Transcriptional Activation of the Human Leptin Gene in Response to Hypoxia OF HYPOXIA-INDUCIBLE FACTOR 1*

In addition to having a major role in energy homeostasis, leptin is emerging as a pleiotropic cytokine with multiple physiological effector functions. The recently discovered proangiogenic activity of leptin suggested the hypothesis that its production might be regulated by hypoxia, as are other angiogenic factors. To examine this proposal, the expression of leptin protein and mRNA was measured and found to be markedly up-reg-ulated in response to ambient or chemical hypoxia (upon exposure to desferrioxamine or cobalt chloride), an effect that requires intact RNA synthesis, suggesting a transcriptional mechanism. Transient transfection of cultured cells with deletion constructs of the leptin gene promoter linked to a reporter gene revealed a functional hypoxia response element (HRE) located at position –116 within the proximal upstream region. This putative HRE harbors a characteristic 5 (cid:1) -RCGTG-3 (cid:1) core motif, a hallmark of hypoxia-sensitive genes and recognized by the hypoxia-inducible factor 1 (HIF1), which consists of a HIF1 (cid:1) /HIF (cid:2) heterodimer. Constructs harboring this –116/HRE supported reporter gene expression in response to hypoxia but not when mutated. Expression of HIF1 (cid:1)

In addition to having a major role in energy homeostasis, leptin is emerging as a pleiotropic cytokine with multiple physiological effector functions. The recently discovered proangiogenic activity of leptin suggested the hypothesis that its production might be regulated by hypoxia, as are other angiogenic factors. To examine this proposal, the expression of leptin protein and mRNA was measured and found to be markedly up-regulated in response to ambient or chemical hypoxia (upon exposure to desferrioxamine or cobalt chloride), an effect that requires intact RNA synthesis, suggesting a transcriptional mechanism. Transient transfection of cultured cells with deletion constructs of the leptin gene promoter linked to a reporter gene revealed a functional hypoxia response element (HRE) located at position -116 within the proximal upstream region. This putative HRE harbors a characteristic 5-RCGTG-3 core motif, a hallmark of hypoxia-sensitive genes and recognized by the hypoxia-inducible factor 1 (HIF1), which consists of a HIF1␣/HIF␤ heterodimer. Constructs harboring this -116/HRE supported reporter gene expression in response to hypoxia but not when mutated. Expression of HIF1␣ cDNA in normoxic cells mimicked hypoxia-induced reporter gene expression in cells cotransfected with the wild type leptin -116/HRE construct but not with the mutant. Gel shift assays with a 32 P-labeled leptin promoter -116/HRE probe and nuclear extracts from hypoxia-treated cells indicated binding of the HIF1␣/␤ heterodimer, which was blocked with an excess of unlabeled -116/HRE probe or a HIF1-binding probe from the erythropoietin gene enhancer. Taken together, these observations demonstrate that the leptin gene is actively engaged by hypoxia through a transcriptional pathway commonly utilized by hypoxia-sensitive genes.
Physiological mechanisms that ensure an appropriate level of oxygen (O 2 ) delivery to tissues have evolved in complex multicellular organisms. Virtually all cells are capable of sensing changes in O 2 tension (pO 2 ) and respond adaptively when the O 2 demand exceeds supply, a condition referred to as hypoxia (1). Hypoxia can develop as a result of ischemia resulting from hypoperfusion, either as a pathological condition or as a transient physiological event (1). Under chronic conditions of hypoxia, typical adaptation responses generally include changes in the expression of genes encoding molecules that facilitate O 2 delivery or by activating metabolic pathways that do not require O 2 , thus maintaining energy homeostasis when O 2 availability is limited (1,2). For example, hypobaric hypoxia leads to a classical response characterized by increased red blood cell mass formation after induction of the erythropoietin (Epo) 1 gene, whose expression is elevated markedly under these conditions (3,4). In addition, the vasodilators nitric oxide and carbon monoxide are generated by the catalytic activity of inducible nitric oxide and heme oxygenase-1, respectively; expression of the genes encoding these enzymes is induced readily after a reduction in pO 2 (5,6). Likewise, up-regulation of the vascular endothelial growth factor gene occurs in response to local vascular hypoxia, which leads to vigorous angiogenesis and vasodilation, such as in tumors (7,8) or during the healing of wounds (9). Finally, the hypoxia-induced metabolic shift from oxidative phosphorylation to glycolysis as the main source of ATP serves to illustrate another classic homeostatic mechanism in response to O 2 deprivation (10). Hence, increased expression of GLUT1 glucose transporter (11) or glycolytic enzymes such as aldolase A (12), enolase 1 (12), lactate dehydrogenase A (12,13), and phosphoglycerate kinase 1 (14) ensues rapidly after hypoxia to facilitate this metabolic adaptation.
Underlying the changes in expression of these erythropoietic, vasoactive, or enzymatic molecules are regulatory mechanisms that modulate transcription of the corresponding genes as well as the rate of mRNA degradation (15,16). A crucial component in the induction of hypoxia-regulated genes is the transcription factor hypoxia-inducible factor 1 (HIF1), which activates transcription by binding to a specific cis-acting regulatory sequence referred to as hypoxia response element (HRE), a hallmark of hypoxia-sensitive target genes (2,17,18). The HIF1 protein complex consists of a heterodimer composed of HIF1␣ and HIF1␤ subunits (17,18). Each of these subunits contains basichelix-loop-helix motifs and other functional domains required for DNA binding, protein heterodimerization, and transactivation of target genes (2,15,16). The HIF1␣ subunit, an 826amino acid protein of 93 kDa, is a member of the PAS protein family by virtue of its homology to Drosophila melanogaster genes encoding the period (Per) and single-minded (Sim) transcription factors and to the mammalian arylhydrocarbon receptor nuclear translocator protein (ARNT), which is now recog-nized to be the HIF1␤ subunit (17). The HIF1␤/ARNT subunit can also heterodimerize with the arylhydrocarbon receptor (AHR) forming HIF1␤⅐AHR complexes, which mediate the transcriptional xenobiotic response (19). However, under conditions of hypoxia, it is the HIF1␣/␤ heterodimer that is translocated to the nucleus where it binds to HREs in target genes, thereby causing activation of transcription (15,16). The HRE consists of the pentanucleotide core consensus sequence 5Ј-RCGTG-3Ј, which can be located in the 5Ј-or 3Ј-flanking regions as well as within the introns of hypoxia-inducible genes (16).
