BNP (brain-type natriuretic peptide) is a cardiac hormone with systemic haemodynamic effects as well as local cytoprotective and antiproliferative properties. It is induced under a variety of pathophysiological conditions, including decompensated heart failure and myocardial infarction. Since regional hypoxia is a potential common denominator of increased wall stretch and myocardial hypoperfusion, we investigated the direct effects of hypoxia on BNP expression, and the role of the HIF (hypoxia-inducible transcription factor) in BNP regulation. Using an RNase protection assay we found a strong hypoxic induction of BNP mRNA expression in different cell lines and in cultured adult rat cardiomyocytes. Systemic hypoxia and exposure to 0.1% CO induced BNP expression in the rodent myocardium in vivo, although this was at a lower amplitude. BNP promoter-driven luciferase expression increased 10-fold after hypoxic stimulation in transient transfections. Inactivation of four putative HREs (hypoxia-response elements) in the promoter by site-directed mutagenesis revealed that the HRE at −466 nt was responsible for hypoxic promoter activation. A functional CACAG motif was identified upstream of this HRE. The HIF-1 complex bound specifically and inducibly only to the HRE at −466 nt, as shown by EMSA (electrophoretic mobility-shift assay) and ChIP (chromatin immunoprecipitation). siRNA (small interfering RNA)-mediated knockdown of HIF-1α, but not HIF-2α, interfered with hypoxic BNP mRNA induction and BNP promoter activation, confirming that BNP is a specific HIF-1α target gene. In conclusion, BNP appears to be part of the protective program steered by HIF-1 in response to oxygen deprivation. Induction of BNP may therefore contribute to the potential benefits of pharmacological HIF inducers in the treatment of ischaemic heart disease and heart failure.
BNP (brain-type natriuretic peptide) is one of three members of the natriuretic peptide family which is predominantly expressed in the adult heart. Like the ANP (atrial natriuretic peptide) it has natriuretic, diuretic and vasodilating properties [1,2] and acts by increasing intracellular cGMP (cyclic guanosine monophosphate) levels via the activation of the pGC (particulate guanylyl cyclase) . In addition to systemic haemodynamic actions, BNP has antimigratory, antifibrotic and cytoprotective effects in the heart [4–6], which are suggested to constitute a counter-regulatory system to attenuate the development of cardiac fibrosis in vivo .
BNP synthesis and secretion is induced by pathophysiological conditions such as left ventricular hypertrophy, myocardial infarction and heart failure [8–10]. Myocardial stretch is a potent inducer of BNP . The contribution of tissue hypoxia as a stimulus for the induction of BNP is less clear, since most experimental approaches investigated myocardial ischaemia [12,13] or long-term hypoxic exposure of animals [14,15], which are associated with far more complex pathophysiological alterations than hypoxia alone. These include changes in haemodynamics and ventricular mass, which in turn can affect BNP gene expression. Hypoxic perfusion of isolated hearts increased BNP protein release, but the effect of hypoxia at the transcriptional level was not investigated . In cardiac myocyte preparations, the effect of hypoxia was less clear: freshly prepared porcine myocytes exhibited a less pronounced induction of a premature BNP mRNA under anoxia compared with the ischaemic myocardium in vivo . Moreover, a recent study did not observe a hypoxic upregulation of BNP mRNA in rat and human cardiomyocytes . It is therefore unclear whether hypoxia constitutes a direct and sufficient stimulus for BNP gene induction in vitro.
Mammalian cells respond to low oxygen with an increased expression of several hypoxia-inducible genes. The master regulator of this cellular adaptation to hypoxia is the HIF (hypoxia-inducible factor), a heterodimeric transcription factor which consists of the constitutively expressed β-subunit termed HIF-1β (or ARNT, aryl hydrocarbon receptor nuclear translocator) and one of two alternative oxygen-regulated α-subunits, HIF-1α or HIF-2α. In hypoxia, the α/β heterodimeric HIF complex induces transcription through interaction with the HREs (hypoxia-response elements) in the regulatory regions of target genes such as EPO (erythropoietin), HO-1 (haem oxygenase 1), VEGF (vascular endothelial growth factor) and iNOS (inducible nitric oxide synthase) (for review see [19,20]). Under normoxia, the HIF-α subunits are rapidly hydroxylated at two conserved prolyl residues by specific prolyl hydroxylases [PHD1, PHD2 and PHD3 (PHD, prolyl hydroxylase domain enzyme)] 1, 2 and 3 [21,22], which enables interaction with the pVHL (von Hippel-Lindau protein) which is the recognition component of the E3 ubiquitin ligase complex that targets HIF-α for proteasomal destruction . Stimulation of HIF target genes has been shown to limit infarct size and ameliorate ventricular function in experimental models of myocardial ischaemia . We have previously demonstrated that both HIF-α isoforms were stabilized in the hypoxic or infarcted myocardium, and that HIF target genes were induced in the perinecrotic areas, which may limit hypoxic damage [25,26]. Whether BNP was part of this HIF-mediated response was not investigated. Gene expression profiling studies [27,28] suggested an involvement of HIF-1 in BNP expression, but since these studies were based on overexpression of stable HIF mutants, the role of hypoxia and the individual contribution of the two HIF-α isoforms remained unknown.
