ABSTRACT
Transcription of the erythropoietin (epo) gene is regulated in response to tissue hypoxia. In this study we show that constructs containing 117 bp of the epo promoter sequence cloned upstream of a luciferase reporter, respond to hypoxia when transfected into the human hepatoma cell line, Hep3B. The sequence -61 to -45 (EP17) relative to the transcription start of the murine epo gene imparted an ~4-fold induction of reporter gene expression due to hypoxia. Internal deletion of EP17 resulted in loss of induction by hypoxia without altering basal expression of the 117 bp epo promoter reporter construct. Mutagenesis studies showed that the bases at positions -53, -59, from -49 to -51 and from -55 to -57 are essential for hypoxic induction. The EP17 sequence is required for the 3' enhancer element of the epo gene to be maximally functional. Gel shift and UV cross-linking experiments showed the presence in Hep3B nuclear extracts, of two protein factors with approximate molecular weights of 52 kDa and 25 kDa that bind to EP17. Introduction of specific mutations in the EP17 region that abolish induction by hypoxia, also eliminated the binding of one or both of these factors. These experiments demonstrate a role for the proximal region of the epo promoter in hypoxic induction of the epo gene.
The glycoprotein, erythropoietin (epo) is the primary regulator of blood cell formation in mammals (
1
). It stimulates the proliferation and differentiation of erythroid progenitor cells (
2
,
3
). In adult animals it is mainly produced in kidneys in response to reduced tissue oxygenation
while in fetal life the liver is the major site of production (
4
,
5
). In spite of active investigation, the mechanisms by which specific cells can recognize relative hypoxia and respond by increased transcription of the epo gene are still not clarified.
Transcriptional regulation of gene expression is, in part, at the level of
initiation and is due to interaction of sequence-specific DNA binding proteins with the regulatory elements located in the
promoter, the 5' and 3' untranslated and intronic regions of the eukaryotic genes. Transcriptional activation involves complex and often cooperative interactions of various factors and gene
expression is probably determined at least in part by the presence or activity of specific factors in a particular cell type (
6
-
8
). The binding activity or abundance of these factors may be regulated by
extracellular signals. Since transcription of the epo gene is affected by tissue oxygenation, it offers an excellent model for the study of the
factors that are sensitive to extracellular cues.
The human hepatoma cell line, Hep3B, which can synthesize and secrete epo in a
regulated manner
in vitro
(
9
), has proven to be a useful system for investigating the transcriptional
regulation of the epo gene (
10
). During the last few years, using these cells, gene transfer experiments with
deletions of the promoter and 3' untranslated regions of the epo gene have identified the presence of a
cis
-acting element in the promoter region and an enhancer element in the 3' untranslated region (
11
-
17
).
Deletion and mutation experiments have established a 117 bp promoter sequence
and a 50 bp 3' enhancer sequence as minimal sequences necessary for the epo gene to
respond to hypoxia (
11
,
12
,
15
). Hypoxic induction of the epo gene was shown to involve a cooperative
interaction of the promoter and enhancer binding factors (
11
).
In vitro
DNA binding studies using nuclear extracts from a number of different cell lines
have shown the presence of a 120 kDa hypoxia-inducible factor (HIF-1) that binds to the 5' portion of the enhancer (
15
). The induction of HIF-1 was shown to be dependent on protein phosphorylation (
16
) and the recent cloning of the HIF-1 cDNA showed that it is a basic helix-loop-helix-PAS heterodimer (
18
). The possible roles of hepatic nuclear factor 4 and of the COUP family of
proteins in tissue-specific and hypoxia-inducible expression of the epo gene have also been demonstrated (
19
).
We have previously shown that the region from -61 to -45 (EP17) relative to the transcription start site of the mouse epo gene
binds to both a protein and a ribonucleoprotein in kidney nuclear extracts. The
RNA component of the latter is down-regulated in response to hypoxia, suggesting a functional (possibly
repressor) role for this component in epo transcription in kidneys (
20
). In the present study we demonstrate the functional importance of EP17 in
hypoxic induction of epo gene transcription in Hep3B cells. Our data also
suggest that this sequence of the promoter region cooperates with the
previously identified 3' enhancer element. In addition, we present evidence for two factors
binding to EP17 in nuclear extracts of Hep3B cells exposed to normal O
2
level and to hypoxia.