We have recently reported that the hormone leptin exhibits robust angiogenic activity when assayed by in vitro and in vivo experiments (20). This unexpected biological activity of leptin is in contrast with the prevailing view of its function, which portrays leptin as a key regulator of body weight primarily through its extensively documented central nervous systemmediated effects on food intake and energy expenditure (21). However, a more complex scheme of leptin action has subsequently emerged involving other peripheral physiological systems in addition to central regulation of appetite and body weight (21,22). Consistent with an important role in angiogenesis, it is conceivable that leptin expression might also be modulated by physiological cues that typically regulate angiogenic factors. Therefore, we wanted to investigate the hypothesis that expression of the leptin gene is induced by hypoxia through transcriptional machinery and mechanisms commonly used by genes encoding products involved in angiogenesis or oxygen homeostasis.
Here we show that leptin expression is markedly up-regulated in response to hypoxia, an event that requires intact RNA synthesis. We examine the activity of the leptin gene promoter linked to a reporter gene in transiently transfected cells and demonstrate the existence of a functional HRE, which is required for hypoxic induction and is transactivated by HIF1␣. We also show that the HIF1␣/␤ heterodimer binds to this leptin promoter HRE upon exposure to hypoxia, and we demonstrate that HRE mutations that disable HIF1␣/␤ binding eliminate leptin promoter activation. Taken together, these observations demonstrate that the leptin gene is actively engaged by hypoxia through mechanisms that are common to other hypoxiainducible genes, consistent with the concept of leptin as a bona fide angiogenic factor.

EXPERIMENTAL PROCEDURES
Materials-The monoclonal antibodies against human cell-specific markers (fibroblast, CD31, cytokeratin-1, muscle-specific actin) used for immunostaining and verification of fibroblastic character were purchased from Dako Ltd. (Cambridgeshire, UK). Fetal bovine serum, cell culture medium (RPMI), and reagents were from Invitrogen. Radioimmunoassay kits for detection of human leptin were obtained from Linco Research, Inc. (St. Louis, MO). The pCR2.1/TA vector for direct cloning of PCR products was purchased from Invitrogen, and the pGL3-Basic vector for promoter/luciferase plasmid construction was from Promega (Madison, WI). The unique site-elimination mutagenesis kit, Hybond-N ϩ nylon membranes, horseradish peroxidase-conjugated goat anti-mouse IgG, and enhanced chemiluminescence (ECL) kits and high performance ECL film (Hyperfilm) were all purchased from Amersham Biosciences. The TRI-Reagent for preparing RNA was obtained from Molecular Research Center (Cincinnati, OH). The pSV␤Gal reporter vector was from CLONTECH (Palo Alto, CA), and the reporter lysis buffer and ␤-galactosidase chemiluminescence assay kit were purchased from Promega. The HIF1␣ cDNA expression vector was obtained from Novus Biologicals (Littleton, CO), and the anti-HIF1␣ and anti-HIF1␤ antibodies were from BD Transduction Laboratories (Franklin Lakes, NJ). CoCl 2 and desferrioxamine (DFO) were purchased from Sigma. Polyvinylidene difluoride membranes were from Bio-Rad. The isotopes [␣-32 P]dCTP (3,000 Ci/mmol) and [␥-32 P]ATP (6,000 Ci/mmol) were obtained from PerkinElmer Life Sciences.
Fibroblast Isolation and Cell Culture-For initial experiments, human skin dermal fibroblasts (hSDF) were isolated from explants of human dermis cut into 5-mm 3 pieces and covered with RPMI culture medium, supplemented with 20% fetal bovine serum, 200 units/ml penicillin, and 200 g/ml streptomycin, until fibroblasts began to migrate out of the explants. Residual explants were then removed, and freshly isolated cells were propagated in RPMI medium containing 20% fetal bovine serum, 2 mM L-glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin sulfate. For studies of endogenous leptin expression, cells were kept at low passage (never exceeding 4) and grown to reach 70 -80% confluence. Only freshly isolated cells were grown (never from a frozen stock). The fibroblastic character of the cells isolated was verified by positive stain reaction with anti-human fibroblast (clone 5B5) and negative stain with anti-human CD31 (clone JC/70A), cytokeratin-1 (clone 34␤B4), or muscle specific actin (clone HHF35). For other experiments involving hSDF, normal human dermal fibroblasts were purchased from Clonetics (San Diego, CA) and cultured in fibroblast growth medium supplemented with 10% fetal bovine serum, insulin, and human basic fibroblast growth factor (Clonetics), following the manufacturer's instructions. Only cells from passages 3-12 were used in this study. For some experiments, HeLa cells (ATCC CCL-2) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin, and streptomycin sulfate.
