Prolyl hydroxylase-1 negatively regulates I?B
kinase-?, giving insight into hypoxia-induced
Eoin P. Cummins*, Edurne Berra†, Katrina M. Comerford*, Amandine Ginouves†, Kathleen T. Fitzgerald*,
Fergal Seeballuck*, Catherine Godson*, Jens E. Nielsen*, Paul Moynagh*, Jacques Pouyssegur†,
and Cormac T. Taylor*‡
*Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Belfield, Dublin 4, Ireland; and†Institute of Signaling,
Developmental Biology, and Cancer Research, Centre National de la Recherche Scientifique, Unite ´ Mixte de Recherche 6543, University of Nice,
Centre Antoine Lacassagne, 33 Avenue Valombrose, 06189 Nice, France
Edited by Laurie H. Glimcher, Harvard Medical School, Boston, MA, and approved October 4, 2006 (received for review March 22, 2006)
Hypoxia is a feature of the microenvironment of a growing tumor.
The transcription factor NF?B is activated in hypoxia, an event that
has significant implications for tumor progression. Here, we dem-
onstrate that hypoxia activates NF?B through a pathway involving
activation of I?B kinase-? (IKK?) leading to phosphorylation-
dependent degradation of I?B? and liberation of NF?B. Further-
more, through increasing the pool and/or activation potential of
IKK?, hypoxia amplifies cellular sensitivity to stimulation with
TNF?. Within its activation loop, IKK? contains an evolutionarily
conserved LxxLAP consensus motif for hydroxylation by prolyl
hydroxylases (PHDs). Mimicking hypoxia by treatment of cells with
siRNA against PHD-1 or PHD-2 or the pan-prolyl hydroxylase
sion of PHD-1 decreases cytokine-stimulated NF?B reporter activ-
activity of NF?B. Hypoxia increases both the expression and activ-
ity of IKK?, and site-directed mutagenesis of the proline residue
(P191A) of the putative IKK? hydroxylation site results in a loss of
hypoxic inducibility. Thus, we hypothesize that hypoxia releases
repression of NF?B activity through decreased PHD-dependent
hydroxylation of IKK?, an event that may contribute to tumor
development and progression through amplification of tumori-
genic signaling pathways.
a drop in pO2(1), a condition that should prove rate limiting for
tumor growth. However, growth of the tumor is paradoxically
involving activation of the hypoxia-specific transcriptional regula-
tor, hypoxia inducible factor-1 (HIF-1), which regulates the expres-
erythropoesis (2). Furthermore, hypoxia also activates NF?B (3), a
transcription factor important in the promotion and progression of
tumor development and survival (4). The net result of HIF-1 and
NF?B activation by hypoxia is increased tumor oxygenation and
survival/growth, respectively. Thus, inhibition of these pathways
represents a potentially important therapeutic window of oppor-
tunity in the treatment of cancer.
The mechanisms by which HIF-1 is activated in hypoxia are
relatively well understood (5). HIF-1? is constitutively synthesized
at a high level in normoxia, but its level is repressed by members of
prolyl hydroxylases (PHDs). Three PHD isoforms have been de-
scribed to date (PHD-1, PHD-2, and PHD-3). Oxygen-dependent
modification of specific proline residues within consensus LxxLAP
motifs (P402 and P564) in HIF-1? by these enzymes, primarily the
PHD-2 isoform, results in the targeting of HIF-1? for ubiquitina-
tion through an E3 ligase complex initiated by the binding of the
uring cancer progression, a state of hypoxia occurs as the
developing tumor outgrows the local blood supply, leading to
Von Hipple Lindau protein (pVHL) and subsequent proteasomal
domain of HIF-1? by Factor Inhibiting HIF (FIH), an asparagine
hydroxylase, represents a second mechanism of oxygen-dependent
repression through inhibition of transactivation (6). Similar mech-
anisms exist for HIF-2? (5). The hypoxic sensitivity of the HIF
pathway is achieved by the absolute requirement of PHD enzymes
for molecular oxygen as a cosubstrate. Inhibition of this pathway in
hypoxia with the resultant stabilization and transactivation of
HIF-? subunits represents a paradigm for oxygen sensing and
hypoxia-responsive alterations in gene expression (7).