Given the increasingly recognized significance of low oxygen as a regulator of cell function and its potential role in BNP regulation, we decided to study the direct effects of hypoxia and the activation of the HIF system on BNP expression in vitro and in vivo. We provide evidence herein that BNP mRNA expression is directly regulated by molecular oxygen and that this regulation is mediated by interaction of endogenous HIF-1α with one of four potential HREs in the proximal 5′ regulatory region of the human BNP gene.
MATERIALS AND METHODS
Cells and reagents
Hep3B and HepG2 human hepatoma cells were purchased from ATCC (American Type Culture Collection, Manassas, VA, U.S.A.), mouse fibroblastoid preadipocyte 3T3L1 cells were a gift from Dr M. Praus (Institut für Laboratoriumsmedizin und Pathobiochemie, Charité Berlin, Medizinische Fakultät der Humboldt Universität zu Berlin, Germany) and MEF (murine embryonic fibroblasts) were a gift of Dr G. Semenza, Johns Hopkins University School of Medicine, Baltimore, MD, U.S.A. HKC-8 human proximal tubular epithelial cells were provided by Dr L. Racusen, Johns Hopkins University School of Medicine, Baltimore, MD, U.S.A. Cell culture reagents were from PAA, Invitrogen and Biochrom. DP (2,2′dipyridyl) was purchased from ICN. All other chemicals were from Sigma. The plasmid 6xHRE/tk/luc was a gift from Dr P. Ratcliffe, Henry Wellcome Building for Molecular Physiology, University of Oxford, U.K., and HIF-2α antibody PM9 was from Dr P. Maxwell, Hammersmith Hospital, London, U.K.
Cell lines were grown in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% FCS (foetal calf serum), 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin. HKC-8 cells were cultured in DMEM/Ham's F-12 containing 5 μg/ml insulin, 5 μg/ml transferrin and 5 ng/ml sodium selenite. For the analyses of BNP mRNA induction in vitro, cells were exposed to 1% O2 in a Jouan IG 750 water-jacketed incubator (Thermo Electron) or stimulated with the hypoxia mimetic DP at a concentration of 100 μM for 16 h.
All animal experiments were approved by local government authorities (Government of Mittelfranken) and conform with the Guide for the Care and Use of Laboratory Animals as published by the U.S. National Institutes of Health (NIH Publication No. 85-23, revised 1996). Three 6-week-old male Sprague–Dawley rats and three 6-week-old C57BL/6 mice were exposed to 0.1% CO or to 8% O2 in a hypoxia workbench (In Vivo 400; Ruskin Technology). After 10 h, the animals were killed by cervical dislocation and tissue was frozen in liquid nitrogen for RNA analysis.
Isolation and hypoxic stimulation of myocytes
For myocyte isolation, male Wistar rats were anaesthetised and the heart was quickly excised and placed into ice-cold Tyrode's solution (138 mM NaCl, 4 mM KCl, 0.9 mM CaCl2, 1 mM MgCl2, 0.33 mM NaH2PO4, 10 mM glucose, 1 mM EGTA, 10 mM Hepes and 1 μM insulin, pH 7.30). The ascending aorta was cannulated, the heart was mounted on a Langendorff apparatus and ventricular myocytes were isolated as described previously [29,30]. Briefly, the heart was perfused with Tyrode's solution for 5 min at 37 °C. Perfusion pressure was 75 mmHg, and all solutions were equilibrated with 100% O2. The hearts were then perfused with Tyrode's solution containing collagenase [type CLS II, 160 units/ml (Biochrom)] and protease (type XIV, 0.5 units/ml), and finally with modified storage solution [130 mM NaCl, 5.4 mM KCl, 1 mM CaCl2, 0.5 mM MgCl2, 0.4 mM NaH2PO4, 22 mM glucose, 5.8 mM NaHCO3, 25 mM Hepes, 1 μM insulin and 1 mg/ml BSA (Merck), pH 7.40]. Ventricular myocytes were isolated by mincing the tissue and gentle mechanical agitation. After a preplating step to eliminate fibroblasts, they were plated in the storage solution containing antibiotics. Cells were kept for 6 h at 37 °C and 5% CO2 before being exposed to hypoxia (1% O2) for 10 h.
