Hypoxia-inducible Factor 1 Transactivates the Human Leptin Gene Promoter*

Increased placental leptin has been demonstrated in preeclampsia, a pregnancy disorder associated with placental hypoxia. This suggests that leptin gene expression is enhanced in response to oxygen deficiency in this organ. In support of this hypothesis, we have previously shown that hypoxia activates the leptin promoter in trophoblast-derived BeWo cells. Hypoxia-inducible factor 1 (HIF-1) is a heterodimeric HIF-1 (cid:1) /HIF-1 (cid:2) complex that regulates the transcription of hypoxia-responsive genes. To test whether this factor is involved in hypoxia-induced leptin promoter activation, BeWo cells were transiently transfected with a HIF-1 (cid:1) expression vector. Exogenous HIF-1 (cid:1) markedly increased luciferase reporter activity driven by the leptin promoter when HIF-1 (cid:2) was co-expressed in the same cells. This effect was similar to that elicited by CoCl 2 , an agent known to stabilize endogenous HIF-1 (cid:1) . These data suggest that HIF-1 (cid:1)

Leptin, originally identified as a satiety factor secreted by adipose tissue, is also produced by the placenta in humans (1). Placental leptin mRNA (2) and protein (3) are markedly increased in preeclampsia, a disorder associated with maternal hypertension, reduction in placental blood flow, and placental hypoxia (4). These observations have led to the proposal that the leptin gene could be induced by hypoxia. In support of this hypothesis, we have previously shown that gene expression and leptin release were increased in trophoblast-derived BeWo cells in response to various conditions of natural or chemical hypoxia. Moreover, the human leptin promoter was activated by hypoxia in these cells (5).
Hypoxia-inducible factor 1 (HIF-1) 1 is a transcription factor of major importance in the cellular response to oxygen deficiency. HIF-1 comprises HIF-1␣ and HIF-1␤ subunits, which both belong to the basic-loop-helix-PAS protein family (for review, see Ref. 6). The HIF-1␤ subunit is constitutively expressed. By contrast, HIF-1␣ is maintained at a low level in normoxic cells through proteasomal degradation of the protein.
The von Hippel-Lindau tumor supressor protein is a component of the complex that targets HIF-1␣ for polyubiquitination and degradation (7). Two recent observations indicate that von Hippel-Lindau protein binds to HIF-1␣ when a proline residue at codon 564 is hydroxylated (8,9). Hydroxylation of HIF-1␣ is controlled by a Fe 2ϩ -dependent hydroxylase activity that is inhibited by decreased oxygen. This mechanism accounts for HIF-1␣ stabilization in hypoxic cells, allowing nuclear translocation and dimerization with HIF-1␤. Stabilization of HIF-1␣ is also induced by chelating or substituting Fe 2ϩ with desferrioxamine and cobalt chloride (CoCl 2 ), respectively. This provides a molecular mechanism accounting for the ability of these agents to mimic the effect of hypoxia in experimental cell systems.
The present study was designed to test whether HIF-1 is involved in hypoxia-induced activation of the human leptin gene promoter in the placental BeWo cells. To investigate this, the transcriptional activity of HIF-1 was manipulated by overexpressing the wild-type or dominant negative form of HIF-1␣. Our data provide evidence that induction of leptin promoter activity by hypoxia is mediated by HIF-1, through a HIF-1 consensus binding site (HRE) located at Ϫ116 in the proximal promoter. This study adds the leptin gene to the list of hypoxiainducible genes regulated by this transcription factor.

EXPERIMENTAL PROCEDURES
Cell Culture-The human choriocarcinoma cell line BeWo was obtained through American Type Culture Collection (Manassas, VA). The cells were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal calf serum and antibiotics in a humidified ambient atmosphere with 5% CO 2 at 37°C.
Plasmids and Constructs-The 5Ј-deleted constructs containing various lengths of the human leptin promoter sequences upstream of the luciferase reporter gene have been described previously (5). Promoter fragments are designated according to their length in bp (p(bp)luc), relative to the transcription start site described in Ref. 10. Expression * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Mutagenesis-A sequence contained within 0.146 kb of the leptin promoter sequence in the p(146)luc construct was mutated by using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Two distinct mutations were generated. The motif 5Ј-GCACGTCG-3Ј spanning Ϫ121 to Ϫ114 was replaced by 5Ј-GCTTAATG-3Ј or by 5Ј-GCAAAACG-3Ј. The generated plasmids were designated p(146)luc mut1 and p(146)luc mut2 , respectively. Each mutation was verified by direct sequencing, and two plasmid preparations isolated from distinct clones were tested in transfection.