For experiments requiring ambient hypoxia conditions, cells were incubated in a 5% CO 2 humidified atmosphere connected to a source of nitrogen. The oxygen level was adjusted by mixing in a balanced tension of N 2 through a gas inlet valve controlled by a model 110 Proox programmable oxygen sensor (Reming Bioinstruments, Redfield, NY), set to either 21% O 2 for normoxia or the indicated levels of oxygen (5-0.5%) for hypoxia. For experiments requiring chemical hypoxia, cells were incubated in the presence of 150 mM CoCl 2 or 100 mM DFO for the indicated periods of time.
Measurement of Leptin Levels in Conditioned Medium-Culture medium from subconfluent monolayers of fibroblasts exposed to the indicated levels of oxygen (hypoxia or normoxia) was collected and stored immediately at -80°C until further use. The concentration of leptin was measured using a radioimmunoassay kit for human leptin, essentially following the manufacturer's instructions. Duplicate samples were serially diluted 2-fold falling within the sensitivity range of the assay kit (0.05-10 ng/ml), using human recombinant leptin as a calibration standard.
Reverse Transcription-PCR (RT-PCR) and Northern Blot Analysis-Fibroblasts exposed to normoxic or hypoxic conditions were harvested, and total cellular RNA was extracted by a modified acid-guanidinium thiocyanate-phenol-chloroform method (TRI-Reagent). Reverse transcription and PCR amplification of the full-length leptin coding region in the mRNA transcript (Ref. 23; GenBank accession no. U43653) were conducted using the Titan One Tube RT-PCR Kit (Roche Molecular Biochemicals) in the presence of 0.5 g of total RNA, a forward primer (5Ј-CCATCCTGGGAAGGAAAATG-3Ј), and a reverse primer (5Ј-CCCT-TAACGTAGTCCTTGCAG-3Ј). The resulting PCR products were loaded onto a 1% agarose gel and visualized by ethidium bromide staining. A hybridization probe for human leptin was generated by RT-PCR using 1 g of total RNA prepared from human adipose tissue and the same primers mentioned above. The resulting 526-bp RT-PCR product was ligated directly into a pCR2.1/TA vector, and the presence of the insert was then verified by automated DNA sequencing, using the fluorescently labeled dideoxynucleotide chain termination method (24). For probe preparation, the cDNA insert was excised by digestion with EcoRI, gel purified, and then radiolabeled with [␣-32 P]dCTP using a random primer labeling kit (Stratagene, La Jolla, CA).
For Northern blot analysis, equal amounts of fibroblast total RNA (13 g/lane) were electrophoresed on 1% agarose and formaldehyde gels and transferred to nylon membranes (Hybond-N ϩ ). Hybridization was performed in 5ϫ SSC, 10ϫ Denhardt's, and 1% SDS, for 16 h at 60°C. Membranes were washed twice in 2ϫ SSC, 1% SDS for 30 min at 60°C and once in 0.2ϫ SSC at 22°C before exposure to autoradiography.
Promoter/Reporter Plasmids and Transient Transfections-A genomic DNA fragment of the human leptin gene (23) containing ϳ2.9 kb of 5Ј-flanking region was prepared by PCR amplification of human DNA using the primers 5Ј-AAGGATGGAGAGGCCCTAGTG-3Ј (forward) and 5Ј-CTTGCAACCGTTGGCGCTGCG-3Ј (reverse). A 2.9-kb PCR amplicon product was obtained and cloned into a pCR2.1/TA vector. The insert in the resulting construct was then isolated and fully verified by automated DNA sequencing. From this construct, a series of 5Ј-nested deletion mutants was generated by restriction enzyme digestion with HindII, NsiI, ScaI, StuI, SacII, and XhoI, which cleave respectively at positions -2643, -2041, -1687, -983, -375, and -172, preceding the start site of transcription. Each fragment was subcloned upstream of the firefly luciferase reporter gene in a pGL3-Basic vector and designated accordingly. For creating the p(-63)/LUC deletion construct, PCR amplification of the 2.9-kb human genomic DNA fragment was conducted in the presence of primers spanning the region between -63 and ϩ28. The forward primer 5Ј-AATCGctcgagCGGGGCAGTT-GCGCAAGTTG-3Ј and the reverse primer 5Ј-ACTGaagcttGCAACCGT-TGGCGCTGCG-3Ј were used, where each primer is designed to contain a 5Ј-overhang restriction site (lowercase underlined), XhoI for the forward primer and HindIII for the reverse primer, thus allowing for directional insertion of the amplified product into pGL3-Basic vector. PCRs were conducted for 30 cycles in a total volume of 50 l using a program consisting of denaturation at 94°C for 30 s, annealing at 52°C for 30 s, and extension at 72°C for 1 min, using a GeneAmp PCR System 9700 thermocycler (Applied Biosystems). Mutations of the -116/ HRE were introduced into the p(-172)/LUC by the unique site elimination method (25), using the mutagenic oligonucleotide 5Ј-GCAGCCGC-CCGGtAtGTCGCTACCCTGAGG-3Ј, thus replacing C by T within the -116/HRE core motif (lowercase underlined). The sequence of the resulting construct was verified by automated DNA sequencing.
Plasmids were transfected into cultured hSDF by lipofection using the FuGENE 6 reagent (Roche Molecular Biochemicals), according to the manufacturer's instructions. Fibroblasts were seeded in 24-well plates at a density of 6 ϫ 10 4 cells/well and transfected with 0.6 g of each reporter plasmid. Transfection efficiencies were normalized by cotransfection with 0.1 g of pSV␤Gal reporter gene. After 24 h, cells were incubated under normoxic or hypoxic conditions for the times indicated (generally 12-16 h), and total lysates prepared from each well were added to 0.2 ml of reporter lysis buffer (Promega). Cell extracts were analyzed for luciferase and ␤-galactosidase activity using a chemiluminescence assay kit and read in a model 1450-024 multiwell plate luminometer (PerkinElmer Life Sciences). For some experiments, 0.6 g of a HIF1␣ cDNA expression plasmid was cotransfected with the indicated leptin promoter/reporter plasmids, and cell lysates were prepared and processed for reporter gene expression activity as described above.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared essentially as described (26). Briefly, cells were harvested, collected by centrifugation, and exposed to a hypotonic buffered solution (10 mM Hepes pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, and a protease inhibitor mixture). Nuclei were recovered by centrifugation and extracted for 1 h with a high salt buffer (20 mM Hepes pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 5% glycerol, and a protease inhibitor mixture). The resulting extracts were collected after centrifugation of nuclei, snap frozen, and kept at -70°C until used.