RNA preparation and RNase protection assay
The tissue was homogenized using an Ultraturrax homogenizer (IKA) and total RNA was extracted with RNAzol B (Biozol) and analysed by an RNase protection assay essentially as described previously . Riboprobe templates for human, murine and rat BNP mRNA were generated by PCR using oligonucleotides based on published sequences (human: NM002521, nt 232–387; mouse: NM008726, nt 136–313; rat: NM031545, nt 118–289) and cloned into pcDNA3 (Invitrogen) using generated 5′KpnI and 3′XhoI restriction sites (for primers, see Supplementary Table 1 at http://www.BiochemJ.org/bj/409/bj4090233add.htm). Quantification of signal intensities was performed using a phosphoimager (FujiBAS 2000, Kodak). Results given represent the means of three independent experiments ±S.D.
Cloning of human (h)BNP promoter luciferase reporter plasmids and generation of HRE and CACAG mutants
The human BNP 5′-flanking sequence (GenBank® accession no. U34833) was analysed for putative HRE using the MatInspector software at www.genomatix.de. A 2027 bp and a 967 bp promoter fragment including exon 1 were amplified from human genomic DNA (primers are detailed in Supplementary Table 1) and cloned into the pGL2 basic luciferase vector (Promega) yielding the plasmids hBNP-Prom(L) and hBNP-Prom respectively. Four BNP-promoter luciferase constructs with mutations in each of the four putative HREs or the CACAG motif were generated by site-directed mutagenesis using the hBNP-Prom (967 bp) plasmid as template and Combi-Pol DNA Polymerase Mix (Invitek). For primer sequences, see Supplementary Table 2 at http://www.BiochemJ.org/bj/409/bj4090233add.htm. Successful cloning of all reporter constructs was verified by sequencing.
Luciferase reporter gene assays
Hep3B cells were cultured in 24-well plates to 50% confluency and transfected with 300 ng/well of the respective luciferase reporter plasmid and 30 ng of a pCMV-β-galactosidase expression vector using the Fugene® transfection reagent (Roche). Transfected cells were stimulated for 18 h with hypoxia (1% O2) or 100 μM DP. For HIF-α knockdown analyses, the plasmid-transfected cells were co-transfected with siRNAs (small interfering RNAs) as described below. Luciferase activities were normalized to the respective β-galactosidase expression. All transfections were performed in triplicate. Results represent the means of three independent experiments±S.D.
HIF-1α and HIF-2α siRNA sequences were the same as described previously  and were synthesized by Xeragon/Qiagen. As a negative control we used an siRNA targeting the firefly luciferase coding sequence . Cells were transfected with siRNA (final concentration 200 nM) at 60–75% confluency by the use of Oligofectamine in serum-free Optimem® medium (both Invitrogen) according to the manufacturer's protocol. Medium was changed 14 h post-transfection and the cells were stimulated with 100 μM DP for 18 h.
Preparation of cell lysates and immunoblotting
Protein extraction and immunoblotting were performed as described previously . Briefly, after siRNA transfection and stimulation with DP for 8 h, Hep3B and HKC-8 cells were lysed in 8 M urea, 10% glycerol, 10 mM Tris/HCl (pH 6.8), 1% SDS, 5 mM dithiothrietol and protease inhibitors Complete® (Roche). Protein (100 μg) was separated on 8% polyacrylamide gels, transferred onto PVDF membranes (Millipore) and incubated with antibodies against HIF-1α (Affinity Bioreagents) and HIF-2α (PM 9). Subsequently, blots were exposed to HRP (horseradish peroxidase)-conjugated secondary antibodies (Dako) and signals were visualised by chemiluminescence (SuperSignal West Dura Extended®; Pierce).
EMSA (electrophoretic mobility-shift assay)
Hep3B cells were exposed to normoxia, 1% O2 or to 100 μM DP and nuclear extracts were prepared using a modified Dignam protocol as described previously . Nuclear protein (5 μg) was incubated with 105 c.p.m. of a [32P]-labelled 24 bp double-stranded oligonucleotide probe (see Supplementary Table 3 at http://www.BiochemJ.org/bj/409/bj4090233add.htm). For supershift assays, the binding reaction was preincubated with 0.5 μg of an anti-HIF-1α monoclonal antibody (Transduction Laboratories) for 30 min before addition of the labelled probe.
ChIP (chromatin immunoprecipitation)
Hep3B cells were exposed to 100 μM DP overnight and then treated with 1% formaldehyde for 15 min at 20 °C. Cells were washed and harvested in cold PBS, lysed in 50 mM Tris/HCl (pH 8), 2 mM EDTA, 1% SDS and protease inhibitors Complete® and sonicated to shear the DNA to fragments of approximately 600 bp. One tenth of the lysate was stored as input control. To the remaining cell lysate, four volumes of 1.25x RIPA buffer [50 mM Tris/HCl (pH 7.5), 0.1% sodium deoxycholate, 0.625% Nonidet P40, 2 mM EDTA, 200 mM NaCl and protease inhibitors Complete®) was added. The diluted lysate was then precleared with Protein A–agarose for 1 h at 4 °C. The supernatant was split into two aliquots, and one half was incubated with unspecific rabbit IgG and the other with 2 μg of a anti-HIF-1α or anti-HIF-2α rabbit polyclonal antibody (Novus Biologicals) for 1 h. Then, Protein A–agarose, preblocked with BSA and Herring sperm DNA, was added and incubation was continued for a further 4 h at 4 °C. Beads were washed for 2×15 min in RIPA buffer, 1×15 min in high salt RIPA buffer (1 M NaCl), and 2×5 min with 10 mM Tris/HCl (pH 8.0) and 1 mM EDTA.
