Erythropoietin production: Molecular mechanisms of the antagonistic actions of cyclic adenosine monophosphate and interleukin-1

FEBS Letters 580 (2006) 3153–3160

Erythropoietin production: Molecular mechanisms of theantagonistic actions of cyclic adenosine monophosphate and interleukin-1

Chimedtseren Batmunkh, Jochen Krajewski, Wolfgang Jelkmann, Thomas Hellwig-Burgel*

Institute of Physiology, University of Luebeck, Ratzeburger Allee 160, D-23538 Luebeck, Germany

Received 21 March 2006; accepted 20 April 2006

Available online 2 May 2006

Edited by Veli-Pekka Lehto

Abstract Erythropoietin (Epo) mRNA expression is suppressedby interleukin 1 (IL-1). Cyclic adenosine monophosphate(cAMP) can increase Epo mRNA and Epo protein levels inIL-1 treated HepG2 cells to some extent. To identify molecularmechanisms of this reaction we investigated three transcriptionfactors (NF-jB, GATA-2 and HIF-1) that control the Epo gene.Western blot analyses and electrophoretic mobility shift assays(EMSAs) revealed that IL-1 strongly activated NF-jB, whichis a likely suppressor of the Epo promoter. Treatment of the cellswith dibutyryl-cAMP (Bt2-cAMP) inhibited the activation ofNF-jB by IL-1. Bt2-cAMP increased GATA-2 DNA binding.Since GATA-2 is a suppressor of the Epo promoter, GATA-2activation was unlikely to cause the increase of Epo mRNAexpression in IL-1 treated cells. Furthermore, Western blots,EMSAs and reporter gene studies showed that Bt2-cAMP waswithout effect on the hypoxia-inducible transcription factorHIF-1. Thus, NF-jB is probably the primary transcription fac-tor by which cAMP counteracts the inhibition of Epo geneexpression by IL-1.� 2006 Federation of European Biochemical Societies. Publishedby Elsevier B.V. All rights reserved.

Keywords: Cytokines; Cyclic adenosine monophosphate;Erythropoietin; Interleukin 1; NF-jB

1. Introduction

The erythropoietin (Epo) gene possesses several regulatory

DNA elements. GATA-2, a zinc finger transcription factor,

binds to the GATA motif located in the �30 region relative

to the transcriptional initiation site of the Epo promoter,

thereby inhibiting Epo gene expression [1,2]. Furthermore,

the 5 0 region of the Epo gene contains binding sites for NF-

jB [3]. The 50 bp hypoxia-responsive element (HRE) of the

3 0 enhancer contains a hypoxia-inducible factor (HIF) binding

site (HBS), a CACA sequence, and a direct repeat of two ste-

roid hormone receptor-binding half-sites separated by two

base pairs (DR-2 element) [4]. Hypoxia-inducible-factor-1

(HIF-1) is a crucial transcription factor for the hypoxic induc-

Abbreviations: ACD, anemia of chronic disease; Bt2-cAMP, dibutyryl-cyclic adenosine monophosphate; EMSA, electrophoretic mobilityshift assay; Epo, erythropoietin; HIF-1, hypoxia-inducible factor-1;HRE, hypoxia-response element; IL, interleukin; TNF, tumor necrosisfactor

*Corresponding author. Fax: +49 451 5004171.E-mail address: [email protected] (T. Hellwig-Burgel).

0014-5793/$32.00 � 2006 Federation of European Biochemical Societies. Pu

doi:10.1016/j.febslet.2006.04.069

tion of the Epo gene [5]. Activated HIF-1 is composed of two

subunits, HIF-1a and HIF-1b [6], with HIF-1a being unstable

in the presence of O2 (for Refs. see [7,8]).

We have recently shown that the proinflammatory cytokines

interleukin 1 (IL-1) and tumor necrosis factor-a (TNF-a) acti-

vate GATA-2 and NF-jB [9]. IL-1 and TNF-a have proved to

inhibit Epo mRNA expression and Epo synthesis in the human

hepatoma cell lines, Hep3B and HepG2 [10–12] and in rat kid-

neys [11,13]. On the other hand, cyclic adenosine monophos-

phate (cAMP) has been reported to stimulate Epo synthesis

in hepatic and renal cell lines [14,15] and to counteract the

inhibitory action of IL-1 and TNF-a on Epo synthesis [12].

The present study aimed at investigating molecular mecha-

nisms by which cAMP prevents the suppression of Epo mRNA

expression and Epo production by IL-1. Results obtained by

Western blotting, electrophoretic mobility shift assays, repor-

ter gene studies, Epo mRNA quantification and Epo immuno-

assay suggested that cAMP restores Epo production in IL-1

treated HepG2 cells by preventing NF-jB mobilization rather

than through modulating GATA-2 or HIF-1 signaling.

