin, aprotinin and 200 mg/ml phenyl-methyl-
sulfonyl-fluoride. Lysates were cleared by centrifugation at
16 000 gfor 20 min at 4 1C and analyzed by SDS–PAGE and
autoradiography. Proteins were analyzed by immunoblotting
using primary antibodies indicated above.
RNA interference
Stable cell lines for shRNA knockdowns were generated by
infection with the lentiviral vector pLKO.1-puro carrying
shRNA sequence for scrambled (Addgene, Cambridge, MA,
USA) or OGT (Sigma). VSVG-pseudotyped lentivirus was
generated by the cotransfection of 293-T packaging cells with
10 mg of DNA and packaging vectors as described previously
(Rubinson et al., 2003). Control-scrambled shRNA sequence
used was: CCTAAGGTTAAGTCGCCCTCGCTCTAGCGA
GGGCGACTTAACCTT. OGT shRNA sequence used was:
for OGT-1, GCCCTAAGTTTGAGTCCAAATCTCGAGATT
TGGACTCAAACTTAGGGC and for OGT-2, GCTGAGCA
GTATTCCGAGAAACTCGAGTTTCTCGGAATACTGCTC
AGC. Cells were infected and selected as described previously
(Reginato et al., 2003).
3D morphogenesis assay and indirect immunofluorescence
Assays were performed as described previously (Reginato
et al., 2005). Briefly, 5 10
3
MCF-10A-ErbB2 or MDA-MB-
231 cells, in respective media containing 2% Matrigel, were
placed in an 8-well chamber slide (BD Biosciences) coated with
50 ml of Matrigel. The number of cells was counted in two
chambers at indicated time points and the mean of each
determined. Immunofluorescence of 3D structures was per-
formed as described previously (Reginato et al., 2005) using
antibodies to integrin-a5 and integrin-a6 then stained with
40,6-diamidino-2-phenylindole. Fluorescent secondary anti-
bodies coupled with Alexa-Fluor dyes (Molecular Probes,
Carlsbad, CA, USA) were used. Confocal analysis was
performed by using the Leica DM6000B Confocal Microscope
(Leica). Images were generated using the Leica Imaging
Software (Wetzlar, Germany) and converted to Tiff format.
Orthotopic xenograft model
MDA-MB-231 cells were infected with lentivirus carrying
control (scramble) shRNA, OGT-1 and OGT-2 shRNA
constructs, as described above. After washes and resuspension
in Hank’s buffered salt solution (Mediatech, Inc., Manassas,
VA, USA), 1.5 10
6
cells in 0.1 ml containing 20% Matrigel
were injected subcutaneously through a 27½ gauge needle into
the fourth inguinal mammary fat pad pair of each 4–6-week-
old female athymic nude Nu/Nu mouse (Charles River,
Wilmington, MA, USA). For each individual, control shRNA
cells were injected into the right gland and OGT cells into the
contralateral gland. After injection, tumors were measured
weekly along and perpendicular to the longest dimension using
digital calipers (Fowler Co., Inc., Newton, MA, USA). Tumor
volumes were calculated as V¼(length) (width)
2
0.52.
After 8 weeks, tumors were excised, weighed and photo-
graphed. Tumors were then flash-frozen in liquid N
2
for
western blot analysis. Frozen tumor samples were mechani-
cally disrupted and resuspended in ice-cold RIPA buffer and
lysed (described above) for immunoblotting.
Conflict of interest
The authors declare no conflict of interest.
Acknowledgements
This study was supported by the Department of Defense, Breast
Cancer Research Program Concept Award: BC086596 to MJR
and Synergistic Idea Award: BC074374 to MJR and KV.
OGT required for breast tumorigenesis
SA Caldwell et al
10
Oncogene
References
Bektas N, Haaf A, Veeck J, Wild PJ, Luscher-Firzlaff J, Hartmann A
et al. (2008). Tight correlation between expression of the Forkhead
transcription factor FOXM1 and HER2 in human breast cancer.
BMC Cancer 8: 42.
Boehmelt G, Wakeham A, Elia A, Sasaki T, Plyte S, Potter J et al.
(2000). Decreased UDP-GlcNAc levels abrogate proliferation
control in EMeg32-deficient cells. EMBO J 19: 5092–5104.
Buse MG, Robinson KA, Marshall BA, Hresko RC, Mueckler MM.
(2002). Enhanced O-GlcNAc protein modification is associated with
insulin resistance in GLUT1-overexpressing muscles. Am J Physiol
Endocrinol Metab 283: E241–E250.
Cheng X, Hart GW. (2001). Alternative O-glycosylation/O-phosphor-
ylation of serine-16 in murine estrogen receptor beta: post-
translational regulation of turnover and transactivation activity.
J Biol Chem 276: 10570–10575.
