| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ARTICLES |
Institut für Biochemie und Molekulare Zellbiologie, Georg-August-Universität, Humboldtallee 23, D-37073 Göttingen, Germany
Address all correspondence and requests for reprints to: Thomas Kietzmann, M.D., Institut für Biochemie und Molekulare Zellbiologie, Humboldtallee 23, D-37073 Göttingen, Germany.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
expression vectors. Furthermore, it was found by gel shift and
supershift assay that HIF-1 and USF-1 or USF-2 could bind to the
GlcPKRE. Our findings implicate that the cross-talk between
oxygen and glucose might have a fundamental role in the regulation of
several physiological and pathophysiological processes. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Hypoxia via the transcription factor hypoxia inducible factor 1 (HIF-1) appears to be a major signal for the regulation of genes involved in embryonic development, vascularization, tumorigenesis and glycolytic metabolism (3, 4, 5, 6, 7).
A special feature of liver metabolism is that it takes place in different areas of the liver acinus. Based on the blood supply the liver acinus represents the smallest functional unit of the liver and extents from the upstream periportal area to the downstream perivenous area (8). Due to metabolism and elimination, respectively, concentration gradients of substrates such as oxygen and hormones are formed during a single passage of blood through the liver (9, 10, 11, 12, 13). The oxygen tension is about 65 mmHg in the periportal area and falls to about 35 mmHg in the perivenous zone. This gradient in oxygen was considered to be a key regulator for the zonal expression of the genes of carbohydrate-metabolizing enzymes (9, 12, 13, 14). The zonal expression then in turn contributes to a different content of key enzymes and catalytic capacities. This was the basis for the model of metabolic zonation (9, 10, 11, 12, 13, 14).
Accordingly, glucose release from glycogenolysis and gluconeogenesis with the key regulatory enzyme phosphoenolpyruvate carboxykinase (PCK1) takes place preferentially in the more oxygenated periportal area; conversely, glucose uptake for glycogen synthesis and glycolysis with the key enzymes glucokinase and L-PK occurs mainly in the less aerobic perivenous area. By microdissection it was found that the L-PK enzyme activity was higher in the perivenous zone (15), where glucose uptake takes place. Moreover, it is known that glucose acted as a very potent activator of the L-PK gene expression and it appeared that the glucose-dependent activation of the L-PK gene is mediated via the transcription factor USF (upstream stimulating factors) binding to the L4 (-168/-145) element in the L-PK promoter consisting of two imperfect palindromic E-boxes (16, 17, 18). Thus, it might be possible that both oxygen and glucose might contribute to the zonated L-PK expression. Therefore, it was the aim of this study to investigate the influence of oxygen and glucose on L-PK gene expression in primary cultured rat hepatocytes in more detail.
We found that in cultured rat hepatocytes perivenous
pO2 induced L-PK gene expression. High glucose
enhanced L-PK gene expression predominantly under periportal
pO2. Transient transfections of primary rat
hepatocytes with L-PK promoter luciferase gene constructs showed that
the modulation by O2 of the glucose-dependent
L-PK gene activation is mediated within the minimal glucose responsive
L-PK gene promoter. Inside the footprinted L4 site the two E-boxes
designated the glucose response element (GlcRE) of the L-PK promoter
showed high identity to a hypoxia response element (HRE) binding the
hypoxia inducible factor-1 (HIF-1). Our transient transfection assays
demonstrated that the glucose response element within the L-PK promoter
can function as a low affinity HRE and that the modulation by
O2 of the glucose-dependent induction of the L-PK
gene was mediated by an interplay between the transcription factors USF
and HIF-1. This could be confirmed in gel shift assays. Thus, it
might be possible that the GlcRE can also function as a HRE and that
the reduced induction by glucose under perivenous
pO2 of the L-PK gene is triggered via an
interplay of HIF-1 at the glucose response element, thereby disturbing
USF complexes. The results of the present study reveal new insights in
the regulation of glucose responsive genes via a cross-talk between the
signals oxygen and glucose.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Animals
Male Wistar rats (200–260 g) were kept on a 12 h day/night
rhythm (light from 0700–1900 h) with free access to water and food
(fed rats). During starvation rats were starved for 60 h from the
beginning of the light period at 0700 to 1900 h on the third day
(fasted rats). Rats were anesthetized with pentobarbital (60 mg/kg BW)
before preparation of hepatocytes between 0800–0900 h. All animal
studies were conducted in accord with the NIH Guide for the Care and
Use of Laboratory Animals.