Hep3B cells (ATCC HB-8064) were grown at 37oC in 5% CO
2
95% air in [alpha] minimal Eagle's medium (Sigma Chemical Co., St. Louis, MO) supplemented with 10% fetal bovine serum (Hyclone Inc., Logan, UT). Penicillin and streptomycin were added at 10 U/ml and 5 [mu]g/ml respectively. The skeletal muscle cell line Sol8 was grown to
confluency in DMEM supplemented with 20% fetal bovine serum and penicillin (10 U/ml), streptomycin (5 [mu]g/ml), and neomycin (100 [mu]g/ml). Myogenic differentiation of these cells was induced by lowering
the serum concentration to 10% as described earlier (
21
).
The plasmids [Delta]H and [Delta]18 were generously provided by Dr H. Franklin Bunn and contain the 385 and 117 bp promoter sequence of the
human epo gene cloned upstream of a luciferase reporter. Internal deletion
mutations and point mutations of the 17 bp sequence (EP17) in [Delta]18 were generated by polymerase chain reaction (PCR) as described (
22
). The reamplified product was cloned into the promoterless luciferase reporter
vector, pX2, for transfection and into pBluescript (Stratagene, La Jolla, CA)
for sequence determination. The enhancer element located in the 3' untranslated region of the epo gene was amplified by PCR using human
genomic DNA as a template. The PCR primers flanked the 3448-3549 region of the epo gene and contained
Bam
HI and
Sac
I cloning sites. The amplified enhancer fragment was cloned upstream of both [Delta]18 and of the internal deletion mutant plasmid. Sequences of all deletion and site specific mutants were confirmed by dideoxy DNA sequencing (
23
) using Sequenase version 2.0 (USB, Cleveland, OH).
One day before transfection, the Hep3B cells were harvested and plated at a density of 1 * 10
6
cells per 100 mm diameter plate. Fifteen micrograms of test DNA and 5 [mu]g of pCMV [beta]-galactosidase DNA were prepared as calcium phosphate precipitates and the
cells were transfected as described (
24
). All transfection experiments were performed in triplicate. The cells were then incubated at 37oC in an atmosphere of 5% CO
2
, 95% air (18% O
2
) for 14-16 h. The medium was changed and the cells incubated for an additional 48
h. For hypoxic exposure parallel dishes were incubated for 48 h in an
atmosphere of 2% O
2
, 5% CO
2
and 93% N
2
.
The cells were harvested and lysed in lysis buffer (25 mM glycylglycine; 15 mM
MgSO
4
, 4 mM EGTA [ethylene glycol-
bis
([beta]-aminoethyl ether)-
N
,
N
,
N
1
,
N
1
-tetracetic acid], 1 mM DTT, 1% Triton X 100). The protein concentration of the cell lysate was measured using a Bio-Rad protein assay reagent (BioRad, Richmond, CA). Chemiluminescense
assay for luciferase activity determination was performed in duplicate with
equal amounts of proteins (50-100 [mu]g) using an assay kit from Promega (Madison, WI). All experiments
were repeated 4-6 times with three different plasmid preparations to ensure reproducibility. Luciferase activities were corrected for variation in transfection efficiencies as determined by assaying the cell extracts for [beta]-galactosidase activity (
25
).
Hep3B cells were grown to confluence and exposed to normoxic and hypoxic
conditions for 48 h, as described above. Nuclear extracts were prepared by the
method of Abmayr and Workman (
26
), with the following modifications: the high salt buffer contained 20% glycerol
and 30 mM KCl and the dialysis buffer contained 25% glycerol and 20 mM KCl. The
extracts were dialyzed for 6 h at 4oC. The dialysate was quick frozen in liquid N
2
and stored in aliquots at -70oC. The nuclear extract from differentiated Sol8 cells was prepared
essentially as described above for Hep3B cells.