Cross-linking was reversed and DNA was eluted by resuspending pellets in 200 μl SDS/NaCl/DTT buffer (62.5 mM Tris/HCl (pH 6.8), 200 mM NaCl, 2% SDS, 10 mM DTT) with incubation at 65 °C overnight. DNA was prepared by phenol/chloroform extraction, precipitated, and dissolved in 10 mM Tris/HCl (pH 8.0). PCR reactions were carried out using a primer pair which flanked HRE 2 of the BNP promoter (for primer sequences, see Supplementary Table 4 at http://www.BiochemJ.org/bj/409/bj4090233add.htm) and yielded a PCR product of 194 bp. CA9 (carbonic anhydrase IX) was used as a positive control, and β-actin, which does not contain a HRE, was the negative control. PCRs were run with 34 cycles.
Results were analysed by Student's t-test. A P level <0.05 was considered significant.
Induction of BNP expression in different human and animal cell lines in response to hypoxia and DP
Previous reports indicated that cardiac cells in culture apparently lose the ability of hypoxic induction of BNP expression . We therefore screened a number of immortalized cell lines for basal and hypoxia-inducible BNP expression, including Hep3B and HepG2 cells, cervical carcinoma HeLa cells, HKC-8 cells and 3T3L1 cells. To investigate the hypoxic regulation of BNP mRNA expression, cells were exposed to 1% O2 or the iron chelator and chemical HIF inducer DP for 16 h. In all cell lines except HeLa, which did not express BNP mRNA at detectable levels, BNP mRNA expression was strongly induced by both stimuli (Figure 1A). Moreover, in HKC-8 and Hep3B cells, BNP mRNA induction by DP was already visible after 4 h of stimulation (Figure 1B).
BNP expression is induced by hypoxia and hypoxia mimetics in different cell lines and in rodent hearts in vivo
Upregulation of BNP mRNA by hypoxia in cultured adult rat cardiomyocytes
Although BNP mRNA abundance in the heart is high in vivo, a hypoxic upregulation of BNP mRNA in cultured neonatal cardiomyocytes in vitro could previously not be demonstrated [18,27]. Freshly isolated adult cardiomyocytes from the left ventricle of rat hearts were exposed to 1% O2 for 10 h. In four independent experiments, BNP mRNA expression was moderately but significantly induced (1.7±0.3-fold against normoxic cells, P<0.05; Figure 1C), which indicates that the response of the BNP gene to hypoxia is detectable, but operates at a lower amplitude compared with tumour cell lines. Cultured primary rat and porcine cardiac fibroblasts did not express BNP mRNA at levels detectable by RNase protection (results not shown).
Exposure to 0.1% CO and systemic hypoxia induces BNP expression in vivo
To elucidate whether systemic hypoxia, in the absence of any ischaemic injury to the heart, is sufficient to induce BNP mRNA levels in vivo, rats and mice were exposed to 0.1% CO or 8% O2 for 10 h. CO exposure induces tissue hypoxia by blocking the oxygen transport capacity of red blood cells and leads to a robust HIF induction in vivo . BNP mRNA expression was upregulated 3.4±0.1-fold and 1.7±0.3-fold after CO exposure in rat and mouse hearts respectively (n=3, P<0.05; Figure 1D). In rats exposed to 8% O2 a less marked increase of cardiac BNP mRNA abundance was detectable (1.4±0.1-fold, n=3, P<0.05; see Supplementary Figure 1E at http://www.BiochemJ.org/bj/409/bj4090233add.htm).
