Differential Transcriptional Regulation of the Two Vascular Endothelial Growth Factor Receptor Genes

Vascular endothelial growth factor (VEGF) and its two endothelial cell-specific receptor tyrosine kinases, Flk-1/KDR and Flt-1, play a key role in physiological and pathological angiogenesis. Hypoxia has been shown to be a major mechanism for up-regulation of VEGF and its receptors in vivo . When we exposed human umbilical vein endothelial cells to hypoxic conditions in vitro , we observed increased levels of Flt-1 expression. In contrast, Flk-1/KDR mRNA levels were unchanged or slightly repressed. These findings suggest a differential transcriptional regulation of the two receptors by hy- poxia. To identify regulatory elements involved in the hypoxic response, promoter regions of the mouse Flt-1 and Flk-1/KDR genes were isolated and tested in conjunction with luciferase reporter gene. In transient transfection assays, hypoxia led to strong transcriptional activation of the Flt-1 promoter, whereas Flk-1/ KDR transcription was essentially unchanged. Promoter deletion analysis demonstrated a 430-bp region of the Flt-1 promoter to be required for transcriptional activation in response to hypoxia. This region includes a heptamer sequence matching the hypoxia-inducible fac-tor-1 (HIF) consensus binding site previously found in other hypoxia-inducible genes such as the VEGF gene and erythropoietin gene. We further narrowed down the element mediating the hypoxia response to a 40-base pair sequence including the putative HIF binding site. We show that this element acts like an enhancer, since it activated transcription irrespective of its location or orientation in

The growth of new blood vessels (angiogenesis) is essential for embryonic development and other physiologic processes such as bone remodeling, wound healing, and ovarian cycle (1,2). Angiogenesis is also a critical component of tumors, inflammatory arthritis, intraocular neovascular syndromes, and other disorders (3,4). The search for potential regulators of angiogenesis led to a number of candidates (aFGF, basic fibroblast growth factor, transforming growth factor-␣, transform-ing growth factor-␤, etc.) (5). Among those, VEGF 1 and its two receptors, Flt-1 and Flk-1/KDR have been shown to be crucially involved in physiological and pathological regulation of blood vessel growth (6). Recently, it has been shown that oxygen tension plays a major role in the regulation of VEGF gene expression (7)(8)(9). VEGF mRNA expression is rapidly and reversibly induced by exposure to low oxygen conditions in a variety of normal and transformed cells. A 47-bp regulatory element located about 1 kb upstream to the VEGF transcription initiation site was found to be involved in the activation of VEGF transcription in hypoxic cells. This element includes a binding site for the transcription factor hypoxia-inducible factor-1 (HIF-1) (10). When reporter constructs containing the VEGF sequences that mediate hypoxia inducibility were cotransfected with expression vectors encoding HIF-1 subunits, reporter gene transcription was much greater than that observed in cells transfected with the reporter alone, both in hypoxic and normoxic conditions (11). HIF-1 has been shown to be involved also in the regulation of the human and mouse erythropoietin (EPO) genes (12,13) as well as other hypoxia inducible genes such as the glycolytic enzymes (14,15).
Hypoxia has been proposed to play an important role also in the regulation of VEGF receptor gene expression. Exposure of rats to acute or chronic hypoxia led to pronounced up-regulation of both Flt-1 and Flk-1/KDR genes in the lung vasculature (16). Also, Flk-1/KDR and Flt-1 mRNAs were substantially up-regulated throughout the heart following myocardial infarction in the rat (17). Furthermore, Flt-1 and Flk-1/KDR mRNAs are markedly up-regulated in ischemic regions of tumors such as glioblastoma multiforme (8,18,19). However, in vitro studies have yielded conflicting findings. Although Thieme et al. (20) have shown that hypoxia increases VEGF receptor number by 50% in cultured bovine retinal capillary endothelial cells, the levels of Flk-1/KDR mRNA appeared to be down-regulated. Also, while an up-regulation of Flt-1 mRNA in response to hypoxia was found in cultured pericytes (21) or in microvessels in skin explants (22), others failed to detect a similar upregulation of Flt-1 in human umbilical vein endothelial (HUVE) cells (23). Furthermore, Brogi et al. (24) reported that the Flk-1/KDR mRNA is not directly induced by hypoxia in HUVE cells or in microvascular endothelial cells. It has been suggested that the in vivo up-regulation of Flk-1/KDR receptor expression is mediated by a paracrine factor released by ischemic tissues (24) or by post-transcriptional mechanisms such as increased mRNA stability (23).
