Carcinogenesis vol.30 no.10 pp.1768–1775, 2009
doi:10.1093/carcin/bgp196
Advance Access publication August 20, 2009
Nutlin-3, an Hdm2 antagonist, inhibits tumor adaptation to hypoxia by stimulating the
FIH-mediated inactivation of HIF-1a
Yoon-Mi Lee, Ji-Hong Lim, Yang-Sook Chun1, Hyo-Eun
Moon2, Myung Kyu Lee3, L.Eric Huang4 and Jong-Wan
Park
Department of Pharmacology, 1Department of Physiology and 2Department of
Neurosurgery, Ischemic/Hypoxic Disease Institute, Seoul National University
College of Medicine, 28 Yongon-dong, Chongno-gu, Seoul 110-799, Korea,
3
Omics and Integration Research Center, Korea Research Institute of
Bioscience and Biotechnology, 52 Eoundong, Yuseong, Daejeon 305-333,
Korea and 4Department of Neurosurgery, University of Utah School of
Medicine, Salt Lake City, UT 84132, USA
To whom correspondence should be addressed. Tel: þ82 2 740 8289;
Fax: þ82 2 745 7996;
Email: [email protected]
The interplay among hypoxia-inducible factor 1-alpha (HIF-1a),
p53 and human orthologue of murine double minute 2 (Hdm2)
has been introduced as a key event in tumor promotion and angiogenesis. Recently, nutlin-3, a small-molecule antagonist of
Hdm2, was demonstrated to inhibit the HIF-1-mediated vascular
endothelial growth factor production and tumor angiogenesis.
Yet, the mechanism by which nutlin-3 inhibits HIF-1 is an open
question. We here addressed the mode-of-action of nutlin-3 with
respect to the HIF-1a–p53–Hdm2 interplay. The effect of nutlin-3
on HIF-1a function was examined by reporter analyses, immunoprecipitation and immunoblotting. Nutlin-3 downregulated HIF-1a,
which occurred p53-dependently but von Hippel-Lindauindependently. On the contrary, nutlin-3 blunted the hypoxic
induction of vascular endothelial growth factor by inactivating
HIF-1 even in p53-null cells. The C-terminal transactivation domain (CAD) of HIF-1a was inactivated by nutlin-3, and furthermore, the factor-inhibiting hypoxia-inducible factor (FIH)
hydroxylation of Asn803 was required for the nutlin-3 action. In
terms of protein interactions, Hdm2 competed with FIH in CAD
binding and inhibited the Asn803 hydroxylation both in vivo and in
vitro, which facilitated p300 recruitment. Moreover, nutlin-3 reinforced the FIH binding and Ans803 hydroxylation by inhibiting
Hdm2. In conclusion, Hdm2 functionally activates HIF-1 by inhibiting the FIH interaction with CAD, and the Hdm2 inhibition
by nutlin-3 results in HIF-1 inactivation and vascular endothelial
growth factor suppression. The interplays among HIF-1a, Hdm2,
FIH and p300 could be potential targets for treating tumors
overexpressing HIF-1a.
Introduction
The p53 tumor suppressor protein is activated in response to diverse
stresses and transactivates numerous genes encoding proteins that
govern DNA repair, cell cycle arrest and apoptosis (1). Besides its
role as a transcription factor, p53 further controls cellular responses to
stress by interacting with various signaling molecules (2). In half of
the human tumors, the TP53 gene is either deleted or inactivated, and
the ectopic expression of p53 induces death or senescence of the
tumor cells (3). Moreover, even in the tumors harboring the wild-type
TP53 gene, p53 expression was often found to be suppressed due to
Abbreviations: aa, amino acid; CAD, C-terminal transactivation domain; CT,
C-terminal; EPO, erythropoietin; FIH, factor-inhibiting hypoxia-inducible factor; HA, hemagglutinin; Hdm2, human orthologue of murine double minute 2;
HIF-1a, hypoxia-inducible factor 1-alpha; NAD, N-terminal transactivation
domain; NES, nuclear export signal; NLS, nuclear localization signal; RT–
PCR, reverse transcription–polymerase chain reaction; SDS, sodium dodecyl
sulfate; VEGF, vascular endothelial growth factor; VHL, von Hippel-Lindau.
impairment of the p53-murine double minute 2 (Mdm2) autoregulatory feedback loop. The expression of human orthologue of murine
double minute 2 (Hdm2) is upregulated by p53 through the p53responsive promoter, and in turn Hdm2 ubiquitinates and destabilizes
p53 (4). Indeed, Hdm2 overexpression was found in many human
tumors, and by so doing, p53 was not accumulated even under genotoxic stress, resulting in continuous proliferation of tumor cells (5).
Therefore, the p53–Hdm2 interaction has been generally considered
as the very promising target for cancer therapy, and consequently
selective small-molecule antagonists of Hdm2, nutlins were recently
developed as anticancer agents. Nutlins disrupt the Hdm2 binding to
p53 by directly targeting the p53-binding pocket on the surface of
Hdm2, which in turn forcefully stabilizes p53 and activates p53dependent cell death pathways (6). In xenograft models of human
tumors, oral treatment of nutlin-3 showed an anticancer effect comparable with intravenous injection of doxorubicin (7).
Hypoxia-inducible factor 1-alpha (HIF-1a) belongs to the basichelix-loop-helix Per/Arnt/Sim family and plays a central role in tumor
adaptation to hypoxia by transactivating a variety of genes, such as
vascular endothelial growth factor (VEGF) and glycolytic enzymes
(8). HIF-1a forms the HIF-1 transcription factor complex by binding
with aryl hydrocarbon receptor nuclear translocator and provides
the prime transactivation domains for messenger RNA synthesis
(9). Under normoxic conditions, HIF-prolyl hydroxylases hydroxylate
HIF-1a. The hydroxylated HIF-1a is ubiquitinated by the von HippelLindau protein (pVHL) and subsequently degraded via 26S
proteasomes (10). Functionally, HIF-1a is also regulated by factorinhibiting hypoxia-inducible factor (FIH). FIH hydroxylates Asn803
in the C-terminal transactivation domain (CAD) of HIF-1a and inhibits p300 co-activator binding to HIF-1a, which leads to HIF-1
inactivation (11,12). Under hypoxic conditions, both proline and asparagine hydroxylation are inhibited due to limited oxygen substrate,
and then HIF-1a is stabilized and activated.