Chu IM, Hengst L, Slingerland JM. (2008). The Cdk inhibitor p27 in
human cancer: prognostic potential and relevance to anticancer
therapy. Nat Rev Cancer 8: 253–267.
Comer FI, Hart GW. (1999). O-GlcNAc and the control of gene
expression. Biochim Biophys Acta 1473: 161–171.
Comer FI, Vosseller K, Wells L, Accavitti MA, Hart GW.
(2001). Characterization of a mouse monoclonal antibody
specific for O-linked N-acetylglucosamine. Anal Biochem 293:
169–177.
Dang CV, Semenza GL. (1999). Oncogenic alterations of metabolism.
Trends Biochem Sci 24: 68–72.
Duffy MJ, Maguire TM, Hill A, McDermott E, O’Higgins N. (2000).
Metalloproteinases: role in breast carcinogenesis, invasion and
metastasis. Breast Cancer Res 2: 252–257.
Gallagher SM, Castorino JJ, Wang D, Philp NJ. (2007).
Monocarboxylate transporter 4 regulates maturation and trafficking
of CD147 to the plasma membrane in the metastatic breast cancer
cell line MDA-MB-231. Cancer Res 67: 4182–4189.
Gartel AL. (2008). FoxM1 inhibitors as potential anticancer drugs.
Expert Opin Ther Targets 12: 663–665.
Gatenby RA, Gillies RJ. (2004). Why do cancers have high aerobic
glycolysis? Nat Rev Cancer 4: 891–899.
Gross BJ, Kraybill BC, Walker S. (2005). Discovery of O-GlcNAc
transferase inhibitors. J Am Chem Soc 127: 14588–14589.
Guertin DA, Sabatini DM. (2007). Defining the role of mTOR in
cancer. Cancer Cell 12: 9–22.
Han I, Kudlow JE. (1997). Reduced O glycosylation of Sp1 is
associated with increased proteasome susceptibility. Mol Cell Biol
17: 2550–2558.
Hart GW, Housley MP, Slawson C. (2007). Cycling of O-linked beta-
N-acetylglucosamine on nucleocytoplasmic proteins. Nature 446:
1017–1022.
Housley MP, Rodgers JT, Udeshi ND, Kelly TJ, Shabanowitz J, Hunt
DF et al. (2008). O-GlcNAc regulates FoxO activation in response
to glucose. J Biol Chem 283: 16283–16292.
Kawauchi K, Araki K, Tobiume K, Tanaka N. (2009). Loss of p53
enhances catalytic activity of IKKbeta through O-linked beta-
N-acetyl glucosamine modification. Proc Natl Acad Sci USA 106:
3431–3436.
Laoukili J, Alvarez-Fernandez M, Stahl M, Medema RH. (2008).
FoxM1 is degraded at mitotic exit in a Cdh1-dependent manner.
Cell Cycle 7: 2720–2726.
Liu F, Iqbal K, Grundke-Iqbal I, Hart GW, Gong CX. (2004).
O-GlcNAcylation regulates phosphorylation of tau: a mechanism
involved in Alzheimer’s disease. Proc Natl Acad Sci USA 101:
10804–10809.
Liu M, Dai B, Kang SH, Ban K, Huang FJ, Lang FF et al.
(2006). FoxM1B is overexpressed in human glioblastomas and
critically regulates the tumorigenicity of glioma cells. Cancer Res 66:
3593–3602.
Lubas WA, Hanover JA. (2000). Functional expression of O-linked
GlcNAc transferase Domain structure and substrate specificity.
J Biol Chem 275: 10983–10988.
Ma RY, Tong TH, Cheung AM, Tsang AC, Leung WY, Yao KM.
(2005). Raf/MEK/MAPK signaling stimulates the nuclear trans-
location and transactivating activity of FOXM1c. J Cell Sci 118:
795–806.
Macauley MS, Whitworth GE, Debowski AW, Chin D, Vocadlo DJ.
(2005). O-GlcNAcase uses substrate-assisted catalysis: kinetic
analysis and development of highly selective mechanism-inspired
inhibitors. J Biol Chem 280: 25313–25322.
Major ML, Lepe R, Costa RH. (2004). Forkhead box M1B
transcriptional activity requires binding of Cdk-cyclin complexes
for phosphorylation-dependent recruitment of p300/CBP coactiva-
tors. Mol Cell Biol 24: 2649–2661.
Marshall S. (2006). Role of insulin, adipocyte hormones, and nutrient-
sensing pathways in regulating fuel metabolism and energy homeo-
stasis: a nutritional perspective of diabetes, obesity, and cancer. Sci
STKE 2006: re7.
Marshall S, Bacote V, Traxinger RR. (1991). Discovery of a metabolic
pathway mediating glucose-induced desensitization of the glucose
transport system Role of hexosamine biosynthesis in the induction
of insulin resistance. J Biol Chem 266: 4706–4712.