2. Materials and methods

2.1. Cell culturesThe human hepatoma cell line HepG2 was purchased from the Ger-

man Collection of Microorganisms and Cell Cultures (DSMZ, Braun-schweig). Cells were grown in RPMI 1640 medium (Gibco, Karlsruhe,Germany) supplemented with 10% fetal calf serum (FCS; Gibco).Medium for the HepG2 derivatives HRG1, IjBa-HepG2 andIjBaM-HepG2 [9,16] was supplemented with 50 lg/ml G418,100 IU/ml penicillin and 100 lg/ml streptomycin (PAA Laborato-ries, Colbe, Germany). Recombinant human IL-1b was a gift fromCiba-Geigy (Basel, Switzerland). Dibutyryl-cAMP (Bt2-cAMP) wasobtained from Sigma (Taufkirchen, Germany). Cell cultures weremaintained at 37 �C in a humidified atmosphere containing 5% CO2

and subcultivated two to three times a week. Cells received fresh med-ium the day before the experiments. For study of hypoxia, cells wereplaced in a humidified atmosphere containing 3% O2, 5% CO2 and bal-anced N2 (Heraeus incubators, Hanau, Germany). In order to excludecytotoxic effects of IL-1b and Bt2-cAMP, the colorimetric tetrazoliumsalt/formazan method was applied, which is based on the reduction of3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT;Sigma) to purple formazan in living cells. Confluent HepG2 culturesin 96-well dishes were used for these studies as described [17].

2.2. Nuclear protein extractionNuclear proteins were extracted as described in detail elsewhere

[9,18]. HepG2 cells were washed with ice-cold phosphate-buffered sal-ine (PBS), scraped off and collected. For Western blot and electropho-retic mobility shift assay (EMSA) analysis of NF-jB and GATA-2, thecell suspensions were centrifuged at 4000 · g and 4 �C for 5 min. Pel-leted cells were resuspended in 350 ll buffer containing 10 mM HEPES

blished by Elsevier B.V. All rights reserved.

Fig. 1. (A) Epo concentrations in cell culture supernatants of HepG2cells exposed to 3% O2 without drugs (HOX) or with 50 lM Bt2-cAMP(HOX + cAMP), 300 pg/ml IL-1 (HOX + IL-1) or the combination ofIL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 24 h. *P 6 0.05compared to the respective cultures without Bt2-cAMP, mean + SD,n = 5, Student’s t-test. (B) Epo mRNA levels quantified by real-timeRT-PCR of extracts of HepG2 cells exposed to 3% O2 without drugs(HOX) or with 50 lM Bt2-cAMP (HOX + cAMP), 300 pg/ml IL-1(HOX + IL-1) or the combination of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 8 h. Expression levels normalized to L28 and related toHOX were calculated with the DDCT method. *P 6 0.05 comparedto the respective cultures without Bt2-cAMP, mean + SD, n = 6,Student’s t-test.

3154 C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160

(pH 7.9), 1.5 mM MgCl2 and 10 mM KCl and placed on ice for15 min. For cell lysis 25 ll of 10% NP-40 solution was added andmixed vigorously for 30 s. Nuclei were collected by centrifugationand resuspended in 60 ll buffer containing 20 mM HEPES (pH 7.9),420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA and 25% glycerol.The suspension was gently agitated on a shaking platform on ice for20 min. Nuclei were centrifuged at 12500 · g and 4 �C for 5 min andthe extracts stored at �80 �C.

For Western blot and EMSA analysis of HIF-1, cell suspensionswere centrifuged at 800 · g and 4 �C for 5 min. Cell pellets werewashed with 2 ml buffer A (10 mM Tris (pH 7.8), 1.5 mM MgCl2,10 mM KCl) and subsequently resuspended in 1 ml buffer A andplaced on ice for 30 min. Nuclei were collected by centrifugation andresuspended in 100 ll buffer C (10 mM KCl, 20 mM Tris (pH 7.8),1.5 mM MgCl2, 20% glycerol) by gentle up and down pipetting. Nucleiwere pelleted by centrifugation at 13000 · g and 4 �C for 30 min.Supernatants were stored at �80 �C. Immediately before use, buffersA and C were supplemented with 2 lg/ml aprotonin, 10 lg/ml leupep-tin, 20 lg/ml pepstatin, 1 mM sodium ortho-vanadate, 0.5 mM ben-zamidine, 2 mM levamisole, 10 mM b-glycerophosphate, 0.5 mMdithiothreitol (DDT) and 0.4 mM phenylmethylsulfonyl fluoride(PMSF). Protein concentrations were determined by the Bradfordmethod [19] with bovine serum albumin (BSA) as standard.

2.3. Western blottingSamples (25 lg of nuclear extracts) were run on denaturing 10% (for

NF-jB subunits) or denaturing 7.5% (for HIF-1a) polyacrylamide gelsand transferred to nitrocellulose-membranes (Hybond; Amersham,Biosciences). Transfer efficiency was verified by staining with 2% Pon-ceau S and/or immunodetection of SP1 or b-actin. Membranes wereblocked overnight at 4 �C in 5% skim milk/PBS and then incubatedwith the respective antibody at room temperature for 2 h. Antibodiesanti-NF-jB-p50, anti-NF-jB-p65, anti-SP1, anti-b-actin were fromSanta Cruz (Heidelberg, Germany) and anti-HIF-1a was from BDBiosciences (Heidelberg, Germany). All antibodies were used in a dilu-tion of 1:1000 in 5% skimmed milk in PBS. For detection, matchedhorseradish peroxidase conjugated secondary antibodies in a 1:2000dilution (all from DAKO, Hamburg, Germany) and enhanced chemi-luminescence substrate (Amersham, Freiburg, Germany) were used.

