© 1999 Macmillan Magazines Ltd
letters to nature
20 MAY 1999
Neurosci. 9, 448–459 (1997).
5. Xu, Q., Alldus, G., Hlder, N. & Wilkinson, D. G. Expression of truncated Sek-1 receptor tyrosine
kinase disrupts the segmental restriction of gene expression in the Xenopus and zebraﬁshhindbrain.
Development 121, 4005– 4016 (1995).
6. Xu, Q. & Wilkinson, D. G. Eph-related receptors and their ligands: mediators of contact dependent
cell interactions. J. Mol. Med. 75, 576– 586 (1997).
7. Flanagan, J. G. &Vanderhaeghen, P.The ephrins and Eph receptors in neural development. Annu. Rev.
Neurobiol. 21, 309–345 (1998).
8. O’Leary, D. D. M. & Wilkinson, D. G. Eph receptors and ephrins in neural development. Curr. Opin.
Neurobiol. 9, 65–73 (1999).
9. Holland, S. J. et al. Bidirectional signalling through the Eph-family receptor Nuk and its transmem-
brane ligands. Nature 383, 722–725 (1996).
10. Bruckner, K., Pasquale, E. B. & Klein, R. Tyrosine phosphorylation of transmembrane ligands for Eph
receptors. Science 275, 1640–1643 (1997).
11. Gale, N. W. et al. Eph receptors and ligands comprise two major speciﬁcity subclasses, and are
reciprocally compartmentalised during embryogenesis. Neuron 17, 9–19 (1996).
12. Eph Nomenclature Committee. Uniﬁed nomenclature for Eph family receptors and theirligands, the
ephrins. Cell 90, 403–404 (1997).
13. Ellis, C. et al. A juxtamembrane autophosphorylation site in the Eph family receptor tyrosine kinase,
Sek, mediates high afﬁnity interaction wi th p59fyn. Oncogene 12, 1727–1736 (1996).
14. Theil, T. et al. Segmental expression of the EphA4 (Sek-1) gene is under direct transcriptional control
of Krox-20. Development 125, 443–452 (1998).
15. Woo, K. & Fraser, S. E. Order and coherence in the fate map of the zebraﬁsh nervous system.
Development 121, 2595– 2609 (1995).
16. Townes, P. L. & Holfreter, J. Directed movements and selective adhesion of embryonic amphibian
cells. J. Exp. Zool. 128, 53– 120 (1955).
17. Steinberg, M. S. Does differential adhesion govern self-assembly processes in histogenesis? Equili-
brium processes and the emergence of a hierarchyamong populations of embryonic cells. J. Exp. Zool.
173, 395–434 (1970).
18. Nose, A., Nagafuchi, A. & Takeichi, M. Expressed recombinant cadherins mediate cell sorting in
model systems. Cell 54, 993–1001 (1988).
19. Friedlander, D. R., Mege, R. M., Cunningham,B. A. & Edelman, G. M. Cell sorting-out is modulated
by both the speciﬁcity and amount of different cell adhesion molecules. Proc. Natl Acad. Sci. USA 86,
20. Godt, D. & Tepass, U. Drosophila oocyte localization is mediated by differential cadherin-based
adhesion. Nature 395, 387–391 (1998).
21. Gonzalez-Reyes, A. & St Johnston, D. The Drosophila A-P axis is polarised by the cadherin-mediated
positioning of the oocyte. Development 125, 3635– 3644 (1998).
22. Meima, L. et al. AL-1-induced growth cone collapse of rat cortical neurons is correlated with REK7
expression and rearrangement of the actin cytoskeleton. Eur. J. Neurosci. 9, 177–188 (1997).
23. Zisch, A. H. et al. Tyrosine phosphorylation of L1 family adhesion molecules: implication of the Eph
kinase Cek5. J. Neurosci. Res. 47, 655– 665 (1997).
24. Lumsden,A. &Keynes, R. Segmentalpatterns ofneuronal development inthe chick hindbrain. Nature
337, 424–428 (1989).
25. Heyman, I., Kent, A. & Lumsden, A. Cellular morphology and extracellular space at rhombomere
boundaries in the chick embryo hindbrain. Dev. Dynamics 198, 241– 253 (1993).
26. Xu, Q., Hlder, N., Patient, R. & Wilson, S. W.Spatially regulated expression of three receptor tyrosine
kinase genes during gastrulation in the zebraﬁsh. Development 120, 287–299 (1994).
27. Irving, C., Nieto, M. A., DasGupta, R., Charnay, . & Wilkinson, D.G. Progressive spatial restriction of
Sek-1 and Krox-20 gene expression during hindbrain segmentation. Dev. Biol. 173, 26–38 (1996).