The EMSA double-stranded probes were based on the sequence of the human leptin gene promoter contained in p(-172)/LUC, which harbors the proximal HRE at -116 in a reverse orientation toward the 5Ј-end of the insert. Specifically, the sequence of the EMSA probes used was: 5Ј-GCCGCCCGGCACGTCGCTACCCTG-3Ј for the wild type (Lep HRE ) and 5Ј-GCAGCCGCCCGGtAtGTCGCTACCCTGAGG-3Ј for the mutant (Lep HRE-mut ), where the sequence corresponds to the sense strand and mutations are shown in lowercase and italics. For some experiments, a 21-bp double-stranded oligonucleotide probe based on the 3Ј-enhancer of the Epo gene (27), which contains a HIF1 binding HRE site, was used and had the sequence 5Ј-GCCCTACGTGCTGCCTCGCAT-3Ј. The annealed, double-stranded leptin promoter probes were labeled with [␥-32 P]ATP using T4 polynucleotide kinase (28), and the resulting radiolabeled probes were then purified on NucTrap columns (Stratagene). Binding reactions were carried out in a total volume of 20 l containing 4 g of nuclear extract protein, 20 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM MgCl 2 , 0.5 mM EDTA, 5 mM dithiothreitol, 5% glycerol (v/v), and 250 ng of poly(dI⅐dC). After preincubation for 5 min at room temperature, radiolabeled probe (50,000 cpm, 1 ng) was added, and the incubation continued for 15 min. For gel supershift assays, 1 g of anti-HIF1␣ antibody alone or in combination with 1 g of anti-ARNT (HIF1␤) antibody was added to the complete binding reaction mixtures, and the incubation continued for 60 min at room temperature. For competition experiments, a 10 -100-fold molar excess of the indicated unlabeled probes was preincubated with nuclear extract and poly(dI⅐dC) prior to the addition of 32 P-labeled probe. The DNA⅐protein complexes were then resolved by electrophoresis onto 5% nondenaturing polyacrylamide gels with 0.5% TBE buffer (45 mM Tris, pH 8.3, 45 mM boric acid, 1 mM EDTA). Gels were dried, and radioactivity was detected by bioimaging using a Cyclone PhosphorImaging System and the Optiquant software package (Packard Instruments Co., Downers Grove, IL).
Immunoblot Analysis-Nuclear extracts were prepared from normoxia or hypoxia-treated cells, and proteins were resolved on a SDSpolyacrylamide gel (10 g/lane) and then transferred onto polyvinyli-dene difluoride membranes. Membranes were blocked with 5% nonfat dry milk in TBST (10 mM Tris-HCl, pH 8, 0.15 M NaCl, 0.05% Tween 20) and incubated at 4°C overnight with 1 g/ml anti-HIF1␣ or anti-HIF1␤ antibody. Blots were washed and incubated with horseradish peroxidase-conjugated goat anti-mouse IgG for 60 min at room temperature. Immune complexes were detected by ECL followed by exposure to high performance chemiluminescence film (Hyperfilm).

Rationale for the Use of Fibroblasts in Leptin Expression
Studies-A large body of experimental evidence has been collected showing that adipocytes represent the major source of leptin, although additional, non-adipocyte sites of leptin synthesis have also been uncovered more recently, including placenta, skeletal muscle, stomach fundic mucosa, and mammary epithelium (21,22). In the course of independent studies on the role of angiogenesis in wound healing, the discovery was made that skin dermal fibroblast-like cells actively express leptin upon inflicting incisional, full-thickness skin wounds in mice, an observation supported by immunohistochemical, in situ hybridization and quantitative RT-PCR data. 2 Although the possible significance of this observation is beyond the scope of the present manuscript, we surmised that dermal fibroblasts might actively engage in leptin synthesis in response to the hypoxia conditions known to develop in the wound environment, most likely as a result of hypoperfusion ischemia (29,30). Therefore, hSDF were used primarily throughout the experiments described herein as a non-adipocyte cellular system with which to study possible regulation of leptin expression by hypoxia. However, for certain experiments, HeLa cells were also utilized as a convenient source of nuclear proteins for verification of gel shift assay results (see below).