Recently, the molecular interaction among HIF-1a, p53 and Hdm2
has been introduced as a key event in tumor promotion. HIF-1a and
p53 affect each other reciprocally; namely, HIF-1a stabilizes p53 by
blocking the Hdm2-mediated p53 ubiquitination (13) but p53 destabilizes HIF-1a by promoting the Hdm2-mediated HIF-1a ubiquitination (14). According to this hypothesis, Hdm2 may function as
a negative regulator of HIF-1a. On the other hand, several recent
reports demonstrated that Hdm2 directly associates with HIF-1a
and then positively regulates HIF-1a expression and activation. Then,
if p53 and Hdm2 control HIF-1a in the opposite way, how is HIF-1a
regulated during p53 activation? Despite substantial expression of
Hdm2 by p53, HIF-1a is invariably found to be downregulated during
p53 activation, suggesting that p53 dominantly regulates HIF-1a.
Therefore, the Hdm2-mediated HIF-1a activation may manifest in
tumor cells lacking activated wild-type p53.
Until today, the detail interactions among HIF-1a, p53 and Hdm2
and their consequences in tumor promotion and angiogenesis
remained confusing and controversial. Then, how do Hdm2 antagonists affect HIF-1-mediated adaptation to hypoxia? Several recent
reports demonstrated that nutlin-3 inhibited HIF-1a and subsequently
inhibited VEGF production and angiogenesis in tumors (15–17). Yet,
the molecular mechanism underlying the HIF-1a inhibition remains
ambiguous. In the present study, we examined the mode-of-action of
nutlin-3 with respect to the HIF-1a–p53–Hdm2 interplay. Nutlin-3
downregulated HIF-1a in p53-positive cells, but also functionally
inactivated HIF-1a even in p53-negative cells. Of two mechanisms,
the latter mainly contributed to VEGF suppression by nutlin-3. Mechanistically, Hdm2 competed with FIH for binding to HIF-1a CAD and
then inhibited the hydroxylation of Asn803, leading to p300 recruitment. Nutlin-3 reinforced the FIH-mediated inactivation of HIF-1a by
Ó The Author 2009. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]
1768
HIF-1 repression by nutlin-3
inhibiting the interaction between Hdm2 and CAD. The inhibition of
the Hdm2–CAD interaction could be a potential strategy for treating
tumors that show aggressive behaviors by highly expressing HIF-1a.
Materials and methods
Reagents and antibodies
Culture media and fetal calf serum were purchased from Invitrogen (Carlsbad,
CA) and other chemicals including nutlin-3 from Sigma–Aldrich (St Louis,
MO). A polyclonal antibody against HIF-1a was raised in rabbits against
human HIF-1a [amino acids (aa) 418–698] (18), and a monoclonal antibody
against hydroxylated Asn803 of HIF-1a was a generous gift from Dr Myung
Kyu Lee (KRIBB, Daejeon, Korea) (19). Antibodies against aryl hydrocarbon
receptor nuclear translocator, p53, b-tubulin, Gal4 (DNA-binding domain) and
horseradish peroxidase-conjugated secondary antibodies were purchased from
Santa Cruz Biotechnology (Santa Cruz, CA). Anti-hemagglutinin (HA)-tag
and anti-Flag-tag antibodies were obtained from Roche Applied Science
(Indianapolis, IN) and Sigma–Aldrich, respectively.
Cell culture
HCT116 (colon cancer) and Hep3B (hepatoma) cell lines were obtained from the
American Type Culture Collection (Manassas, VA), and VHL-null and VHL
wild-type RCC4 (kidney cancer) cell lines from the European Collection of
Cell Cultures (London, UK). An HCT116 (p53/) cell line was generously
given by Dr Deug Y.Shin (Dankook University College of Medicine). Cells
were cultured in Dulbecco’s modified Eagle’s medium or a-modified Eagle’s
medium, supplemented with 10% heat-inactivated fetal calf serum. Gas tensions in the O2/CO2 incubator (Vision Science Co., Korea) were 20% O2/5%
CO2 for normoxic incubation or 1% O2/5% CO2 for hypoxic incubation.
Expression plasmids and transfection
The cDNA of Hdm2 (NM_002392) was cloned by reverse transcription–
polymerase chain reaction (RT–PCR) using Pfu DNA polymerase and inserted
into the pcDNA-Flag expression vector by blunt-end ligation. To remove nuclear localization signal (NLS) or nuclear export signals (NESs) from Hdm2,
we replaced the NLS sequence ‘RKRHK’ with ‘AKAHK’ (designated as
DNLS) or the NES sequence ‘LSFDESLAL’ with ‘LSFDESAAA’ (designated
as DNES) by site-directed mutagenesis. HA-tagged HIF-1a, HA-tagged p300,
erythropoietin (EPO)-enhancer luciferase, VEGF-promoter luciferase, Gal4CAD (aa 776–826 of HIF-1a), Gal4-CAD N803A mutant and VP16-p300 CH1
plasmids were constructed by Dr Eric Huang (University of Utah) (20). HAtagged FIH plasmid was constructed as described previously (21). To suppress
endogenous FIH-1 and Hdm2, synthesized silencing RNA duplexes were
obtained from Invitrogen. RNA sequences targeting FIH-1 and Hdm2 are
5#-CAGUUGCGCAGUUATAGCUUC-3# and 5#-GCCACAAAUCUGAUAGUAU-3#, respectively. Cells at 40% confluence in dishes (60 mm diameter)
were transfected with plasmids or silencing RNAs using the calcium phosphate
or Lipofectamine reagent (Invitrogen).