Acknowledgements. We thank N. Gale and G. Yancopoulos for ephrin clones, M. Henkemeyer for the
EphB2 clone and S. Fraser, J. P. Vincent, R. Krumlauf and P. Trainor for discussions. This work was
supported by the MRC, an EC Biotechnology grant and an EMBO Fellowship to G.M.
Correspondence and requests for materials should be addressed to D.G.W. (e-mail: firstname.lastname@example.org.
The tumour suppressor
protein VHL targets
hypoxia-inducible factors for
Patrick H. Maxwell*, Michael S. Wiesener*,
Gin-Wen Chang*, Steven C. Clifford†, Emma C. Vaux‡,
Matthew E. Cockman‡, Charles C. Wykoff‡,
Christopher W. Pugh‡, Eamonn R. Maher†
& Peter J. Ratcliffe*‡
*Wellcome Trust Centrefor Human Genetics, WindmillRoad, OxfordOX3 7BN,UK
†Section of Medical and Molecular Genetics, Department of Paediatrics and Child
Health, University of Birmingham, Birmingham B15 2TT, UK
‡Institute of Molecular Medicine, John Radcliffe Hospital, Oxford OX3 9DS, UK
Hypoxia-inducible factor-1 (HIF-1) has a key role in cellular
responses to hypoxia, including the regulation of genes involved
in energy metabolism, angiogenesis and apoptosis1–4. The a
subunits of HIF are rapidly degraded by the proteasome under
normal conditions, but are stabilized by hypoxia5. Cobaltous ions
or iron chelators mimic hypoxia, indicating that the stimuli may
interact through effects on a ferroprotein oxygen sensor6,7. Here
we demonstrate a critical role for the von Hippel-Lindau (VHL)
tumour suppressor gene product pVHL in HIF-1 regulation. In
VHL-defective cells, HIF a-subunits are constitutively stabilized
and HIF-1 is activated. Re-expression of pVHL restored oxygen-
dependent instability. pVHL and HIF a-subunits co-immunopre-
cipitate, and pVHL is present in the hypoxic HIF-1 DNA-binding
complex. In cells exposed to iron chelation or cobaltous ions,
HIF-1 is dissociated from pVHL. These ﬁndings indicate that the
interaction between HIF-1 and pVHL is iron dependent, and
that it is necessary for the oxygen-dependent degradation of
HIF a-subunits. Thus, constitutive HIF-1 activation may underlie
the angiogenic phenotype of VHL-associated tumours. The
pVHL/HIF-1 interaction provides a new focus for understanding
cellular oxygen sensing.
Enhanced glucose metabolism and angiogenesis are classical
features of cancer8,9, involving upregulation of genes that are
normally induced by hypoxia. In addition to stimulation by the
hypoxic microenvironment10, genetic alterations contribute to these
effects8,9. A striking example is von Hippel-Lindau (VHL) disease, a
hereditary human cancer syndrome predisposing sufferers to highly
angiogenic tumours11. Constitutive upregulation of hypoxically
inducible messenger RNAs encoding vascular endothelial growth
factor (VEGF) and glucose transporter 1 (GLUT-1) in these tumour
cells is correctable by re-expression of pVHL. A post-transcriptional
mechanism has been proposed12,13. We studied the involvement of
pVHL in oxygen-regulated gene expression using ribonuclease
protection analysis of two VHL-deﬁcient renal carcinoma lines,
RCC4 and 786-O. Eleven genes encoding products involved in
glucose transport, glycolysis, high-energy phosphate metabolism
and angiogenesis were examined; nine are induced by hypoxia in
other mammalian cells and two (LDH-B and PFK-M) are repressed
by hypoxia. None of these responses was seen in the VHL-defective
cell lines. Responses to hypoxia were restored by stable transfection
of a wild-type VHL gene, with effects ranging from a rather modest
action of hypoxia (PFK-L and LDH-B) to substantial regulation
(Fig. 1 shows results for RCC4 cells; similar effects were seen in 786-
O cells, data not shown). These results indicate that the previously
described upregulation of hypoxia-inducible mRNAs in VHL-
defective cells12,13 extends to a broad range of oxygen-regulated
genes and involves a constitutive ‘hypoxia pattern’ for both posi-
tively and negatively regulated genes.