2.4. Electrophoretic mobility shift assayNF-jB and GATA-2 EMSAs were performed as described in [9] and

HIF-1 EMSAs as in [18]. Oligonucleotide sequences were as follows:NF-jB sense: 5 0-AGT TGA GGG GAC TTT CCC-30; NF-jB anti-sense: 5 0-GCC TGG GAA AGT CCC CTC-3 0; GATA-2 sense:5 0-CAC ACA TGC AGA TAA CAG CCC CGA CC-3 0; GATA-2antisense: 5 0-GGT CGG GGC TGT TAT CTG GAT GTG TG-3 0;HIF-1 sense: 5 0-TTC CTG CAC GTA CAC ACA AAG CGC ACGTAT TTC-3 0; HIF-1 antisense: 5 0-GAA ATA CGT GCG CTT TGTGTG TAC GTG CAG GAA-3 0. Oligonucleotides (MWG, Ebersberg,Germany) were labeled with T4 polynucleotide kinase (MBI Fermen-tas, St. Leon-Rot, Germany) in the presence of c-[32P]-ATP (NEN,Koln, Germany). Protein–DNA complexes were resolved by electro-phoresis in non-denaturing 6% polyacrylamide gels. For supershiftanalyses, anti-NF-jB-p50, NF-jB-p65 (all from Santa Cruz, Heidel-berg) and anti-HIF-1a (BD Biosciences) were used.

2.5. Luciferase assaysHRG-1 cells, which are HepG2 cells stably transfected with a hypox-

ia responsive luciferase plasmid [16], were used for monitoring HIF-1activity in reporter gene studies. Cells were grown to �40% confluenceon 24-well plates with 0.5 ml medium per well. Experimental periodswere 24 h. Thereafter, cells were washed with PBS and lysed with pas-sive lysis buffer (Promega, Mannheim, Germany). Luminescence wasmeasured with a Micro Lumate (LB 96P; Berthold Technologies,Bad Wildbad, Germany).

2.6. Enzyme-linked immunoassay (ELISA)Parental HepG2, IjBa-HepG2 or IjBaM-HepG2 cells were seeded

in 24-well plates and grown to 90–95% confluence. For study, cultureswere washed thoroughly with fresh medium and incubated under hyp-oxic conditions for 24 h. Epo concentrations were measured in cell cul-ture supernatants by commercial ELISA (Medac, Wedel, Germany).

2.7. Reverse transcription and polymerase chain reaction (RT-PCR)Total RNA was isolated according to Chomczynski and Sacchi [20]

or with the ABI Prism� 6100 NucleicAcid PrepStation (Applied Bio-systems, Darmstadt, Germany). One microgram of total RNA was re-verse transcribed into cDNA using oligo (dT) primers. Target cDNAswere quantified by real-time RT-PCR on an ABI 7000 Sequence detec-tion system (Applied Biosystems) by means of either a commercialSYBR green PCR kit (for the detection of L28; Eurogentec, Seraing,Belgium) or Assays-on-Demand (for the detection of Epo; AppliedBiosystems, part number 4331182). Relative expression levels were cal-culated using the DDCT-method (normalized to L28 and related to thenormoxic control). L28 primer sequences were: L28-for-48: 5 0-ATGGTC GTG CGG AAC TGC T-3 0 and L28-rev-149: 5 0-TTG TAGCGG AAG GAA TTG CG-3 0.

2.8. StatisticsResults are shown as means + standard deviations (SD). Student’s t-

test was applied to compare mean values in cultures without and withBt2-cAMP. A significant difference was assumed at P 6 0.05.

3. Results

IL-1 (300 pg/ml) reduced the amount of immunoreactive

Epo produced in hypoxically incubated (3% O2) HepG2

cultures (Fig. 1A). The addition of Bt2-cAMP (50 lM) partly

C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160 3155

prevented the IL-1 induced suppression of Epo production.

Real time RT-PCR measurements showed that IL-1 treated

cells expressed only 50% of the amount of Epo mRNA in hyp-

oxic control cells after 8 h of incubation (Fig. 1B). Bt2-cAMP

moderately increased Epo mRNA levels in IL-1 treated cells.

Measurements by MTT assay proved that neither IL-1 nor

Bt2-cAMP were cytotoxic when added to the cultures for 24 h.

Immunoblotting of nuclear extracts of HepG2 cells revealed

that IL-1 stimulated the translocation of NF-jB p50 and p65

subunits (Fig. 2A). Bt2-cAMP slightly weakened the induction

of NF-jB by IL-1. Clearly, IL-1 increased NF-jB DNA-bind-

Fig. 2. (A) Western blot of p50 and p65 NF-jB subunits in nuclear extractshypoxic HepG2 cells treated with 300 pg/ml IL-1 (HOX + IL-1), 10 lM Bt2-for 1 h were subjected to SDS–PAGE. Abbreviation. unsp., unspecific signal.binding in nuclear extracts from HepG2 cells maintained at 20% O2 (NOX) orml IL-1 (HOX + IL-1), 10 lM Bt2-cAMP (HOX + cAMP) or the combinatiop65, DNA-binding of p65/p65 homodimers; p65/p50, DNA-binding of p65/punspecific DNA-binding; free probe, unbound oligonucleotides. (C) Reporreporter plasmid and cultivated at 20% O2 (NOX) or 7% O2 (HOX) for 2(NOX + IL-1, HOX + IL-1), 50 lM Bt2-cAMP (NOX + Bt2-cAMP, HOX +1 + Bt2-cAMP, HOX + IL-1 + Bt2-cAMP) for 22 h. Luciferase activity in culthe respective culture without Bt2-cAMP, mean + SD, n = 4, Student’s t-testNF-jB-responsive element; SV40, minimal promoter of the Simian virus 40;signal.