Using real time RT-PCR technology, we found an up-regulation of the Flt-1 expression compared with the Flk-1/KDR in HUVE cells. Sequence analysis of the mouse and human Flt-1 promoter revealed a heptamer element highly homologous to the HIF consensus sites present in the 5Ј-region of mouse and human VEGF genes and the 3Ј-enhancer of the human EPO gene. So far, such a homologous element could be found neither in the human nor in the mouse Flk-1 promoter sequences available.
In the present study, we provide evidence that a 40-bp sequence including such an element is sufficient to confer hypoxia inducibility when tested in a heterologous promoter context and revealed enhancer-like features. These differences in the regulation further emphasize the different roles played by these receptors in mediating VEGF biological responses.

EXPERIMENTAL PROCEDURES
Cloning of 5Ј-Flanking Region of the Mouse Flt-1 and Flk-1 Genes-A 129Sv/ev male genomic library in GEM-11 vector (Promega, Madison, WI), generated by partial Sau3A digestion, was screened with a set of overlapping oligonucleotides covering the 5Ј-end of the leader sequences of the mouse Flt-1 and Flk-1 gene, respectively (25). Oligonucleotide probes used to isolate the mouse Flt-1 gene were mFlt-1 probe 1 and probe 2. For the mouse Flk-1 gene, oligonucleotide mFlk-1 probe 1 and probe 2 were radioactively labeled after annealing in a Klenow fill-in reaction (DNA Polymerase I, Large (Klenow) fragment (25). For sequences of oligonucleotides, see Table I. All enzymes used were from New England Biolabs (Beverly, MA), unless otherwise indicated. 16 pmol of each oligonucleotide was present in 20 l of 10 mM Tris-HCl, 10 mM MgCl 2 , 50 mM NaCl, 1 mM dithiothreitol, heated for 10 min at 55°C, and allowed to cool for 10 min at room temperature. The mixture was made 100 M for dGTP, dTTP, and dATP. 2 units of Klenow fragment were then added, and the total volume was adjusted to 50 l. The labeling reaction occurred at 37°C for 30 min in the presence of 50 Ci of [ 32 P]dCTP (3000 mCi/mmol, Amersham Corp. Three overlapping but not identical clones of each gene were isolated. The restriction maps of these genes were generated using the partial restriction method (25) (data not shown). Fragments were cloned into pBluescript KS (Stratagene, La Jolla, CA). A 3.5-kb NcoI fragment of the Flt-1 gene and a 2.1-kb PstI fragment from the Flk-1 phages were subcloned into pGEM5 or pSK to generate pGEM5Flt-1(NcoI) and pSKFlk-1(PstI), respectively. Sequencing reactions were performed in an automatic sequencer (model 373A, Applied BioSystems, Foster City, CA). Both strands of the Flt-1 promoter including the region from Ϫ3181 to ϩ352 and the region between Ϫ1824 and ϩ148 of the Flk-1 promoter were sequenced by cycle sequencing. Sequence analysis was done using the Sequencher 3.1 program (Gene Codes Corporation, Ann Arbor, MI).
Cell Cultures-Primary cultures of HUVE cells were obtained from Clonetics (San Diego, CA). Cells were maintained in the presence of endothelial growth medium (EGM; Clonetics). EGM consists of endothelial basic medium plus 0.1 ng/ml recombinant human EGF, 10 g/ml hydrocortisone, 500 g/ml gentamycin, 500 ng/ml amphotericin B, 12 g/ml bovine brain extract, and 2% fetal bovine serum. 24 h before exposure to hypoxia, cells were trypsinized and plated on gelatinized 10-cm culture dishes to a density of 8.8 ϫ 10 3 cells/cm 2 . Dishes were then floated with preanalyzed gas mixture for 20 min and kept in 0% O 2 , 5% CO 2 , and 95% N 2 at 37°C for various durations. Oxygen concentration in the incubators was monitored by a portable oxygen analyzer (Teldyne Brown Engineering, San Leandro, CA).
Hep3B cells (ATCC number HB-8064) were grown in minimal essential medium with Earle's salts with nonessential amino acids, without glutamine (Life Technologies), complemented with 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin.