As a number of these genes (VEGF, GLUT-1, AK-3, ALD-A,
PGK-1, PFK-L and LDH-A) contain hypoxia-response elements
(HREs) which bind HIF-1 and/or show altered expression in cells
lacking HIF-1 (refs 2, 14 and references therein), this survey of
expression in VHL-defective cells prompted us to look for effects of
pVHL on HIF-1 and HRE function. RCC4 cells were co-transfected
with reporter plasmids which did or did not contain HREs from the
mouse phosphoglycerate-kinase-1 or erythropoietin (Epo) enhan-
cers, and either the VHL-expression plasmid pcDNA3-VHL or an
empty vector. pVHL suppressed HRE activity in normoxic cells and
restored induction by hypoxia (Fig. 2a). Similar results were
obtained by sequential stable transfection of RCC4 cells with an
HRE reporter followed by pcDNA3-VHL (data not shown). HIF-1
itself was examined by electrophoretic mobility shift assay (EMSA),
which showed a constitutive HIF-1 DNA-binding species in VHL-
deﬁcient RCC4 cells, with restoration of the normal hypoxia-
inducible pattern in RCC4 cells stably transfected with pcDNA3-
VHL (RCC4/VHL) (Fig. 2b). In other cells, HIF-1 activation by
hypoxia involves a large increase in HIF-1aabundance from low
basal levels in normoxia1,15. Western blotting of whole-cell extracts
showed that RCC4 cells express constitutively high levels of both
HIF-1aand a related molecule, HIF-2a(also termed EPAS-1, HRF,
HLF and MOP2) which is normally regulated in a similar way16
(Fig. 2c). Constitutively high levels of these proteins were found in
eight other VHL-defective cell lines, in contrast to the renal
© 1999 Macmillan Magazines Ltd
letters to nature
20 MAY 1999
carcinoma line Caki-1 (which expresses pVHL normally17) and a
wide range of previously reported cell lines (Fig. 2d and Supple-
mentary Information S2). Certain VHL-defective cells (for example,
786-O, KTCL140) expressed HIF-2aat a high constitutive level but
did not express detectable HIF-1a(Fig. 2c, d and Supplementary
Information S2). Examination of stable transfectants of RCC4 and
786-O cells demonstrated that expression of the wild-type, but not
truncated, VHL gene restored regulation of HIF a-subunits by
oxygen without affecting the levels of mRNA encoding either
subunit (data not shown).
To investigate the role of pVHL in HIF-1 regulation we tested for
interactions between HIF a-subunits and pVHL using a combina-
tion of hypoxia and/or proteasomal blockade to induce HIF a-
subunits. Anti-pVHL immunoprecipitates of extracts from protea-
somally blocked RCC4/VHL cells, but not RCC4 cells, contained
both HIF-1aand HIF-2a(Fig. 3a). Similar results were obtained
with hypoxia in the absence of proteasomal blockade. In the inverse
reactions, immunoprecipitating antibodies to HIF-2aor HIF-1a
co-precipitated pVHL, although a smaller proportion of the total
was captured (Fig. 3b). Anti-pVHL immunoprecipitations also
demonstrated the interaction in HeLa cells, which express pVHL
normally (Fig. 3c). We next tested whether pVHL is incorporated in
the HIF-1 DNA-binding complex. Addition of antibodies raised
against pVHL (but not control antibodies) to nuclear extract from
RCC4/VHL cells and HeLa cells produced a change in mobility,
which was not observed with VHL-defective RCC4 cells (Fig. 3d).
HIF-1 migrated as two species. Only the slower-mobility HIF-1
species was shifted by anti-pVHL, whereas both species were shifted
Figure 1 Effect of pVHL on oxygen-regulated gene expression. mRNA analysis of
RCC4 cells and stable transfectant expressing pVHL (RCC4/VHL). N, normoxia;
H, hypoxia (1% O
, 16 h). VEGF, vascular endoth elial growth factor; GLUT-1,
glucose transporter 1; AK-3, adenylate kinase 3; TGF-b1, transforming growth
factor-b1; ALD-A, aldolase A; PGK-1, phosphoglyceratekinase 1; PFK, phos-
phofructokinase; LDH, lactate dehydrogenase. U6 small nuclear (sn) RNA was
used as an internal control. Also illustrated are two genes not inﬂuenced by VHL
status or hypoxia; nuclear respiratory factor 1 (NRF-1) and b-actin. Amount of RNA
analysed is detailed in Table S1 of the Supplementary Information.