ing (homodimers and heterodimers; Fig. 2B). Bt2-cAMP alone

did not alter the pattern of NF-jB DNA-binding. However,

Bt2-cAMP strongly blocked IL-1 induced NF-jB DNA-bind-

ing (Fig. 2B).

To further prove the importance of NF-jB in modulating

the effects of Bt2-cAMP, NF-jB dependent reporter-gene as-

says were performed. IL-1 induced luciferase activity in nor-

moxia as well as in hypoxia. Bt2-cAMP did not alter NF-jB

dependent luciferase activity in the absence of IL-1, whereas

Bt2-cAMP greatly blocked the IL-1 effect (Fig. 2C). However,

hypoxia alone was sufficient to slightly induce NF-jB depen-

(25 lg) from normoxic (20% O2: NOX), hypoxic (3% O2: HOX) andcAMP (HOX + cAMP) or with a combination of IL-1 and Bt2-cAMPEqual loading was verified by detection of b-actin. (B) NF-jB DNA-3% O2 (HOX) for 1 h. Hypoxic cultures were also treated with 300 pg/n of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP). Abbreviations. p65/50 heterodimers; p50/p50, DNA-binding p50/p50 homodimers; unsp.,

ter gene assay on HepG2 cells transiently transfected with a NF-jB4 h. After a medium renewal cells were treated with 100 pg/ml IL-1Bt2-cAMP) or the combination of IL-1 and Bt2-cAMP (NOX + IL-

tures exposed to normoxia (NOX) was set to 1. *P 6 0.05 compared to. (D) Schematic drawing of the used NF-jB reporter construct. NRE,luciferase, luciferase gene of Photinus pyralis; AAA, SV40 late poly(A)

3156 C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160

dent luciferase activity. A schematic drawing of the used repor-

ter plasmid is seen in Fig. 2D.

Epo production was compared in genetically engineered

HepG2 cells that either overexpressed IjBa (IjBa-HepG2)

or expressed a dominant/negative isoform of IjBa with two

Ser/Ala point-mutations (Ser 32 and Ser 36; IjBaM-HepG2).

The IjBa-HepG2 cells should behave like the parental HepG2

cell line, whereas the IkBaM-HepG2 line should be less able to

activate NF-jB upon IL-1 stimulation. IL-1 treatment reduced

the rate of Epo production in IjBa-HepG2 but not in IjBaM-

HepG2 cultures (Fig. 3A). Bt2-cAMP partly prevented the

reduction of Epo production in IL-1 treated IjBa-HepG2

cells. To further characterize the genetically modified HepG2

lines, phosphorylated IjBa proteins were quantified. Immuno-

blotting of whole cell extracts in response to IL-1 revealed

strong phosphorylation of IjBa proteins in parental HepG2

and in IjBa-HepG2 cells, while IjBaM-HepG2 cells showed

only weak phosphorylation of IjBa proteins (Fig. 3B). Fur-

thermore, overexpression of the mutated form of IjBa de-

creased NF-jB DNA-binding. This effect was most

prominent for p50/p50 homodimers.

Fig. 3. (A) Epo concentrations in cell culture supernatants of IjBa-HepG2 a50 lM Bt2-cAMP (HOX + cAMP), 300 pg/ml IL-1 (HOX + IL-1) or the c*P 6 0.05 compared to the respective cultures without Bt2-cAMP, mean + SDby Western blot analysis. HepG2, IjBa-HepG2 and IjBaM-HepG2 cellsmicrograms of whole cell extracts were subjected to SDS–PAGE. Cells were stverified by detection of b-actin. (C) EMSA for NF-jB DNA-binding activityFive micrograms of nuclear extracts from non-stimulated normoxic (20% O2

were used for EMSAs. Binding reactions were run on a 6% non-denaturatingp50/p50, homodimer of p50 subunits; sp. comp, specific competitor; unsp.Specificity of signals was verified by supershift analysis with anti-p50 and anp65). Antibodies were added to the binding reactions prior to the labelled o

Hypoxic incubation of HepG2 cells resulted in a loss of

the GATA-2 DNA-binding seen on normoxic incubation

(Fig. 4A and B). IL-1 restored GATA-2 DNA-binding un-

der hypoxic conditions. Bt2-cAMP administration also in-

creased GATA-2 DNA-binding in hypoxic cells. IL-1 and

Bt2-cAMP did not exert additive effects. Specificity was dem-

onstrated by adding unlabelled probe to the binding reaction

(Fig. 4C).

