Role of NF-kB in the erythropoietin decrease induced by

Document Sample
Role of NF-kB in the erythropoietin decrease induced by Powered By Docstoc
					The FASEB Journal express article 10.1096/fj.02-0168fje. Published online September 5, 2002.

Inhibition of erythropoietin gene expression signaling
involves the transcription factors GATA-2 and NF-κB
Katia La Ferla, Christian Reimann, Wolfgang Jelkmann, and Thomas Hellwig-Bürgel

Institute of Physiology, University of Luebeck, Luebeck, Germany

Corresponding author: Wolfgang Jelkmann, Institute of Physiology, University of Luebeck,
Ratzeburger Allee 160, 23538 Luebeck, Germany. E-mail:


The anemia of chronic inflammatory and malignant diseases is partly due to impaired synthesis
of the hormone erythropoietin (Epo). The proinflammatory cytokines interleukin-1 (IL-1) and
tumor necrosis factor α (TNF-α) suppress in vitro Epo gene expression and Epo protein
secretion. However, the molecular mechanisms of this inhibition are poorly understood. The
human Epo promoter and the 5' flanking region contain several recognition sequences for
transcription factors acting either positively or negatively. Herein, we investigated the roles of
the transcription factors GATA-2 and NF-κB in the modulation of Epo gene expression by IL-1β
and TNF-α in the human hepatoma cell line HepG2. Electrophoretic mobility shift assays
revealed increased GATA-2 and NF-κB DNA binding in cells treated with IL-1β or TNF-α.
Reporter gene assays with a sequence from the Epo promoter in front of the firefly luciferase
gene showed that the cytokines reduced Epo reporter gene activity. Functional inactivation of
GATA-2 and NF-κB by oligo-decoy techniques prevented the inhibition of Epo production by
IL-1β and TNF-α. In HepG2 cells stably transfected with a dominant-negative form of IκBα, the
activation of NF-κB was inhibited, while Epo mRNA levels and Epo secretion increased. Thus,
both GATA-2 and NF-κB seem to be involved in the suppression of Epo gene expression by IL-
1β and TNF-α in vitro and may be responsible for impaired Epo synthesis in inflammatory
diseases in vivo.

Key words: cytokines • anemia • interleukin 1 • tumor necrosis factor α • promoter

T      he hormone erythropoietin (Epo) is essential in the proliferation and differentiation of
       erythrocytic progenitor cells in bone marrow and in the maintenance of a steady supply of
       red blood cells. Epo is produced mainly in the kidney and the liver. Epo synthesis is
increased in response to tissue hypoxia, being regulated by a negative feedback mechanism
involving oxygen tension. Abnormally low plasma Epo concentrations are measured in patients
with malignant, infectious, or autoimmune diseases (1, 2). The pathogenesis of their anemia,
known as anemia of chronic disease (ACD), involves several factors, including impaired iron
turnover and, most importantly, reduced erythropoiesis. Administration of recombinant human
Epo can improve ACD, whereas iron administration is much less effective.
Epo mRNA expression and Epo protein secretion are induced by hypoxia in the human hepatoma
cell lines HepG2 and Hep3B (3). In both cell lines, tumor necrosis factor α (TNF-α) and
interleukin 1β (IL-1β) cause a dose-dependent decrease of hypoxia-induced Epo gene expression
and Epo secretion (4–6). Furthermore, IL-1 has been shown to inhibit murine erythropoiesis in
vitro and in vivo (7, 8) and to suppress renal Epo gene expression in rats (9). The most important
transcription factor for hypoxic stimulation of Epo production is the hypoxia-inducible factor 1
(HIF-1). Because IL-1β and TNF-α activate HIF-1 and increase HIF-1 DNA binding, HIF-1
cannot be the mediator of the cytokine-induced Epo repression (10). Apart from the hypoxic
response element, the Epo promoter and the 5' flanking region contain other potentially
regulatory elements. The minimal DNA sequence required for hypoxic basal transcription, a
region from bp -32 to -1 in respective to the transcriptional startpoint, includes a binding site for
the GATA transcription factors (11). The GATA motif is highly conserved among species and
located in a region where a TATA motif is present in many other genes (12). GATA-2 has been
reported to inhibit Epo gene expression (12–14). In addition, the 5' flanking region of the Epo
gene contains several binding sites for NF-κB (15). One of these is adjacent to the minimal
hypoxia-responsive element of the Epo promoter. NF-κB transcription factors are common
mediators of proinflammatory cytokines and can modulate the expression of many genes
involved in inflammation.