NF?B-like HIF-1 is a prosurvival transcription factor implicated
in tumorigenesis through increasing the expression of genes that
inhibit apoptosis and growth arrest in premalignant cells and
promote tumor progression through production of cytokines (4).
The mechanisms of NF?B activation have been best characterized
for their role in inflammation in response to a host of proinflam-
matory ligands (e.g., TNF?, IL-1, and lipopolysaccharide). A
complex sequence of events resulting from receptor occupation by
these ligands triggers cascades involving a diverse array of adaptor
molecules and enzymes that are ligand-specific (8). There is,
it is clear that hypoxia activates NF?B, the signaling pathways
remain unclear. In the current study, we sought to interrogate the
signaling pathways leading to NF?B activation in hypoxia. We
investigated whether similar oxygen-sensing mechanisms that exist
for HIF-1 may also exist for NF?B. Specifically, we investigated a
role for the PHDs.
Hypoxia Activates NF?B. Consistent with previous studies, hypoxia
activated NF?B (10, 11). Hypoxia stimulated a 1.9 ? 0.1- and 5.9 ?
1.7-fold increase in basal NF?B-dependent transcriptional activity
at 24 and 48 h, respectively, as measured by the NF?B-dependent
luciferase reporter assay (Fig. 1A). Furthermore, this event is
preceded by increased levels of nuclear NF?B (as demonstrated by
p65 binding to immobilized oligonucleotides containing the con-
Author contributions: E.P.C., K.M.C., C.G., P.M., J.P., and C.T.T. designed research; E.P.C.,
E.B., K.M.C., A.G., K.T.F., F.S., and J.E.N. performed research; J.E.N. contributed new
reagents/analytic tools; E.P.C., E.B., F.S., P.M., and J.P. analyzed data; and E.P.C. and C.T.T.
wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
Freely available online through the PNAS open access option.
Abbreviations: HIF, hypoxia inducible factor; PHD, prolyl hydroxylase.
‡To whom correspondence should be addressed at: Conway Institute of Biomolecular and
Biomedical Research, School of Medicine and Medical Science, College of Life Sciences,
University College Dublin, Belfield, Dublin 4, Ireland. E-mail: email@example.com.
© 2006 by The National Academy of Sciences of the USA
November 28, 2006 ?
vol. 103 ?
sensus NF?B response element) detectable 4 h after exposure to
Fig. 1B). Next, cells were exposed to graded hypoxia (21%, 10%,
5%, 3%, and 1% O2) in separate ambient atmospheric hypoxia
chambers. Using oxygen quenching oxymetry, we determined
extracellular oxygen tensions at the level of the monolayer to be
105 ? 2.9, 67 ? 1.6, 23 ? 1.0, 9 ? 1.0, and 2 ? 0.2 mmHg (1
mmHg ? 133 Pa), respectively (Fig. 1C). DNA binding activity of
p65 and HIF-1 in nuclear fractions was measured in the same cell
lysates and was related to oxymetry readings. Whereas the peak of
p65 DNA binding in hypoxia occurred at a cell pO2of 9.2 ? 0.9
lower pO2value (1.7 ? 0.2 mmHg; Fig. 1E).
TNF?-Induced NF?B Nuclear Binding Is Enhanced in Hypoxia. To
determine possible functional consequences for our observa-
to a stimulus of the NF?B pathway, TNF?. HeLa cells were
exposed to a preconditioned hypoxic medium (1% O2for 1 h)
before treatment with low doses of TNF? (0.01–0.1 ng/ml for
1 h). Nuclear lysates were prepared, and DNA-binding assay was
carried out. As predicted, TNF?-dependent activation of the
NF?B pathway was significantly enhanced in cells in hypoxia
(P ? 0.05; Fig. 2).
Hypoxia Activates NF?B Through IKK. The IKK complex is the
convergence point for many diverse NF?B-activating stimuli in-
dependent NF?B activation involves this pathway. Cells were
exposed to a preconditioned hypoxic medium (1% O2), and the
phosphorylation status of IKK?/? was determined by Western blot
analysis. Hypoxia results in a rapid phosphorylation of the IKK?/?
complex detectable after 5 min, which is temporally upstream of
I?B? phosphorylation on S32 and S36 (maximal after 15 min) and,
in turn, is temporally upstream of I?B? degradation (Fig. 3A). The
transient nature of I?B? phosphorylation is consistent with reports
(12). Critically, these measurements were made in the same cell
lysates. Thus, hypoxia likely activates NF?B through IKK-
dependent mechanisms. Abolition of HIF-1 expression in HeLa
cells using siRNA did not alter hypoxia-induced I?B? phosphory-
lation (Fig. 3B), indicating that hypoxia-induced activation of the
NF?B pathway is independent of HIF-1?.