Synthetic oligonucleotides were purified by electrophoresis on 15% denaturing
polyacrylamide gels and then subjected to chromatography on Sep-Pak C
18
columns (Waters Associates, Milford, MA) (
27
). For use as competitors or probes in gel mobility shift assays, complementary
strands of each oligonucleotide were denatured in 10 mM Tris-HCl (pH 7.5), 1 mM EDTA at 85oC for 3 min, annealed by slow cooling at room temperature and
purified on a 15% polyacrylamide gel. Double stranded oligonucleotide probes
were 5' end labelled with [[gamma]
32
P]dATP using T
4
polynucleotide kinase (United States Biochemical Corp., Cleveland, OH). The following double stranded oligonucleotides (oligos) were used in this study. Oligo EP17
(CCCCCACCCCCACCCGC) corresponds to the region from -61 to -45 of the murine epo gene. The mutant oligos were: mut 1 (CC
Electrophoretic mobility shift assays were performed using nuclear extracts of
Hep3B (normoxic and hypoxic) and Sol8 cells. Binding reactions were carried out
in a total volume of 25 [mu]l containing 50 000 c.p.m. (0.1-1 ng) of the radiolabelled probe, 5-10 [mu]g of total protein of the nuclear extract and 2 [mu]g of double stranded poly (dI[middot]dC) as non-specific competitor. The binding buffer consisted of 10 mM Tris-HCl (pH 7.4), 100 mM NaCl, 0.1 mM EGTA, 0.5 mM
dithiothreitol, 0.3 mM MgCl
2
, 8% glycerol and 0.5 mM phenylmethylsulfonyl fluoride. Binding was performed at 25oC for 20 min. Putative, unlabelled competitor oligos were added at 25-100-fold molar excess over the labelled probe for 3-5 min prior to addition of the probe. Binding reaction
mixtures with purified SP1 (3 ng) from HeLA cells contained 5 [mu]g of bovine serum albumin (BSA) and 0.5 [mu]g of the double stranded poly (dI[middot]dC) nonspecific competitor. The reaction mixtures were loaded on 5% polyacrylamide gels (44:1, acrylamide/bis-acrylamide) containing 0.5* Tris-borate-EDTA buffer and electrophoresed with 0.5* Tris-borate-EDTA electrode
buffer (0.045 M Tris base, 0.045 M boric acid, 0.001 M EDTA). Electrophoresis
was carried out at 150 V at 4oC (
27
), the gels were fixed in 10% acetic acid, dried and exposed to Kodak XAR-5 film.
Double stranded probes were radiolabelled as described above. Binding reactions
were replicated and multiple identical reactions were run on the same
polyacrylamide gel. Following electrophoresis the wet gels were exposed to UV irradiation (254 nm) for 20 min and the gel
exposed to X-ray film overnight. The regions corresponding to protein-DNA complexes were cut from the gel and placed in a l.5 ml Eppendorf tube. The cross-linked complexes were eluted from the polyacrylamide by crushing the gel in
protein sample buffer [10% (vol/vol) glycerol, 0.25% (wt/vol) sodium dodecyl
sulfate (SDS), 62.5 mM Tris (pH 6.8), 2.5% (vol/vol) [beta]-mercaptoethanol, and 0.025% (wt/vol) bromophenol blue], supplemented
with 0.2 M NaCl. The slurry was incubated at 37oC for 2 h and at 95oC for 2 min and spun through a Schleicher and Schuell Centrex spin
filter, packed with glass wool. The filtrate was run on an SDS 12.5%
polyacrylamide electrophoresis gel with Rainbow molecular weight markers
(Amersham, Arlington Heights, IL).
The role of the EP17 sequence in the 5' untranslated region of the epo promoter in the regulation of epo gene
expression was studied in Hep3B cells transiently transfected with luciferase
reporter constructs containing different sub-fragments of the human epo upstream sequence and compared with the
promoterless control plasmid pX2. As shown in Figure
1
, following hypoxic stimulation, cells transfected with the plasmid [Delta]H [385 bp of epo 5' flanking sequence linked to the luciferase reporter] showed 4-6-fold greater activation of luciferase than those with
pX2. When the plasmid, [Delta]18, containing 117 bp of the epo 5' flanking sequence was transfected into Hep3B cells there was a
similar increase in luciferase activity in response to hypoxia. To determine whether EP17 plays a functional role in the response to hypoxia, an internal
deletion of the EP17 sequence in the [Delta]18 plasmid was constructed ([Delta]EP[Delta]18). With Hep3B containing this construct hypoxia did not
induce increased reporter expression. The loss of response to hypoxia is not
due to a distance effect caused by the deletion in the epo promoter because, as
described later, introduction of specific mutations in the EP17 region resulted
in a similar loss of hypoxia-induced luciferase activity. These data clearly indicate that the EP17 region of the epo promoter participates to some extent in hypoxia-inducible expression of the epo gene. When the EP17 sequence was cloned upstream of the TK promoter driving luciferase expression, no hypoxic induction was seen (Table
1
).