Figure 2 Effect of pVHL on HIF-1 and HRE activity. a, Representative transient
transfections of RCC4 cells with VHL expression vector (+) or empty vector, and
luciferase reporter genes containing no HRE, or HREs from the PGK-1 or
erythropoietin (Epo) genes linked to SV40 or TK promoters. N, normoxia; H,
hypoxia (0.1% O
, 24 h —results were similar with 1% O
). b, EMSA using the Epo
HRE. N, normoxia; H, hypoxia (1% O
, 4 h). In HeLa and RCC4/VHL, HIF-1 is a
doublet (S, slower and F, faster mobility). RCC4 cells contain only faster mobility
HIF-1, with equivalent levels in normoxia and hypoxia. Constitutive binding
species are indicated (C). c, Immunoblots of whole cell extracts for HIF a-
subunits. Upper panels; RCC4 and RCC4/VHL cells (+VHL). Lower panel; 786-O
cells stably transfected with vector alone, a full length VHL gene (+VHL), or a
truncated VHL gene (1-115; +Tr). HIF-1awas not detected in 786-O cells. d,
Immunoblot of UMRC2, UMRC3 and KTCL140 (renal carcinoma lines with VHL
mutations30), Caki-1 (renal carcinoma line expressing wild-type pVHL) and the
hepatoma line Hep3B.
© 1999 Macmillan Magazines Ltd
letters to nature
20 MAY 1999
by anti-HIF-1a(Fig. 3e). Similar results were obtained in other cell
lines (Hep3B, Caki-1, MRC5-V2 and 293; data not shown).
Furthermore, whereas RCC4/VHL, HeLa and other cells contained
both HIF-1 species, RCC4 extracts contained only the faster-
mobility species (Figs 3d, Fig. 2b). Thus, VHL-defective cells lack
the slower-mobility species which is restored by re-expression of
pVHL, and shifted by anti-pVHL. This indicates that the DNA-
binding HIF-1 doublet arises from two species —containing or not
containing pVHL. Combination supershift analysis conﬁrmed that
the slower-mobility species contained both HIF-1aand pVHL
HIF-1 activation by hypoxia is mimicked by cobaltous ions and
iron chelation6,7. We therefore tested whether the pVHL/HIF-1
interaction was regulated by these stimuli. Proteasomal blockade
induces an HIF-1 DNA-binding complex in normoxic cells18;
comparison of this normoxic complex with EMSA of hypoxic
cells with or without proteasomal inhibitors showed a similar
shift and anti-pVHL supershift (Fig. 4a and data not shown).
Together with immunoprecipitation data, this indicates that the
interaction with pVHL occurs in both normoxic and hypoxic cells.
In contrast, EMSA analysis of RCC4/VHL cells treated with cobalt
and the iron chelator desferrioxamine (DFO) demonstrated only
the faster mobility HIF-1. This did not supershift with anti-pVHL,
indicating that the pVHL/HIF-1 complex could not form in cells
exposed to these stimuli. Similar results were obtained in other cell
types and are consistent with hitherto unexplained mobility differ-
ences between previous analyses of HIF-1 from cobalt- or DFO-
versus hypoxia-stimulated cells7, indicating that this is a general
effect. Treatment with DFO 4 h before hypoxia prevented the
formation of the pVHL/HIF-1 complex (Fig. 4b). Addition of
iron chelators could not break the pVHL/HIF-1 complex in vitro,
whereas addition of in vitro-translated wild-type pVHL (but not a
truncated pVHL) could restore the slower-mobility species to
nuclear extracts of proteasomally blocked, normoxic and hypoxic
RCC4 cells, but not to cells treated with DFO or cobalt (Fig. 4c, and
Fig. S1a of Supplementary Information). Immunoprecipitation
studies also indicated that the interaction between HIF-1 and
pVHL is iron-dependent. Whereas both HIF-1aand HIF-2awere
contained in anti-pVHL immunoprecipitates from hypoxic RCC4/
VHL cells, neither was contained in precipitates from DFO- or
cobalt-treated cells (Fig. 4d). The iron-dependent interaction
between HIF a-subunits and pVHL may be direct or indirect.
However, in vitro-translated wild-type pVHL did not bind to an
in vitro-translated HIF-1 DNA-binding complex (Fig. S1b in
Supplementary Information), in contrast to the interaction with
RCC4 extracts (Fig. 4c), indicating that an additional factor or
modiﬁcation of HIF-1 not represented in rabbit reticulocyte lysates
is necessary for the association.