Cytosolic and nuclear extracts
After centrifugation at 1500g for 5 min at 4°C, the cell pellets were resuspended in a buffer A (20 mM Tris, pH 7.8, 1.5 mM MgCl2, 10 mM KCl,
0.2 mM ethylenediaminetetraacetic acid, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, protease inhibitor cocktail and 1 mM Na3VO4) and
cooled on ice. After re-suspended in 0.6% nonyl phenoxylpolyethoxylethanol
40, the lysed cells were centrifuged and the supernatants were collected as the
cytosolic fraction. Nuclear pellets were re-suspended in a buffer B, which was
prepared by adding NaCl (finally 400 mM) and glycerol (finally 5%) to the
buffer A. After launching on shaking incubator at 4°C for 30 min, the pellets
were centrifuged and the supernatant was collected as the nuclear fraction.
Immunoblotting and immunoprecipitation
For immunoblotting, total lysates or fractionated samples were electrophoresed
on 8–12% sodium dodecyl sulfate (SDS)/polyacrylamide gels and transferred
to an Immobilon-P membrane (Millipore, Bedford, MA). The membranes were
sequentially incubated with the primary antibodies (diluted 1:200–1000 in 5%
skim milk, overnight at 4°C) and the secondary antibodies conjugated with
horseradish peroxidase (diluted 1:5000 in 5% skim milk, 2 h at room temperature), and the immune complexes were visualized using the enhanced
luminol-based chemiluminescent-plus kit (GE Healthcare Bio-Sciences,
Piscataway, NJ). For immunoprecipitation, total lysates or fractionated samples were incubated with an antibody for precipitation for 4 h and then incubated with protein A/G-Sepharose beads (GE Healthcare Bio-Sciences)
overnight at 4°C. The precipitated proteins were eluted in the denaturing
SDS sample buffer and subjected to SDS–polyacrylamide gel electrophoresis
and immunoblotting.
In vitro binding and hydroxylation analyses
GST-tagged HIF-1a C-terminal [(CT), aa 577–826] peptide was expressed in
Escherichia coli BL21 cells and purified using GSH-affinity chromatography.
Full-length FIH and His-tagged Hdm2 peptides were purchased from ProSpecTany TechnoGene (Rehovot, Israel) and ProteinOne (Bethesda, MD), respectively. The purities (.92%) of recombinant proteins used were checked by
SDS–polyacrylamide gel electrophoresis and Coomassie Blue R-250 staining.
Purified GST-CT (0.5 lg) was immobilized on GSH beads and further incubated with purified FIH (0.3 lg) or/and His-Hdm2 (0.2 or 0.5 lg) at 30°C for
1 h. Peptides were eluted with an SDS sample buffer and identified by SDS–
polyacrylamide gel electrophoresis and immunoblotting. For the in vitro
Asn803 hydroxylation assay, GST-CT (0.1 lg) was incubated with FIH
(0.1 lg), His-Hdm2 (0.05 or 0.2 lg) or 20 lM nutlin-3 at 30°C for 1 h in
a reaction buffer containing 40 mM Tris–HCl, pH 7.4, 4 mM ascorbic acid,
4 mM 2-oxoglutarate, 1.5 mM FeSO4, 10 mM KCl and 3 mM MgCl2. The
hydroxylation at the Asn803 residue was analyzed by immunoblotting using an
anti-Asn803 (OH) antibody.
Reporter assays
Cells were co-transfected with 0.5–2 lg each of reporter (EPO-enhancer,
VEGF-promoter or Gal4-promoter luciferase) plasmid and cytomegalovirusb-gal plasmid using calcium phosphate or Lipofectamine reagent. The final
DNA concentrations were adjusted by adding pcDNA. After stabilized for
24 h, cells were incubated under either normoxic or hypoxic conditions, in
the absence or in the presence of nutlin-3 for 16 h, and then lysed to determine
luciferase and b-gal activities.
Semiquantitative RT–PCR
The messenger RNA levels were quantified using a highly sensitive semiquantitative RT–PCR, as described previously (18). Total RNAs, extracted using
TRIZOL (Invitrogen), were reverse transcribed at 46°C for 20 min and the
cDNAs were amplified over 16–18 PCR cycles (94°C for 30 s, 53°C for 30 s
and 70°C for 30 s) in a 20 ll of reaction mixture containing 5 lCi of
[a-32P]dCTP. The PCR products were electrophoresed on a 4% polyacrylamide
gel, and dried gels were autoradiographed. Primers for human VEGF-A,
aldolase-A, lactate dehydrogenase A, phosphoglycerate kinase 1, enolase 1
and b-actin were designed as described previously (18). The primer sequences
for human Bcl-2/adenovirus E1B 19-kDa-interacting protein 3 is 5#-GATACCCATAGCATTGGAGA-3# and 5#-CTAAATTAGGAACGCAGCAT-3#.
VEGF assay
VEGF protein levels in the medium were quantified using a human VEGF
enzyme-linked immunosorbent assay kit provided by Invitrogen, according
to the manufacturer’s instruction. The conditioned media obtained from
cultured HCT116 cells were incubated with VEGF antibody-coated microtiter
plates. After several washes, a second incubation was performed with an
enzyme-linked VEGF antibody.
Statistics
All data were analyzed using Microsoft Excel 2002 software, and results are
expressed as means and standard deviations. We used the unpaired Student’s
t-test to compare reporter activities between control and nutlin-3 treatment
groups. Differences were considered significant when P value was ,0.05.
All statistical tests were two-sided.
Results and discussion
p53-dependent and VHL-independent action of nutlin-3 in HIF-1a
suppression
In a recent report, the anti-angiogenic activity of nutlin-3 was evaluated both in cultured endothelial cells and in Matrigel sponges (16).