This study was intended to investigate in vitro the influence of proinflammatory cytokines on
GATA-2 and NF-κB activation in relation to Epo synthesis. Effects of TNF-α and IL-1β on
GATA-2 and NF-κB DNA binding were studied in HepG2 cells. Reporter gene assays and
oligo-decoy experiments (16, 17) were carried out as additional approaches to study the
involvement of GATA-2 and NF-κB in Epo gene expression.


Cell cultures

The human hepatoma cell line HepG2 was purchased from the German Collection of
Microorganisms and Cell Cultures (DSMZ Braunschweig). Cells were grown in RPMI 1640
medium (Invitrogen, Carlsbad, CA), supplemented with penicillin (100 U/ml), streptomycin (100
µg/ml,) and 10% fetal calf serum (FCS) in a humidified incubator at 37°C. For preparation of
nuclear extracts, we cultured cells with serum- and antibiotic-free medium for 24 h. For hypoxia
experiments, we placed cells in an incubator at 3% O2 (Heraeus incubators, Kendro, Newtown,
CT). Recombinant human TNF-α was a gift from BASF/Knoll (Ludwigshafen, Germany) and
recombinant human IL-1β from Ciba-Geigy (Basel, Switzerland).

Nuclear protein extraction

Nuclear extracts were prepared according to established protocols (18, 19) with minor
modifications. In brief, at the end of the incubation period, cells were kept on ice, washed once
with ice-cold phosphate-buffered saline (PBS), scraped off, and collected. The cell suspensions
were centrifuged at 1000g for 5 min at 4°C. The cell pellets were resuspended in 300 µl of low-
salt buffer (10 mmol/l HEPES, pH 7.9; 1.5 mmol/l MgCl2; 10 mmol/l KCl). After incubation on
ice for 10–15 min, cell membranes were destroyed by adding 25 µl of a 10% solution of NP-40
and vigorously mixing for 30 s. Nuclei were collected by centrifugation and resuspended in 100
µl high-salt buffer (20 mmol/l HEPES, pH 7.9; 420 mmol/l NaCl; 1.5 mmol/l MgCl2; 0.2 mmol/l
EDTA; 25% glycerol) and incubated on ice for 20 min on a shaking platform. Buffers were
supplemented with 2 µg/ml aprotinin, 10 µg/ml leupeptin, 20 µg/ml pepstatin, 1 mmol/l sodium-
ortho-vanadate, 0.5 mmol/l benzamidine, 2 mmol/l levamisole, 10 mmol/l β-glycerophosphate,
0.5 mmol/l DTT, and 0.4 mmol/l PMSF just before use. Nuclei were centrifuged at 12500g at
4°C for 10 min. Supernatants were shock-frozen and stored at –80°C. Protein concentrations
were determined by the Lowry method with the commercially available DC protein assay kit
(BioRad Laboratories, Hercules, CA).

Electrophoretic mobility shift assay (EMSA)

For EMSAs, the following synthetic oligonucleotides were used: for NF-κB 5'-
AGTTGAGGGGACTTTCCCAGGC-3'                  and     the      complementary     strand    5´-
GCCTGGGAAAGTCCCCTCAACT-3',                          for            GATA-2                5'-
CACACATGCAGATAACACCCCCGACC-3'                   and     the   complementary     strand   5'-
GGTCGGGGCTGTTATCTGCATGTGTG-3'. Single-stranded oligonucleotides were labeled
with T4-polynucleotide kinase (MBI Fermentas) and γ [32P]-ATP. A twofold molar excess of
unlabeled complementary oligonucleotides was annealed, and double-stranded oligonucleotides
were purified on spin columns (Micro Bio-Spin P30, BioRad).

Binding reactions were performed for 30 min on ice in 20 µl buffer (1 mmol/l MgCl2; 0.5 mmol/l
M EDTA; 0.5 mmol/l DTT; 50 mmol/l NaCl; 10 mmol/l Tris-HCl, pH 7.5; 4% glycerol)
containing 5 µg protein, 1 µg poly(dI:dC), and 15,000 cpm labeled oligonucleotide. Binding
complexes were resolved by electrophoresis in vertical nondenaturing 6% polyacrylamide gels,
using 0.3 × TBE as running buffer. Gels were dried, and binding complexes were quantitated by
phosphoimager analysis. For supershift experiments, 2 µg of anti-NF-κB p-50 H-119 (Santa
Cruz Biotechnology, Santa Cruz, CA), 5 µg of anti-NF-κB p-65 (Rel A) (Biomol, Stressgen
Biotechnologies, San Diego, CA), and 1.6 µg of anti-GATA-2 H116 (Santa Cruz Biotechnology)
were incubated with binding reaction mixtures for 1 h on ice before the labeled probe was added.