In agreement with others (
11
), we have found that the enhancer element located in the 3' untranslated region of the epo gene is 5-fold more active in conjunction with the epo promoter than when
cloned upstream of a heterologous promoter. It therefore suggests that factors
binding to the enhancer region must, in some way, interact with those binding
to the promoter region. To determine whether this interaction is mediated via
factors binding to the EP17 region, three additional constructs were made and
the reponse with respect to induction by hypoxia was compared with [Delta]18. These three were [Delta]18Enh where the 101 bp enhancer element was cloned upstream of [Delta]18, [Delta]EP[Delta]18Enh where the enhancer was cloned upstream
of the EP17 deletion mutant and TKenh where the same enhancer element was
cloned upstream of the thymidine kinase promoter driving luciferase expression.
Each of these DNAs in combination with pCMV [beta]-galactosidase was transfected into Hep3B cells which were then exposed to normal levels of O
2
or to low O
2
and luciferase activity, normalized to [beta]-galactosidase activity, determined. As shown in Figure
1
B, [Delta]18Enh showed ~18-fold induction of luciferase activity in response to hypoxia.
In contrast the magnitude of enhancer activity was drastically reduced (4-fold induction) when the EP17 sequence was deleted in the same construct ([Delta]EP[Delta]18Enh) (Fig.
1
B). There was only a 5-fold induction in luciferase activity with the heterologous promoter, TKenh (Table
1
). These data suggest that the C rich EP17 region is necessary but not sufficient
for the contribution of the promoter to the response to hypoxia.
These transfection experiments indicate that the EP17 region of the epo promoter is functionally important for the hypoxia-inducible expression of the epo gene. Electrophoretic mobility shift assays were
performed to identify nuclear proteins that may bind to this region. Extracts
were prepared from Hep3B cells grown under normoxic and hypoxic conditions. A
synthetic double stranded oligonucleotide corresponding to the EP17 sequence of
the epo promoter was radiolabelled and used as a probe. As shown in Figure
2
, this sequence binds to two nuclear factors (C1 and C2) in extracts from both normoxic and hypoxic Hep3B nuclei (lanes 2 and 8). Each of the EP17 binding factors was sequence specific, because the
binding was competed for by excess unlabeled specific oligonucleotides EP17
(Fig.
4
, lane 4) as well as by EP31, a DNA sequence from -31 to -61 in the epo gene which includes EP17 (Fig.
2
, lanes 3 and 11) but not by the nonspecific oligonucleotides EP22,
corresponding to -80 to -101 region of the epo gene (Fig.
2
, lanes 4 and 9), or the polylinker region of the pX2 vector (Fig.
2
, lanes 6 and 12). Rough quantitation showed that the C1 and C2 components were
not significantly induced by hypoxia.
We have identified a DNA element, EP17, at -61 to -45 relative to the transcription start site of the mouse epo gene, that is
involved in hypoxia-regulated epo gene expression in Hep3B cells. Transfections of Hep3B cells
with a human 117 bp epo promoter/luciferase construct conferred 4-fold induction in reporter activity due to hypoxia. Internal deletion or
point mutations in EP17 resulted in loss of response to hypoxia but the basal expression remained unaltered, suggesting that factors binding to the EP17 element are activators of epo gene transcription in hypoxic cells.
These data are in agreement with earlier reports documenting the role of the
promoter region in hypoxia-induced regulation of the epo gene (
11
,
12
,
31
). Two earlier reports (
13
,
32
) in contrast failed to show the presence of an hypoxia-responsive element in the 5' flanking region.