Normally, HIF a-subunits are targeted for rapid degradation in
normoxic cells by a proteasomal mechanism operating on an
internal oxygen-dependent-degradation (ODD) domain5. Our
data suggest that pVHL might be required for this process—a
possibility which would be consistent with recent data that pVHL
forms a multiprotein complex (containing Cul-2 and elongins B
and C) which has homology with ubiquitin-ligase/proteasome-
targeting complexes in yeast19,20. When cells were switched from
Figure 3 Association of pVHL with HIF-1. a, Immunoblots for HIF a-subunits (2a,
1a) of IG32 (VHL ip) and control immunoprecipitates (using VG-7be) of RCC4/VHL
(VHL+) and RCC4 (VHL−) cells exposed (4 h) to normoxia or hypoxia (1% O
with or without proteasomal inhibition (PI+). Aliquots of selected input lysates
were also immunoblotted. b, Immunoprecipitation of RCC4/VHL (VHL+) and
RCC4 (VHL−) extracts with polyclonal antibodies to HIF a-subunits or normal
rabbit immunoglobulin (control ip) followed by immunoblotting for pVHL (V). A
cross-reacting species arose from the HIF-2aantibody (asterisk). c, Immuno-
precipitation of HeLa extracts with IG32 (VHL
) or pAb419 (control) followed by
immunoblotting for HIF a-subunits. d, Anti-pVHL supershifts. IG32 (VHL Ab+) or
VG-7be (Control Ab+) was added to binding reactions of nuclear extracts from
normoxic or hypoxic (1% O
, 4 h; H+) cells. Anti-pVHL supershifted (SS) the slower
HIF-1 species (S) in HeLa and hypoxic RCC4/VHL cells. No supershift was seen
with RCC4 cells, which lack pVHL. e, Anti-pVHL, anti-HIF-1aand combination
supershifts in HeLa cells. Anti-pVHL (IG32, +VHL Ab) supershifted slower mobility
HIF-1. Anti-HIF-1a(clone 54) supershifted both components (SS). Addition of both
antibodies ‘super-super-shifted’ (SSS) slower mobility HIF-1.
© 1999 Macmillan Magazines Ltd
letters to nature
20 MAY 1999
hypoxia to normoxia with addition of cycloheximide, HIF a-
subunits decayed with a half-life of about 5 min in wild-type VHL
transfectants, compared with ,60 min in the VHL-defective RCC4
and 786-O cells, thus conﬁrming a strong effect of pVHL on stability
(Fig. 5a). Moreover, functional studies of Gal4 chimaeras contain-
ing the HIF-1aODD domain demonstrated a striking dependence
of the isolated ODD domain on pVHL (Fig. 5b).
These experiments deﬁne a function for pVHL in the regulation
of HIF-1. Given recent demonstrations of the importance of HIF-1
in tumour angiogenesis3,21, constitutive HIF-1 activation is clearly
consistent with the angiogenic phenotype of VHL disease. Whether
it is a sufﬁcient explanation for oncogenesis is less clear. HIF-1
mediates gene activation not only by hypoxia, but also by growth
factors such as insulin and insulin-like growth factor-1 (ref. 22).
HIF-1 targets such as molecules involved in enhanced glucose
metabolism and angiogenesis8,9,14 are classically upregulated (by
different mechanisms) in many forms of cancer, supporting an
important role in tumour progression, although their role in the
initiation of oncogenesis is less clear. pVHL is probably a multi-
functional protein which could have other tumour suppressor
actions11,23,24. One possibility is that other gene products could be
targeted in a similar manner to HIF a-subunits; however, compari-
son of anti-pVHL immunoprecipitates from metabolically labelled
RCC4/VHL cells with and without proteasomal inhibitors has so far
demonstrated only two species: HIF-1aand HIF-2a(G.-W.C.,
As both pVHL and HIF-1 are widely expressed, it is likely that the
physiological role of pVHL in HIF-1 regulation is general. It is not
yet clear whether pVHL has actions on oxygen-regulated gene
expression other than through HIF-1. Stabilization of hypoxia-
inducible mRNAs has been reported in VHL-deﬁcient cells12,13. This
might represent an independent action of pVHL. However, regula-
tion of these RNAs is commonly abolished in HIF-1-deﬁcient
cells3,25, so the mRNA stability factors could lie downstream of
Figure 5 Effect of pVHL on HIFastability and ODD domain function. a, Western
analysis of HIF a-subunit stability in cells lacking VHL (RCC4, 786-O) and stable
transfectants expressing pVHL (+VHL). Cells were incubated in hypoxia (4 h),
then moved to normoxia (time 0) with addition of cycloheximide (100 mM) and
harvested up to 80 min later. b, Representative functional assay of the HIF-1aODD
domain. Hep3B or RCC4 cells were transfected with Gal4 reporter pUAS-tk-Luc,
and either pGalVP16 (upper panel) encoding the Gal4 DNA-binding domain fused
to the VP16 activation domain, or pGala344-698VP16 (lower panel) which includes
HIF-1aamino acids 344– 698 (containing the entire ODD domain5). RCC4 cells
were co-transfected with pcDNA3 (−), pcDNA3-VHL (VHL) or pcDNA3-VHL.103FS
(TrVHL). After transfection, cells were divided for 24 h incubation in normoxia (N)
or hypoxia (H; 0.1% O
). Corrected luciferase counts are shown, normalized to the
normoxic value with pGalVP16 or pGalVP16+pcDNA3. The HIF-1adomain con-
fers suppression and hypoxic regulationin Hep3B cells but not RCC4 cells, where
re-expression of pVHL restores these properties.