The human BNP promoter is activated by hypoxia and contains a functional HRE at −466 nt
Sequence analysis of the human BNP promoter revealed six putative HREs (hypoxia-response elements) within a 2 kb region upstream of the transcriptional start site (Table 1). Hep3B cells were transiently transfected with the 2027 bp construct hBNP-Prom(L) and stimulated with 1% O2 or 100 μM DP for 18 h. Hypoxia and DP increased reporter activity 2.3±0.2-fold and 2.5±0.1-fold respectively (P<0.05 against unstimulated, Figure 2A). In contrast, the activity of the 5′-deleted 967 bp construct hBNP-Prom, which still contained the four putative HREs, was increased 10.3±1.3-fold and 11±1.5-fold after hypoxia and DP stimulation respectively (P<0.05 against unstimulated; Figure 2B), indicating that the functional HREs are located in the proximal region and that the putative negative-regulatory regions are present in the more distant regions. To determine which of the putative HREs was essential for the hypoxic induction, we inactivated each of the four HIF binding sequences by site-directed mutagenesis, leaving the other three HREs of the 5′-deleted promoter construct intact. Mutation of the BNP-HRE 2 at −466 nt (ACGTGC→ATTTGC, HIF binding site in bold) significantly reduced the increase in reporter gene activity by hypoxia from 10.3-fold to 3.0±0.3-fold and by DP from 11-fold to 3.7±1.1-fold (P<0.05 against wild-type for both stimuli; Figure 2C and 2D). Mutation of BNP-HRE 1 (CCACGG→CTATTC) reduced the induction of reporter activity after stimulation with hypoxia from 11±1.5-fold to 6.0±0.8-fold (P<0.05 against wild-type), but the induction by DP was not significantly affected. Mutation of the other two HREs (HRE 3: TCACGT→TTATTT and HRE4: CCACGT→CTATTC) did not change either the hypoxic or the DP-stimulated induction of luciferase activity (Figure 2C and 2D). Thus mutational analysis of the four putative HREs in the proximal BNP promoter revealed that the hypoxic induction was mainly dependent on the HRE at −466 nt.
|Location||Sequence 5′→3′||Direction||Ancillary sequence|
|−1158 nt||G GCGTG AT||Sense||No|
|−877 nt||C ACGTG GA||Sense||No|
|BNP-HRE 1||−626 nt||TC CACGC A||Antisense||No|
|BNP-HRE 2||−466 nt||T ACGTG CG||Sense||CACAG (−492 nt)|
|BNP-HRE 3||−363 nt||CT CACGT C||Antisense||No|
|BNP-HRE 4||−147 nt||CC CACGTC||Antisense||No|
|hEpo-HRE||T ACGTG CT||CACAG|
|hVEGF-HRE||T ACGTG GG||AACAG|
|Location||Sequence 5′→3′||Direction||Ancillary sequence|
|−1158 nt||G GCGTG AT||Sense||No|
|−877 nt||C ACGTG GA||Sense||No|
|BNP-HRE 1||−626 nt||TC CACGC A||Antisense||No|
|BNP-HRE 2||−466 nt||T ACGTG CG||Sense||CACAG (−492 nt)|
|BNP-HRE 3||−363 nt||CT CACGT C||Antisense||No|
|BNP-HRE 4||−147 nt||CC CACGTC||Antisense||No|
|hEpo-HRE||T ACGTG CT||CACAG|
|hVEGF-HRE||T ACGTG GG||AACAG|
Hypoxia activates human BNP promoter reporter constructs
The endogenous HIF-1 complex binds to the HRE at −466 nt of the human BNP promoter
Gel-shift analyses were performed to investigate whether the endogenous HIF-1 complex binds to the functional HRE at −466 nt. When the [32P]-labelled BNP-HRE 2 probe was incubated with nuclear extracts from hypoxic or DP-stimulated Hep3B cells, a binding complex appeared which was not observed when extracts from normoxic cells were used or the HIF-binding sequence was mutated (Figure 3A). No specific binding complex was observed with probes for HRE 1, HRE 3 and HRE 4 (results not shown). Addition of a monoclonal anti-HIF-1α antibody to the binding reaction resulted in a supershifted complex, which demonstrated that HIF-1α is a component of the binding activity induced by hypoxia (Figure 3B).
Identification of the functional HIF-1 binding site in the human BNP promoter using EMSA and ChIP
To investigate further if HIF-1α binds to the native BNP promoter in vivo, ChIP was performed. Immunoprecipitation of chromatin complexes prepared from DP-stimulated Hep3B cells with a HIF-1α antibody led to an marked enrichment of a BNP promoter fragment spanning HRE 2 (Figure 3C, upper panel). In contrast, chromatin complexes from unstimulated Hep3B cells and chromatin exposed to non-specific rabbit IgG yielded only weak bands after PCR amplification, presumably due to residual non-specific binding of DNA to Protein A–agarose beads. Similar to BNP, a HRE-spanning fragment of the CA9 promoter was enriched by the HIF-1α antibody (Figure 3C, middle panel). As a negative control, a β-actin gene sequence lacking a functional HRE was amplified, which was not enriched in DP-stimulated cell lysates by the HIF-1α antibody (Figure 3C, lower panel). To test whether HIF-2α occupied the BNP-HRE, ChIP was also performed with a HIF-2α antibody. The BNP promoter and β-actin gene fragment were not enriched by PCR (Figure 3D, upper and lower panels), but surprisingly the CA9 promoter fragment was (Figure 3D, middle panel). However, this result is in keeping with a recent study by Lau et al. , who also found selective enrichment of the HRE-containing CA9 promoter fragment in chromatin from hypoxic Hep3B cells after precipitation with a HIF-2α antibody, despite CA9 being a well-characterized HIF-1α target. These findings emphasize that ChIP does not necessarily allow conclusions concerning the function of the respective HIF-α isoform on the gene of interest. Rather, it indicates that the transcription factor is in close proximity to, or a part of, the transcriptional complex. However, EMSA experiments clearly demonstrated the binding of HIF-1α protein to HRE 2. Thus, taken together, these results corroborated the findings of the reporter assays and further support (1) that BNP is a direct transcriptional target of HIF-1, and (2) that hypoxic promoter activation is mediated by the HRE 2.