In the present study, when the EP17 region was placed upstream of a heterologous
promoter, there was no induction by hypoxia, suggesting that it must interact
with other elements in the epo promoter region. It is interesting to note that
a DNA element (P-1) located adjacent to the EP17 element (from -65 to -117) was shown to bind to factors in Hep3B nuclear extracts
and to regulate the hypoxia-induced increase in epo gene transcription (
11
). Neither the P-1 (
11
) nor the EP17 (the present study) elements were found active on a heterologous
promoter. Further when EP17 was either deleted or point mutated in the 117 bp
epo promoter/reporter construct, any induction contributed by the P-1 element was lost. These data suggest a possible interaction of the P-1 and EP17 binding factors. Our data, in addition, suggest that EP17 acts cooperatively with the enhancer element located ~120 bp 3' to polyA addition site of the human epo gene, because
when EP17 is deleted, the hypoxia inducibility confered by the enhancer element
is not different from that found with a heterologous promoter.
Gel shift and UV cross-linking experiments with Hep3B nuclear extracts show that two proteins of
apparent molecular weights of 52 kDa and 25 kDa interact with EP17. The factors
C1 and C2 that bind to EP17 are not induced by hypoxia. Further, we have no
evidence to suggest that these factors are modified by hypoxia because in gel-shift assay the mobility pattern of the complexes are not altered by
exposure of cells to low oxygen. The increased transactivation of the epo
promoter/reporter constructs seen in the transfection experiments is possibly
achieved via factor-factor interaction with the epo promoter and enhancer element, that binds
to a hypoxia-inducible factor, HIF-1 (
15
). Such an interaction may be responsible for the EP17 role in hypoxia-induced reporter expression shown in Figure
1
. If so, the interaction may be transient since we see no evidence for it in
mobility shift experiments possibly because of dissociation of the protein-protein complexes during electrophoresis. Unstable protein-protein interaction was demonstrated for a homeodomain protein, Phox, and the serum response factor (SRF), for target
sequence element, SRE.
In this study Phox-1 could modulate the functional response of SRF and the rate of SRF-SRE association/dissociation in the absence of its own target
binding sequence and without altering the mobility of the SRF-SRE complex in mobility shift assay (
33
).
With the recent cloning of HIF-1 cDNA (
18
), it may now be possible to study the interaction of the EP17 DNA-protein
complex with HIF-1.
Gel shift analysis with mutated EP17 indicated that in each of the two CAC
sequences the flanking Cs are essential for binding to C1 and C2 but that the A
can be replaced by G without affecting binding. Cross-linking studies, however, show that the apparent molecular weights of C1
(52 kDa) and C2 (25 kDa) are different from those of SP1 (130 kDa) or of an SP1
related transcription factor (82 kDa) that bind to the CAC sequence (
34
,
35
).
The mechanisms by which hypoxia affects epo gene transcription are unclear. Reports from various laboratories indicate that there are several
points in the epo gene where
trans
-acting factors could regulate epo gene transcription. For example, the
enhancer element is shown to bind to at least two transcription factors, one of
which (HIF-1) is induced in response to hypoxia (
15
). The protein factor binding to the P-1 element is shared by the enhancer element and is not induced by hypoxia
(
11
). The two protein factors binding to EP17 described in this study are not
induced by hypoxia and their molecular weights do not correspond to the
previously described factors involved in regulation of epo gene transcription. Based on these studies we propose that the induction of HIF-1 due to hypoxia may allow a cooperative interaction of various constitutively expressed factors interacting with the 3' region of the enhancer element, the P-1 element and the EP17 region. These interactions may lead to
the strong induction of epo gene transcription following exposure of Hep3B
cells to hypoxia.
This work was supported by grants HL30121 and HL21676 from the National Heart,
Lung, and Blood Institute and by a gift from Kirin-Amgen, Inc. We are grateful to Paul Mungai for his extensive technical assistance throughout this work. We also wish to
acknowledge Smilija Jakovcic for critical review of the manuscript, Edward Des Jardins for supplying purified SP1, H. Franklin Bunn (Harvard
Medical School) for providing [Delta]18pX2 and [Delta]H clones, and Betty Kniaz for excellent secretarial assistance.
*To whom correspondence should be addressed. Tel: +1 312 702 1348; Fax: +1 312
702 0439; Email: egoldwas@midway.uchicago.edu
+
Present address: The Heart Institute for Children, Christ Hospital Medical
Center, Oak Lawn, IL 60453, USA
REFERENCES
Return