Transient Transfection by Electroporation and Luciferase Assay with Hep3B Cells-For transient expression in Hep3B cells, 0.5 g of test DNA and 0.5 g of reference template were added to each well of six-well plates. Cells were at a density of 0.5 ϫ 10 6 /well. Plasmid DNA was prepared by using commercial kits (Qiagen, Santa Clara, CA) and introduced into cells by electroporation with a Gene Pulser (Bio-Rad, Richmond, CA) at 260 V and 960 microfarads following the manufacturer's instructions. Quadruplicate electroporations were pooled and split into six-well plates with 5 ml of medium in each well. Cells were then allowed to recover for 24 h in a 5% CO 2 , 95% air incubator at 37°C. Following medium change, triplicate wells were placed in a hypoxic (0% O 2 , 5% CO 2 ) or in a normoxic (5% CO 2 , air) incubator for the indicated duration of time. Cell extracts were generated by incubation in 500 l of 1 ϫ passive lysis buffer solution (dual luciferase assay; Promega) at room temperature for 15 min and frozen at Ϫ70°C. 10 l of the extracts were analyzed in a luminometer (TD-20e, Turner Designs, Inc., Mountain View, CA) using the reagents provided in the kit. Light production was measured for 15 s, and results were expressed as relative light units (RLU). The mean RLU was corrected with the signals obtained from the reference constructs. The relative luciferase activity (mean Ϯ S.E.) was calculated as luciferase (RLU)/Renilla luciferase (RLU).
Transient Transfection by Lipofection and Luciferase Assay with HeLa Cells-1 g of test DNA and 0.1 g/well CMV-RLluciferase control vector (dual luciferase assay; Promega) were used for transient transfection of 0.2 ϫ 10 6 HeLa cells/well in six-well plates. Immediately before the addition of the DNA-liposome complex, cells were washed twice with phosphate-buffered saline. The plasmid DNA mix was then added to a 15-ml Falcon tube containing 0.1 ml of Opti-MEM1 and mixed. 10 l of Lipofectin (Life Technologies, Inc.) were added to another 15-ml tube containing 0.1 ml of Optim-MEM1 and mixed. Plasmid DNA and Lipofectin were mixed and incubated at room temperature to allow the complex to form. After 15 min, the DNA-liposomal complex was diluted with 0.8 ml of Opti-MEM1 (Life Technologies, Inc.), mixed, and layered gently on top of the cells. Cells were then incubated at 37°C for 14 -16 h. After removal of the DNA-liposome mixture, 3 ml of complemented medium was added 24 h before exposure to hypoxia. For final luciferase assay, cells were lysed in 200-l passive lysis buffer (Promega). Cell lysates were then transferred into a 1.5-ml microcentrifuge tube and cleared by centrifugation at 10,000 rpm for 1 min at 4°C. Luciferase and Renilla luciferase activity were measured by mixing 20 l of extract with 100 l of luciferase assay buffer and the subsequent addition of 100 l of Stop and Glow solution, respectively.
The constructs mFlt-HIE50, HIE40, HIE30, HIE50 -234A, hEPO HIE, and hVEGF HIE were made by ligation of the corresponding oligonucleotide set into the SalI site 2.8 kb upstream of the SV40 promoter ("enhancer" position) or into the XhoI site ("promoter" position) of the pGL2SV40prom vector (Promega). For oligonucleotide sequences, see Table I. Phosphorylation of 50 pmol of the corresponding oligonucleotide set was carried out in 70 mM Tris-HCl (pH 7.6), 10 mM MgCl 2 , 5 mM dithiothreitol, 5 mM rATP (Pharmacia Biotech Inc.) and 10 units of T4 polynucleotide kinase (New England Biolabs, Beverly, MA) for 1 h at 37°C in a total volume of 20 l. The reaction mixture was heated for 10 min at 65°C and slowly cooled to room temperature. 0.1 l of this mix was used in ligation reactions.
Constructs mFlt-1HIE264 and mFlt-1HIE100 were generated by PCR (25) of pGEM5Flt-1(NcoI) using the oligonucleotides mFlt-1HIE264.F and mFlt-1HIE264.R and oligonucleotides mFlt-1HIE100.F and mFlt-1HIE100.R, respectively. The PCR products were subcloned into pSK to generate pSKmFlt-1HIE264 and pSKmFlt-1HIE100, respectively. The inserts were cut by BamHI and SalI and cloned in "enhancer" position in pGL2prom, as above described. To clone these inserts in "promoter" position in pGL2SV40prom, they were cut with XhoI and SacI and then cloned into the XhoI site of the reporter plasmid. All constructs were analyzed by restriction digestion analysis and partial sequencing.