To evaluate the nutlin-3 activity in a different system, we loaded
nutlin-3 on the chick embryo chorioallantoic membranes and also
found that nutlin-3 effectively inhibited the capillary formation in
the membrane (supplementary Figure 1 is available at Carcinogenesis
Online). Therefore, nultin-3 is likely to have anti-angiogenic activity
in vivo as well as in vitro. To understand the mechanism by which
nultin-3 has anti-angiogenic activity, we examined the effects of
nutlin-3 on the HIF-1 activation and VEGF expression. Previously,
it was reported that p53 directly interacts with HIF-1a and by so doing
it promotes the Hdm2-mediated ubiquitination and proteasomal
degradation of HIF-1a (14). Nutlin-3 stabilizes p53 by blocking the
Hdm2–p53 interaction, and the increased p53 secondarily induces
Hdm2 expression by targeting the Hdm2 promoter (4). Since it
1769
Y.-M.Lee et al.
upregulates both p53 and Hdm2, nutlin-3 might stimulate the p53- and
Hdm2-mediated degradation of HIF-1a. In the present study, the
hypoxic levels of HIF-1a were higher in p53/ HCT116 cells than
those in p53þ/þ HCT116 cells, suggesting that p53 promotes HIF-1a
degradation. Also, nutlin-3 increased p53 and Hdm2 levels and decreased HIF-1a levels in p53-positive cells. However, nutlin-3 failed
to suppress HIF-1a in p53-negative cells (Figure 1A). To rule out the
possible involvement of pVHL in nutlin-3-induced suppression of
HIF-1a, VHLþ/þ and VHL/ RCC4 cells were treated with nutlin-3. As expected, HIF-1a was stabilized in VHL/ cells even
under normoxic conditions. However, nutlin-3 reduced HIF-1a expression with increased p53 and Hdm2, regardless of the presence
of pVHL (Figure 1B). These results support the previous reports
suggesting that nutlin-3 degrades HIF-1a in p53-dependent manner.
p53-independent action of nutlin-3 in VEGF and glycolytic enzymes
suppression
As the hypoxic induction of VEGF is a representative action of HIF-1,
we examined whether nutlin-3 blunts the VEGF response to hypoxia
p53-dependently. VEGF secretion increased during hypoxia and was
attenuated by nutlin-3 in p53þ/þ HCT116 cells. Unexpectedly,
nutlin-3 inhibited the level of secreted VEGF under either normoxic
or hypoxic conditions in both p53/ and p53þ/þ HCT116 cells
(Figure 1C). Furthermore, nutlin-3 inhibited the hypoxic inductions of
VEGF and glycolytic enzyme messenger RNAs regardless of the
expression of p53 (Figure 1D). To exclude the possibility that such
the effects of nutlin-3 resulted from cytotoxicity, we performed an
MTT analysis and found that nutlin-3 at 10 and 20 lM concentrations
did not affect cell viability (supplementary Figure 2 is available at
Carcinogenesis Online). These results suggest that the HIF-1dependent hypoxic gene regulation is attenuated by nutlin-3 via some
mechanisms other than the p53-dependent HIF-1a downregulation.
Nutlin-3 functionally inactivates HIF-1 p53-independently
To examine the transcriptional activity of HIF-1, we assayed the
activities of the EPO-enhancer and the VEGF-promoter, both of
which contain the HIF-1-targeted, hypoxia response elements.
Nutlin-3 significantly attenuated the hypoxic activation of HIF-1 in
both reporters, and these HIF-1-inhibitory actions of nutlin-3 occurred
in both p53/ and p53þ/þ cells (Figure 2A and B). To clarify that
the nutlin-3 inactivation of HIF-1 is attributed to the functional repression of HIF-1a, we analyzed the activity of HIF-1a CAD in two
p53-defective cell lines, p53/ HCT116 and Hep3B. Since the Gal4
(DNA-binding domain)-HIF-1a (CAD) chimeric protein level is not
affected by hypoxia or by nutlin-3 treatment (data not shown), the
Gal4-promoter assay using this protein specifically reflects the CAD
activity. In both cell lines, the CAD was activated under hypoxic
conditions and the hypoxic activation was noticeably inhibited by
nutlin-3 (Figure 2C). However, nutlin-3 did not affect the activity of
the CAD-N803A mutant, which is not regulated by FIH owing to lack
of Asn803 (supplementary Figure 3 is available at Carcinogenesis
Online). These results suggest that nutlin-3 inhibits the function of
HIF-1a CAD. In addition to CAD, HIF-1a has another transcriptional
activation domain in aa 531–575, which is named N-terminal
Fig. 1. Effects of nutlin-3 on HIF-1a expression and activity. (A) p53-dependent HIF-1a suppression. p53þ/þ or p53/ HCT116 cells were incubated under
normoxic or hypoxic (16 h) conditions in the absence or presence of nutlin-3 and lysed in a denaturing SDS sample buffer. Proteins were analyzed by
immunoblotting using specific antibodies. (B) pVHL-independent HIF-1a suppression. VHLþ/þ or VHL/ RCC4 cells were subjected to hypoxia and indicated
proteins were analyzed, as described in (A). (C) p53-independent reduction of VEGF production. p53þ/þ or p53/ HCT116 cells (2 105 cells) were cultured
in the 100 mm dish and incubated under normoxic or hypoxic conditions for 16 h. Nutlin-3 was administered into the media 1 h prior to hypoxic incubation. VEGF
was analyzed in collected media using an enzyme-linked immunosorbent assay kit and its concentration was calculated according to the curve plotted by standard
VEGF. Each bar represents the mean ± SD obtained from four separate experiments. P ,0.05 versus untreated normoxic or hypoxic control. (D) p53-independent
repression of HIF-1 target genes. p53þ/þ or p53/ HCT116 cells were treated with nutlin-3 and incubated under hypoxic conditions for 16 h. VEGF-A,
aldolase-A, lactate dehydrogenase A (LDH-A) and b-actin messenger RNAs were isolated and analyzed by semiquantitative RT–PCR and autoradiography.
1770
HIF-1 repression by nutlin-3
Fig. 2. Nutlin-3 represses HIF-1a CAD p53-independently. To examine the effect of nutlin-3 on transcriptional activity of HIF-1, EPO-enhancer (A) and VEGFpromoter (B) activities were analyzed using the luciferase reporter system. Luciferase reporter vector containing EPO-enhancer or VEGF-promoter was
co-transfected with b-gal plasmid into p53þ/þ or p53/ HCT116 cells. After 16 h incubation, luciferase activities were measured using a luminometer and were
normalized to b-gal activity. Results (means ± SDs, n 5 4) are presented as relative values versus the normoxic control. P , 0.05 versus the normoxic control;
#P , 0.05 versus the hypoxic control. (C) To examine whether nutlin-3 inhibits HIF-1 FIH-dependently, the Gal4-CAD plasmid was co-transfected with
Gal4-luciferase reporter plasmid and b-gal plasmid into p53/ HCT116 or p53-defective Hep3B cells. After stabilized for 24 h, the cells were incubated under
normoxic or hypoxic conditions for 16 h in the absence or presence of nutlin-3. Luciferase activities (means ± SDs, n 5 4) were normalized to b-gal activities and
presented as relative values versus the normoxic activity of wild-type Gal4-CAD. P , 0.05 versus the normoxic control; #P , 0.05 versus the hypoxic control.