Plasmid construction

Epo reporter plasmids were constructed using the Kpn I and Sac I restriction enzyme sites within
the PGL3-prom luciferase plasmid (Promega, Madison, WI). DNA inserts were synthetic
oligonucleotides spanning the sequence from bp -38 to -14 of the Epo promoter
(CACACATGCAGATAACAGCCCCGAC) containing either the GATA element (pGATAwt)
or the GATA element mutated to TATA (pGATAmut). Constructs were verified by sequence


HepG2 cells (5 × 106) in 400 µl RPMI 1640 medium were transiently transfected with 50 µg of
the previously mentioned constructs by electroporation (250 V, 950 µF) with a Gene Pulser 2
(BioRad). The transfected cells were allowed to grow for 24 h. After medium renewal, cells were
stimulated with cytokines. Luciferase activity was determined after 24 h. All results obtained for
firefly luciferase were normalized to total cellular protein.

NF-κB-responsive and -nonresponsive HepG2 cells were generated by transfection with either
pCMV-IκBα or pCMV- IκBα-M (Clontech, Palo Alto, CA) by electroporation. pCMV- IκBα-
transfected cells overexpress IκBα, whereas pCMV-IκBαM-transfected cells overexpress a
mutated form of IκBα, with a serine-alanine mutation at residues 32 and 36. This form is not
phosphorylated by Iκ-kinases and remains bound to NF-κB, leading to inhibition of NF-κB
activation. After electroporation, cells were allowed to grow for 48 h and selected in medium
containing 0.5 mg/ml of G418.

Oligo-decoy experiments

HepG2 cells were seeded at subconfluence in 96-well plates in medium with FCS. Double-
stranded phosphorothioate backbone oligonucleotides with binding sites for either NF-κB (NF-
κB-wt) or GATA-2 (GATA-2-wt) or with no-functional binding sites (NF-κB-mut, GATA-2-
mut), respectively, were added to cell cultures to a final concentration of 3 mmol/l for 24 h. The
sequences of oligonucleotides were NF-κB-wt 5'-AGTTGAGGGGACTTTCCCAGGC-3', NF-
κB-mut             5'-AGTTGAGCCTGAACCCCCAGGC-3';                           GATA-wt              5'-
ACGCACACATGCAGATAACAGCCCCGACCC-3',                                    GATA-mut                  5´-
ACGCACCACTGCAGGGATCAGCCCCGACCC-3' and the respective complementary strands.
After medium renewal, cells were exposed to hypoxia alone or in combination with IL-1β. After
12 h of hypoxic incubation, medium was collected and assayed for secreted Epo. The efficiency
of oligonucleotide uptake was proven with FITC-labeled oligonucleotides and fluorescence
microscopy. About 30–40% of cells showed FITC fluorescence after the 24 h uptake period (data
not shown). Epo concentrations were related to total cellular protein determined by the Bradford

Assay of Epo

Cells were seeded at a density of 5 × 106/well in 6-well plates or 1 × 105/well in 96-well plates
and incubated at 3% O2 for 12 h or 24 h as specified. We determined Epo concentrations in cell
culture supernatants by ELISA, using a commercial kit (Medac).

RNA isolation and Northern blot

Total RNA was isolated with the guanidinium thiocyanate-phenol-chloroform method (20). Total
RNA (15 µg) was separated on a 0.7 mol/l formaldehyde/1% agarose gel and vacuum transferred
onto a nylon membrane (Biodyne A membrane, Pall Corporation, East Hills, NY). Epo probe
was obtained by reverse transcriptase-polymerase chain reaction (RT-PCR) of RNA from
hypoxic HepG2 cells with Epo-specific primers as previously described (10). The probe was
labeled with α [32P]-dCTP and a random prime DNA labeling kit (MBI Fermentas).
Hybridization was carried out in 50% formamide, 1× Denhardt’s, and 0.2% sodium dodecyl
sulfate (SDS) at 42°C for 24 h. The membrane was washed with 0.2× SSC/0.1% SDS at 60°C for
1 h, sealed in a plastic bag, and analyzed by phosphoimaging. The membrane was rehybridized
with a 18 S rRNA probe for normalization of the Epo signals.


Results are given as the mean and SD. Statistical analyses were performed using the methods of
Student, Dunnett, or Bonferroni as appropriate. P was considered significant at the 5% level.