BNP induction is dependent on HIF-1α but not on HIF-2α
Previous reports demonstrated that target gene specificity of the HIF-α isoforms can be overridden by overexpression of their subunits . We therefore performed siRNA knockdown of endogenous HIF-1α or HIF-2α prior to DP stimulation to determine the relative contribution of either HIF-α isoform to BNP induction. In Hep3B cells, siRNA transfection substantially reduced mRNA expression of HIF-1α and HIF-2α (Figure 4A). Efficient knockdown of HIF-α-proteins was confirmed by immunoblotting (Figure 4B). Knockdown of HIF-1α reduced BNP mRNA levels significantly to 58±10% of DP-induced, control siRNA-transfected levels (P<0.05 against siLuc+DP; Figure 4A). siRNA against HIF-2α did not affect BNP mRNA expression (88±20% of control levels, no significant change). In HKC-8 cells, a reduction of BNP mRNA induction to 42±10% of DP-stimulated control values was also observed after knockdown of HIF-1α (P<0.05 against siLuc+DP; Figure 4C) but not after knockdown of HIF-2α (82±25%, no significant change). Furthermore, the activities of hBNP-promoter constructs also responded to HIF-1α knockdown in Hep3B cells. siRNA against HIF-1α reduced the DP-induced activity of the 967 bp BNP promoter construct from 9±0.5-fold to 2.5±0.1-fold (P<0.05 against siRNA luc, Figure 4D), whereas siRNA against HIF-2α had no effect (to 10.4±1.4-fold, no significant change).
siRNA-mediated knockdown of HIF-1α reduces BNP expression and BNP promoter-driven reporter gene activity
To further support the notion that hypoxic BNP regulation is mediated by HIF-1α, we used MEFs expressing or devoid of HIF-1α. BNP was only inducible in HIF-1α positive MEFs, but not in the HIF-1α−/− cell line (Figure 4E), confirming that BNP mRNA induction was dependent on HIF-1α, but not HIF-2α.
A functional CACAG motif upstream of the HRE at −466 nt is required for full activation of the human BNP promoter
Sequence analysis revealed the presence of a CACAG motif 27 bp upstream of the HRE at −466 nt. In a previous study this sequence motif, which can also be termed HAS (HIF-1 ancillary sequence), was identified downstream of the functional HRE in the EPO enhancer . To elucidate the contribution of this motif to BNP promoter function we tested the activation of a hBNP promoter reporter plasmid with a mutated HAS (GCACAGC→GAATAAC) in transiently transfected Hep3B cells. Mutation reduced the induction of promoter activity by DP from 6.2±1.2-fold to 3.3±0.7-fold (wild-type against CACAG mutant, P<0.05; Figure 5A). This was still significantly higher than the reduction observed after mutation of the HRE at −466 nt (1.1±0.04-fold; P<0.05). Mutation of both the CACAG and the HRE did not further change the promoter activation by DP (1.4±0.4-fold; HRE+CACAG mut) compared with 1.1±0.04-fold (HRE mut, no significant change; Figure 5A). Binding of nuclear proteins to the HAS was then investigated by EMSA. As depicted in Figure 5(B), specific binding of nuclear proteins was independent of DP stimulation and mutation abolished the binding complex. We conclude that the CACAG sequence upstream of the HRE at −466 nt is necessary for the full hypoxic response of the BNP promoter in Hep3B cells, indicating that a multipartite organization of the HRE, which is observed in other HIF-regulated genes, is also present in the BNP-HRE.
A CACAG motif upstream of the BNP-HRE is required for maximum hypoxic hBNP promoter activity
BNP has widely been recognized as a reliable and sensitive clinical marker of heart failure. Previous studies suggested that BNP levels are influenced by hypoxia, but the underlying molecular mechanisms remained largely elusive. In the present study, we demonstrate that hypoxia directly stimulates BNP mRNA induction and that this induction is HIF-dependent. We identified the functional HRE at −466 nt in the 5′-flanking region of the BNP gene and demonstrate that binding of endogenous HIF-1α, but not of HIF-2α, is sufficient for the hypoxic transactivation of BNP. These findings add novel insights into the molecular regulation of BNP gene expression and might have considerable clinical implications.