Induction of Leptin Expression by Hypoxia-To determine whether the expression of leptin might be regulated in response to a reduction in O 2 tension, freshly prepared hSDF in culture were exposed to an atmosphere of normoxia (21% oxygen) or the indicated levels of hypoxia (5% oxygen or less) for 48 h. The quantity of leptin secreted into the culture medium was then determined by radioimmunoassay using antibodies specific for human leptin (see "Experimental Procedures"). As shown in Fig. 1, exposure of dermal fibroblasts to escalating levels of hypoxia (from 5 to 0.5% oxygen) leads to a marked increase in the amount of leptin produced and secreted by these cells into the culture medium. For example, exposure of fibroblasts to hypoxia at 1% O 2 gives rise to an increase in the level of immunoreactive leptin in the culture medium by Ͼ8-fold, from Ͻ0.05 ng/ml (undetectable) to 0.4 ng/ml (see Fig. 1). In addition, expression of leptin was also assessed by Northern blot analysis using total cellular RNA prepared at various times of hypoxia. After size fractionation of RNA by agarose/ formaldehyde gel electrophoresis, RNA was transferred to nylon membranes, hybridized to a 32 P-labeled human leptin cDNA probe, and exposed to autoradiography. As illustrated in Fig. 2A, hypoxic treatment of fibroblasts causes a rapid induction in leptin mRNA, already detectable after 2 h of hypoxia and apparently reaching completion after 12 h. Importantly, this hypoxia-induced up-regulation in leptin mRNA requires ongoing RNA synthesis because the effect is abolished by treatment of fibroblasts with actinomycin D prior to a short hypoxic exposure for 2 h, as shown in Fig. 2B. Thus, expression of leptin is vigorously induced in response to hypoxia in skin dermal fibroblasts. Because this up-regulation requires an intact RNA synthesis machinery, it appears that the effect depends on active transcription (most likely of the leptin gene) as a prerequisite for full hypoxia-induced leptin expression.
Given the crucial role of HIF1␣ in the transcriptional activation of target genes in response to hypoxia, experiments were conducted to determine whether expression of this protein might be increased in hSDF upon exposure to reduced O 2 tension. As expected, hypoxia results in markedly elevated levels of HIF1␣ detected by immunoblot analysis of nuclear extracts from treated fibroblasts (Fig. 2C). In contrast, expression of HIF1␤ was not affected by this treatment (Fig. 2C, right  panel). Acute up-regulation of HIF1␣ is a well known effect of hypoxia in other systems, and it reflects primarily a drastic reduction in the degradation rate of the protein by the proteosomal pathway (16,31,32). Furthermore, chemical hypoxia produced by treatment of hSDF with CoCl 2 or DFO under conditions of normal atmospheric O 2 tension also results in dramatic induction of HIF1␣, an effect that is indistinguishable from that observed upon ambient hypoxia (Fig. 2C, left and  middle panels). These findings are consistent with classical observations indicating that cobaltous ions and iron chelators mimic the stabilization effect of hypoxia upon HIF1␣, suggesting the involvement of a specific ferroprotein oxygen sensor in this process (33)(34)(35)(36). As expected, cells exposed to chemical hypoxia also exhibit notable induction of leptin expression as demonstrated by detection of a 526-bp RT-PCR product originating specifically from leptin mRNA present in hypoxic cells but not in untreated cells (Fig. 2D). Taken together, these results suggested that HIF1␣ might be an important mediator  2. Effect of hypoxia upon the expression of leptin and HIF1 proteins. A, cultured hSDF were kept under normoxia (N) or exposed to a 1% O 2 hypoxia environment for the times indicated, and then total RNA was prepared and fractionated by agarose gel electrophoresis. After capillary transfer to nylon membranes, the blots were hybridized with a 32 P-labeled, full-length human leptin cDNA probe, washed, and exposed to autoradiography. For comparison, the relative migration of leptin mRNA present in human adipocytes (Ad) is shown in the right panel. The ethidium bromide staining pattern is shown in the lower panel to verify equal loading of RNA samples in each lane. B, cells were cultured and left intact or treated with 5 g/ml actinomycin D for 2 h before exposure to either a normoxia (N) or 1% O 2 hypoxia environment for 2 additional h. RNA was then prepared and processed as described above for A. Hyp, cells exposed to hypoxia only; Hyp ϩ Act D, cells pretreated with actinomycin D and then exposed to hypoxia. C, cells maintained in normoxia (N) or exposed to either 1% O 2 ambient hypoxia (N 2 ) or chemical hypoxia with CoCl 2 or DFO for 16 h were collected, and total soluble nuclear extracts were then prepared as described under "Experimental Procedures." Proteins were resolved by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and developed for immunoblot analysis using antibodies against HIF1␣ (left and middle panels) or HIF1␤ (right panel). For comparison, extracts from HeLa cells exposed to hypoxia or not were analyzed by immunoblotting with respect to their content of HIF1␣ using the appropriate antibodies (middle panel). D, cells were kept under normoxia or exposed to ambient (N 2 ) or chemical hypoxia with the indicated agent (CoCl 2 or DFO) for 16 h, and then total RNA was prepared and subjected to RT-PCR as described under "Experimental Procedures." PCR products were resolved by agarose gel electrophoresis and visualized by ethidium bromide staining. The arrow indicates the position of the expected 526 bp band from amplification of the leptin mRNA transcript with respect to molecular weight markers on the left. of the induction of leptin expression observed in hSDF exposed to hypoxia.