Induction of NF-κB DNA binding by IL-1β and TNF-α

Normoxic (20% O2) and hypoxic (3% O2) HepG2 cells stimulated with IL-1β (300 pg/ml) or
TNF-α (10 ng/ml) for 1 h showed increased NF-κB DNA-binding (Fig. 1A). Hypoxia alone did
not induce p50/p65 heterodimer DNA binding. A time-course study with IL-1β showed that
p50/p65 heterodimer formation was maximal between 30 min and 1 h, whereas the signals
decreased after 2 and 4 h of incubation (Fig. 1B). Binding assays were also performed with
extracts from cells stably transfected with pCMV- IκBα-M, a dominant-negative form of IκBα.
pCMV-IκBα-M-transfected cells showed a marked decrease in NF-κB DNA binding compared
with pCMV-IκBα-transfected cells or to nontransfected HepG2 cells (Fig. 1C). Preincubation of
the reaction mixture with a polyclonal anti-p50 antibody induced supershifts of the specific
signals in extracts from IL-1β-treated nontransfected cells, whereas preincubation with a
polyclonal anti-p65 antibody prevented p50/p65 and p65/p65 DNA binding.

Effects of NF-κB inhibition on Epo production

Northern blots confirmed a decrease of Epo mRNA in hypoxic HepG2 cells upon treatment with
IL-1β or TNF-α for 24 h. The cytokines also decreased Epo mRNA levels in HepG2 cells
transfected with pCMV-IκBα. However, Epo mRNA levels were higher in IL-1β- or TNF-α-
treated cells transfected with pCMV-IκBα-M than in pCMV-IκBα-transfected cells treated with
the cytokines (Fig. 2).

Epo levels increased about twofold in culture supernatants of pCMV-IκBα-M compared with
pCMV-IκBα-transfected HepG2 cells following 24 h incubation at 3% O2, in both the presence
and the absence of cytokines (Fig. 3).

Oligo-decoy experiments showed that IL-1β failed to reduce Epo production when NF-κB
function was blocked by treating cells with a NF-κB binding oligonucleotide (NF-κB-wt). In
contrast, the NF-κB decoy-oligonucleotide with the mutated NF-κB binding site (NF-κB-mut)
did not prevent the inhibition of Epo production by IL-1β (Fig. 4).

Induction of GATA-2 DNA binding by IL-1β and TNF-α

HepG2 cells showed strong GATA-2 DNA binding in normoxia (Fig. 5A), which was reduced
upon exposure of the cells to hypoxia. Stimulation with IL-1β or TNF-α for 4 h led to an
increase in GATA-2 DNA binding in hypoxia. The specificity of signals seen in EMSAs was
demonstrated by competition experiments and by preincubation of the binding reaction mixtures
with a specific antibody (Fig. 5B).

Influence of IL-1β and TNF-α on Epo promoter activity

Luciferase activity of normoxic cells transfected with pGATAmut was 1.7-fold greater than in
pGATAwt-transfected cells. In hypoxia, luciferase activity was increased compared with
normoxia in both pGATAwt and pGATAmut cells. Stimulation of hypoxic pGATAmut-
transfected cells with IL-1β induced a 2.6-fold increase in luciferase activity compared with
pGATAwt IL-1β-treated cells. A similar difference was seen when the effect of TNF-α was
studied in pGATA-mut vs. pGATA-wt cells (Fig. 6).

Oligo-decoy experiments showed that with respect to Epo production, IL-1β induced only a
moderate decrease when GATA-2 function was blocked by treating cells with a GATA binding
oligonucleotide (GATA-2-wt). In cells treated with an ineffective mutated oligonucleotide
(GATA-2-mut), GATA-2 function was maintained and HepG2 cells produced little Epo upon IL-
1β treatment (Fig. 7).


The proinflammatory cytokines IL-1β and TNF-α are central mediators of immune responses.
They have several biological activities in common, including the ability to induce anemia by
inhibiting erythropoiesis. IL-1β and TNF-α suppress Epo production in vitro (4–6, 21). Our
study suggests that this effect is mediated by the transcription factors NF-κB and GATA-2 in
HepG2 cells.

IL-1β and TNF-α induced NF-κB activation in normoxia and hypoxia, whereas hypoxia alone
did not activate NF-κB. To investigate a possible involvement of NF-κB activation in the
control of Epo production, we inhibited NF-κB function by stable transfection of HepG2 cells
with a dominant-negative form of the inhibitory protein IκBα, which is usually present in
HepG2 cells (22). Cells expressing the mutant form of IκBα showed higher rates of Epo
production when treated with cytokines compared with cells expressing the normal form of
IκBα, suggesting that NF-κB activation is involved in the pathway leading to the down-
regulation of Epo production by IL-1β and TNF-α. These results were confirmed by RNA
hybridization experiments showing decreased mRNA levels in native HepG2 cells treated with
IL-1β or TNF-α and in cells expressing the normal form of IκBα.