Cell culture experiments
Liver cells were isolated by collagenase perfusion. Cells
(1 x 106 per dish) were maintained under
standard conditions in an atmosphere of 16% O2,
79% N2, and 5% CO2 (by
vol.) in medium M 199 containing 0.5 nM insulin added as a
growth factor for culture maintenance, 100 nM dexamethasone
required as a permissive hormone and 4% FCS for the initial 4 h
of culture. After 4 h cells were cultured in serum-free M 199 at
16% O2 (mimicking arterial
O2 tensions) or 8% O2
(mimicking venous O2 tensions) with additions as
described in the legends to figures. The O2
values take into account the O2 diffusion
gradient from the media surface to the cells (19).
The cell line HepG2 was cultured in MEM supplemented with 10% FCS under the same conditions as described above.
RNA preparation and Northern analysis
Total RNA was prepared from 3 x 106
cells as described (20). Twenty micrograms of RNA were
denaturated by formaldehyde and used in a typical Northern experiment.
Digoxigenin (DIG)-labeled antisense L-PK and ß-actin (ß-AC) RNA
served as hybridization probes. Respective complementary DNAs (cDNAs)
of L-PK (886 bp; nucleotide 0–886; GenBank Accession No. X05684) and
ß-Act (550 bp; nucleotide 69–618; EMBL: HSA C07) were cloned into
pBS and transcribed into antisense RNA. ß-Act antisense RNA was
synthesized by using T7 RNA polymerase and DIG RNA labeling mixture.
L-PK antisense RNA was synthesized from pBS-PK containing a 886 bp
PstI/XbaI cDNA fragment (21) by
using T3 RNA polymerase and DIG RNA labeling mixture. Hybridizations
were carried out with 50 ng/ml transcript at 68 C according to the
DIG-nucleic acid detection kit (Roche Molecular Biochemicals, Mannheim, Germany). Detection of hybrids was
performed by enzyme-linked immunoassay using an anti-digoxigenin
alkaline phosphatase conjugate. Hybrids were visualized via
chemiluminescence with CSPD. Luminescent blots were exposed to
Hyperfilm-MP (Amersham Pharmacia Biotech, Braunschweig,
Germany) and quantified with a videodensitometer (Biotech Fischer,
Reiskirchen, Germany).
Plasmid constructs
The construct -183PK-LUC was constructed by excision of the
-183 PK-promoter sequence from -183PK/CAT which was a gift from Axel
Kahn (22), with SacI and subsequent ligation
into the SacI sites of pGL3-basic (Promega Corp., Mannheim, Germany).
The plasmids pGL3-GlcPKRE, pGL3-mGlcPKRE, and pGL3-HEPORE were constructed by inserting double-stranded oligonucleotides containing three glucose response elements with a 5'-BglII and a 3'-SacI site or containing three hypoxia response elements with a 5'- and 3'-BglII site into pGL3-promoter (Promega Corp., Mannheim, Germany), in which the luciferase gene is under the control of the SV40 promoter (GlcPKRE3SV40-LUC; mGlcPKRE3SV40-LUC; HEPORE3SV40-LUC). All constructs were verified by sequencing in both directions. The GlcPKRE sequence is from the L-PK gene, GenBank Accession No. X05684. Inside the footprinted L4 site (-168/-145) (16, 17) of the L-PK promoter the two imperfect palindromic E-Boxes (-165/-149) were designated the glucose response element of the L-PK promoter (GlcPKRE). The HEPORE sequence is from the erythropoietin (EPO) gene, GenBank Accession No. X02158 (23).
Two 17 bp (-165/-149) elements responsible for the induction of the L-PK gene by glucose are shown in capital letters (17) and are separated from the third by an EcoRI site (lower case). The mutated base pairs within the GlcPKRE are bold and in lower case letters.
Two 8 bp (3457–3464) elements responsible for the induction of the EPO gene by hypoxia are in italics and shown in capital letters and are separated from the third by an EcoRI site (lower case). Two 9 bp flanking regions are also shown in capital letters (23).
GlcPKRE oligonucleotide
5'CACGGGGCACTCCCGTG CACGGGGCACTCCCGTG gaattc CACGGGGCACTCCCGTG 3'
mGlcPKRE oligonucleotide
5'CACGGGGCACTCCCaaa CACGGGGCACTCCCaaa gaattc CACGGGGCACTCCCaaa 3'
HEPORE oligonucleotide
5'CCTACGTGCTGTCACAG CCTACGTGCT gaattc TACGTGCTCACAG 3'
The plasmid pcDNAHIF-1 containing the full length rat HIF-1
cDNA under the control of the cytomegalovirus promoter (CMV) was
constructed as described (24). The HIF-1 mutant
(pcDNAmHIF-1) was constructed from pcDNAHIF-1 by removing AS
1–51 belonging to the basic-helix-loop-helix domain, which is
responsible for DNA binding with restriction by SacI. All
constructs were verified by sequencing in both directions. The
pcDNAHIF-1ß was a gift from Oliver Hankinson (25).