Transcriptional Activation of the Leptin Gene Promoter by Hypoxia-To test directly the hypothesis that hypoxia causes a transcriptional activation of the leptin gene, a fragment of human genomic DNA encompassing ϳ2.9-kb of 5Ј-flanking sequence (23) was subcloned upstream of a luciferase reporter gene. Inspection of the nucleotide sequence within this region revealed the existence of multiple 5Ј-RCGTG-3Ј core motif HREs, putatively identified as potential HIF1 binding sites. The relative position and orientation of these elements are indicated at the top of Fig. 3, in reference to specific restriction enzyme digestion sites that were used to create truncations of the promoter region, thereby producing a series of nested deletion constructs (Fig. 3). Thus, a battery of seven different constructs was produced from positions -2912 (full-length fragment), -2643 (HindII), -2041 (NsiI), -1687 (ScaI), -983 (StuI), -375 (SacII), and -172 (XhoI), all extending 28 bp past the start site of transcription. An additional construct consisting of a fragment from -63 to ϩ28 was generated by PCR amplification employing appropriate flanking primers (see "Experimental Procedures") to create p(-63)/LUC, which bears the shortest segment of 5Ј-flanking region. Each of these leptin promoter luciferase plasmids was transiently transfected into hSDF that were subsequently subjected to chemical hypoxia (or not) for a period of 16 h. 3 Total soluble cell extracts were prepared, and the level of luciferase activity was then examined by chemiluminescence. As shown in Fig. 3, it is evident that exposure to hypoxia results in a significant increase in the activity of luciferase measured in extracts derived from cells transfected with any of the promoter/reporter constructs, except for that containing the shortest segment of upstream sequence, i.e. p(-63)/ LUC (see Fig. 3). Thus, regulatory element(s) presumably residing in the intervening region comprised between positions -2912 (full-length promoter construct) and -63 seem(s) to be The restriction enzyme digestion sites indicated were used to produce 5Ј-nested deletion mutants, giving rise to the constructs shown and designated by the length of 5Ј-flanking region that they contain (the shortest construct, p(-63)/LUC, was generated by PCR amplification of the appropriate genomic region (see "Results")). Each of these promoter/reporter constructs was transiently transfected into cultured hSDF and subsequently exposed to chemical hypoxia with DFO (or not) for 16 h. Total soluble extracts were then prepared, and the level of luciferase activity was determined by chemiluminescence. Results (mean Ϯ S.E.) are expressed relative to the activity observed with the promoterless construct pGL3-Basic, and they represent data from three independent experiments. capable of supporting an ϳ2.5-fold induction of reporter gene activity in response to hypoxia. Because this effect is no longer present in cells transfected with p(-63)/LUC, it would appear that a putative regulatory element required for hypoxia-induced reporter gene expression resides in a short segment of 91 bp, located between positions -172 and -63. However, this conclusion must be interpreted with caution given the short distance to the start site of transcription in p(-63)/LUC, which may fail to support basal transcription adequately as indicated by a level of reporter activity indistinguishable from that achieved with a promoterless luciferase construct (Fig. 3). Therefore, an alternative approach involving the use of sitespecific mutants of p(-172)/LUC was used to assess the existence and functional significance of putative HRE elements within this short segment (-172 to -63) of the leptin gene promoter.
Identification of a Functional HRE in the Proximal Promoter Region of the Leptin Gene-To determine whether a functional HRE exists within the leptin proximal promoter segment represented in p(-172)/LUC, specific point mutations were introduced into the core sequence of the putative HRE site contained in this construct (see "Experimental Procedures"). Fibroblasts were transiently transfected with either the wild type or mutant promoter/reporter construct and then kept under normoxia or exposed to chemical hypoxia by treatment with CoCl 2 or DFO. As shown in Fig. 4, although the wild type p(-172)/ LUC HRE construct supports hypoxic induction of luciferase activity by 2.5-3-fold, this activation is abrogated markedly when the mutant construct, i.e. p(-172)/LUC HRE-mut , is transfected. Although it is evident that treatment with either CoCl 2 or DFO effectively mimics hypoxic induction of luciferase reporter gene expression, this activation does not occur when the HRE mutant construct is used (Fig. 4, left panel). Thus, it appears that hypoxic transcriptional induction of the leptin promoter requires an intact HRE located between -116 and -121. Sitespecific mutations introduced into the core sequence of this motif block hypoxic induction, presumably as a result of failure of HIF1␣ to bind to the mutated HRE site (see below).
To determine whether HIF1␣ is capable of transactivating reporter gene expression directly through the proximal HRE site, transient cotransfection experiments were performed under normoxic conditions. As illustrated in Fig. 4 (right panel), transfection of the HIF1␣ cDNA results in significant transactivation of luciferase activity when cells are cotransfected with the wild type p(-172)/LUC HRE construct but not with the mutant p(-172)/ LUC HRE-mut plasmid. This result supports the notion that HIF1␣, operating through an intact HRE motif in the proximal upstream region, is a transactivator of the leptin gene promoter.
Binding of the HIF1 Complex to the -116/HRE Site in the Proximal Promoter Region-To determine whether the proximal HRE site at -116 is capable of interacting with a HIF1␣/␤ heterodimer, EMSA experiments were conducted. Cells exposed to normoxia or hypoxia were harvested, lysed, and nuclear extracts were prepared and incubated with a 32 P-labeled double-stranded 24-mer probe based on the sequence of the leptin promoter region between -127 and -105, which contains the -116/HRE (see "Experimental Procedures"). The resulting DNA⅐protein complexes were then resolved by nondenaturing gel electrophoresis and visualized after exposure to autoradiography. As shown in Fig. 5A, hypoxia causes the appearance of a distinct complex with a retarded electrophoretic migration  5. EMSA analysis of the leptin promoter -116/HRE site in cells exposed to hypoxia. A, nuclear extracts prepared from hSDF kept under normoxia or exposed to hypoxia for 16 h were incubated with a 32 P-labeled Lep HRE gel shift probe. The resulting complexes were resolved by nondenaturing gel electrophoresis, and their location was revealed by phosphorimager detection. Complexes observed in extracts from cells under normoxia (N, lane 1) or hypoxia conditions (Hyp, lanes 2-11) are indicated by the arrows. Binding reactions were carried out in the presence of a 10-, 50-, and 100-molar excess of unlabeled homologous probe (Lep HRE , lanes 3-6), the same probe with site-specific mutations within the HRE site (Lep HRE-mut , lanes 6 -8), or a related probe based on the HRE site of the Epo gene enhancer (Epo, lanes 9 -11; see "Experimental Procedures"). B, for comparison, HeLa cells were treated and nuclear extracts processed in a manner identical to that described for A. The relative positions of equivalent complexes are also indicated (N and Hyp). that appears as a single band (lane 2). This complex can be eliminated completely by competition with an excess of the homologous oligonucleotide (lanes 3-5) but not with a probe in which the -116/HRE has been mutated (lanes 6 -8). Importantly, another probe based on the HRE motif present in the Epo gene enhancer (27), which is known to interact with HIF1 (37), is also an effective competitor (lanes 9 -11). For comparison, nuclear extracts prepared from HeLa cells exposed to ambient hypoxia for 16 h were used for equivalent EMSA experiments, yielding virtually identical results (Fig. 5B). Thus, other cells known to undergo an equivalent induction of HIF1␣ protein in response to hypoxia (i.e. HeLa) also support formation of the same DNA⅐protein complexes when using a leptin promoter probe containing the -116/HRE site.