Transient Transfection-LipofectAMINE (Invitrogen)-mediated transfection of BeWo cells was performed as described previously (5). Briefly, 1 day before transfection, cells were plated into 35-mm 6-well dishes and transfected with 500 ng/well luciferase reporter construct and 60 ng/well pRSV-␤-galactosidase expression vector to normalize for transfection efficiency. In some experiments, expression vectors encoding HIF-1␤, HIF-1␣ wild-type, or HIF-1␣ mutant cDNA were co-transfected with the reporter constructs, as indicated in the figure legends. The total amount of transfected DNA was kept constant by adding the appropriate amount of pcDNA3 empty vector. Five h after transfection, the medium was changed, and the cells were cultured for 24 h in serum-free medium with or without addition of 100 M CoCl 2 (Sigma). Transfections were performed at least in triplicate. In each experiment, individual data were calculated as the mean of triplicates and expressed as the ratio of luciferase to ␤ galactosidase activity measured in the same cell lysate, as described previously (5).
Statistical Analysis-Statistical analysis was performed using Student's t test for paired data, with significance defined as p Ͻ 0.05.

RESULTS
Effect of Increasing HIF-1 Transcriptional Activity-An initial series of experiments was conducted to test whether increasing HIF-1 transcriptional activity would activate the leptin gene promoter in BeWo cells. Because transcriptional activation by HIF-1 depends on the amounts of HIF-1␣ available for heterodimerization with the constitutively expressed partner HIF-1␤, we increased the cellular level of HIF-1␣ by the use of the pcDNA3-HA-HIF-1␣ expression vector. Previous experiments have shown that transient transfection of a fibroblast cell line with this plasmid allowed the detection of HIF-1␣ protein under normoxia (12). The effect of HIF-1␣ overexpression on the luciferase activity driven by 1.872 kb of the human leptin promoter was tested. As shown in Fig. 1, overexpression of HIF-1␣ moderately increased luciferase activity (by 2-fold) compared with cells expressing the reporter only. However, when cells were co-transfected with HIF-1␣ and HIF-1␤, re- porter gene expression increased markedly compared with cells transfected with either subunit individually. CoCl 2 treatment induced a stimulatory effect on leptin promoter activity, which was very similar to that elicited by HIF-1␣ overexpression in the absence or presence of added HIF-1␤. Moreover, when cells overexpressing HIF-1␣ were treated with CoCl 2 , luciferase activity did not double, indicating that the effects of CoCl 2 and HIF-1␣ were not fully additive. These data support the implication of HIF-1␣/HIF-1␤ dimers in the transactivation of the leptin gene promoter. Because overexpression of HIF-1␤ markedly enhanced luciferase activity in the presence of HIF-1␣ and/or CoCl 2 , it is likely that low endogenous HIF-1␤ levels restrict HIF-1 transcriptional activity in these experimental conditions. Thus, in the next series of experiments, the cells were systematically transfected with the HIF-1␤ expression vector.
Deletion Analysis of the 5Ј-Flanking Region of the Leptin Gene-To determine the promoter region mediating activation by exogenous HIF-1␣ or CoCl 2 , BeWo cells were transfected with reporter constructs containing various lengths of the leptin gene promoter region. For each construct, the fold increase in luciferase activity elicited by either HIF-1␣ overexpression, CoCl 2 treatment, or a combination of both was determined over non-stimulated cells. As mentioned above, HIF-1␤ was rou-tinely transfected, whatever the experimental condition. A similar pattern of reporter gene expression was observed for several constructs containing up to 0.146 kb of the leptin gene 5Ј-flanking region (Fig. 2). Both HIF-1␣ and CoCl 2 individually stimulated luciferase activity by 5-7-fold. The effect of the two stimuli in combination was always greater than that elicited by HIF-1␣ or CoCl 2 alone. However, when combined, these effects were never fully additive. In contrast to all other deleted constructs, the p(116)luc reporter vector was unresponsive to HIF-1␣ and/or CoCl 2 . This analysis revealed that the first 146 bp of the leptin promoter harbor a sequence responsive to CoCl 2 and exogenous HIF-1␣, which is missing or disrupted in the p(116)luc construct.