HIF-1a was not detectable in nuclear extracts from norm-

oxic cells. IL-1 treatment resulted in moderately increased

HIF-1a levels (Fig. 5A and B). Hypoxia led to a strong accu-

mulation of nuclear HIF-1a, which was further increased by

IL-1 treatment. Bt2-cAMP did not affect HIF-1a protein

amounts under any condition studied. HIF-1 DNA-binding

complexes were not detectable in nuclear extracts from norm-

oxic cells (Fig. 5C and D). Strong HIF-1 DNA-binding was

seen in cells exposed to hypoxia for 4 h. IL-1 increased this ef-

fect. Bt2-cAMP did not alter HIF-1 DNA binding (Fig. 5C and

D). Furthermore, Bt2-cAMP did not alter HIF-1 dependent

luciferase activity in hypoxic or in IL-1 treated cultures

(Fig. 6).

nd IjBaM-HepG2 cells incubated at 3% O2 (HOX) and treatment withombination of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 24 h., n = 5, Student’s t-test. (B) Detection of phosphorylated IjBa proteins

were exposed to the experimental conditions for 45 min. Twenty-fiveimulated with 300 pg/ml IL-1 or were left untreated. Equal loading wasin HepG2, IjBa-HepG2 and IjBaM-HepG2 cells in response to IL-1.NOX) cells and normoxic cells stimulated with 300 pg/ml IL-1 for 1 h

PAA gel. Abbreviations. p65/p50, heterodimer of p65 and p50 subunits;comp, unspecific competitor; free probe, unbound oligonucleotides.

ti-p65 antibodies (HepG2 + IL-1 + anti-p50 and HepG2 + IL-1 + anti-ligonucleotides.

Fig. 4. (A) GATA-2 DNA-binding in nuclear extracts (5 lg) from normoxic HepG2 cells (20% O2: NOX), hypoxic cells (3% O2: HOX) and hypoxiccells treated with 300 pg/ml IL-1 (HOX + IL-1), 10 lM Bt2-cAMP (HOX + cAMP) or with the combination of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 4 h. Abbreviations. GATA-2, GATA-2 specific DNA-binding; unsp., unspecific DNA-binding; free probe, unbound oligonucleotides.(B) Relative GATA-2 DNA-binding was determined by densitometric evaluation of the signals, with signal intensities of the normoxic samples beingset to 1. *P 6 0.05 compared to the respective control cultures without Bt2-cAMP, mean + SD, n = 5, Student’s t-test. (C) Specificity of GATA-2DNA-binding was verified by competition experiments with unlabeled specific and unspecific oligonucleotides (50-fold excess of unlabelledoligonucleotides).

C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160 3157

4. Discussion

Insufficent Epo production is not only common in end-stage

renal disease but also in chronic infections, autoimmune dis-

eases and malignancies, thereby contributing to the anemia

of chronic disease (ACD) [21,22]. IL-1 and TNF-a have been

shown to reduce Epo mRNA expression and Epo secretion

in the human hepatoma cell lines HepG2 and Hep3B [10–12]

and to inhibit Epo formation in rat kidneys in vivo and

in vitro [11,13]. On the other side, stimulation of Epo produc-

tion has been observed on the addition of cAMP analogues

(concentration range 10–1000 lM) to hypoxic human hepa-

toma cell cultures in absence of cytokines [12,23]. In the pres-

ent study, up to 50 lM Bt2-cAMP was applied, which proved

to be untoxic for HepG2 cultures. Evidence has been also pro-

vided earlier that cAMP can restore Epo secretion in HepG2

cells treated with IL-1 or TNF-a [12]. Along these lines the

present real-time RT-PCR measurements revealed increased

Epo mRNA levels following the addition of Bt2-cAMP to

IL-1 treated HepG2 cells.

The present findings suggest that NF-jB is the most impor-

tant transcription factor with respect to the antagonistic effect

of cAMP on the IL-1 repressed production of Epo. Oligo-de-

coy and reporter gene studies from our group have shown that

NF-jB inhibits the Epo promoter [9], although there is a re-

port to the opposite [24]. Gene suppression by NF-jB has been

explained by competition between NF-jB and other transcrip-

tion factors for the co-factors p300/CBP [25,26]. The transcrip-

tional co-activators p300/CBP are critically involved in

hypoxia-induced Epo expression [4]. The present studies

showed that IL-1 activated NF-jB DNA-binding. This activa-

tion was prevented, when the cells were treated with Bt2-

cAMP. The role of NF-jB during hypoxia is subject of contro-

versial discussions. It has been reported that NF-jB is acti-

vated by hypoxia in Jurkat T cells, HT29 and HepG2 cells

[27,28] although there is a report on the opposite [29]. Rupec

and Baeuerle [29] demonstrated that NF-jB is strongly acti-

vated upon reoxygenation and that within minutes. This might

be a critical point regarding the outcome of the experiment.