(D) To examine whether or not nutlin-3 inhibits the activity of HIF-1a NAD, P53/ HCT116 cells were incubated under normoxic or hypoxic conditions for 16 h
in the absence or presence of nutlin-3, and messenger RNAs of the NAD-dependent genes (phosphoglycerate kinase 1, Bcl-2/adenovirus E1B 19-kDa interacting
protein 3 (BNIP3) and enolase 1] were analyzed by sensitive RT–PCR and autoradiography.
transactivation domain (NAD). NAD has transcriptional activity even
in moderate hypoxia because it is not controlled by FIH. In contrast,
CAD is inactivated by FIH in moderate hypoxia (22). Recently, it has
been reported that NAD and CAD express different genes in response
to the hypoxic gradient (23). Given this report, phosphoglycerate
kinase 1, enolase 1 and Bcl-2/adenovirus E1B 19-kDa-interacting
protein 3 were expressed specifically by NAD. To examine whether
nutlin-3 also inactivates NAD, we analyzed these NAD-regulated
genes. Although these genes were induced in response to hypoxia,
nutlin-3 could not inhibit the hypoxic induction (Figure 2D). Of two
transactivation domains, only CAD is likely to be targeted by nutlin-3.
Hdm2 is involved in the nutlin-3-mediated HIF-1 repression
Despite the physical interaction between Hdm2 and HIF-1a (15), the
consequence of the interaction remains controversial. Ravi et al. (14)
inhibited Hdm2 using a dominant-negative mutant and demonstrated
that Hdm2 mediated the p53-induced degradation of HIF-1a. In recent
two reports, however, ectopically expressed Hdm2 increased the intracellular level of HIF-1a in hypoxia and augmented the VEGF-promoter activity (24,25). Therefore, we checked the role of Hdm2 in
HIF-1a regulation in EPO-enhancer and Gal4 reporter systems. In
both reporters, Hdm2 expression significantly enhanced the luciferase
activities and also recovered the reporter activities repressed by nutlin3 (Figure 3A and B). These results suggest that Hdm2 acts as a positive
regulator of HIF-1a CAD. The CAD is known to require p300 co-
activator for its full activation, and the CH1 domain of p300 directly
binds to CAD (20). To examine whether Hdm2 and nutlin-3 functionally affect the CAD–p300 interaction, we performed a mammalian
two-hybrid assay using Gal4-CAD, CH1-VP16 and Gal4 reporter
plasmids. The CAD–p300 (CH1) binding was enhanced in hypoxia
and further increased by Hdm2 expression. The binding was inhibited
by nutlin-3 and fully recovered by Hdm2 expression (Figure 3C).
Interplay among HIF-1a, Hdm2, FIH and p300
To examine the effects of Hdm2 and nutlin-3 on the p300 recruitment by
HIF-1a, we precipitated p300 with anti-p300 antibody and found that
the co-precipitation of HIF-1a with p300 was enhanced by Hdm2 expression. Also, the co-precipitation was inhibited by nutlin-3 treatment
and recovered by Hdm2 expression (Figure 4A). Next, we examined the
role of endogenous Hdm2 in the CAD–p300 interaction using Hdm2silencing RNA. The CAD–p300 binding was enhanced by knockingdown FIH but noticeably attenuated by knocking-down Hdm2 (Figure
4B). To understand the mechanism underlying the Hdm2-enhanced
p300 binding to CAD, FIH was co-expressed with CAD, p300 and/
or Hdm2. FIH was found to bind to CAD, which was accompanied
with the dissociation between CAD and p300 (Figure 4C, lanes 1–3).
However, Hdm2 expression reduced the FIH binding to CAD and recovered the CAD–p300 binding. This effect of Hdm2 was abolished by
nutlin-3 treatment (Figure 4C, lanes 4 and 5). Since FIH is known to
block the CAD–p300 binding by hydroxylating Asn803 of CAD, we
1771
Y.-M.Lee et al.
Fig. 3. Involvement of Hdm2 in nutlin-3 inhibition of HIF-1. EPO-enhancer luciferase/b-gal (A) or wild-type Gal4-CAD/Gal4-luciferase/b-gal (B) plasmids
were co-transfected with the Flag-Hdm2 plasmid into p53/ HCT116 cells. After 16 h hypoxia, luciferase and b-gal activities (means ± SDs, n 5 4) were
measured. (C) Mammalian two-hybrid assay for CAD–p300 binding. p53/ HCT116 cells were co-transfected with plasmids of Gal4 [DNA-binding
domain (DBD)]-CAD (1 lg), Gal4-promoter luciferase reporter (1 lg) and VP16-p300 CH1 (0.5 lg) or Flag-Hdm2. The cells were incubated under
normoxic or hypoxic conditions for 16 h in the absence or presence of 20 lM nutlin-3 and then lysed to determine luciferase activity. Results are presented as
relative values versus the normoxic control and are plotted as means ± SDs of four experiments.
analyzed the hydroxylation state of Asn803 using a specific antibody.
FIH expression stimulated the hydroxylation of Asn803, which verifies
the specificity of the antibody against hydroxylated Asn803. The FIHinduced Asn803 hydroxylation was attenuated by Hdm2 expression
and was resumed by nutlin-3 (Figure 4D, left panel). Moreover, the
Asn803 hydroxylation by endogenous FIH was also stimulated by
nutlin-3 treatment (Figure 4D, right panel).