PHD Inhibition Stimulates NF?B Activity. Prolyl hydroxylases are
central to oxygen-sensing pathways leading to HIF-1? activation
as a result of their absolute requirement upon molecular oxygen
as a cofactor for enzymatic activity (2). The above data led us to
the hypothesis that decreased PHD activity in hypoxia may also
underlie NF?B activation. To further test this, we examined
components of the NF?B pathway for the presence of the
previously described, conserved LxxLAP motif for proline hy-
droxylation in HIF-1? (13). Whereas p65, p50, I?B?, and
NEMO (IKK?) are without this sequence, both IKK? and IKK?
contain the motif (Fig. 4A). This motif is evolutionarily con-
served between humans, mice, and zebrafish. Interestingly, this
motif is adjacent to the sites of phosphorylation of IKK?
(177/181) on the activation loop, making it likely accessible to
modification by enzymes such as PHDs.
(B) Nuclear extracts from HeLa cells exposed to preconditioned hypoxic medium (1% O2; 4 h) demonstrate increased NF?B DNA binding. (C) HeLa cells exposed
to graded hypoxia (21–1% O2; 24 h) demonstrate decreased extracellular pO2values as measured by fluorescence quenching oxymetry and increased NF?B and
HIF-1? DNA binding activity (D and E).
additional hour) demonstrates enhanced NF?B-nuclear binding when com-
pared with normoxic controls.
NF?B DNA-binding assay in HeLa cells exposed to preconditioned
Cummins et al.
November 28, 2006 ?
vol. 103 ?
no. 48 ?
To investigate the possibility that PHDs are playing a role in
NF?B activation in hypoxia, we used PHD isoform-specific
siRNA (14). Using this approach, we effectively knocked down
each of the PHD-1, -2, and -3 isoforms by 73.3%, 80.3%, and
93.6%, respectively. Quantitation of PHD knockdown as mea-
sured by quantitative RT-PCR is demonstrated in Fig. 4B.
Knockdown of each individual PHD isoform was without effect
on expression levels of the other isoforms as measured by
quantitative RT-PCR. HeLa cells transfected with an HIF-1-
dependent luciferase reporter construct demonstrated sensitiv-
ity to PHD-2 knockdown as previously reported (ref. 13 and data
not shown). Interestingly, cells transfected with an NF?B-
dependent reporter construct demonstrated the greatest re-
porter activity in cells deficient in the PHD-1 isoform, indicating
HIF-1 and NF?B pathways (Fig. 4C). Conversely, artificial
transient overexpression of PHD-1 in normoxia causes a de-
crease in TNF?-stimulated NF?B activity (Fig. 4D).
HeLa cells transfected with an NF?B-luciferase reporter con-
struct were treated in normoxia with the pan-hydroxylase inhibitor
fold increase in NF?B reporter activity compared with vehicle-
treated control cells (P ? 0.05, n ? 4; Fig. 4E). To assess whether
course (0–75 min) and demonstrated phosphorylation of S32/36
residues of I?B?, which was maximal after a 30-min exposure (Fig.
4F), thus indicating that PHD inhibition increases IKK activity.
Collectively, these data indicate a repressive role for PHD-1 in
NF?B signaling. Furthermore, PHD inhibition by DMOG (1 mM,
24 h) or PHD1 siRNA caused a significant increase in the expres-
sion of COX-2, an important NF?B-dependent inflammatory
marker (Fig. 4 G and H).
Hypoxia Alters the Cellular Pool of IKK? and IKK??. HIF-1 stability
is controlled by the PHD-dependent hydroxylation of specific
residues in the CODD and NODD regions. The presence of an
HIF-? LxxLAP consensus motif in IKK? led us to examine the
expression of these proteins under hypoxia. In line with previous
studies (15), HeLa cells exposed to graded hypoxia for 24 h
demonstrated increased IKK? protein levels, whereas IKK?/
NEMO (which does not contain the hydroxylation motif) did not
increase (Fig. 5A). We next investigated the impact of hypoxia on
IKK? expression in more physiologically relevant cell lines of
immune and epithelial origin. Similar to HeLa cells, IKK? levels
were increased by graded hypoxia in THP-1 monocytic cells and in
CaCo-2 intestinal epithelial cells in response to instantaneous
hypoxia (1% O2) (Fig. 5B).