Yao and O’Dywer [28], who reported the activation of NF-

jB by hypoxia, allowed their HepG2 cell cultures to reoxygen-

ate after 24 h of hypoxia and thus it is unclear if NF-jB was

activated by hypoxia or reoxygenation. In our hands and using

HepG2 cells as a model, we only see slight differences between

normoxic and hypoxic nuclear NF-jB amounts and activity

(Fig. 2A and B). We cannot exclude that this slight induction

is evoked by reoxygenation, even when attention was payed on

harvesting cells quickly on ice and we therefore favor the

hypothesis that NF-jB is not greatly influenced by hypoxia

alone. In the context of Epo expression one has to take into

account, that the Epo gene is not regulated by one transcrip-

tion factor solely. Instead there are numerous potential bind-

ing sites for different transcription factors in the promoter

and the 3 0 enhancer region of the Epo gene. Epo expression

Fig. 5. (A and B) Western blot analysis of HIF-1a protein nuclear extract (25 lg) from HepG2 cells maintained at 20% O2 (NOX) or 3% O2 withoutdrugs (HOX) or with 300 pg/ml IL-1 (HOX + IL-1), 10 lM Bt2-cAMP (HOX + cAMP) or the combination of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 4 h. Equal loading was verified by detection of Sp1 protein. Relative HIF-1a amounts were determined by densitometry (HIF-1a-specific signal/Sp1-specific signal; hypoxia alone was set to 1). *P 6 0.05 compared to the respective control cultures without Bt2-cAMP, mean + SD,n = 3, Student’s t-test. (C and D) HIF-1 DNA-binding determined by EMSA in nuclear extracts from HepG2 cells maintained at 20% O2 (NOX) or3% O2 (HOX) for 4 h. Hypoxic cells were also treated with 300 pg/ml IL-1 (HOX + IL-1), 10 lM Bt2-cAMP (HOX + cAMP) or the combination ofIL-1 and Bt2-cAMP (HOX + IL-1 + cAMP). Specificity was verified by supershift analysis (HOX + IL-1 + anti HIF-1a) with an anti HIF-1aantibody. Abbreviations. unsp., unspecific DNA-binding; const., constitutive DNA-binding; HIF, HIF-1-specific DNA-binding; free probe, unboundoligonucleotides. The bars represent the densitometrical evaluation of HIF-1 specific signals related to corresponding constitutive signals. Thehypoxic control was set to 1. *P 6 0.05 compared to the respective control cultures without Bt2-cAMP, mean + SD, n = 3, Student’s t-test.

3158 C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160

is the result of an interplay of many different transcription fac-

tors of partly opposite functions. The balance of these tran-

scription factors determine the extent of Epo expression.

Hypoxia, which is the physiological stimulus for increased

Epo production, strongly induces HIF-1, whereas the induc-

tion of NF-jB by moderate hypoxia is, if at all, only very slight

(Fig. 2A and B). In contrast, IL-1 induces NF-jB strongly but

HIF-1 is only moderately activated [9,18]. That means, during

hypoxia there is an overweight of HIF-1 activity and the per-

haps existing slight activity of NF-jB is overrun. The situation

changes when IL-1 induces NF-jB. In this situation there is

high NF-jB activity and as a result the inhibitory action be-

comes prominent.

To further test for an involvement of NF-jB in the inhibi-

tory action of IL-1 on the Epo gene HepG2 cells stably trans-

fected with a dominant-negative form of the inhibitory protein

IjBa were used. In contrast to its effect in wild-type HepG2

cells or in cells expressing the normal form of IjBa, cAMP

did not antagonize that effect of IL-1 on Epo production in

cells expressing the mutated IjBa. The transcriptional activity

of NF-jB is partly dependent on phosphorylation of NF-jB

p65 [30]. It is thought that the transcriptional co-activator

Fig. 6. (A) Hypoxia dependent reporter gene assay with HRG-1 cellscultivated at 3% O2 (HOX) without or with 300 pg/ml IL-1(HOX + IL-1), 10 lM Bt2-cAMP (HOX + cAMP) or with the combi-nation of IL-1 and Bt2-cAMP (HOX + IL-1 + cAMP) for 22 h.Luciferase activity in cultures exposed to hypoxia (HOX) was set to1. *P 6 0.05 compared to the respective control cultures without Bt2-cAMP, mean + SD, n = 6, Student’s t-test. (B) Scheme of the hypoxiadependent luciferase reporter construct. TfHRE: Hypoxia responsiveelement of the transferrin gene, containing two HIF-1 binding sites,SV40: minimal promoter of the Simian virus 40, luciferase: luciferasegene from Photinus pyralis, AAA: SV40 late poly(A) signal.

C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160 3159

p300/CBP associates with NF-jB p65 in response to PKA

mediated phosphorylation [31]. In harmony with the present

findings in HepG2 cells, earlier studies have shown that the

activation of NF-jB by IL-1 is also suppressed by cAMP ele-

vation in human pancreatic cancer cells PaCa-2 [32].

GATA-2 blocks the Epo promoter [1,2]. The present study

showed that GATA-2 DNA-binding was reduced under hyp-

oxic condition in HepG2 cells. IL-1 and TNF-a activate

GATA-2 [9,33]. Importantly, the impaired production of

Epo due to the action of IL-1 or TNF-a is rescued by the

GATA-specific inhibitors K-7174 and K-11706 in vitro and

in vivo [33,34]. Based on the present observation, however,

GATA-2 does not appear to be involved in the effect of cAMP

on IL-1 suppressed Epo production.