Hdm2 inhibits Asn803 hydroxylation by directly competing with FIH
in CAD binding
To confirm the action of Hdm2 in the FIH repression of HIF-1a, we
analyzed the FIH–CAD interaction and the Asn803 hydroxylation
in vitro using recombinant peptides. As expected, FIH directly bound
to HIF-1a CT. However, Hdm2 could also bind to CT and by so doing
dissociated FIH from CT in a dose-dependent manner. Moreover,
nutlin-3 inhibited the Hdm2 binding to CT, which in turn resumed
the FIH–CT interaction (Figure 5A). In the in vitro hydroxylation
assay, the Asn803 hydroxylation was induced by FIH incubation
and this was inhibited by Hdm2 co-incubation. Nutlin-3 treatment
reversed the inhibitory action of Hdm2 on Asn803 hydroxylation
(Figure 5B). Therefore, Hdm2 is likely to directly compete with
FIH in CAD binding and Asn803 hydroxylation.
Subcellular localization of Hdm2 and FIH
Since FIH is present in the cytoplasm, the hydroxylation of HIF-1a
Asn803 is believed to occur in the cytoplasm before HIF-1a enters
into the nucleus. Then to compete with FIH for CAD binding, Hdm2
1772
should coexist with FIH in the cytoplasm. Since Hdm2 is detected
predominantly in the nucleus, it is regarded as a nuclear protein.
However, Hdm2 can shuttle between the nucleus and the cytoplasm
by its NLS and NES sequences (26,27), and the Hdm2 shuttling is
regulated by phosphoinositide 3-kinases/AKT and mitogen-activated
protein kinase pathways (28,29). In the strict sense, therefore, Hdm2
is localized both in the nucleus and in the cytoplasm. Indeed, Hdm2
was also detected in the cytosolic fraction though it was substantially
in the nuclear fraction (Figure 6A). As expected, FIH was identified to
be exclusively expressed in the cytoplasm, and these localizations
were not altered by hypoxia. To understand where Hdm2 interacts
with CAD, we examined co-immunoprecipitation of Gal4-CAD and
Hdm2 in the cytosolic and nuclear fractions. Although both Gal4CAD and Hdm2 levels were obviously lower in the cytosolic fraction
versus the nuclear fraction, the Hdm2–Gal4-CAD complex level was
much higher in the cytosolic fraction (Figure 6B). To re-check the
place that CAD, FIH and Hdm2 interplay, we co-expressed wild-type
Hdm2, Hdm2DNLS confined in the cytoplasm, or Hdm2DNES confined in the nucleus (Figure 6C, upper panel) with the reporter plasmids for CAD activity. Compared with wild-type Hdm2, Hdm2DNLS
significantly enhanced CAD activity, but Hdm2DNES did not (Figure
6C, lower panel). These results suggest that the interplays among
CAD, FIH and Hdm2 occur mainly in the cytoplasm.
Proposed mode-of-action of HIF-1 inhibition by nutlin-3
Based on these results, we here propose a new hypothesis regarding
the mechanism by which nutlin-3 functionally inhibits HIF-1
HIF-1 repression by nutlin-3
Fig. 5. In vitro analyses for the FIH–CAD interaction and the Asn803
hydroxylation. (A) Hdm2 inhibits the FIH binding to CAD. Purified
recombinant GST-tagged HIF-1a CT, full-length FIH and His-tagged
full-length Hdm2 peptides were co-incubated, and GST-CT was pulled-down
using glutathione beads. Pulled-down peptides and input levels were
analyzed by western blotting. (B) Hdm2 inhibits the Asn803 hydroxylation
by FIH. Recombinant GST-CT, FIH and His-Hdm2 peptides were
co-incubated in a hydroxylation buffer 30°C for 1 h, and the
Asn803-hydroxylated GST-CT was analyzed using a monoclonal antibody
recognizing hydroxylated Asn803. Peptide levels in the reaction mixture
were verified by western blotting.
Fig. 4. Competition of Hdm2 with FIH in CAD binding. (A) Effect of Hdm2
expression on p300 binding to HIF-1a. HEK293 cells were transfected with
1 lg each of HA–HIF-1a and HA–p300 or/and Flag-Hdm2 plasmid. After
48 h stabilization, cells were subjected to hypoxia for 16 h in the absence or
presence of nutlin-3. HA–p300 was immunoprecipitated with pre-immune
serum (IgG) or anti-p300 antiserum and protein G/A beads. Precipitated
HA–p300 and co-precipitated HA–HIF-1a were identified using anti-p300
and anti-HIF-1a antibodies, respectively. (B) Effect of Hdm2 knock-down on
p300 binding to HIF-1a. HEK293 cells were transfected with 20 nM each of
FIH-silencing RNA or/and Hdm2-silencing RNA. After stabilized for 48 h,
cells were incubated in hypoxia for 8 h and then harvested. p300 was
immunoprecipitated with anti-p300 and protein G/A beads. Co-precipitated
p300 and HIF-1a were identified using anti-p300 and anti-HIF-1a
antibodies. (C) Interactions among CAD, p300, FIH and Hdm2. HEK293
cells were transfected with 1 lg each of Gal4-CAD, HA–p300, HA–FIH or/
and Flag-Hdm2 plasmid and then subjected to 8 h hypoxia. Gal4-CAD was
immunoprecipitated with anti-Gal4 antiserum and protein G/A beads.
Co-precipitated proteins were identified using specific antibodies.
(D) Hydroxylation of Asn803 in CAD. HEK293 cells were transfected with
1 lg each of Gal4-CAD, HA–FIH or/and Flag-Hdm2 plasmid and then
subjected to 8 h hypoxia in the absence or presence of nutlin-3 (5–20 lM).