In comparison, Epo mRNA levels in cells expressing the mutant form of IκBα were higher
upon cytokine treatment. Furthermore, oligo-decoy experiments revealed an increase in Epo
production in cells with impaired NF-κB function. These data give additional evidence that NF-
κB can be involved in the repression of certain genes, whereas most reports describe activation
of gene transcription by NF-κB. Earlier studies have shown that NF-κB mediates TNF-α-
induced suppression of the genes encoding cytochrome P-450 1A1 (23) and the carcinoma-
associated epithelial cell adhesion molecule (24). This gene suppression has been explained by
competition of NF-κB with other transcription factors for the cofactor p 300/CREB-binding
protein (CBP) (23, 24). The transcription cofactor P300/CBP is also involved in hypoxia-
induced Epo gene expression (25). In addition, we have shown previously that exogenous
cAMP- analogs partly reverse the inhibition of Epo mRNA expression by IL-1β and TNF-α (6).

GATA-2 binds to a specific motif present in the hypoxia-inducible promoter of the Epo gene
(12–14). In addition, nitric oxide is thought to stimulate Epo gene expression by reducing
GATA-2 availability (26), whereas hydrogen peroxide exerts the opposite effect (27). The role of
GATA-2 in the inhibition of Epo production by cytokines has not been investigated previously.
Our results show that GATA-2 DNA binding was decreased in hypoxic vs. normoxic cells,
whereas Epo production increased in hypoxia. Treatment with cytokines restored signals in
EMSAs to a level comparable to those observed under normoxic conditions, suggesting that
GATA-2 may modulate signal transduction pathways of the cytokines reducing Epo production.

To further prove that GATA-2 is involved in the cytokine-induced reduction of Epo synthesis,
we performed reporter gene assays with a construct spanning a sequence from the Epo promoter,
including the GATA element, and investigated the influence of cytokines on luciferase activity.
We also measured luciferase activity in cells transfected with a sequence containing TATA
instead of the GATA element. Cytokines decreased Epo promoter activity in cells transfected
with pGATAwt compared with control cells, whereas Epo promoter activity was increased 2.5-
fold when the TATA element was transfected. Others (12) have shown that a point mutation of
the GATA sequence into TATA enhances transcriptional activity of the Epo promoter even in
normoxia. Thus, the GATA motif in the Epo promoter can avoid binding and activation by
TATA binding proteins, probably explaining the low basal expression of the Epo gene in
normoxia. Our results confirm that the GATA element is a repressor element for the Epo gene.
The observation that its activity is enhanced by IL-1β and TNF-α is a novel finding. Moreover,
oligo-decoy experiments showed an increase in Epo production by cells with impaired GATA-2

It is generally assumed that HIF-1 is the primary transcription factor in the hypoxic induction of
Epo gene expression. HIF-1 binds to the Epo enhancer as a dimer composed of an α- and a β-
subunit (28). In the presence of O2, the α-subunit is rapidly prolyl- and asparaginyl-
hydroxylated, targeted by the von Hippel-Lindau protein/E3 ubiquitation ligase complex, and
degraded by the ubiquitin-proteasome system (29–32). In hypoxia, HIF-1α is enabled to enter
the nucleus and to form the active heterodimer with HIF-1β. We have recently shown that
neither IL-1β nor TNF-α suppresses HIF-1 DNA binding in HepG2 cells. In contrast, IL-1β
increases nuclear HIF-1α protein levels and HIF-1 DNA-binding (10, 33), whereas TNF-α
induces HIF-1 DNA binding only (10). In any case, the cytokines inhibit Epo mRNA expression,
although they stimulate HIF-1 DNA binding.

On the basis of the present findings, we propose that there is a balance between Epo gene
inducing and repressing transcription factors. In hypoxia, the increase in HIF-1 and the decrease
in GATA-2 result in a stimulation of Epo gene transcription. IL-1 and TNF-α may induce HIF-1,
too, but with respect to Epo gene expression, this effect is probably overcome by the
simultaneous activation of GATA-2 and NF-κB.

We are indebted to Dr. Shigehiko Imagawa for the gift of a GATA-2 plasmid, which was used
for the generation of new constructs. This study was supported by a grant from the Deutsche
Forschungsgemeinschaft (SFB 367-C8).