Figure 4 Effect of cobaltous ions and iron chelation on the pVHL/HIF-1
interaction. a, EMSA and supershift analysis of RCC4/VHL cells exposed (4 h)
to normoxia (N), hypoxia (H; 1% O
), DFO (100 mM), cobaltous chloride (Co,
100 mM) or proteasomal inhibition (PI). In lanes 6– 10, IG32 was added (+aVHL). b,
EMSA and supershift analysis of RCC4/VHL cells subjected to hypoxia (H; 1% O
for 8 h) or DFO (100 mM) with hypoxia (D →H; cel ls exposed to DFO for 8 h, 4 h
normoxia, then 4 h hypoxia). c, EMSA of RCC4 cells showing only faster mobility
HIF-1 (lanes 1– 4). Addition of in vitro transcribed/translated pVHL alone (lanes 5–
8, +IVTT), or together with IG32 (lanes 9– 12, +IVTT+aVHL); slower mobility HIF-1
forms and supershifts in cells exposed to normoxia, hypoxia and PI, but not DFO.
d, HIFaimmunoblots of anti-pVHL immunoprecipitates of RCC4/VHL cells.
© 1999 Macmillan Magazines Ltd
letters to nature
20 MAY 1999
HIF-1. Positive and negative effects on HIF-1 activation by iron
chelators, hydrogen peroxide and redox active agents have indicated
that the underlying oxygen sensing mechanism may involve oxygen-
dependent generation of partially reduced oxygen radical species by
redox active iron centre(s)2,6,7,15,26. Such system(s) could act distantly
on a transduction pathway or might directly modify the ODD
domains of HIFathrough a caged radical system, as proposed for
iron sensing in the regulation of IRP2 (ref. 27). Cobaltous ions
activate HIF-1 in inverse relation to iron availability, indicating that
there may be competition for incorporation into a metal centre28.
Our data support a model in which VHL/HIF complexes form in
normoxic cells and target HIFasubunits for destruction. In
hypoxia, degradation is suppressed despite complex formation,
perhaps because a critical targeting modiﬁcation of the HIFa
ODD domain cannot occur without oxygen. Cobaltous ions and
desferrioxamine prevent formation of the VHL complex, providing
a different mechanism for stabilization of HIF-1 and potentially
explaining why activation by these stimuli is relatively resistant to
oxygen and certain other radical-generating processes6,16,29. The iron
dependence of complex formation could indicate that an iron-
containing protein is an essential component of the complex,
perhaps involved in local generation of an oxygen-sensing signal. M
Cells and transfections. 786-O cells expressing pVHL, truncated pVHL
(amino acids 1– 115), or empty vector17 were a gift from W. G. Kaelin. RCC4
cells were a gift from C. H. C. M. Buys. Other RCC lines were provided by M.
Lerman. HeLa and Hep3B cells were fromECACC. RCC4/VHL was obtained by
transfection with pcDNA3-VHL and G418 selection. Cells were plated in
medium lacking G418 24 h before experiments, which were performed on
dishes approaching conﬂuence. Proteasomal inhibition was with
100 mM calpain inhibitor I and 10 mMN-carbobenzoxyl-L-leucinyl-L-norval-
inal. Transient transfections were by electroporation. Transfected cells were
split for parallel normoxic and hypoxic incubation (Napco 7001, Precision
Scientiﬁc). Luciferase reporter gene activity was corrected for transfection
efﬁciency by assay of b-galactosidase expression from the co-transfected
control plasmid pCMV–bGal.
RNA analysis. Total RNAwas extractedand analysed by ribonuclease protection.
Riboprobe details are given in Table S1 of the Supplementary Information.
Plasmid constructions. pCDNA3-VHL contained nucleotides 214–855 of
GenBank accession no. L15409in pcDNA3 (Invitrogen). pcDNA3-VHL.103FS
was made using site-directed mutagenesis to delete nucleotides 522–523. HRE
reporter genes were based on pGL3-basic (Promega) or pPUR (Clontech) and
contained either a minimal SV40 promoter or a minimal (−40 bp) thymidine
kinase promoter linked to a ﬁreﬂy luciferase gene (details of HREs are in
Supplementary Information S3). pGalVP16 encoded the Gal4 DNA-binding
domain (amino acids 1– 147) linked in-frame to the activation domain (amino
acids 410–490) from herpes simplex virus protein 16; pGala344–698VP16
encoded the indicated amino acids of HIF-1abetween those domains. Plasmid
pUAS-tk-Luc contained two copies of the Gal4 binding site linked to a
thymidine-kinase-promoted luciferase reporter gene.