Effect of a Dominant Negative Form of HIF-1␣-To directly evaluate the involvement of HIF-1 in mediating the effect of CoCl 2 on this region of the leptin promoter, BeWo cells were co-transfected with the p(146)luc construct and a dominant negative form of HIF-1␣ (HIF-1␣ DN). This mutant HIF-1␣ heterodimerizes with HIF-1␤ but lacks a DNA binding domain, thereby producing transcriptionally inactive dimers (11). As shown in Fig. 3, HIF-1␣ DN fully inhibited the stimulatory effect of exogenous HIF-1␣ on leptin promoter activity, demonstrating the efficiency of this dominant negative form of HIF-1␣. HIF-1␣ DN also totally abolished the stimulatory effect of CoCl 2 on luciferase activity in cells transfected with the p(146)luc construct. By contrast, the reporter activity arising from the p(116)luc construct was not significantly affected. These data provide direct evidence that endogenous HIF-1␣ is required for CoCl 2 -induced activation of the leptin promoter through a region extending 146 bp upstream of the transcription start site.
Mutational Analysis of a Putative HRE within 0.146 kb of the Leptin Promoter-Sequence analysis revealed the presence of a 5Ј-RCGTG-3Ј HIF-1 binding consensus sequence (HRE) within the 0.146-kb promoter fragment. This putative HRE, located between Ϫ120 and Ϫ116 on the noncoding strand, is disrupted in the p(116)luc construct. To assess its functional importance, the sequence was mutated in the p(146)luc reporter vector. Constructs containing two distinct mutated fragments, p(146)luc mut1 and p(146)luc mut2 , were transfected in BeWo cells, and their capacity to respond to CoCl 2 treatment and HIF-1␣ overexpression was tested. As shown in Fig. 4, the luciferase activity produced by both mutated leptin promoter fragments was not increased by these stimuli. This supports the hypothesis that this HRE consensus sequence is required for hypoxia-mediated induction of leptin promoter activity.
In Vitro Binding of HIF-1␣ and HIF-1␤ to the Ϫ116 Leptin Promoter HRE-To test whether the HRE identified within the proximal region of the human leptin promoter binds the HIF-1 complex composed of HIF-1␣ and HIF-1␤, the two subunits were synthesized in vitro by using the reticulocyte lysate system. Electrophoretic mobility shift assays were performed with the proteins obtained in unprogrammed, HIF-1␣-primed, or HIF-1␤-primed reticulocyte lysates. As shown in Fig. 5, a specific complex with retarded migration appeared exclusively when HIF-1␣-and HIF-1␤-primed lysates were incubated together with the labeled probe containing the intact HRE. No complex was visualized when mutated oligonucleotides were used as the probe (Fig. 5A). In addition, the specific binding of HIF-1␣/HIF-1␤ to the wild-type probe was eliminated by competition with an excess of homologous unlabeled probe, but not with each of the two mutated oligonucleotides (Fig. 5B). These data demonstrate that HIF-1 binds to the consensus HRE present within the proximal region of the leptin gene promoter. DISCUSSION We have shown previously that leptin gene expression is increased by hypoxia in a trophoblast-derived cell line (5). The present study provides functional evidence that HIF-1 mediates this effect via a HIF-1-responsive element located at Ϫ116 in the human leptin promoter. This conclusion is based on results obtained in experiments where the cellular level of HIF-1␣ was altered to induce or inhibit HIF-1 transcriptional activity. Consistent with the implication of HIF-1, exogenous overexpression of HIF-1␣ in the BeWo cells markedly activated the leptin promoter. Moreover, a similar amount of stimulation was produced by CoCl 2 treatment, giving support to the idea that stabilization of endogenous HIF-1␣ by this agent mediates leptin promoter activation. The most compelling evidence for the implication of HIF-1 came from the use of a dominant negative form of HIF-1␣, which totally abolished the effect of CoCl 2 . This demonstrates unequivocally that CoCl 2 -induced leptin promoter activity is driven by increased endogenous HIF-1␣ leading to the activation of HIF-1.