Real Time RT-PCR Analysis-HUVE cells from pooled donors were cultured as described under "Cell Cultures." Cells were initially expanded for 8 -10 days in the presence of EGM. 38 h prior to exposure to hypoxia, cells were split and seeded at a density of 100,000 cells/well in six-well plates in EGM. Immediately prior to hypoxic incubation (0% O 2 , 5% CO 2 , 95% N 2 ), cells were washed once, and then 5 ml of assay medium (endothelial basic medium plus 2% fetal bovine serum, 10 g/ml hydrocortisone, 500 g/ml gentamycin, 0.5 g/ml amphotericin-B) was added to each well. After incubations of various duration, cells were harvested by the STAT 60 method (Tel-Test, Inc., Friends-wood, TX), and total RNA was prepared according to the manufacturer's recommendations. The RNA was dissolved in 50 l of H 2 O, and its concentration was determined by spectrophotometer (A 260 ).
To monitor gene expression, we used real time RT-PCR analysis. This novel approach has been described previously (26,27). Briefly, a gene-specific PCR oligonucleotide primer pair defines the "amplicon." Within the amplicon, an oligonucleotide probe labeled with a reporter fluorescent dye (FAM) at the 5Ј-end and a quencher fluorescent dye (TAMRA) at the 3Ј-end are designed. When the probe is intact, the reporter dye emission is quenched. During the extension phase of the PCR cycle, the nucleolytic activity of the DNA polymerase cleaves the hybridization probe and releases the reporter dye from the probe. Fluorescence intensity produced during PCR amplifications is monitored by the sequence detector directly in the reaction tube ("real time"). A computer algorithm compares the amount of reporter dye emission with the quenching dye emission and calculates the threshold cycle number (C T ), when signals reach 10 times the standard deviation of the base line. It was demonstrated that the calculated C T values are a quantitative measurement for the mRNA levels of various genes tested (27). 100 ng of total RNA was added to a 50-l RT-PCR reaction (PCR-Access, Promega). The reaction master mix was prepared according to the manufacturer's protocol to give final concentrations of 1 ϫ avian myeloblastosis virus/Tfl reaction buffer, 0.2 mM dNTPs, 1.5 mM MgSO 4 , 0.1 unit/ml avian myeloblastosis virus reverse transcriptase, 0.1 unit/l Tfl DNA polymerase, 250 nM concentration of the primers, and 200 nM concentration of the corresponding probe, as described by Gibson et al. (27). Primers and probes for real time PCR analysis of Flt-1 and Flk-1/KDR genes were designed by the Oligo version 4.0 program (National Bioscience, Plymouth, MN), according to Heid et al. (26). For sequences of all oligonucleotides used, see Table I. The primers for the human Flk-1/KDR gene were HUMKDR 2530.F and HUMKDR 3043.R, and the probe was HUMKDR 2872.FP. For Flt-1 analysis, the following primers were used: HSFLT 2689.R and HSFLT 2228.F; the probe was HSFLT 2549.FP. Primers and probes were synthesized at Genentech using conventional nucleic acid synthesis chemistry. The ␤-actin primer and probe (TaqMan ␤-actin detection reagents) were purchased from Perkin-Elmer and Applied Biosystems.
RT-PCR reactions and the resulting relative increase in reporter fluorescent dye emission were monitored in real time by the 7700 sequence detector (Perkin-Elmer). Signals were analyzed by the sequence detector 1.0 program (Perkin-Elmer). Conditions were as follows: 1 cycle at 48°C for 45 min, 1 cycle at 94°C for 2 min, 40 cycles at 94°C for 30 s, 60°C for 1 min, 68°C for 2 min. Data were generated as indicated in the legend to Fig. 1.

RESULTS
To determine whether the VEGF receptors are directly regulated by hypoxia in endothelial cells, primary HUVE cells were incubated in hypoxic conditions (0% O 2 , 5% CO 2 ) and analyzed for VEGF receptor gene expression by real time quantitative RT-PCR analysis (27). In two independent experiments conducted with different HUVE cell preparations derived from different donors, Flt-1 levels were induced 4.2 Ϯ 0.8-fold after 32 h of growth in hypoxia. Expression levels for Flk-1/KDR were unchanged or weakly down-regulated at the same time point. Time course experiments revealed a 2-3-fold stimulation of Flt-1 after 60 h in hypoxia (Fig. 1A), whereas the Flk-1/KDR levels were moderately down-regulated during the same period (Fig. 1B). Initial Northern blot experiments yielded essentially similar results (data not shown).