Cell transfection and luciferase assay
Freshly isolated rat hepatocytes (about 1 x
106 cells per dish) were transfected with
GlcPKRE3SV40-LUC constructs
as described (26): in brief 2.5 µg DNA were precipitated
in 150 µl of transfection buffer, composed of 7.5 µl
CaCl2, 75 µl of 2x HEPES (pH 7.05) and 67.5
µl H2O, and was added to the hepatocytes for
5 h. In HIF-1 or HIF-1ß cotransfection assays 2 µg of the
GlcPKRE3SV40-LUC constructs
were transfected together with either 250 ng pcDNAHIF-1 and/or 250
ng pcDNAHIF-1ß and filled to 500 ng with an empty vector. After
5 h, the medium was changed and the cells were cultured under
arterial pO2 for 19 h. Then, medium was
changed again and the cells were further cultured for 24 h under
arterial pO2 (16% O2) or
venous pO2 (8% O2).
Western blot analysis
Primary cultured hepatocytes or confluent HepG2 cells in 960 mm
dishes were washed with ice-cold 0.9% NaCl and the dishes were frozen
for 10 min at -20 C. The cells were scraped in 5 ml 0.9% NaCl and
whole-cell extracts were prepared with urea extraction buffer as
described previously (27). Protein content in the
supernatant was determined using the Bradford method. Thirty to 50 µg
of protein dissolved in 27 µl SDS gel sample buffer
(28), were loaded onto a 10% or for HIF-1 a 7.5% SDS
gel and after electrophoresis blotted onto nitrocellulose
membranes. Nonspecific binding sites were blocked with blocking buffer
(10 mM Tris/HCl (pH 7.5), 100 mM NaCl, 0.1%
Tween 20, 10% milk powder). Blots were incubated with the primary
monoclonal mouse antibody against rat L-PK as described
(29) in a 1:100 dilution, with the primary mouse
antibody against human HIF-1 (Affinity, Exeter) in a 1:250 dilution,
with the primary rabbit antibody against human HIF-1ß (Affinity BioReagents, Inc., Grünberg, Germany) in a 1:500 dilution
or with the primary rabbit antibody against golgi membrane (GM)
(Bioscience, Göttingen, Germany) in a 1:8000 dilution in blocking
buffer overnight at 4 C. Washing was performed with blocking buffer
without milk powder. The secondary antibody was an antimouse IgG HRP or
an antirabbit IgG HRP (Santa Cruz Biotechnology, Inc.,
Santa Cruz, CA) and used in a 1:2000 dilution for 1 h. After
washing for 30 min, the ECL Western blotting system (Amersham Pharmacia Biotech) was used for detection. Under these
conditions L-PK was visible as a band of 60 kDa, HIF-1 as a band of
about 120 kDa, HIF-1ß as a band of about 95 kDa and GM as a band of
94 kDa.
Immunohistochemistry
Liver tissue of normal fed or fasted rats was embedded in
paraffin. Then 5 µm sections were prepared. Paraffin was removed from
liver sections by xylene. The sections were rehydrated with descending
ethanol concentrations and finally equilibrated for 10 min in TBS [50
mM Tris/HCl (pH 7.5), 150 mM NaCl].
Subsequently, the sections were treated with 100 µl Pepsin (0.4%
wt/vol Pepsin, 0.01 N HCl) for 5 min at 37 C and washed
again in TBS for 10 min. To prevent unspecific binding of the primary
antibody the sections were blocked with 20% FCS in TBS for 30 min at
37 C. The primary diluted antibody (1:50 L-PK; 1:100 GS (glutamine
synthetase; Affinity, Exeter) as a perivenous marker) was applied in
0.1% BSA in TBS overnight at 37 C. The peroxidase-conjugated secondary
antibodies were used in a 1:100 dilution with 0.1% BSA in TBS for 30
min at 37 C. After washing sections were incubated with a 1:10 dilution
of diaminobenzidine (Pierce Chemical Co., Rockford, IL) in
1x peroxidase buffer (Pierce Chemical Co.) for 10 min.
The reaction was stopped in TBS and the sections were photographed and
covered with DePex (Serva, Heidelberg, Germany).
Preparation of nuclear extracts
The binding of the glucose responsive complex to the L4 element
of the L-PK gene in vitro appeared not to be regulated by
either glucose or hormones; it was then further shown by in
vivo footprint experiments that the binding to the L4 element was
also not regulated by fasting or refeeding a carbohydrate rich diet
(16, 35). Thus, HepG2 cells were used as a source for
nuclear extracts containing USF or HIF.