Because IKK? has been described as the primary positive
regulator of NF?B activity in inflammatory processes and as the
molecular link between inflammation and cancer (16–18), we
focused our subsequent studies on this isoform. To assess the
role of the proline residue within the LxxLAP motif of IKK?
(P191), we used a site-directed mutagenesis strategy to mutate
P191 to an alanine residue. HeLa cells were transiently trans-
fected with wild-type IKK? or P191A IKK? DNA. Forty-eight
hours after transfection, cells were exposed to normoxia or
preconditioned hypoxic medium (3% O2for 2 h; pO2? 19.9
mmHg). Cytosolic extracts were prepared, and IKK? expression
levels were compared by Western blot analysis. Cells transfected
with the wild-type vector demonstrated elevated levels of IKK?
in hypoxia. Cells transfected with the P191A IKK? vector
demonstrated a loss of hypoxic inducibility (Fig. 5C). This is
suggestive of a key role for P191 in the hypoxic regulation of
Interestingly, digital residue replacement modeling indicates
that a modeled IKK peptide containing the LxxLAP motif has
the characteristic boomerang structure of the HIF-1 CODD
sequence, which is also subject to prolyl hydroxylation by PHDs
(19). Furthermore, based on the crystal structure of the HIF-1
CODD peptide interaction with VHL (19), we were able to
model an IKK?-pVHL interaction without having to drastically
modify the pVHL structure (data not shown).
Coimmunoprecipitation studies demonstrate an interaction be-
tween VHL and IKK? in normoxic cells (Fig. 5D); however, this
does not appear to result in ubiquitination or proteasomal degra-
dation of IKK? (data not shown). Furthermore, coimmunoprecipi-
tation studies reveal the interaction between PHD1 and both IKK
? and ? isoforms. Although this indirectly indicates that IKK is
hydroxylated, further critical experimentation using mass spectro-
metric analysis will be necessary to directly demonstrate hydroxy-
lation. A distinct possibility is that inhibition of IKK activity by
hydroxylation of P191 may be due to steric hindrance of the
phosphorylation of the nearby residues S177 and S181 within the
may prevent activation through the blockage of the serine phos-
phorylation residues on the activation loop. Thus, we propose that
hydroxylation represses IKK? activity by altering protein expres-
sion or by preventing its activation through phosphorylation on
PHD enzymes are critical in sensing and transducing the HIF-
dependent transcriptional response to hypoxia, an event clearly
demonstrating an absolute requirement for molecular oxygen as a
cosubstrate (7, 20). Thus, PHDs are true oxygen/hypoxia sensors.
HIFs with NF?B also demonstrating hypoxic sensitivity and im-
pacting the resultant hypoxic transcriptome (3). However, to date,
(5–120 min) demonstrate temporally sequential activation of IKK?/?, phos-
phorylation of I?B?, and degradation of I?B?. (B) HeLa cells transfected with
siRNA against HIF-1? or nontarget siRNA exposed to hypoxia (1% O2; 24 h)
demonstrate effective HIF-1? knockdown. siRNA against HIF-1? does not
prevent hypoxia-induced phosphorylation of I?B?.
Hypoxia activates NF?B through the IKK complex. (A) HeLa cells
www.pnas.org?cgi?doi?10.1073?pnas.0602235103Cummins et al.
our understanding of PHD involvement in hypoxic signaling has
been restricted mainly to the HIF-1 pathway. Because of its role in
tumor progression, NF?B has important pathophysiologic implica-
tions in cancer, a condition involving significant tissue hypoxia.