Sequence analysis reveals the presence of several putative FIG. 5. In vitro binding of HIF-1␣ and HIF-1␤ on the HRE located at ؊116 in the leptin promoter. A, radiolabeled oligonucleotides corresponding to wild-type (wt) or mutated (mut1 and mut2) HRE located at Ϫ116 in the leptin promoter were incubated with 2 l of in vitro-translated HIF-1␣, 2 l of in vitro-translated HIF-1␤, or both as indicated. Unprogrammed reticulocyte lysate was used as control and also to keep the total volume of lysate at 4 l. B, competition assays were carried out by incubating the radiolabeled wt probe with in vitro-translated HIF-1␣ and HIF-1␤ and cold wt or mutated probes in 5-, 10-or 50-fold molar excess, as indicated. These autoradiograms are representative of four independent electrophoretic mobility shift assays.
HREs within the first 1.872 kb of the human leptin promoter. We have previously observed that two regions containing 1.87 and 1.20 kb of the promoter respectively conferred high and relatively lower responsiveness to hypoxia (5). These data suggested to us that a distal HRE located at Ϫ1.83 kb in the promoter could mediate the effect of hypoxia. However, this hypothesis was not confirmed by subsequent experiments performed in BeWo cells overexpressing HIF-1␤. Indeed, we show here that both CoCl 2 -and HIF-1␣-induced activation of the leptin promoter are of a similar magnitude for each 5Ј-deleted fragment extending from 1.872 to 0.146 kb. Therefore, it is possible that distinct hypoxia responsiveness was coincidentally associated with promoter length, although it cannot be excluded that low levels of endogenous HIF-1␤ have been instrumental in this effect. In the present study, deletion analysis and site-specific mutagenesis clearly implicate the most proximal HRE of the leptin promoter in HIF-1 responsiveness.
These observations add the human leptin gene to a list of genes activated by hypoxia via the HIF-1 pathway. After its initial discovery as a satiety factor, leptin has been subsequently implicated in a variety of functions, some of which are altered in response to decreased oxygen availability. For example, leptin has been shown to exert a potent proangiogenic effect in experimental systems in vitro and in vivo (13,14). During preeclampsia, in which placental hypoxia is a prominent feature, enhancing leptin production could be part of a compensatory response aimed at developing new vessels. Consistent with the idea that adipose leptin is also induced by hypoxia, we have recently observed that leptin gene expression is increased in human PAZ6 adipose cells in response to cellular hypoxia (15) and in the adipose tissue of rats submitted to hypobaric hypoxia. 2 If leptin also exerts a proangiogenic effect in this tissue, it can be anticipated that a local effect of leptin would be to stimulate vascularization during normal or pathological adipose tissue growth. Interestingly, the angiogenic capacity of adipose tissue has been used clinically to promote wound healing and revascularization of ischemic tissues (16,17). This effect could be mediated, at least in part, by leptin, in concert with other angiogenic factors such as vascular endothelial growth factor (18). Besides the placenta, leptin is produced in several non-adipose tissues, including the stomach (19). Consistent with a stimulatory effect of a local hypoxic environment produced at wound sites, leptin gene expression is increased in gastric ulcers (20,21). This suggests that leptin might participate in the mechanisms leading to ulcer healing in the stomach because the hormone has been shown to promote skin wound re-epithelialization (22,23). These observations favor the idea that up-regulation of leptin production by hypoxia is physiologically relevant not only in the placenta but also in other leptin-producing tissues, including adipose tissue.
Hypoxia is not the only condition that stabilizes HIF-1␣ and activates HIF-1 transcriptional activity. Several hormones and growth factors, including insulin and insulin-like growth factor I (24), angiotensin II, thrombin and platelet-derived growth factor (11), and, more recently, endothelin-1 (25), have been shown to increase the level of HIF-1␣ in various cell types. In addition, inflammatory cytokines, such as interleukin 1␤ and tumor necrosis factor ␣, also induce HIF-1 activity in normoxic cells (26 -28). It would be of interest to know whether these factors could regulate leptin gene expression in some cell types by activating hypoxia-independent HIF-1 pathways. Interest-ingly, this could be the case in the hypoxic placenta during preeclampsia, in which increased inflammatory cytokine production has been described (29,30).
In conclusion, the data presented here are consistent with the leptin gene being a genuine hypoxia-inducible gene. Moreover, they show that hypoxia mediates increased leptin gene expression via HIF-1␣ and HIF-1-dependent transcriptional activity, as described for several other genes regulated by low oxygen availability.