HIF-1 is the primary transcription factor in the hypoxic

induction of Epo expression. cAMP response element-binding

protein (CREB) binding protein 1 (CBP-1) has been demon-

strated to bind to the HIF-1 DNA recognition site in HeLa

cells [23,35]. However, in the present study HIF-1 protein lev-

els and HIF-1 DNA-binding were not affected by Bt2-cAMP in

hypoxic and IL-1 treated HepG2 cells. Thus, in this model Bt2-

cAMP was unlikely to stimulate Epo production through the

HIF-1 DNA recognition site.

Reportedly, cAMP stimulates Epo synthesis in human renal

carcinoma and hepatoma cell lines [14,15,36], although drugs

activating the adenylate cyclase/cAMP/PKA pathway fail to

produce a major Epo mRNA increase in rat kidneys during

short term (2 h) perfusion experiments [37]. In a study on hu-

man volunteers the b2-adrenergic receptor agonist fenoterol,

which leads to an endogenous cAMP increase, proved to stim-

ulate Epo production in vivo [38]. It remains to be investigated

whether a beneficial value can be assigned to adenylate cyclase

activating drugs in patients suffering from impaired Epo syn-

thesis due to high levels of proinflammatory cytokines.

Taken together, our results suggest that the elevation of

intracellular cAMP inhibited NF-jB activity thereby partially

restoring cytokine-repressed Epo production. A better under-

standing of the molecular mechanism of the counteraction be-

tween cAMP and IL-1 inhibited Epo production will be

important to design new therapeutic options to treat patients

with ACD.

Acknowledgements: We are grateful to Ms. Gabriele Huck for excellenttechnical assistance. Financial support was provided by the GermanResearch Society (DFG, GRK-288) and the Mongolian Academy ofSciences.

References

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

[2] Imagawa, S., Yamamoto, M. and Miura, Y. (1997) Negativeregulation of the erythropoietin gene expression by the GATAtranscription factors. Blood 89, 1430–1439.

[3] Lee-Huang, S., Lin, J.J., Kung, H.F., Huang, P.L., Lee, L.and Huang, P.L. (1993) The human erythropoietin-encodinggene contains a CAAT box, TATA boxes and othertranscriptional regulatory elements in its 5 0 flanking region.Gene 128, 227–236.

[4] Ebert, B.L. and Bunn, H.F. (1999) Regulation of the erythropoi-etin gene. Blood 94, 1864–1877.

[5] Semenza, G.L. and Wang, G.L. (1992) A nuclear factor inducedby hypoxia via de novo protein synthesis binds to the humanerythropoietin gene enhancer at a site required for transcriptionalactivation. Mol. Cell Biol. 12, 5447–5454.

[6] 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 het-erodimer regulated by cellular O2 tension. Proc. Natl. Acad. Sci.USA 92, 5510–5514.

[7] Metzen, E. and Ratcliffe, P.J. (2004) HIF hydroxylation andcellular oxygen sensing. Biol. Chem. 385, 223–230.

[8] Semenza, G.L. (2004) Hydroxylation of HIF-1: oxygen sensing atthe molecular level. Physiology (Bethesda) 19, 176–182.

[9] La Ferla, K., Reimann, C., Jelkmann, W. and Hellwig-Burgel, T.(2002) Inhibition of erythropoietin gene expression signalinginvolves the transcription factors GATA-2 and NF-jB. FASEB J.16, 1811–1813.

[10] Faquin, W.C., Schneider, T.J. and Goldberg, M.A. (1992) Effectof inflammatory cytokines on hypoxia-induced erythropoietinproduction. Blood 79, 1987–1994.

[11] Jelkmann, W., Pagel, H., Wolff, M. and Fandrey, J. (1992)Monokines inhibiting erythropoietin production in human hep-atoma cultures and in isolated perfused rat kidneys. Life Sci. 50,301–308.

[12] 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 geneexpression. Eur. J. Biochem. 226, 335–340.

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

[14] Sherwood, J.B., Burns, E.R. and Shouval, D. (1987) Stimulationby cAMP of erythropoietin secretion by an established humanrenal carcinoma cell line. Blood 69, 1053–1057.

[15] Nakashima, J., Brookins, J., Beckman, B. and Fisher, J.W. (1991)Increased erythropoietin secretion in human hepatoma cells byN6-cyclohexyladenosine. Am. J. Physiol. 261, C455–C460.

[16] Stiehl, D.P., Jelkmann, W., Wenger, R.H. and Hellwig-Burgel, T.(2002) Normoxic induction of the hypoxia-inducible factor-1a by

3160 C. Batmunkh et al. / FEBS Letters 580 (2006) 3153–3160

insulin and interleukin-1b involves the phosphatidylinositol 3-kinase pathway. FEBS Lett. 512, 157–162.

[17] Wolff, M. and Jelkmann, W. (1993) Effects of chemotherapeuticand immunosuppressive drugs on the production of erythropoi-etin in human hepatoma cultures. Ann. Hematol. 66, 27–31.

[18] Hellwig-Burgel, T., Rutkowski, K., Metzen, E., Fandrey, J. andJelkmann, W. (1999) Interleukin-1b and tumor necrosis factor-astimulate DNA binding of hypoxia-inducible factor-1. Blood 94,1561–1567.