Hypoxia-activated NF?B is demonstrable by the nuclear accu-
mulation of the p65 subunit as well as by reporter gene assay and
correlates with decreased cellular pO2. Furthermore, hypoxia-
induced NF?B activity is preceded by temporally sequential IKK
activation, I?B phosphorylation, and I?B degradation, indicating
that hypoxia activates NF?B through increased IKK activity. The
IKK-dependence of hypoxia-induced NF?B activity is consistent
with most other stimuli of this pathway. Although previous studies
have hypothesized roles for tyrosine phosphorylation (21), reactive
oxygen species generation (22), and p42/44 or PI-3-kinase (23) in
mediating NF?B responses in hypoxia, none of these theories
clearly demonstrates a direct relationship between decreased mo-
lecular oxygen and NF?B activation such as through altered PHD
activity. Our model proposes that while hypoxia activates NF?B it
also increases the sensitivity of the NF?B pathway to activation
through proinflammatory stimuli, such as cytokines. Thus, the
importance of this pathway may be greatest where hypoxia occurs
against the backdrop of increased inflammatory activity [such as in
may heavily affect tumor-promoting gene expression.
Alternative signaling targets for proline hydroxylation in the
transcriptional response to hypoxia are attractive in developing
the theory that hydroxylating enzymes are central to cellular
oxygen sensing (7, 20). In the current study, we demonstrate that
IKK?, a critical regulator of NF?B activity, contains a conserved
prolyl hydroxylation consensus site homologous with the two
conserved sites within the oxygen-dependent degradation do-
main (ODD) of HIF-1 (LxxLAP). Like HIF-1?, the pool of
IKK? in cells is increased in hypoxia; however, unlike HIF?, the
absolute expression of IKKs is not determined by the presence
of oxygen because basal expression is clearly detectable. Thus, it
appears that hypoxia is more a modulator of IKK expression and
activation than an absolute determinant. This may be due to the
presence of just a single LxxLAP motif in IKK? rather than
the two present in HIF-1?. A recent study has demonstrated the
requirement of a leucine residue within 10 residues of the
LxxLAP motif for hydroxylation of the consensus motif (25).
Two leucine residues reside closely downstream of the LxxLAP
motif in IKK?. Importantly, the putative site of hydroxylation
(P191) on IKK? resides just 10 residues from the known
S181 is positioned in the vicinity of P191 in the well conserved
protein kinase activation loop, and P191 itself is predicted to be
positioned in the substrate binding cleft. Experimental studies
(26) of a prototypic protein kinase (the cAMP-dependent pro-
tein kinase, PKA) show that the conformation and dynamics of
the activation loop are known to be linked with the phosphor-
ylation of S181. Thus, it is possible that hydroxylation of P191
alters the conformation of the activation loop, making phos-
phorylation difficult or ineffective (i.e., the kinase might be
phosphorylated but inactive) and/or impairs the binding of the
protein kinase substrate. This would cause a decrease in IKK?
(and subsequently NF?B) activity. Similarly, it is also conceiv-
able that phosphorylation renders hydroxylation more difficult
cells were cotransfected with isoform-specific PHD siRNAs (20 nM), the reporter vector (pNF?B-LUC), and 100 ng of a ?-galactosidase construct. As a control for
measured (37). Results are expressed as the fold induction over control. (D) HeLa cells were transiently cotransfected with pEGLN2-FLAG (PHD1; 0.15–0.3 ?g) or
empty pcDNA (0.15–0.3 ?g) vector and pNF?B-LUC. After transfection, cells were treated with TNF? (0.1 ng/ml) for 24 h. Whole-cell lysates were prepared, and
a luciferase assay was carried out. Results are protein-normalized RLU values expressed as fold over basal luciferase activity. (E) HeLa cells were transiently
(H) Three PHD isoforms were knocked down in HeLa cells as described above, and Cox-2 expression was determined by Western blot analysis.
PHDs suppress NF?B activity. (A) IKK? and IKK? but not NEMO (not shown) contain conserved LxxLAP motifs. (B) HeLa cells transiently transfected with
Cummins et al.
November 28, 2006 ?
vol. 103 ?
no. 48 ?
because the hydroxylation site in active IKK? could be masked
by the substrate. Additionally, the activation loop would be less
dynamic and hence less accessible to PHDs.