To verify that the hypoxia-inducible EMSA band corresponds to a HIF1 complex, gel supershift experiments were conducted using specific polyclonal antibodies directed against HIF1␣ or HIF1␤. As shown in Fig. 6A, both anti-HIF1␣ and anti-HIF1␤ antibodies are capable of producing a conspicuous supershift band (lanes 3 and 4), accompanied by complete disappearance of the original EMSA band observed under hypoxic conditions (lane 2). Furthermore, the addition of both antibodies results in a highly retarded complex exhibiting a slower mobility than that achieved by each antibody alone (lane 5). In contrast, the use of an irrelevant antibody does not alter the mobility of the hypoxia-induced EMSA band (compare lanes 2 and 6). Again, for comparison purposes, nuclear extracts from HeLa cells exposed to hypoxia were used in the same manner as hSDF extracts with identical qualitative results, although the relative abundance of the complexes is not equivalent (Fig. 6B). Taken together, these findings indicate that hypoxia induces activation of the leptin gene promoter as a result, at least in part, of recruitment and binding of an intact heterodimer HIF1␣/␤ to the -116/HRE site within the proximal promoter region of the leptin gene. 4

DISCUSSION
In this report, we show for the first time that the leptin gene is transcriptionally activated in response to hypoxia through a mechanism that involves binding of the heterodimer HIF1␣/␤ to a functional HRE site located within the proximal promoter region. There are several key observations leading to this conclusion. First, exposure of human skin dermal fibroblasts in culture to a hypoxic environment causes a marked increase in the level of immunoreactive leptin secreted in the conditioned medium. This elevation is accompanied by a corresponding increase in the cellular content of leptin mRNA, an effect that requires intact RNA synthesis. Second, nested deletion analysis of the 5Ј-upstream region uncovers at least one functional HRE site located at -116 which is required for a full transcriptional response to hypoxia and which carries a 5Ј-RCGTG-3Ј core consensus sequence. Third, under normoxic conditions, treatment with CoCl 2 or DFO, agents that are commonly used to mimic induction of hypoxia-responsive genes through stabilization effects on HIF1␣, leads to activation of reporter gene expression driven by p(-172)/LUC HRE , which contains the proximal HRE at -116; site-specific mutations introduced into this HRE disable hypoxic induction. Fourth, cotransfection with a cDNA encoding HIF1␣ transactivates reporter gene expression from p(-172)/LUC HRE under normoxia conditions, but not when the HRE mutant p(-172)/LUC HRE-mut construct is used. Finally, EMSA experiments employing nuclear extracts from cells exposed to hypoxia show that both HIF1␣ and HIF1␤ bind as a heterodimer to the intact -116 HRE locus, but not when it has been mutated. Furthermore, a HIF1-binding probe (based on the sequence of the HRE site of the Epo gene enhancer) blocks binding of HIF1 to the -116/HRE. Taken together, these observations indicate that transcription of the leptin gene is stimulated by hypoxia through a mechanism that involves binding of heterodimeric HIF1␣/␤ to the proximal HRE site located at -116 within the leptin gene promoter.
Our recent discovery (20) (and also by others (38)) of the novel proangiogenic activity of leptin primarily led us to investigate whether its expression might be regulated by hypoxia in a fashion similar to that of many other angiogenic factors. In this regard, previous studies had shown that leptin is produced 4 It would appear that a transcriptional response alone is insufficient to account for the robust level of leptin expression observed upon exposure of fibroblasts to hypoxia. Although hypoxia-induced leptin promoter activation reveals a 3-fold induction at best, it is evident that the magnitude of the increase in the level of secreted leptin, or leptin mRNA induction, occurs to a much higher extent (Figs. 1 and 2). Nonetheless, we propose that the transcriptional activation of the leptin promoter described herein contributes in part to the hypoxic response. in placental trophoblasts (39), and its circulating levels are increased in women with severe preeclampsia (40), an acute hypertensive condition associated with diminished uteroplacental blood flow and placental hypoxia. Furthermore, the human choriocarcinoma cell line BeWo produces leptin in culture, and its rate of secretion is also augmented upon exposure to hypoxia (41), although the pathophysiologic significance of placental production of leptin is unknown at present. Efforts to understand the basis for transcriptional regulation of leptin gene expression in placenta have revealed upstream nuclear protein-binding, cis-acting elements between -1885 and -1830 which support differential expression in placental trophoblasts but not in adipocytes (41). Close inspection of the sequence within this segment reveals two sets of direct repeats, where the 3Ј-half region may provide binding sites for nuclear proteins as mutations introduced in this region cause loss of promoter activity (41). An independent study confirmed expression of leptin in choriocarcinoma cell lines, including JEG-3 and JAR cells, as well as the existence of the aforementioned upstream placental enhancer whose activity is required for leptin expression in placental cells but not in adipocytes or HeLa cells (42). Thus, although these studies generally illustrate mechanisms that regulate expression of the leptin gene in non-adipocytes (i.e. placental trophoblastic cells), they do not address whether hypoxia per se alters the rate of transcription of the leptin gene.