Nuclear extracts were prepared by modification of a standard protocol (30) with buffers A and C containing 0.5 mM dithioerythritol (DTE; Sigma, Deisenhofen, Germany), 0.4 mM phenylmethylsulfonyl fluoride (Serva), 2 µg of leupeptin per ml (Roche), 2 µg of pepstatin per ml (Roche), 2 µg of aprotinin per ml (Bayer Corp., Leverkusen, Germany), 1 mM sodium vanadate (Sigma) and the complete protease inhibitor cocktail tablets (Roche). Briefly, confluent HepG2 cells in 10-cm dishes were cultured at 8% O2 or 16% O2 in the presence or absence of 25 mM glucose for 24 h, then the cells were washed twice with ice-cold 0.9% NaCl, scraped into 300 µl NaCl, and pelleted by centrifugation at 2000 rpm for 2 min at 4 C in an Sorvall high-speed centrifuge RC 5B (SS 34 rotor). The cell pellet was resuspended in 4 packed cell volumes (PCV) of buffer A [10 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 10 mM KCl], and incubated on ice for 10 min. The cell suspension was Dounce homogenized with a type-B pestle, and the nuclei were pelleted by centrifugation at 2300 rpm for 10 min, resuspended in 3 PCV of buffer A, and pelleted once again by centrifugation at 14500 rpm for 20 min. The nuclei were resuspended in 3 PCV of buffer C [20 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 0.42 mM NaCl, 0.2 mM EDTA, 20% glycerol], and incubated at 4 C with gentle agitation for 30 min. Nuclear debris was pelleted by centrifugation for 30 min at 14500 rpm. The supernatant was dialyzed against a 50-fold volume of buffer D [20 mM HEPES-KOH (pH 7.9), 0.1 mM KCl, 0.2 mM EDTA, 20% glycerol] for 4 h. The dialysate was centrifuged for 10 min at 14500 rpm, and aliquots of the supernatant were frozen in liquid N2 and stored at -70 C. Protein concentration was determined by the Bradford method with BSA standards.
Electrophoretic mobility shift assay (EMSA)
The sequence of the GlcPKRE
oligonucleotide used for the EMSA is shown in Fig. 9
. Equal amounts of
complementary oligonucleotides were annealed, labeled by 5'-end
labeling with [
-32P]ATP (Amersham Pharmacia Biotech, Braunschweig, Germany) and T4 polynucleotide
kinase (MBI, St. Leon-Rot) and purified with the Nucleotide Removal Kit
(QIAGEN, Hilden, Germany). Binding reactions were carried
out in a total volume of 20 µl containing 250 mM KCl, 5
mM MgCl2, 5.5 mM EDTA,
25% glycerol, 10 µg of nuclear extract, 100 ng poly d(I-C) and 5
mM DTE. After preincubation for 10 min at 4 C, 1 µl of
the labeled probe (104 cpm) was added and the
incubation was continued for an additional 45 min. For supershift
analysis 1–2 µl monoclonal mouse antibody against human HIF-1
(Novus Biologicals, Littleton, CO), polyclonal rabbit antibody
against human USF-1 or polyclonal rabbit antibody against mouse USF-2
(Santa Cruz Biotechnology, Inc.) was added to the EMSA
reaction and incubated at 4 C for 2 h.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). In livers of fed rats the
zonation of L-PK protein expression was diminished (Fig. 1): L-PK
protein was found in the perivenous and also in the periportal region
compared with livers of fasted rats (Fig. 1). It appeared that the
levels of L-PK protein in the perivenous zone were not changed in the
fed state suggesting that feeding might induce L-PK protein expression
mainly in the periportal area (Fig. 1). In a serial section the
perivenous marker glutamine synthetase (31, 32) was
localized distal perivenously in the last two cell layers around the
central veins (Fig. 1). Thus, in livers of fasted rats the L-PK protein
was present in the perivenous area of the liver acinus where the
O2 tension is low (9, 12, 13, 14) and in
fed rats L-PK protein was detected additionally in the periportal area
of the liver acinus. Thus oxygen and glucose might have a direct
influence on the zonated L-PK gene expression. Therefore, in the next
experiments the influence of oxygen and glucose on L-PK messenger RNA
(mRNA) L-PK protein expression was investigated in primary rat
hepatocyte cultures.
|
).
Increasing glucose concentrations from 5 mM up to 50
mM enhanced L-PK mRNA expression only under arterial
pO2. There was no induction by glucose of L-PK
mRNA under perivenous pO2 (Fig. 2A). Twenty-five
millimolar concentration of glucose induced L-PK mRNA about 7-fold and
50 mM about 12-fold under arterial
pO2. A glucose concentration of 25 mM
was chosen for further experiments. The induction of L-PK mRNA by
glucose was in line with previous observations (21).
|
). Thus, L-PK gene expression was induced by venous pO2 and by glucose predominantly under periportal pO2.