 1. Means, R. T., and Krantz, S. B. (1992) Progress in understanding the pathogenesis of the
    anemia of chronic disease. Blood 80, 1639–1647

 2. Jelkmann, W. (1998) Proinflammatory cytokines lowering erythropoietin production. J.
    Interferon Cytokine Res. 18, 555–559

 3. Goldberg, M. A., Glass, G. A., Cunningham, J. M., and Bunn, H. F. (1987) The regulated
    expression of erythropoietin by two human hepatoma cell lines. Proc. Natl. Acad. Sci. U. S.
    A. 84, 7972–7976

 4. Faquin, W. C., Schneider, T. J., and Goldberg, M. A. (1992) Effect of inflammatory
    cytokines on hypoxia-induced erythropoietin production. Blood 79, 1987–1994

 5. Jelkmann, W., Pagel, H., Wolff, M., and Fandrey, J. (1992) Monokines inhibiting
    erythropoietin production in human hepatoma cultures and in isolated perfused rat kidneys.
    Life Sci. 50, 301–308

 6. Fandrey, J., Huwiler, A., Frede, S., Pfeilschifter, J., and Jelkmann, W. (1994) Distinct
    signaling pathways mediate phorbol-ester-induced and cytokine-induced inhibition of
    erythropoietin gene expression. Eur. J. Biochem. 226, 335–340

 7. Johnson, C. S., Keckler, D. J., Topper, M. I., Braunschweiger, P. G., and Furmanski, P.
    (1989) In vivo hematopoietic effects of recombinant interleukin-1 alpha in mice:
    stimulation of granulocytic, monocytic, megakaryocytic, and early erythroid progenitors,
    suppression of late-stage erythropoiesis, and reversal of erythroid suppression with
    erythropoietin. Blood 73, 678–683

 8. Schooley, J. C., Kullgren, B., and Allison, A. C. (1987) Inhibition by interleukin-1 of the
    action of erythropoietin on erythroid precursors and its possible role in the pathogenesis of
    hypoplastic anaemias. Br. J. Haematol. 67, 11–17

 9. Frede, S., Fandrey, J., Pagel, H., Hellwig, T., and Jelkmann, W. (1997) Erythropoietin gene
    expression is suppressed after lipopolysaccharide or interleukin-1 beta injections in rats.
    Am. J. Physiol. 273, R1067–R1071

10. Hellwig-Burgel T., Rutkowski, K., Metzen, E., Fandrey, J., and Jelkmann, W. (1999)
    Interleukin-1β and tumor necrosis factor-α stimulate DNA binding of hypoxia-inducible
    factor-1. Blood 94, 1561–1567
11. Blanchard, K. L., Acquaviva, A. M., Galson, D. L., and Bunn, H. F. (1992) Hypoxic
    induction of the human erythropoietin gene: cooperation between the promoter and
    enhancer, each of which contains steroid receptor response elements. Mol. Cell. Biol. 12,

12. Tsuchiya, T., Okada, M., Ueda, M., and Yasukochi, Y. (1997) Activation of the
    erythropoietin promoter by a point mutation from GATA to TATA in the -30 region. J.
    Biochem. (Tokyo) 121, 193–196

13. Imagawa, S., Izumi, T., and Miura, Y. (1994) Positive and negative regulation of the
    erythropoietin gene. J. Biol. Chem. 269, 9038–9044

14. Imagawa, S., Yamamoto, M., and Miura, Y. (1997) Negative regulation of the
    erythropoietin gene expression by the GATA transcription factors. Blood 89, 1430–1439

15. Lee, H. S., Lin, J. J., Kung, H. F., Huang, P. L., Lee, L., and Huang, P. L. (1993) The
    human erythropoietin-encoding gene contains a CAAT box, TATA boxes and other
    transcriptional regulatory elements in its 5' flanking region. Gene 128, 227–236

16. Bielinska, A., Shivdasani, R. A., Zhang, L. Q., and Nabel, G. J. (1990) Regulation of gene
    expression with double-stranded phosphorothioate oligonucleotides. Science 250 , 997–

17. Iwase, R., Namba, M., Yamaoka, T., and Murakami, A. (1997) Gene regulation by decoy
    approach (I): synthesis and properties of photo-crosslinked oligonucleotides. Nucleic Acids
    Symp. Ser. 203–204

18. Schreiber, E., Matthias, P., Muller, M. M., and Schaffner, W. (1989) Rapid detection of
    octamer binding proteins with 'mini-extracts', prepared from a small number of cells.
    Nucleic Acids Res. 17, 6419

19. Andrews, N. C., and Faller, D. V. (1991) A rapid micropreparation technique for extraction
    of DNA-binding proteins from limiting numbers of mammalian cells. Nucleic Acids Res.
    19, 2499