In a more recent study, a multifold promoter activation in response to hypoxia was shown by using deletion mutants of the 5Ј-flanking region of the leptin gene linked to a reporter gene, spanning a region between -2.38 and -0.17 kb (43). It is noteworthy that in that report report hypoxic induction appears to be more significant when longer constructs extending up to -1.87 kb of upstream region are used, leading the authors to speculate whether this effect might be mediated by HIF1 binding (43). This hypothesis, although not proven directly, finds support in the existence of a putative 5Ј-RCGTG-3Ј HRE with a distal location at -1832 (and contained within our p(-2041)/LUC construct), which is almost identical to the HIF1binding HRE present in the 3Ј-flanking region enhancer of the human Epo gene. However, because a significant level of hypoxic induction still remains with shorter promoter/reporter constructs, it is not immediately obvious which element(s) could be involved. This is particularly the case considering that there are at least seven other putative HREs downstream from the -1832 HRE site (see Fig. 3) as well as other functional motifs that have been shown to participate in activation of hypoxia-regulated genes (see below). Our results show that the relative stimulation caused by hypoxia is not significantly different between p(-2041)/LUC (which contains the upstream HRE site at -1832) and p(-1687)/LUC (which does not); furthermore, site-specific mutations that disable binding of HIF1 to this -1832/HRE do not prevent hypoxic induction (results not shown). Because an equivalent level of relative hypoxiadriven promoter activation is observed with p(-172)/LUC, it is apparent that additional, more proximally located cis-acting elements are also involved in this response.
The proximal promoter region of the human leptin gene contains a number of putative binding motifs for transcription factors, some of which have been demonstrated to play an important functional role in the regulation of basal or tissuespecific (adipose) expression of the gene. As shown in Fig. 7, the first 172-bp of upstream sequence includes the HRE at -116 and two neighboring GC boxes (at -99 and -94), flanked by two juxtaposed putative motifs for binding of the activator protein-2 factor. Further downstream, a palindromic CCAAT/enhancer-binding protein (C/EBP) response element is located at -47 followed by a TATA-like box and an additional GC box immediately downstream, preceding the start site of transcription (Fig. 7). A number of studies have established that C/EBP␣, which is an indispensable transcriptional activator of adipocyte genes during preadipocyte differentiation, also functions as a transactivator of the leptin gene promoter by binding to the proximal C/EBP response element located at -47 (44 -47). In addition, analysis of the leptin promoter using clustered point mutants introduced at key sites confirmed not only an obligatory requirement for an intact C/EBP response element, but it also revealed an important role for the GC box centered at -99 and the TATA box element in maintaining a high level of transcriptional activity in transfected rat adipocytes (47). It is of interest to note that a C/EBP consensus site in the promoter of the interleukin-6 gene serves as the binding site for C/EBP␤ and acts as a hypoxia-sensitive enhancer in endothelial cells or lung vasculature through a mechanism that does not involve HIF1 (48). These findings have been verified in the context of mice bearing transgenes in which an intact or mutant C/EBP site (also referred to as NF-IL-6 site) was linked to a lacZ reporter gene and expression assessed after exposure to hypoxia (49). Likewise, Sp1 and Sp3 transcription factors binding in a mutually competitive fashion to a promoter GC-rich element downstream from a GATAA site have been implicated in the hypoxic induction of pyruvate kinase M and ␤-enolase The hypoxia-sensitive HRE at -116 is shown together with the two G residues that were mutated (q). The relative orientations of the HRE and the activator protein-2 sites are denoted by the arrows above their respective boxes. The sequence shown here was from Ref. 23, GenBank accession no. U43653. genes in muscle (50). Thus, although additional mechanisms for transcriptional activation of hypoxia-sensitive genes exist in addition to the HIF1-mediated pathway, our results indicate that site-specific mutations into the -116/HRE are sufficient to prevent binding of HIF1␣/␤ and to block the hypoxic or CoCl 2 / DFO-induced activation of the leptin promoter completely, thereby implicating HIF1 as a major effector in this response.
A growing list of sites of leptin production (and its receptor) by cells other than adipocytes is consistent with an increasingly recognized role for leptin as a pleitotropic molecule vastly involved not only in energy homeostasis but also in reproductive function, hematopoiesis, immune regulation, angiogenesis, and wound healing (21,22). We (20) and others (38) have demonstrated previously that leptin acts as a potent angiogenic factor on endothelial cells in vitro and in vivo, possibly serving distinct purposes in the various tissues in which it is produced. For example, in adipose tissue, leptin might behave as a locally active paracrine angiogenic factor that maintains an appropriate balance between blood supply and fat depot size (20). Leptin produced in placental trophoblasts may act in concert with other placental angiogenic factors to regulate placental vascularization and transplacental exchange (51). During wound healing, leptin production may be required for supporting angiogenesis, 2 which is necessary for granulation tissue formation (52), or perhaps for additional functions including keratinogenesis (53,54) or local production of angiogenic cytokines such as vascular endothelial growth factor, basic fibroblast growth factor, and transforming growth factor-␤ (9, 55, 56). One commonality among all of these different scenarios in which leptin is produced may be the establishment of a hypoxic environment that necessitates the appropriate underlying machinery to provide an attendant physiological adaptation response. The findings reported herein are consistent with the angiogenic character of leptin whose modulation by hypoxia is thus common to other angiogenic factors. Although mechanisms other than HIF1-mediated promoter activation exist to up-regulate expression of hypoxia-sensitive genes, binding of the HIF1␣/␤ heterodimer to the leptin promoter constitutes at least one important component. Further studies will be necessary to evaluate the possibility that hypoxia may also modulate the stability of leptin mRNA as well as the role of HREs and additional promoter elements in hypoxia-induced leptin gene transcription in other cells.