Time course of L-PK mRNA induction by venous pO2 and by
glucose under arterial pO2
The increase of L-PK mRNA by venous
pO2 was already maximal (450%) after 4 h
incubation under venous pO2. Then L-PK mRNA
remained at this level up to 24 h (Fig. 3). In the presence of 25 mM
glucose and arterial pO2 the increase of L-PK
mRNA reached maximal levels (800%) already after 8 h. This
enhancement remained constant for another 16 h (Fig. 3).
|
). Thus, L-PK gene activation by venous
pO2 and by glucose occurred mainly on the
transcriptional level. The protein synthesis inhibitor cycloheximide
(CHX) also caused an inhibition of venous
pO2-dependent L-PK mRNA induction within 6 h
(Fig. 4). This indicated that the venous
pO2-dependent as well as the glucose-dependent
induction of L-PK mRNA was also dependent at least partially on ongoing
protein synthesis of e.g. hypoxia inducible factor-1 (HIF-1)
or USF. Thus, the present findings would be in line with previous
reports showing that CHX blocked the induction by hypoxia of other
genes (i.e. of HIF-1-dependent) coding for glycolytic
enzymes (22).
|
).
Thus, the modulation by O2 of the
glucose-dependent L-PK gene activation was mediated within the minimal
glucose responsive L-PK gene promoter.
|
). Hepatocytes transfected
with luciferase gene constructs containing three mutated glucose
response elements in front of the SV40 promoter
(mGlcPKRE3SV40-LUC) showed
a Luc activity similar to that in hepatocytes transfected with SV40
promoter luciferase constructs (SV 40-Luc) that was neither modulated
by O2 nor by glucose (Fig. 5B). Thus, the glucose
response element within the L-PK gene promoter was sufficient to confer
the modulation by O2 of the glucose-dependent
induction of the L-PK gene expression. Reciprocally, in hepatocytes
transfected with luciferase gene constructs containing three hypoxia
response elements in front of the SV40 promoter
(HEPORE3SV40-LUC) venous
pO2 induced Luc activity by about 2-fold (Fig. 5B). A concentration of 25 mM glucose
prevented the modulation by venous pO2 and
induced Luc activity in hepatocytes transfected with
HEPORE3SV40-LUC to the same
levels of about 3-fold under arterial and venous
pO2 (Fig. 5B). These results supported a
cross-talk between the signals hypoxia and glucose.
Reduction of the glucose-dependent activation of L-PK promoter
luciferase gene constructs by hypoxia inducible factor-1 (HIF-1)
The GlcRE within the promoter of the L-PK gene (-165/-149)
(17), which was responsible for the induction by glucose,
revealed high identity to the binding site of HIF-1. Sequence
comparison between the HIF-1 consensus binding site and the glucose
response element within the L-PK promoter showed only a 1-bp mismatch
between the first E-box of the GlcRE and the HIF-1 binding site. This
was also found for the second E-box (Fig. 7A). Therefore, it might be
possible that HIF-1 can interfere with the glucose-dependent-induction
of the L-PK gene or that glucose can influence HIF-1. Under venous
pO2, HIF-1 was present with an about 2-fold
higher level compared with arterial pO2 and under
25 mM glucose the induction of the HIF-1 protein by
venous pO2 was prevented (Fig. 6). In contrast, glucose did not
influence the HIF-1ß protein expression (Fig. 6). Cotransfection of
the L-PK promoter construct -183PK-LUC or of the
GlcPKRE3SV40-LUC with
expression vectors for HIF-1, HIF-1ß and a HIF-1 mutant lacking
aa (amino acid) 1–51 belonging to the DNA binding domain were
performed. In hepatocytes transfected with the L-PK promoter construct
-183PK-LUC cotransfection of HIF-1 reduced the maximal 12-fold
glucose-dependent induction of Luc activity by about 50%, so that the
glucose-dependent induction of Luc activity was only about 6-fold (Fig. 7B). Cotransfection of -183PK-LUC with
the HIF-1 mutant did not impair the glucose-dependent induction of
Luc activity (Fig. 7B).
|
|
). Cotransfection of
GlcPKRE3SV40-LUC with the
HIF-1 vector reduced the glucose-dependent induction of Luc activity
again by about 50%, whereas cotransfection of vectors for HIF-1ß or
vectors for the HIF-1 mutant had no significant influence (Fig. 7C).