The role of IKK? in oncogenesis and inflammation is well
established, but there is emerging evidence that IKK? may have
antiinflammatory functions (27, 28). We also observed an increase
in IKK? in hypoxia (data not shown). The observation that both
the possibility of a counterbalancing mechanism where a p65-
mediated response could be resolved over a course of hypoxia.
in activation of NF?B-dependent signaling in normoxia. Mutation
of the hydroxylated proline residues of HIF-1? results in a loss of
hypoxic inducibility (29). Similarly, mutation of proline residue 191
in IKK?, which resides within the consensus sequence, results in a
PHD-2, NF?B activation appears to be more sensitive to silencing
of the PHD-1 isoform, suggesting somewhat differential regulation
to HIF-1, which is predominantly regulated by PHD-2. These data
through an inhibition of protein hydroxylation, resulting in evasion
of degradation in a similar manner to HIF-1. Indeed, it is worth
noting that a number of studies have demonstrated that the E3
pVHL suppresses NF?B activity in normoxia (30–32). However,
although IKK? interacts with VHL in normoxia, we found no
kinetics of IKK? activation in hypoxia suggests the likelihood that
VHL masking of the activation loop.
Thus, hypoxia increases the expression and activation of IKK,
leading to increased sensitivity of cells to inflammatory stimuli
such as cytokines. This has important implications in cancer
where elevated levels of proinflammatory cytokines coexist with
hypoxia in the microenvironment of the tumor (33). Further-
more, this is consistent with previous studies demonstrating
synergy between hypoxia and inflammatory cytokines in the
activation of cells (10, 34).
activity, increases the activity of cellular IKK?, an event that
amplifies the cellular capacity for response to cytokines. Fur-
thermore, we believe that such events may directly impact on
tumor development through enhanced expression of genes that
promote tumor development and growth.
Cell Culture and Hypoxia. HeLa, CaCo-2, and THP-1 cells were
placed in one of four hypoxia chambers (Coy Laboratories, Grass
Lake, MI; Ruskin Technologies, Leeds, U.K.) allowing the estab-
lishment of graded, humidified, ambient, atmospheric hypoxia of
21%, 10%, 5%, 3%, and 1% O2with 5% CO2and a balance of N2
in all cases. Extracellular pO2measurements were made by using
fluorescence quenching oxymetry (Oxylite-2000; Oxford Optronix,
Oxford, U.K.). Hypoxia did not induce apoptosis or necrosis (data
instantaneous hypoxia was achieved by exposure of cells in hypoxia
chambers to preconditioned media.
Western Blot Analysis. Cytoplasmic or whole-cell lysates were sep-
arated by SDS/PAGE, transferred to nitrocellulose membranes,
and immunoblotted as described (35). IKK?, IKK?, and phospho-
IKK? (S180)/IKK? (S181) polyclonal antibodies (Cell Signaling),
IKK?/NEMO polyclonal antibody (Santa Cruz Biotechnology),
I?B? polyclonal antibody (Upstate Biotechnology), FLAG poly-
hypoxia (1% O2) for 0–6 h. Whole-cell extracts were immunoblotted for IKK?, NEMO, and ?-actin where indicated. (C) HeLa cells were transiently transfected with 1
?g of wild-type IKK? or P191A mutant IKK?. Forty-eight hours after transfection, the cells were maintained in normoxia (N; 21% O2) or exposed to preconditioned
by using densitometric analysis (n ? 4). (D) pVHL was immunoprecipitated from whole HeLa cell extracts (500 or 1,000 ?g of total protein), and immunoprecipitates
were immunoblotted for IKK? and hexokinase (similarly sized negative control). (E) NETN lysates from HeLa cells transiently overexpressing PHD-1-FLAG were
immunoprecipitated by using a specific anti-FLAG resin. Immunoprecipitates were immunoblotted for IKK?, IKK?, FLAG, and hexokinase.
www.pnas.org?cgi?doi?10.1073?pnas.0602235103 Cummins et al.
clonal antibody (Sigma), Hexokinase polyclonal antibody (Biogen- Download full-text
esis), Cox-2 polyclonal antibody (Cayman Chemical), phospho-
I?B? monoclonal antibody (Cell Signaling), and HIF-1?
monoclonal antibody (BD Transduction Laboratories) were used.
NF?B-DNA Binding Assays. Nuclear lysates for DNA-binding assays
were prepared as per the manufacturer’s instructions (TransAm
kit; ActiveMotif). The nuclear extract was incubated with an
immobilized oligonucleotide on a 96-well plate (containing a
specific transcription factor binding site as indicated) and de-
tected by the transcription factor ELISA.