20. Chomczynski, P., and Sacchi, N. (1987) Single-step method of RNA isolation by acid
    guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162, 156–159

21. Jelkmann, W., and Hellwig-Buergel, T. (1999) Tumor necrosis factor p55 receptor (TNF-
    RI) mediates the in vitro inhibition of hepatic erythropoietin production. Exp. Hematol. 27,

22. Han, Y., Meng, T., Murray, N. R., Fields, A. P., and Brasier, A. R. (1999) Interleukin-1-
    induced nuclear factor-kappaB-IkappaBalpha autoregulatory feedback loop in hepatocytes.
    A role for protein kinase calpha in post-transcriptional regulation of IκBα resynthesis. J.
    Biol. Chem. 274, 939–947
23. Ke, S., Rabson, A. B., Germino, J. F., Gallo, M. A., and Tian, Y. (2001) Mechanism of
    suppression of cytochrome P-450 1A1 expression by tumor necrosis factor-alpha and
    lipopolysaccharide. J. Biol. Chem. 276, 39638–39644

24. Gires, O., Kieu, C., Fix, P., Schmitt, B., Munz, M., Wollenberg, B., and Zeidler, R. (2001)
    Tumor necrosis factor alpha negatively regulates the expression of the carcinoma-
    associated antigen epithelial cell adhesion molecule. Cancer 92, 620–628

25. Ebert, B. L., and Bunn, H. F. (1999) Regulation of the Erythropoietin gene. Blood 94,

26. Tarumoto, T., Imagawa, S., Ohmine, K., Nagai, T., Higuchi, M., Imai, N., Suzuki, N.,
    Yamamoto, M., and Ozawa, K. (2000) N(G)-monomethyl-L-arginine inhibits
    erythropoietin gene expression by stimulating GATA-2. Blood 96, 1716–1722

27. Tabata, M., Tarumota, T., Ohmine, K., Furukawa, Y., Hatake, K., Ozawa, K., Hasegawa,
    Y., Mukai, H., Yamamoto, M., and Imagawa, S. (2001) Stimulation of GATA-2 as a
    mechanism of hydrogen peroxide suppression in hypoxia-induced erythropoietin gene
    expression. J. Cell. Physiol. 186, 260–267

28. Wang, G. L., Jiang, B. H., Rue, E. A., and Semenza, G. L. (1995) Hypoxia-inducible factor
    1 is a basic-helix-loop-helix-PAS heterodimer regulated by cellular O2 tension. Proc. Natl.
    Acad. Sci. USA 92, 5510–5514

29. Jaakkola, P., Mole, D. R., Tian, Y. M., Wilson, M. I., Gielbert, J., Gaskell, S. J.,
    Kriegsheim, A., Hebestreit, H. F., Mukherji, M., Schofield, C. J., Maxwell, P. H., Pugh, C.
    W., and Ratcliffe, P. J. (2001) Targeting of HIF-α to the von Hippel-Lindau ubiquitylation
    complex by O2-regulated prolyl hydroxylation. Science 292, 468–472

30. Ivan, M., Kondo, K., Yang, H., Kim, W., Valiando, J., Ohh, M., Salic, A., Asara, J. M.,
    Lane, W. S., and Kaelin Jr., W. G. (2001) HIFα targeted for VHL-mediated destruction by
    proline hydroxylation: implications for O2 sensing. Science 292, 464–468

31. Lando, D., Peet, D. J., Whelan, D. A., Gorman, J. J., and Whitelaw, M. L. (2002)
    Asparagine hydroxylation of the HIF transactivation domain: a hypoxic switch. Science
    295, 858–861

32. Hewitson, K. S., McNeill, L. A., Riordan, M. V., Tian, Y. M., Bullock, A. N., Welford, R.
    W., Elkins, J. M., Oldham, N. J., Bhattacharya, S., Gleadle, J. M., Ratcliffe, P. J., Pugh, C.
    W., and Schofield, C. J. (2002) Hypoxia inducible factor (HIF) asparagine hydroxylase is
    identical to factor inhibiting HIF (FIH) and is related to the cupin structural family. J Biol
    Chem 277, 26351–26355

33. Stiehl, D. P., Jelkmann, W., Wenger, R. H., and Hellwig-Bürgel, T. (2002) Normoxic
    induction of the hypoxia-inducible factor-1α by insulin and interleukin-1β involves the
    phosphatidylinositol 3-kinase pathway. FEBS Letters 512, 157–162