Cotransfection of
GlcPKRE3SV40-LUC with the
HIF-1 expression vector together with the HIF-1ß expression vector
also decreased the glucose-dependent induction of Luc activity by about
50% (Fig. 7C). Thus, in primary rat hepatocytes the reduction of the
glucose-induced L-PK expression under venous pO2
was mediated by interaction of HIF-1 with the GlcRE within the L-PK
promoter.
It seems likely that in the absence of glucose HIF-1 can bind to the
GlcRE and can take part in the induction of L-PK gene expression by
venous pO2. To support this, cotransfections of
GlcPKRE3SV40-LUC,
HEPORE3SV40-LUC as a
positive control or with SV40-LUC as a negative control with different
amounts of a HIF-1 expression vector were performed (Fig. 8B). Cotransfection of hepatocytes with
GlcPKRE3SV40-LUC and
increasing amounts of the HIF-1 expression vector showed a
concentration-dependent induction of Luc activity (Fig. 8B). It was
found that the induction of Luc activity started when 0.5 µg of
HIF-1 were cotransfected, whereas cotransfection of 0.25 µg of
HIF-1 did not induce Luc activity. Cotransfection of 0.5 µg
HIF-1 enhanced Luc activity about 1.7-fold and 4 µg HIF-1 about
3.5-fold (Fig. 8B).
|
expression vector resulted in an about 2.5-fold enhanced Luc
activity when 0.25 µg of HIF-1 vector were used (Fig. 8B). After
cotransfection of 0.5 µg HIF-1 Luc activity was induced by about
3.5-fold; 4 µg HIF-1 vector induced Luc activity by about
11.5-fold (Fig. 8B). Cotransfection of hepatocytes with either
GlcPKRE3SV40-LUC or
HEPORE3SV40-LUC and the
HIF-1 mutant expression vector did not induce the Luc activity (Fig. 8C). The small increase in SV40 transfected cells with increasing
amounts of the HIF-1 expression vector was in line with previous
studies (34). These results indicated that HIF-1 was
able to mediate the induction of gene expression at the glucose
response element of the L-PK gene even in the absence of glucose. Thus,
the glucose response element could also act as a low affinity hypoxia
response element.
Binding of HIF-1, USF-1, and USF-2 to the glucose response
element in the L-PK promoter
The binding of HIF-1 and USF to the GlcRE within the L-PK promoter
was examined by EMSA using nuclear extracts from HepG2 cells.
Three major retarded bands could be detected under arterial and venous
pO2 (Fig. 9).
Beside the fastest and the slowest complex that were constitutively
present the formation of a complex migrating in the middle was enhanced
with nuclear extracts from cells cultured under venous
pO2 (Fig. 9). Thus, it seemed likely that the
hypoxia induced DNA-protein complex contained HIF-1. To test for the
presence of HIF-1 an antibody against HIF-1 was included in the
binding reaction. Addition of the HIF-1 antibody to the EMSA
reaction inhibited the formation of the hypoxia-dependent DNA complex
and led to a supershifted complex (Fig. 9). It was already known that
USF could bind the GlcRE within the L-PK promoter as a complex
consisting of USF-1 and USF-2 (16, 35). The slowest
complex binding to the GlcPKRE appeared to
contain USF because an antibody against USF-1 or
USF-2 prevented formation of this complex (Fig. 9).
Furthermore, addition of either the antibody against USF-1 or the
antibody against USF-2 reduced also the formation of the hypoxia
inducible complex (Fig. 9). Addition of an HIF-1 antibody together
with an USF-1 or an USF-2 antibody lead to a supershifted complex and
to nondetectable HIF-1 and USF protein complexes (Fig. 9). The same
pattern was observed using nuclear extracts from cells treated for
24 h with 25 mM glucose (data not shown). These data
indicated that HIF-1 as well as USF-1 and USF-2 bound to the GlcRE
within the L-PK promoter confirming that the reduction of the
glucose-induced L-PK expression under venous pO2
was mediated by interaction of HIF-1 with the GlcRE within the L-PK
promoter. Coimmunoprecipitation assays revealed that a
heterodimerization between HIF-1 and USF does not occur. Therefore,
it might be that in the presence of glucose and low oxygen a HIF-1
complex consisting of HIF-1 and HIF-1ß binds to the GlcRE within
the L-PK promoter and replaces the binding of the USF complex thereby
reducing the glucose-dependent induction of L-PK gene expression under
low oxygen.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) in primary rat
hepatocytes. The modulatory role of O2 is also in
line with previous observations in which insulin activated the
glycolytic key enzyme gene glucokinase maximally under perivenous
pO2 (4) and, reciprocally, glucagon
activated the gluconeogenic key enzyme gene PCK1 maximally under
periportal pO2 (20, 37). Thus, the
positive modulation by venous pO2 of the L-PK
expression was in accordance with the perivenous zonation of L-PK
protein (Fig. 1) and with the perivenous zonation of glycolysis in rat
liver.