In vivo increased BNP mRNA abundance was previously reported after prolonged moderate hypobaric hypoxia  or myocardial ischaemia . However, the experimental settings used in these studies elicited complex pathophysiological responses, such as ventricular hypertrophy or altered cardiac wall tensions, which extend vastly beyond the effects of hypoxia alone. Our in vitro approach provides strong evidence that hypoxia is a direct and sufficient stimulus to induce BNP expression via binding of the endogenous HIF-1 complex to a functional HRE in the BNP promoter. In contrast with a previous report on primary neonatal cardiac cells, which did not show BNP induction in hypoxia [18,27], we demonstrate that five different non-cardiac cell lines as well as primary adult rat cardiomyocytes exhibited significant hypoxia-induced BNP expression. In addition, hypoxia was a sufficient stimulus to activate BNP promoter constructs in the transformed cell lines. Although BNP expression in vivo is limited to the heart, basic regulatory mechanisms such as the HIF-dependent gene induction appear to be well conserved in transformed cells of different origins. Potential inhibitors of gene expression may be absent or downregulated and HIF transcriptionally upregulated in these cells, which makes them a valuable tool to study basic mechanisms of hypoxic gene regulation. In contrast, the amplitude of BNP mRNA induction in primary rat cardiomyocytes and in rodent hearts in vivo was lower than in the transformed cells. The reasons for this reduced BNP gene activation remain to be determined, but potent inhibitors of HIF activation such as PHD 3  and CITED2 (cAMP-responsive element-binding protein-binding protein/p300-interacting transactivator with ED-rich tail 2)  are expressed at high levels in the heart and might contribute to a reduced HIF response of this tissue in vivo and in cultured cardiomyocytes in vitro. In support of this notion, a significant hypoxic response, including induction of classical HIF target genes and a vast increase in BNP mRNA levels, could only be elicited in cultured human foetal cardiac cells by infection with an adenoviral expression vector for a stable HIF-1α/VP16 chimaeric protein, but not by hypoxia or infection with an adenoviral expression vector for wild-type HIF-1α . Moreover, crosstalk with non-myoctes may be needed for maximum BNP upregulation under hypoxic conditions and/or by mechanical stress . In the present study, we therefore also investigated cultured rat and porcine cardiac fibroblasts, but they did not express BNP mRNA at detectable levels (results not shown).
Finally, rather than hypoxia, ischaemia is the most frequent trigger of cardiac BNP induction. Thus, reactive oxygen species or inflammatory mediators may possibly amplify the hypoxic response of the BNP gene in cardiac cells by activation of transcription factors which act in an additive or synergistic manner with HIF.
The in vitro findings of an attenuated BNP response to hypoxia in cultured cardiac cells was confirmed in the present study in vivo using short-term exposure of rats to hypoxia (8% O2) or 0.1% CO. Interestingly, the amplitude of BNP induction was lower in the hearts of hypoxic animals than in the CO-treated animals, which may, in part, be attributable to a more pronounced decline in blood pressure and heart rate in the hypoxic group compared with the CO-treated group (results not shown). This possibly counteracted BNP induction through a reduced cardiac workload. In addition, we have previously shown that CO exposure is a stronger stimulus for HIF target gene activation in vivo than hypoxia [26,35].
Our results define HIF-1α as a novel and potentially important mediator of increased BNP expression in addition to the known effect of myocyte stretch . Interestingly, volume overload and mechanical strain have been shown to induce HIF-1α protein and the HIF target gene VEGF in vivo [41,42]. It is possible that a regionally reduced oxygen supply generated by increased wall stress activates the HIF system which in turn could amplify the BNP response. Mechanical stress can also directly induce HIF-1α in cultured vascular smooth muscle cells . Therefore an increased BNP level may be regarded as an indicator of regional myocardial hypoxia or activation of the HIF system even before the onset of clinical symptoms or reduced coronary blood flow.