Generation of IKK? Mutant by Site-Directed Mutagenesis. The
P191A IKK? mutant was generated by using a site-directed
mutagenesis kit described previously (36). Plasmid constructs
were verified by using DNA sequencing. Primers designed for
IKK? were as follows: forward, 5?-CCCTGCAGTACCTGGC-
CGCAGAGCTACTGGAGC-3?; reverse, 5?-GCTCCAGTAG-
Transient Transfection. Transient transfection of plasmid DNA
was performed in HeLa cells (50–70% confluent) by using
FuGENE 6 transfection reagent (Roche) according to the
manufacturer’s instructions. After transfection, cells were ex-
posed overnight at 37°C (5% CO2). Medium was replaced, and
cells were exposed to experimental treatment as described.
Reporter Assay. Cells were transfected with an NF?B-luciferase
construct (Stratagene Cis-Reporting Systems) with a synthetic
promoter containing a concatomer of five NF?B response
elements. Cell lysates were prepared by using a 1? luciferase
lysis buffer (Promega). Substrate was added to the cell lysate,
and luciferase activity was measured in a luminometer. Exper-
iments were carried out in duplicate or triplicate, and luciferase
values were normalized to cotransfected CMV Renilla luciferase,
?-Gal, or protein control values.
RNA Interference by siRNA. HeLa cells were grown to 40–50%
confluency and transfected with 5 nM specific siRNA against
HIF-1? or control nontarget siRNA (Dharmacon) by using a
Lipofectamine 2000 transfection reagent, according to the man-
ufacturer’s instructions. Cells were maintained in antibiotic-free
media for 48 h after transfection to achieve maximal knockdown
of the target gene. RNA interference by siRNA against PHD
isoforms was carried out as described (14) and quantified by
quantitative RT-PCR (Applied Biosystems).
Coimmunoprecipitation Studies. Whole HeLa cell lysates were pre-
pared in a lysis buffer as described above. Lysate was precleared
with 50 ?l of protein A/G PLUS agarose beads (Santa Cruz
Biotechnology) and rotated for 1 h at 4°C. Five hundred micro-
grams and 1,000 ?g of lysate were incubated with 2 ?g of an
anti-VHL antibody (BD Pharmingen) and brought to a final
volume of 650 ?l with lysis buffer. The lysis buffer (650 ?l) ?
anti-VHL antibody as well as the nonimmunoprecipitated whole-
cell extracts were included as controls (?Ab, ?Ab, and ?Ctrl,
respectively). Samples were rotated for 2 h at room temperature.
Thirty microliters of protein A/G PLUS agarose beads was added
times in lysis buffer. The supernatant was discarded, and the pellet
was resuspended in reducing sample buffer (65 ?l) and boiled for
10 min before electrophoresis.
HeLa cells were transiently transfected with pEGLN2-FLAG.
Whole-cell lysates were prepared in NETN buffer [20 mM
Tris?HCl (pH 8), 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet
P-40, protease inhibitor mixture]. Five hundred micrograms and
1,000 ?g of protein lysate were incubated with 30 ?l of anti-
FLAG agarose affinity gel (Sigma) and brought to a final volume
of 500 ?l with an NETN buffer. NETN buffer alone (Lysis) was
incubated with the beads as a control, in addition to a nonim-
munoprecipitated NETN lysate (?Ctrl). Samples were rotated
overnight at 4°C. Samples were centrifuged at 17,968 ? g for 6
min at 4°C. Beads were washed three times in NETN buffer. The
supernatant was discarded, and the pellet was resuspended in
reducing sample buffer (65 ?l) and boiled for 10 min before
Statistical Analysis. All data are presented as mean ? SEM for n
independent experiments. Statistical significance was evaluated
by using one-way ANOVA or Student’s t test for unpaired or
paired data where an asterisk corresponds to P ? 0.05.
We thank Anne Marie Griffin for expert technical assistance. pEGLN2
plasmids were a kind gift from Dr. William Kaelin (Harvard Medical
School, Boston, MA). This work was supported by grants from Science
Foundation Ireland, the Health Research Board of Ireland, the Well-
come Trust, the Government of Ireland Programme for Research in
Third Level Institutions, Centre National de la Recherche Scientifique,
and La Ligue Nationale Contre le Cancer.