Because the difference in glucose concentration from the periportal to
the perivenous area is rather shallow the predominant L-PK gene
activation by glucose under periportal pO2 would
result in a diminution of the zonation so that the L-PK protein can be
found to be expressed beside in the perivenous zone also in the
periportal region as observed in the livers of fed rats (Fig. 1). In
the fasted state, as in between meals, the glucose-dependent induction
of the L-PK gene expression in the periportal area is no longer present
(Fig. 1) so that the L-PK gene activation by venous
pO2 would be predominant and lead to the
perivenous zonation of the enzyme as it was shown in fasted rats (Fig. 1). Glucose plays not only an important role in the regulation of L-PK
gene expression (1, 18, 33, 38), it is also known to
induce the expression of glucagon receptor mRNA in hepatocytes
(39) and the lipogenic enzyme genes acetyl-CoA carboxylase
(40) and fatty acid synthase (41, 42, 43) in
liver and adipose tissue (44). In this study, the
glucose-dependent induction of the L-PK gene occurred predominantly
under periportal pO2 (Figs. 2 and 5). A
concentration of 25 mM glucose increased L-PK mRNA levels
about 7-fold and protein levels about 5-fold within 24 h at 16%
O2 (Fig. 2). Because L-PK is a key enzyme of
glycolysis and is linked to glucose uptake the induction of L-PK by
glucose was in line with expectations and with previous studies
(33). Accordingly, L-PK mRNA levels were high in livers of
fed rats, whereas L-PK mRNA levels were low in livers of fasted rats
(45).
To support that the observed modulation by O2 of
the L-PK gene occurs on the transcriptional level and involves probably
factors which are generated via de-novo protein synthesis such as HIF-1
experiments in the presence of actinomycin D and cycloheximide were
performed. Both inhibitors prevented the venous
pO2-dependent and the glucose-dependent L-PK mRNA
induction within 8 h (Fig. 4). Because both inhibitors act in a
general manner by suppressing either the all-over transcription or
protein synthesis, other gene products that indirectly regulate LPK
expression might be also be affected. Thus, care has to be taken into
account for the interpretation of these results. However, the data are
in line with the findings that the L-PK gene activation by glucose
occurred mainly on the transcriptional level (17) and that
the induction by venous pO2 involves at least
partially ongoing protein synthesis of e.g. hypoxia
inducible factor-1 (HIF-1). Thus, the present findings would be in line
with previous reports showing that CHX blocked the induction by hypoxia
of other genes (i.e. of HIF-1-dependent) coding for
glycolytic enzymes (23).
In the 5'-flanking region of the L-PK gene, the L4 box contains a
GlcPKRE (glucose/insulin response element: GIRE)
that is formed by two palindromic E-boxes (-165 -CACGGG-160; -154
-CCCGTG-149) (16, 17) differing from the CACGTG E-box
consensus by only one nucleotide (18, 46). A similar
glucose response element has been described in the 5'-flanking region
of the Spot14 gene (47). The L-PK GlcRE acted as binding
site for the basic-helix-loop-helix transcription factors USF-1 and
USF-2 (upstream stimulating factor) (16, 35, 48) that
mainly conferred the glucose response to the L-PK gene
(49). Transient transfections in mhAT3F hepatoma cells
(49) and experiments with USF-2 -/- mice revealed that
USF-2 proteins were required for a normal transcriptional response of
the L-PK and Spot14 gene to glucose (50). Beside USF
proteins several other factors appeared to take part in the glucose
response complex of the L-PK gene. One negative regulating factor is
chicken ovalbumin upstream promoter-transcription factor II (COUP-TFII)
(46), whereas an as yet unidentified factor had an
additional positive effect on the glucose response of the L-PK gene
(51). In this study, the GlcRE of the L-PK promoter was
demonstrated to be responsible for the induction by glucose of the L-PK
gene and also to be responsive for O2 (Fig. 8)
that modulated the glucose-dependent L-PK activation (Fig. 5). Because
the expression of the L-PK gene was induced by perivenous
pO2 and by glucose predominantly under arterial
pO2 (Fig. 2), a possible interference between the
induction by hypoxia and glucose was investigated. Therefore, the GlcRE
within the L-PK promoter was compared with the hypoxia responsive
element (HRE), the binding site for HIF-1; the GlcRE revealed high
homology with the HRE (Fig. 7A). Due to the sequence similarity of the
GlcRE with the HRE and the fact that the transcription factor USF, as
mediator of the response to glucose, and HIF-1, as mediator of the
response to hypoxia, belong to the family of basic-helix-loop-helix
proteins (52, 53) an interaction between HIF-1 and USF at
the GlcRE seemed to be possible. Indeed, the present transfection
assays with the L-PK promoter luciferase constructs displayed first
evidence for an interaction between HIF-1 and the GlcRE of the L-PK
promoter. HIF-1 is a heterodimeric transcription factor consisting of
the O2-sensitive HIF-1 subunit and the
HIF-1ß subunit. HIF-1ß is the constitutive expressed subunit and
also known as the arylhydrocarbon-receptor nuclear translocator (ARNT).