[19] Bradford, M.M. (1976) A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72, 248–254.

[20] Chomczynski, P. and Sacchi, N. (1987) Single-step method ofRNA isolation by acid guanidinium thiocyanate–phenol–chloro-form extraction. Anal. Biochem. 162, 156–159.

[21] Jelkmann, W. (1998) Proinflammatory cytokines lowering eryth-ropoietin production. J. Interferon Cytokine Res. 18, 555–559.

[22] Weiss, G. and Goodnough, L.T. (2005) Anemia of chronicdisease. N Engl. J. Med. 352, 1011–1023.

[23] Kvietikova, I., Wenger, R.H., Marti, H.H. and Gassmann, M.(1997) The hypoxia-inducible factor-1 DNA recognition site iscAMP-responsive. Kidney Int. 51, 564–566.

[24] Figueroa, Y.G., Chan, A.K., Ibrahim, R., Tang, Y., Burow,M.E., Alam, J., Scandurro, A.B. and Beckman, B.S. (2002) NF-jB plays a key role in hypoxia-inducible factor-1-regulatederythropoietin gene expression. Exp. Hematol. 30, 1419–1427.

[25] Ke, S., Rabson, A.B., Germino, J.F., Gallo, M.A. and Tian, Y.(2001) Mechanism of suppression of cytochrome P-450 1A1expression by tumor necrosis factor-alpha and lipopolysaccha-ride. J. Biol. Chem. 276, 39638–39644.

[26] Gires, O., Kieu, C., Fix, P., Schmitt, B., Munz, M., Wollenberg,B. and Zeidler, R. (2001) Tumor necrosis factor alpha negativelyregulates the expression of the carcinoma-associated antigenepithelial cell adhesion molecule. Cancer 92, 620–628.

[27] Koong, A.C., Chen, E.Y. and Giaccia, A.J. (1994) Hypoxiacauses the activation of nuclear factor kappa B through thephosphorylation of I kappa B alpha on tyrosine residues. CancerRes. 54, 1425–1430.

[28] Yao, K.S. and O’Dwyer, P.J. (1995) Involvement of NF-kappa Bin the induction of NAD(P)H:quinone oxidoreductase (DT-diaphorase) by hypoxia, oltipraz and mitomycin C. Biochem.Pharmacol. 49, 275–282.

[29] Rupec, R.A. and Baeuerle, P.A. (1995) The genomic response oftumor cells to hypoxia and reoxygenation. Differential activationof transcription factors AP-1 and NF-kappa B. Eur. J. Biochem.234, 632–640.

[30] Vermeulen, L., De Wilde, G., Notebaert, S., Vanden Berghe, W.and Haegeman, G. (2002) Regulation of the transcriptionalactivity of the nuclear factor-kappaB p65 subunit. Biochem.Pharmacol. 64, 963–970.

[31] Zhong, H., Voll, R.E. and Ghosh, S. (1998) Phosphorylation ofNF-kappa B p65 by PKA stimulates transcriptional activity bypromoting a novel bivalent interaction with the coactivator CBP/p300. Mol. Cell 1, 661–671.

[32] Kamthong, P.J., Wu, F.M. and Wu, M.C. (2000) cAMPattenuates interleukin-1-stimulated macrophage colony-stimu-lating factor (M-CSF) expression. Biochem. J. 350 (Pt 1),115–122.

[33] Imagawa, S., Nakano, Y., Obara, N., Suzuki, N., Doi, T.,Kodama, T., Nagasawa, T. and Yamamoto, M. (2003) A GATA-specific inhibitor (K-7174) rescues anemia induced by IL-1beta,TNF-alpha, or L-NMMA. FASEB J. 17, 1742–1744.

[34] Nakano, Y., Imagawa, S., Matsumoto, K., Stockmann, C.,Obara, N., Suzuki, N., Doi, T., Kodama, T., Takahashi, S.,Nagasawa, T. and Yamamoto, M. (2004) Oral administration ofK-11706 inhibits GATA binding activity, enhances hypoxia-inducible factor 1 binding activity, and restores indicators in anin vivo mouse model of anemia of chronic disease. Blood 104,4300–4307.

[35] Kvietikova, I., Wenger, R.H., Marti, H.H. and Gassmann, M.(1995) The transcription factors ATF-1 and CREB-1 bindconstitutively to the hypoxia-inducible factor-1 (HIF-1) DNArecognition site. Nucleic Acids Res. 23, 4542–4550.

[36] Hagiwara, M., Pincus, S.M., Chen, I.L., Beckman, B.S. andFisher, J.W. (1985) Effects of dibutyryl adenosine 3 0,5 0-cyclicmonophosphate on erythropoietin production in human renalcarcinoma cell cultures. Blood 66, 714–717.

[37] Tan, C.C. and Ratcliffe, P.J. (1992) Rapid oxygen-dependentchanges in erythropoietin mRNA in perfused rat kidneys:evidence against mediation by cAMP. Kidney Int. 41, 1581–1587.

[38] Gleiter, C.H., Becker, T., Schreeb, K.H., Freudenthaler, S. andGundert-Remy, U. (1997) Fenoterol but not dobutamineincreases erythropoietin production in humans. Clin. Pharmacol.Ther. 61, 669–676.