1. Harris AL (2002) Nat Rev Cancer 2:38–47.
2. Schofield CJ, Ratcliffe PJ (2004) Nat Rev Mol Cell Biol 5:343–354.
3. Cummins EP, Taylor CT (2005) Pflu ¨gers Arch 450:363–371.
4. Karin M, Greten FR (2005) Nat Rev Immunol 5:749–759.
5. Berra E, Ginouves A, Pouyssegur J (2006) EMBO Rep 7:41–45.
6. Lando D, Peet DJ, Whelan DA, Gorman JJ, Whitelaw ML (2002) Science 295:858–
7. Schofield CJ, Ratcliffe PJ (2005) Biochem Biophys Res Commun 338:617–626.
8. Moynagh PN (2005) J Cell Sci 118:4589–4592.
9. Yamamoto Y, Gaynor RB (2004) Trends Biochem Sci 29:72–79.
10. Taylor CT, Dzus AL, Colgan SP (1998) Gastroenterology 114:657–668.
11. Ryan S, Taylor CT, McNicholas WT (2005) Circulation 112:2660–2667.
12. Hoffmann A, Levchenko A, Scott ML, Baltimore D (2002) Science 298:1241–1245.
13. Huang J, Zhao Q, Mooney SM, Lee FS (2002) J Biol Chem 277:39792–39800.
14. Berra E, Benizri E, Ginouves A, Volmat V, Roux D, Pouyssegur J (2003) EMBO J
15. Chen Y, Shi G, Xia W, Kong C, Zhao S, Gaw AF, Chen EY, Yang GP, Giaccia AJ,
Le QT, Koong AC (2004) Cancer Res 64:7302–7310.
16. Li ZW, Chu W, Hu Y, Delhase M, Deerinck T, Ellisman M, Johnson R, Karin M
(1999) J Exp Med 189:1839–1845.
17. Greten FR, Eckmann L, Greten TF, Park JM, Li ZW, Egan LJ, Kagnoff MF, Karin
M (2004) Cell 118:285–296.
18. Pikarsky E, Porat RM, Stein I, Abramovitch R, Amit S, Kasem S, Gutkovich-Pyest
E, Urieli-Shoval S, Galun E, Ben-Neriah Y (2004) Nature 431:461–466.
19. Hon WC, Wilson MI, Harlos K, Claridge TD, Schofield CJ, Pugh CW, Maxwell PH,
Ratcliffe PJ, Stuart DI, Jones EY (2002) Nature 417:975–978.
20. Kaelin WG (2005) Annu Rev Biochem 74:115–128.
21. Koong AC, Chen EY, Giaccia AJ (1994) Cancer Res 54:1425–1430.
22. Chandel NS, Trzyna WC, McClintock DS, Schumacker PT (2000) J Immunol
23. Zampetaki A, Mitsialis SA, Pfeilschifter J, Kourembanas S (2004) FASEB J 18:1090–
24. Semenza GL (2003) Nat Rev Cancer 3:721–732.
25. Kageyama Y, Koshiji M, To KK, Tian YM, Ratcliffe PJ, Huang LE (2004) FASEB
26. Wu J, Jones JM, Nguyen-Huu X, Ten Eyck LF, Taylor SS (2004) Biochemistry
27. Lawrence T, Bebien M, Liu GY, Nizet V, Karin M (2005) Nature 434:1138–1143.
28. Li Q, Lu Q, Bottero V, Estepa G, Morrison L, Mercurio F, Verma IM (2005) Proc
Natl Acad Sci USA 102:12425–12430.
29. Hagen T, Taylor CT, Lam F, Moncada S (2003) Science 302:1975–1978.
30. An J, Fisher M, Rettig MB (2005) Oncogene 24:1563–1570.
31. An J, Rettig MB (2005) Mol Cell Biol 25:7546–7556.
32. Qi H, Ohh M (2003) Cancer Res 63:7076–7080.
33. Balkwill F, Mantovani A (2001) Lancet 357:539–545.
34. Li X, Kimura H, Hirota K, Kasuno K, Torii K, Okada T, Kurooka H, Yokota Y,
Yoshida H (2005) Kidney Int 68:569–583.
35. Taylor CT, Fueki N, Agah A, Hershberg RM, Colgan SP (1999) J Biol Chem
36. Comerford KM, Leonard MO, Karhausen J, Carey R, Colgan SP, Taylor CT (2003)
Proc Natl Acad Sci USA 100:986–991.
Cummins et al.
November 28, 2006 ?
vol. 103 ?
no. 48 ?