Expressed proteins and Asn803 hydroxylation of Gal4-CAD were analyzed
by western blotting.
accordingly (11). Later, FIH has been characterized as a 2-oxoglutarate-dependent dioxygenase that requires molecular oxygen for
hydroxylating the b-carbon on Asn803 (30). Therefore, in the absence
of oxygen, HIF-1a escapes from the negative regulation of FIH. However, since FIH has a higher affinity for oxygen, its activity is not as
sensitive to the oxygen tension as prolyl hydroxylases are (22). In
moderate hypoxia, HIF-1a becomes stabilized owing to prolyl
hydroxylase inhibition, but it remains inactive because FIH still retains its activity. When the oxygen tension drops below the critical
level for FIH activity, HIF-1a becomes fully activated. Previously,
FIH has been known to have its activity at 1% oxygen (21). In the
present study, we also used 1% oxygen for hypoxia and thus the FIH
activity can be reinforced by nutlin-3. Indeed, this is supported by our
results demonstrating that the FIH expression induced the Asn803
hydroxylation and the p300 dissociation from CAD. However, when
HCT116 cells were exposed to severe hypoxia (0.3% oxygen), both
EPO-enhancer and Gal4-CAD activities were not inhibited by nutlin3 (supplementary Figure 4 is available at Carcinogenesis Online).
Therefore, it is expected that the proposed mode-of-action of nutlin3 occurs only in moderate hypoxia.
(Figure 6D). Namely, under hypoxic conditions, Hdm2 competes with
FIH for HIF-1a–CAD binding and then prevents the hydroxylation
of Asn803, which enhances p300 recruitment by CAD and leads
to transcription-on of hypoxia-induced genes. When treated with nutlin-3, the interaction between Hdm2 and CAD is dissociated. Then,
the chance of FIH–CAD binding increases and Asn803 of CAD becomes hydroxylated, which in turn blocks the p300 recruitment and
leads to transcription-off of hypoxia-induced genes.
Hdm2–HIF-1a binding
Several research groups have reported that HIF-1a forms a complex
with Hdm2. Regarding which domain of HIF-1a binds Hdm2, LaRusch et al. (15) suggested that the N-terminus of HIF-1a is required
for the interaction with Hdm2. Given that Phe19 of p53 participates in
the p53–Hdm2 binding, they replaced Phe37 of HIF-1a with tyrosine
and found that the F37Y mutant was not co-precipitated with Hdm2.
However, we here demonstrated that the CAD (aa 776–826), which is
present at the CT of HIF-1a, was sufficient for Hdm2 binding using
the Gal4-CAD fusion protein. Taken together with these results, it is
possible that the N-terminus and the CT separately interacts with
Hdm2 or that both HIF-1a domains cooperatively work for Hdm2
binding. To further understand the protein interaction, a detail peptide
mapping for Hdm2 binding or a 3D structure analysis of the
HIF-1a–Hdm2 complex should be performed.
FIH activation in hypoxia
Is it possible that FIH have the hydroxylation activity under oxygendeficient conditions? FIH was first identified as an HIF-1a CADinteracting protein by yeast two-hybrid screening and was named
Hdm2–CAD interaction as an anticancer target
HIF-1a has been disclosed to be positively associated with angiogenesis, local invasion, metastasis and treatment failure in cancer diseases
(31). Therefore, HIF-1a is viewed as a tumor marker of poor
1773
Y.-M.Lee et al.
interaction may be a potential target for cancer therapy. Also,
nutlin-3, which disrupts the Hdm2–CAD interaction and stimulates
FIH to inhibit CAD, could be a lead compound in developing antiHIF, anticancer drugs.
Supplementary material
Supplementary Figures 1–4 can be found at http://carcin.oxfordjournals.
org/.
Funding
Korea Research Foundation (2008-E00054); Bone Metabolism
Research Center (R11-2008-023-02001-0) provided by Korea Science
and Engineering Foundation.
Acknowledgements
Conflict of Interest Statement: None declared.
References
Fig. 6. Hdm2–CAD interaction in the cytoplasm and proposed mechanism
underlying HIF-inhibitory action of nutlin-3. (A) Subcellular localization of
Hdm2 and FIH. p53/ HCT116 cells were incubated in normoxia or
hypoxia for 16 h and then homogenated to fractionate cytosolic and nuclear
proteins. Protein levels were analyzed by western blotting. Lamin B and
b-tubulin were blotted as markers for nuclear and cytosolic fractions,
respectively. (B) The Hdm2–CAD interaction occurs mainly in the
cytoplasm. From p53/ HCT116 cells transfected with the Gal4-CAD
plasmid, cytosolic and nuclear proteins were extracted. Gal4-CAD was
immunoprecipitated with pre-immune serum (IgG) or anti-Gal4 antiserum
and protein G/A beads. Co-precipitated Hdm2 were identified by western
blotting with anti-Hdm2 antibody. (C) HIF-1a CAD activation by
cytoplasmic Hdm2. Wild-type Gal4-CAD, Gal4-luciferase and b-gal
plasmids were co-transfected with Flag-Hdm2, Flag-Hdm2DNLS or
Flag-Hdm2DNES plasmid into p53/ HCT116 cells. After 16 h hypoxia,
luciferase and b-gal activities (means ± SDs, n 5 4) were measured.
(D) Proposed mode-of-action of HIF-1 inhibition by nutlin-3. Hdm2
activates HIF-1a by competing FIH for CAD binding (upper panel). Nutlin-3
antagonizes the Hdm2–CAD interaction and FIH inhibits HIF-1a by
hydroxylating Asn803.
prognosis and a promising therapeutic target. To date, many small
molecules have been identified to deregulate the expression of HIF1a, and some of these are currently developed as potential drugs for
cancer therapy (32). Besides HIF-1a suppression, the functional inhibition of HIF-1a might be also a strategy for cancer therapy. Indeed,
chetomin, which inhibits p300 binding to HIF-1a, was identified to
repress the transcriptional activity of HIF-1a and showed the anticancer activity in xenografted human tumors (33). Bortezomib, a proteasome inhibitor used for multiple myeloma therapy, was also identified
to functionally repress HIF-1a by reinforcing the FIH-mediated CAD
inhibition and inhibited angiogenesis in multiple myeloma and solid
tumors (34). In this aspect, the HIF-1a activation by Hdm2–CAD
1774
1. Riley,T. et al. (2008) Transcriptional control of human p53-regulated genes.
Nat. Rev. Mol. Cell Biol., 9, 402–412.
2. Haupt,S. et al. (2003) Apoptosis—the p53 network. J. Cell Sci., 116,
4077–4085.
3. Petitjean,A. et al. (2007) TP53 mutations in human cancers: functional
selection and impact on cancer prognosis and outcomes. Oncogene, 26,
2157–2165.