Cell lysis, immunoblotting and immunoprecipitation. Whole cell extracts
were prepared by homogenization in denaturing conditions and aliquots
immunoblotted for HIF a-subunits with 28b (anti-HIF-1a) and 190b (anti-
HIF-2a) as described16, or using clone 54 (anti-HIF-1a, Transduction Labora-
tories). For immunoprecipitation, lysis was performed in 100 mM NaCl, 0.5%
Igepal CA630, 20 mM Tris-HCl (pH 7.6), 5mM MgCl
and 1 mM sodium
orthovanadate with aprotinin (10 mgml
), ‘Complete’ protease inhibitor
(Boehringer) and 1.0 mM 4-(2-aminoethyl)benzene sulphonyl ﬂuoride for
30 min on ice. After clearance by centrifugation, 120mg aliquots of lysate
were incubated for 2 h at 4 8C with 4 mg afﬁnity-puriﬁed anti-HIF-2apoly-
clonal antibodies (raised against a bacterially expressed fusion protein includ-
ing amino acids 535– 631) or 4mg ammonium sulphate precipitated anti-HIF-
1apolyclonal antibodies (raised against an immunogen including amino acids
530–652) in parallel with normal rabbit immunoglobulin (control), or alter-
natively with 0.7 mg anti-pVHL antibody (IG32, Pharmingen) or control
(antibody to SV40 T antigen, pAb419, a gift from E. Harlow or antibody to
VEGF, VG-7be, a gift from H. Turley). 10 ml conjugated agarose beads pre-
blocked with 20 mg ml
BSA was added and lysates incubated for 2 h with
rocking. Pellets were washed ﬁve times, eluted with sample buffer, and divided
into 2– 6 aliquots for immunoblotting.
Electrophoretic mobility shift and supershift assays. We prepared nuclear
extracts using a modiﬁed Dignam protocol and incubated 5mg (HeLa) or
7.5 mg (RCC4) with a
P-labelled 24-bp oligonucleotide probe (sense strand;
59-GCCCTACGTGCTGCCTCGCATGGC-39) from the mouse Epo 39enhan-
cer as described25. For supershift assays, 0.5 mg IG32, VG-7be (isotype and
subclass matched control for IG32) or clone 54 (anti-HIF-1a) was added and
reactions were incubated for 4 h at 4 8C before electrophoresis. In vitro
transcription translations of pcDNA3-VHL and pcDNA3-VHL.103FS were
done using reticulocyte lysate (Promega); 1 ml of a 1:5 dilution in PBS was
added to binding reactions 2 h before electrophoresis or addition of antibody.
Received 11 March; accepted 13 April 1999.
1. Wang,G. L., Jiang, B.-H., Rue, E. A. & Semenza, G. L. Hypoxia-inducible factor I is a basic-helix-loop-
helix-PASheterodimer regulatedby cellular O
tension.Proc. Natl Acad. Sci. USA 92, 5510– 5514 (1995).
2. Bunn, H. F. & Poyton, R. O. Oxygen sensing and molecular adaptation to hypoxia. Physiol. Rev. 76,
3. Carmeliet, P.et al. Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour
angiogenesis. Nature 394, 485–90 (1998).
4. An,W.G. et al.Stabilizationof wild-type p53by hypoxia-induciblefactor1a.Nature 392, 405– 408(1998).
5. Huang, L. E., Gu, J., Schau, M. & Bunn, H. F. Regulation of hypoxia-inducible factor 1ais mediated by
an oxygen-dependent domain via the ubiquitin-proteasome pathway. Proc. Natl Acad. Sci. USA 95,
6. Goldberg, M. A., Dunning, S. P.& Bunn, H. F. Regulationof the erythropoietin gene: evidence that the
oxygen sensor is a heme protein. Science 242, 1412– 1415 (1988).
7. Wang, G. L. & Semenza,G. L. Desferrioxamine induces erythropoietin gene expression and hypoxia-
inducible factor 1 DNA-binding activity: implications for models of hypoxia signal transduction.
Blood 82, 3610–3615 (1993).
8. Warburg, O. The Metabolism of Tumours (Arnold Constable, London, 1930).
9. Hanahan, D. & Folkman, J. Patterns and emerging mechanisms of the angiogenic switch during
tumorigenesis. Cell 86, 353–364 (1996).
10. Shweiki,D., Itin, A., Soffer, D. & Keshet,E. Vascular endothelialgrowth factor induced by hypoxia may
mediate hypoxia-initiated angiogenesis. Nature 359, 843–845 (1992).