Under low pO2 HIF-1 activated the expression of a
number of genes (3) among them those encoding glycolytic
enzymes such as aldolase A and C (54), lactate
dehydrogenase A and phosphoglycerat kinase 1 (55). The
interaction of the signals glucose and oxygen (hypoxia) was proposed
also from experiments with mouse embryonic stem cells in which
hypoglycemia and hypoxia induced the expression of phosphoglycerate
kinase-1, vascular endothelial growth factor, lactate dehydrogenase and
glucose transporter-1. The induction by hypoxia as well as the
induction by hypoglycemia was abolished in HIF-1-deficient embryonic
stem cells (5, 6). It was also found in this study that
HIF-1 was enhanced expressed under 8% O2
(Fig. 6). 25 mM glucose prevented the induction of HIF-1
protein by venous pO2 (Fig. 6). This is in line
with the finding that in hepatocytes transfected with luciferase gene
constructs containing 3 hypoxia response elements in front of the SV40
promoter (HEPORE3SV40-LUC)
25 mM glucose prevented the induction by venous
pO2 (Fig. 5). The interaction of HIF-1 with the
CACGTG (E-box) binding site was described previously (7, 56, 57, 58). Cotransfection of hepatocytes with
GlcPKRE3SV40-LUC and an
expression vector for HIF-1 reduced the glucose-induced Luc activity
by 50% thus mimicking venous pO2, whereas
cotransfection with an expression vector for HIF-1ß had only a slight
effect (Fig. 7C). Cotransfection of hepatocytes with the HIF-1
vector together with the HIF-1ß vector and
GlcPKRE3SV40-LUC also
decreased the glucose-induced Luc activity (Fig. 7C). However, the
response to venous pO2
(8%O2) alone was not as predominant as the
response to glucose alone at 16% O2 because the
induction of Luc activity in hepatocytes transfected with
GlcPKRE3SV40-LUC at 8%
O2 was only slightly but not significant (Fig. 5). In the presence of higher amounts of HIF-1 Luc activity was
induced (Fig. 8). Thus, the glucose response element might function as
a hypoxia responsive site with a lower affinity for HIF-1 than for
USF indicating that this element is not the only HIF-1 binding site
mediating the induction of the L-PK gene expression by venous
pO2. The specifity of the interaction between
HIF-1 and the GlcPKRE was indicated by the
cotransfection of
GlcPKRE3SV40-LUC with a
HIF-1 mutant lacking aa 1–51 belonging to the DNA binding domain
that did not impair the glucose-dependent induction of Luc activity
(Fig. 7C). This further showed that the reduction of the
glucose-dependent L-PK gene activation is dependent of HIF-1 DNA
binding activity. The binding of HIF-1 and USF was shown in EMSAs
(Fig. 9).
From the present results the following model was proposed in which
HIF-1 might be able to interfere with the glucose response complex. In
the presence of high glucose under arterial pO2
the USF heterodimer binds the GlcPKRE and
mediates the glucose-dependent induction of the L-PK gene. In the
presence of high glucose under venous pO2, it is
likely that a HIF-1 containing complex could replace an USF protein
complex, thus disturbing the glucose responsive complex which results
in a reduction of the glucose-dependent L-PK gene activation. This
would explain why glucose could induce L-PK mRNA and protein
predominantly only under arterial pO2. The
interaction of HIF-1 with the GlcPKRE
might play a role in the induction of the L-PK gene by venous
pO2. In line with the results was the finding
that cotransfection of HIF-1 could induce Luc activity in
GlcPKRE3SV40-LUC
transfected hepatocytes (Fig. 8). Our results for the first time
demonstrate the mutual molecular interaction of the signals oxygen and
glucose which might represent a more common mechanism of oxygen and
glucose dependent gene regulation as in metabolism, tumorigenesis or
embryonic development.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received March 16, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
in hypoxia-mediated apoptosis, cell proliferation and
tumour angiogenesis. Nature 394:485–490[CrossRef][Medline]
.
Blood 92:2260–2268
0.05. NE,
Nuclear extract; I, induced HIF-1