Myatt SS, Lam EW. (2007). The emerging roles of forkhead box (Fox)
proteins in cancer. Nat Rev Cancer 7: 847–859.
Myatt SS, Lam EW. (2008). Targeting FOXM1. Nat Rev Cancer 8: 242.
O’Donnell N, Zachara NE, Hart GW, Marth JD. (2004). Ogt-
dependent X-chromosome-linked protein glycosylation is a requisite
modification in somatic cell function and embryo viability. Mol Cell
Biol 24: 1680–1690.
Park HJ, Costa RH, Lau LF, Tyner AL, Raychaudhuri P. (2008).
Anaphase-promoting complex/cyclosome-CDH1-mediated proteo-
lysis of the forkhead box M1 transcription factor is critical for
regulated entry into S phase. Mol Cell Biol 28: 5162–5171.
Reginato MJ, Mills KR, Becker EB, Lynch DK, Bonni A,
Muthuswamy SK et al. (2005). Bim regulation of lumen formation
in cultured mammary epithelial acini is targeted by oncogenes. Mol
Cell Biol 25: 4591–4601.
Reginato MJ, Mills KR, Paulus JK, Lynch DK, Sgroi DC, Debnath J
et al. (2003). Integrins and EGFR coordinately regulate the
pro-apoptotic protein Bim to prevent anoikis. Nat Cell Biol 5:
733–740.
Roos MD, Su K, Baker JR, Kudlow JE. (1997). O glycosylation of
an Sp1-derived peptide blocks known Sp1 protein interactions.
Mol Cell Biol 17: 6472–6480.
Rubinson DA, Dillon CP, Kwiatkowski AV, Sievers C, Yang L,
Kopinja J et al. (2003). A lentivirus-based system to func-
tionally silence genes in primary mammalian cells, stem cells
and transgenic mice by RNA interference. Nat Genet 33:
401–406.
Shaw RJ. (2006). Glucose metabolism and cancer. Curr Opin Cell Biol
18: 598–608.
Slawson C, Zachara NE, Vosseller K, Cheung WD, Lane MD, Hart
GW. (2005). Perturbations in O-linked beta-N-acetylglucosamine
protein modification cause severe defects in mitotic progression and
cytokinesis. J Biol Chem 280: 32944–32956.
Vosseller K, Wells L, Lane MD, Hart GW. (2002). Elevated
nucleocytoplasmic glycosylation by O-GlcNAc results in insulin
resistance associated with defects in Akt activation in 3T3-L1
adipocytes. Proc Natl Acad Sci USA 99: 5313–5318.
Wang IC, Chen YJ, Hughes D, Petrovic V, Major ML, Park HJ
et al. (2005). Forkhead box M1 regulates the transcriptional
network of genes essential for mitotic progression and genes
encoding the SCF (Skp2-Cks1) ubiquitin ligase. Mol Cell Biol 25:
10875–10894.
Wang Z, Banerjee S, Kong D, Li Y, Sarkar FH. (2007). Down-
regulation of Forkhead Box M1 transcription factor leads to the
inhibition of invasion and angiogenesis of pancreatic cancer cells.
Cancer Res 67: 8293–8300.
Warburg O. (1956). On the origin of cancer cells. Science 123:
309–314.
OGT required for breast tumorigenesis
SA Caldwell et al
11
Oncogene
Wells L, Vosseller K, Hart GW. (2001). Glycosylation of nucleocytoplasmic
proteins: signal transduction and O-GlcNAc. Science 291: 2376–2378.
Wonsey DR, Follettie MT. (2005). Loss of the forkhead transcrip-
tion factor FoxM1 causes centrosome amplification and mitotic
catastrophe. Cancer Res 65: 5181–5189.
Yakes FM, Chinratanalab W, Ritter CA, King W, Seelig S, Arteaga
CL. (2002). Herceptin-induced inhibition of phosphatidylinositol-3
kinase and Akt Is required for antibody-mediated effects on p27,
cyclin D1, and antitumor action. Cancer Res 62: 4132–4141.
Yang WH, Kim JE, Nam HW, Ju JW, Kim HS, Kim YS et al. (2006).
Modification of p53 with O-linked N-acetylglucosamine regulates
p53 activity and stability. Nat Cell Biol 8: 1074–1083.
Yang X, Ongusaha PP, Miles PD, Havstad JC, Zhang F, So WV et al.
(2008). Phosphoinositide signalling links O-GlcNAc transferase to
insulin resistance. Nature 451: 964–969.
Zhu W, Leber B, Andrews DW. (2001). Cytoplasmic O-glycosylation
prevents cell surface transport of E-cadherin during apoptosis.
EMBO J 20: 5999–6007.
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)
OGT required for breast tumorigenesis
SA Caldwell et al
12
Oncogene