4. Mayo,L.D. et al. (2002) The PTEN, Mdm2, p53 tumor suppressoroncoprotein network. Trends Biochem. Sci., 27, 462–467.
5. Vargas,D.A. et al. (2003) Mdm2: a regulator of cell growth and death. Adv.
Cancer Res., 89, 1–34.
6. Vassilev,L.T. (2007) MDM2 inhibitors for cancer therapy. Trends Mol.
Med., 13, 23–31.
7. Vassilev,L.T. et al. (2004) In vivo activation of the p53 pathway by smallmolecule antagonists of MDM2. Science, 303, 844–848.
8. Gordan,J.D. et al. (2007) Hypoxia-inducible factors: central regulators of
the tumor phenotype. Curr. Opin. Genet. Dev., 17, 71–77.
9. Jiang,B.H. et al. (1996) Dimerization, DNA binding, and transactivation
properties of hypoxia-inducible factor 1. J. Biol. Chem., 271, 17771–
17778.
10. Schofield,C.J. et al. (2004) Oxygen sensing by HIF hydroxylases. Nat. Rev.
Mol. Cell Biol., 5, 343–354.
11. Mahon,P.C. et al. (2001) FIH-1: a novel protein that interacts with HIF1alpha and VHL to mediate repression of HIF-1 transcriptional activity.
Genes Dev., 15, 2675–2686.
12. Lando,D. et al. (2002) Asparagine hydroxylation of the HIF transactivation
domain a hypoxic switch. Science, 295, 858–861.
13. An,W.G. et al. (1998) Stabilization of wild-type p53 by hypoxia-inducible
factor 1alpha. Nature, 392, 405–408.
14. Ravi,R. et al. (2000) Regulation of tumor angiogenesis by p53-induced
degradation of hypoxia-inducible factor 1alpha. Genes Dev., 14, 34–44.
15. LaRusch,G.A. et al. (2007) Nutlin3 blocks vascular endothelial growth
factor induction by preventing the interaction between hypoxia inducible
factor 1alpha and Hdm2. Cancer Res., 67, 450–454.
16. Secchiero,P. et al. (2007) Antiangiogenic activity of the MDM2 antagonist
nutlin-3. Circ. Res., 100, 61–69.
17. Binder,B.R. (2007) A novel application for murine double minute 2 antagonists: the p53 tumor suppressor network also controls angiogenesis. Circ.
Res., 100, 13–14.
18. Chun,Y.S. et al. (2001) A new HIF-1 alpha variant induced by zinc
ion suppresses HIF-1-mediated hypoxic responses. J. Cell Sci., 114,
4051–4061.
19. Lee,S.H. et al. (2008) Monoclonal antibody-based screening assay for
factor inhibiting hypoxia-inducible factor inhibitors. J. Biomol. Screen.,
13, 494–503.
20. Gu,J. et al. (2001) Molecular mechanism of hypoxia-inducible factor
1alpha-p300 interaction. A leucine-rich interface regulated by a single
cysteine. J. Biol. Chem., 276, 3550–3554.
HIF-1 repression by nutlin-3
21. Yeo,E.J. et al. (2006) Amphotericin B blunts erythropoietin response to
hypoxia by reinforcing FIH-mediated repression of HIF-1. Blood, 107,
916–923.
22. Pouysségur,J. et al. (2006) Hypoxia signalling in cancer and approaches to
enforce tumour regression. Nature, 441, 437–443.
23. Dayan,F. et al. (2006) The oxygen sensor factor-inhibiting hypoxiainducible factor-1 controls expression of distinct genes through the bifunctional transcriptional character of hypoxia-inducible factor-1alpha. Cancer
Res., 66, 3688–3698.
24. Bárdos,J.I. et al. (2004) Growth factor-mediated induction of HDM2
positively regulates hypoxia-inducible factor 1alpha expression. Mol. Cell.
Biol., 24, 2905–2914.
25. Nieminen,A.L. et al. (2005) Mdm2 and HIF-1alpha interaction in tumor
cells during hypoxia. J. Cell. Physiol., 204, 364–369.
26. Roth,J. et al. (1998) Nucleo-cytoplasmic shuttling of the hdm2 oncoprotein
regulates the levels of the p53 protein via a pathway used by the human
immunodeficiency virus rev protein. EMBO J., 17, 554–564.
27. O’Keefe,K. et al. (2003) Nucleocytoplasmic shuttling of p53 is essential for
MDM2-mediated cytoplasmic degradation but not ubiquitination. Mol.
Cell. Biol., 23, 6396–6405.
28. Ozes,O.N. et al. (2001) A phosphatidylinositol 3-kinase/Akt/mTOR pathway mediates and PTEN antagonizes tumor necrosis factor inhibition of
insulin signaling through insulin receptor substrate-1. Proc. Natl Acad. Sci.
USA, 98, 4640–4645.
29. Jackson,M.W. et al. (2006) Hdm2 nuclear export, regulated by insulin-like
growth factor-I/MAPK/p90Rsk signaling, mediates the transformation of
human cells. J. Biol. Chem., 281, 16814–16820.
30. Koivunen,P. et al. (2004) Catalytic properties of the asparaginyl hydroxylase (FIH) in the oxygen sensing pathway are distinct from those of its
prolyl 4-hydroxylases. J. Biol. Chem., 279, 9899–9904.
31. Semenza,G.L. (2002) HIF-1 and tumor progression: pathophysiology and
therapeutics. Trends Mol. Med., 8, S62–S67.
32. Belozerov,V.E. et al. (2005) Hypoxia inducible factor-1: a novel target for
cancer therapy. Anticancer Drugs, 16, 901–909.
33. Kung,A.L. et al. (2004) Small molecule blockade of transcriptional coactivation of the hypoxia-inducible factor pathway. Cancer Cell, 6, 33–43.
34. Shin,D.H. et al. (2008) Bortezomib inhibits tumor adaptation to hypoxia by
stimulating the FIH-mediated repression of hypoxia-inducible factor-1.
Blood, 111, 3131–3136.
Received March 5, 2009; revised July 4, 2009; accepted July 30, 2009
1775