                                                    Received April 17, 2002; accepted July 15, 2002.
Fig. 1

Figure 1. NF-κB DNA binding in HepG2 cells. A) Electrophoretic mobility shift assays (EMSAs) performed with 5 µg
protein from untreated normoxic cells (lane 1), normoxic cells stimulated with IL-1β (300 pg/ml) or tumor necrosis factor
(TNF)-α (10 ng/ml) (lanes 2 and 3), hypoxic control cells (lane 4), and hypoxic cells stimulated with IL-1β or TNF-α
(lanes 5 and 6) for 1 h (similar results were obtained in three independent experiments). B) Time course of NF-κB DNA
binding in normoxic HepG2 cells stimulated with IL-1β. C) EMSAs performed with 5 µg protein from control and IL-1β-
treated normoxic cells: nontransfected cells (G2, lanes 1 and 2), with pCMV-IκBα (IκBα, lanes 3 and 4) and pCMV-
IκBα-M (M, lanes 5 and 6) transfected HepG2 cells stimulated with IL-1β. Specificity of signals was shown by
incubation of binding reactions for 1 h on ice with polyclonal antibodies against the p50 (lane 7) or p65 (lane 8) subunit of
NF-κB in supershift assays (G2 [S]). Induction ratios are given in the tables below the figure.
Fig. 2

Figure 2. Epo mRNA expression. A) Representative Northern blot of cells incubated under hypoxic conditions (3%
O2) for 24 h in the absence or presence of IL-1β (300 pg/ml) or TNF-α (10 ng/ml). Total RNA (15 µg) from HepG2 cells
(lanes 1–3), from HepG2 cells transfected with pCMV-IκBα (lanes 4–6), and from HepG2 cells transfected with pCMV-
IκBα-M were analyzed. The membrane was rehybridized to a 18 S rRNA probe. B) Mean values and SD of three separate
experiments. Epo mRNA was related to 18 S rRNA. Hypoxic samples were set 100%. *P < 0.05 (Dunnett’s test).
Fig. 3

Figure 3. Epo production in IκBα-transfected cells. HepG2 cells transfected with pCMV-IκBα or with pCMV-IκBα-
M were incubated under hypoxic conditions (3% O2) with or without cytokines (IL-1β, 300 pg/ml; TNF-α, 10 ng/ml) for
24 h. Epo concentrations were related to total cellular protein (mean ±SD of four separate experiments).*P < 0.05
(Student’s t test).
Fig. 4

Figure 4. Effects of NF-κB decoy on Epo production. Double-stranded oligonucleotides (3 mmol/l) with functional
(NF-κB-wt) or mutated binding site (NF-κB-mut) for NF-κB were added to cell cultures. Epo concentrations in culture
supernatants were assessed after 12 h in hypoxia (3% O2) with or without IL-1β (300 pg/ml). Results are means ±SD of
three experiments. *P < 0.05 (Bonferroni’s test).
Fig. 5

Figure 5. GATA-2 DNA binding in HepG2 cells. A) Electrophoretic mobility shift assays (EMSAs) performed with 5
µg protein from untreated normoxic cells (lane 1), normoxic cells stimulated with IL-1β (300 pg/ml, lane 2) or TNF-α (10
ng/ml, lane 3), hypoxic cells (lane 4), and hypoxic cells stimulated with IL-1β or TNF-α (lanes 5 and 6) for 4 h (similar
results were obtained in three independent experiments). B) EMSAs performed with 5 µg protein from untreated
normoxic cells (lane 1), normoxic cells treated with IL-1β (300 pg/ml, lane 2), and normoxic cells preincubated with a
100-fold excess of an AP-1 consensus oligonucleotide (1 pmol) as unspecific competitor (u.c., lane 3) or with the GATA
oligonucleotide (1 pmol) as specific competitor (s.c., lane 4). Preincubation of the binding reaction with an anti-GATA-2
polyclonal antibody (lane 5) prevented GATA-2 DNA-binding. Induction ratios are given in the tables below the figure.
Fig. 6

Figure 6. Reporter gene assay of HepG2 cells transfected with pGATA-wt and pGATA-mut. Cells were stimulated
with IL-1β (300 pg/ml) or TNF-α (10 ng/ml) for 24 h. Luciferase activity was compared with untreated normoxic cells set
100% and normalized to total cellular protein (mean ±SD of three separate experiments).
Fig. 7

Figure 7. Effects of GATA-2 decoy on Epo production. Cells were treated with double-stranded oligonucleotides (3
mmol/l) with a functional binding site (GATA-wt) and a mutated binding site (GATA-mut) for GATA. Epo
concentrations in culture supernatants were assessed after 12 h in hypoxia (3% O2) with or without IL-1β (300 pg/ml).