11. Kaelin, W. G. & Maher, E. R. The VHL tumour-suppressor gene paradigm. TrendsGenet. 14, 423 –426
12. Gnarra, J. R. et al .Post-transcriptional regulation of vascular endothelial growth factor mRNA by the
product of the VHL tumor suppressor gene. Proc. Natl Acad. Sci. USA 93, 10589– 10594 (1996).
13. Iliopoulos, O.,Levy, A. P.,Jiang, C., Kaelin,W. G. Jr & Goldberg, M. A. Negative regulation of hypoxia-
inducible genes by the von Hippel-Lindau protein. Proc. Natl Acad. Sci. USA 93, 10595 –10599 (1996).
14. Dang,C. V.& Semenza,G. L.Oncogenicalterationsof metabolism.TrendsBiochem.Sci. 24, 68– 72 (1999).
15. Huang, L. E., Arany, Z., Livingston, D. M. & Bunn, H. F.Activation of hypoxia-inducible transcription
factor depends primarily on redox-sensitive stabilization of its asubunit. J. Biol. Chem. 271, 32253–
16. Wiesener, M. S. et al. Induction of endothelial PAS domain protein-1 byhypoxia: characterization and
comparison with hypoxia-inducible factor-1a.Blood 92, 2260–2268 (1998).
17. Iliopoulos, O., Kibel, A., Gray, S. & Kaelin, W. G. Jr Tumour supression by the human von Hippel-
Lindau gene product. Nature Med. 1, 822– 826 (1995).
18. Salceda, S. & Caro, J. Hypoxia-inducible factor 1a(HIF-1a) protein is rapidly degraded by the
ubiquitin-proteasome system under normoxic conditions. J. Biol. Chem. 272, 22642–22647 (1997).
19. Pause, A. et al. the von Hippel-Lindau tumor-suppressor gene product forms a stable complex with
humanCUL-2,a memberof the Cdc53family of proteins.Proc.Natl Acad.Sci. USA 94, 2156–2161 (1997).
20. Lonergan, K. M. et al. Regulation of hypoxia-inducible mRNAs by the von Hippel-Lindau tumor
suppressor protein requires binding to complexes containing elongins B/C and Cul2. Mol. Cell Biol.
18, 732– 741 (1998).
21. Maxwell, P. H. et al. Hypoxia inducible factor-1 modulates gene expression in solid tumors and
inﬂuences both angiogenesis and tumor growth. Proc. Natl Acad. Sci. USA 94, 8104 –8109 (1997).
22. Zelzer,E. et al. Insulin induces transcription of target genes through the hypoxia-inducible factor HIF-
1a/ARNT. EMBO J. 17, 5083–5094 (1998).
23. Duan, D.R. et al . Inhibition of transcription elongation by the VHL tumor suppressor protein. Science
269, 1402–1406 (1995).
24. Ohh, M. et al. The von Hippel-Lindau tumor suppressor protein is required for proper assembly of an
extracellular ﬁbronectin matrix. Mol. Cell 1, 959–968 (1998).
25. Wood, S. M. et al. Selection and analysis of a mutant cell line defective in the hypoxia-inducible factor-
1a-subunit (HIF-1a). J. Biol. Chem. 273, 8360–8368 (1998).
26. Fandrey, J., Frede, S. & Jelkmann, W. Role of hydrogen peroxide in hypoxia-induced erythropoietin
production. Biochem. J. 303, 507–510 (1994).
27. Iwai, K. et al. Iron-dependent oxidation, ubiquitination, and degradation of iron regulatory protein 2:
implications for degradation of oxidized proteins. Proc. Natl Acad. Sci. USA 95, 4924– 4928 (1998).
28. Ho, V. T. & Bunn, H. F. Effects of transition metals on the expression of the erythropoietin gene:
further evidence that the oxygen sensor is a heme protein. Biochem. Biophys. Res. Commun. 223, 175–
29. Chandel, N. S. et al. Mitochondrial reactive oxygen species trigger hypoxia-induced transcription.
Proc. Natl Acad. Sci. USA 95, 11715– 11720 (1998).
30. Gnarra, J. R. et al. Mutations of the VHL tumour suppressor gene in renal carcinoma. Nature Genet. 7,
Supplementary information is available on Nature’s World-Wide Web site (http://www.nature.com) or
as paper copy from the London editorial ofﬁce of Nature.
Acknowledgements. We thank W. Kaelin, C. Buys and M. Lerman for cell lines, and N. Proudfoot,
A. Harris, D. Gillespie, J. O’Rourke,Y.-M. Tian and L. Nicholls. Financial support was from the Wellcome
Trust, the Barnes Trust, the Deutsche Forschungsgemeinschaft, the Cancer Research Campaign, Action
Research and the Medical Research Council.
Correspondence and requests for materials should be addressedto P.J.R. (e-mail: email@example.com.