We identified the HRE at −466 nt as the functional cis-acting element mediating the HIF response of the BNP promoter. This result was surprising given the presence of three other putative HIF binding sequences, one in the 967 bp and two in the 2027 bp promoter construct, which did not contribute significantly to the hypoxic activation. The longer construct even showed a lower amplitude of induction upon hypoxic stimulation, potentially due to binding of additional inhibitory factors. The localization of a functional HRE at −466 nt in the BNP promoter is in agreement with a recent study of Luo et al. , but, in contrast with their study, we did not utilize forced overexpression of HIF-1α protein or a normoxically stable HIF-1α/VP 16 chimaeric protein to induce BNP. We demonstrate for the first time that the endogenous hypoxia-induced HIF-1 complex binds to HRE 2 in the BNP promoter in vitro by gel-shift analyses and in vivo by ChIP. This further corroborates our conclusion that the HRE at −466 nt is most important for the hypoxic BNP induction. We also show that the HIF-α isoform mediating the hypoxia response is HIF-1α, whereas HIF-2α has no effect on BNP induction. HIF-1 was initially identified as a hypoxia-inducible transcription factor that bound to the enhancer region of the human EPO gene . Subsequent studies revealed that HIF is the master regulator in a widespread system orchestrating the cellular response of mammalian cells to hypoxia . After the cloning of the second HIF-α subunit, HIF-2α or EPAS1 (endothelial Per-ARNT-Sim domain protein 1) , a number of studies were undertaken to differentiate the functional roles of the two HIF-α subunits [28,32]. Until now, the hypoxic induction of the vast majority of identified HIF target genes could be attributed to HIF-1α, whereas the function of HIF-2α is still not well defined. The question of which HIF-α isoform mediates the hypoxic BNP induction is of particular relevance, because cardiomyocytes are one of the few cell types which accumulate both HIF-α isoforms in hypoxia in vivo . Forced overexpression of the HIF-α subunits results in loss of target gene specificity , therefore this approach does not allow conclusions of which HIF-α isoform mediates the response. In fact, the same HIF-1α/VP 16 chimaeric protein which was used to induce BNP in cardiomyocytes  activated the known HIF-2α target gene EPO in astrocytes, whereas a plasmid expressing stable HIF-1α did not . These findings demonstrate that under certain circumstances the physiological regulation can be over-ridden. We applied RNA interference at BNP promoter constructs and at the endogenous gene to dissect the contribution of the two HIF-α subunits to BNP regulation. These experiments together with the EMSA revealed unambiguously that BNP induction is dependent on HIF-1α and not HIF-2α.
Interestingly, a CACAG sequence motif is present upstream of the only functional HRE at −466 nt. Thus the BNP-HRE exhibits an organization similar to the EPO-HRE  and VEGF-HRE . Mutational analysis in reporter constructs revealed that this CACAG sequence is functional and required for full hypoxic activation of the BNP promoter. The binding of nuclear protein to the CACAG motif was not inducible by DP, which is in keeping with the notion that the binding protein is expressed constitutively. The CACAG motif was previously described to be without effects on BNP promoter activity . Different experimental settings, for example different cell types used for reporter assays, might account for this discrepancy. The results of Luo et al.  were based on cotransfection of a 50 bp BNP-HRE promoter fragment together with HIF-1α or HIF-1α/VP16 expression vectors in HeLa cells. As mentioned above, we were not able to detect BNP mRNA in HeLa cells. We characterized the contribution of the CACAG motif using the hypoxia-inducible 967 bp BNP promoter construct in Hep3B cells. Since these cells express BNP endogenously, our approach may be more suitable to elucidate the regulatory pathways underlying the induction of the endogenous gene.
The involvement of HIF-1α in the hypoxic upregulation of BNP underscores the protective aspect of the HIF response. Several studies indicate that BNP, in addition to its beneficial effects on cardiac workload, may act as an autocrine factor with antiproliferative, antifibrotic  and direct cytoprotective  properties. Thus, HIF-1-mediated BNP induction may be viewed as part of the local defence mechanism of the myocardium that limits hypoxic damage. Like other HIF target gene products such as EPO and HO-1, therapeutic application of BNP or induction by ischaemic preconditioning may have protective effects under conditions of hypoxic injury [24,47]. In this respect, pharmacological inhibition of the HIF PHDs, which stabilize HIF-α and induce HIF target genes in vivo , is a promising strategy because it generates a cellular environment consisting of a plethora of protective factors in contrast with the administration of single target gene products. Recent results confirm the therapeutic principle that normoxic stabilization of HIF-α can ameliorate ischaemic injury via activation of HIF target genes in the kidney . In addition, PHD inhibition was shown to be effective in the heart and reduced cardiac remodelling after myocardial ischaemia . Thus, HIF represents an attractive molecular target in therapeutic approaches aiming at an amelioration of the cellular adaptation to low oxygen supply. Pharmacological HIF inducers may be used for an induction of HIF target genes including BNP to complement the endogenous response in order to limit myocardial tissue damage and prevent excessive cardiac remodelling.
This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft), Bonn, Germany (DFG, SFB 423, WE4275/1-1), the Interdisciplinary Center for Clinical Research (IZKF) at the University of Erlangen-Nuremberg and the Roche Foundation for Anaemia Research, Meggen, Switzerland (RoFAR). We thank Hans Fees and Andrea Kosel for excellent technical assistance.
(human) brain-type natriuretic peptide
carbonic anhydrase IX
Dulbecco's modified Eagle's medium
electrophoretic mobility-shift assay
hypoxia-inducible transcription factor
HIF-1 ancillary sequence
haem oxygenase 1
murine embryonic fibroblast
prolyl hydroxylase domain-enzyme
small interfering RNA
vascular endothelial growth factor