To test if the differential regulation of the VEGF receptor genes is due to transcriptional regulatory regions, promoter regions located upstream of both genes were isolated and tested for their ability to respond to hypoxia in fusion constructs with the luciferase gene.
To isolate the 5Ј-flanking region of the murine Flk-1 and Flt-1 genes, a genomic DNA library from 129Sv/ev mice was screened using probes corresponding to the signal peptide sequence of their respective gene products. For each receptor gene, three independent phage clones were isolated and analyzed by restriction mapping. To exclude the possibility of recombination artifacts, genomic DNA from 129/SvJ mice (Stratagene) was analyzed by Southern blot hybridization and showed identical restriction fragments (data not shown). A 3.4-kb KpnI/NcoI Flt-1 and a 2.1-kb PstI Flk-1/KDR were subcloned into pSK vector (Stratagene) and used for further analysis.
The nucleotide sequence of the KDR/Flk-1 promoter region (Ϫ1829/ϩ148) and the comparison with the human sequence (Ϫ780/ϩ148) (28) are shown in Fig. 2A. Mouse Flt-1 promoter sequences (Ϫ3181/ϩ276) and the comparison with the available human gene sequence (Ϫ1195/ϩ276) (29,30), are shown in Fig. 2B. The sequence comparison between the human and mouse Flt-1 genes revealed 78% similarity in a 1.5-kb promoter region ( Fig. 2A). Flk-1/KDR exhibited a 60% similarity in a 0.9-kb promoter region. Interestingly, comparison of the coding regions between the human and mouse genes reveals a 75% similarity for Flt-1 (31, 32) and 85% for Flk-1/KDR (33, 34). Thus, the 5Ј-regulatory regions of the Flt-1 gene appear to have  Differential Transcriptional Regulation of Flt-1 and KDR an even higher degree of evolutionary conservation than those of KDR.
By further sequence analysis, we identified a putative HIF consensus binding site located at position Ϫ959 of the mouse Flt-1 promoter: ACGTGGAAT. This site matches 7 of 9 nucleotides of the VEGF hypoxia enhancer and 6 of 9 residues of the EPO hypoxia enhancer (Table II). The putative HIF binding site lies in a region that is highly conserved between human and mouse (100% over 33 bp and 91% over 115 bp). Fragments derived from promoters of both genes were cloned in conjunction with the luciferase gene and used in transient transfection experiments. Due to the difficulty of transfecting primary endothelial cell cultures in an efficient and reproducible fashion, we decided to use HeLa or Hep3B cells. These cell lines have been successfully employed in the analysis of other hypoxiainducible genes such as the EPO and VEGF genes (35,36). Transient transfection experiments with luciferase reporter genes under the control of promoter elements of various hypoxia-inducible genes were performed in both cell lines to rule out cell type-specific artifacts of the promoters under hypoxia. Representative results of both cell lines are shown. Flt-1 and Flk-1 promoter fragments were able to drive transcription of the luciferase reporter constructs to about equal levels when tested in Hep3B cells grown under normoxic conditions. However, incubation of transfected Hep3B cells in hypoxia for 34 h led to a 3-8-fold increase in Flt-1-driven luciferase activity, whereas the Flk-1 promoter was not significantly affected (Fig.  3B). These observations indicate that hypoxia-responsive elements are located within the 3.4-kb 5Ј-promoter fragment of the Flt-1 gene. Similar results were found when the constructs were tested in HeLa cells (data not shown). To further define regions of the Flt-1 promoter mediating the hypoxic up-regulation, a series of 5Ј-deletion constructs were generated (Fig. 4A) and transiently transfected into HeLa (Fig. 4B) and Hep3B cells (data not show). Pairwise comparison between cells exposed to hypoxia versus cells grown in normoxic conditions revealed a significant drop between construct Flt-(Ϫ977/ϩ209)luc and Flt-(Ϫ547/ϩ209)-luc. These experiments indicated that a 430-bp fragment located between positions Ϫ977 and Ϫ547 in the Flt-1 promoter, including the putative HIF consensus binding site, contains regulatory elements that mediate, at least partially, transcriptional stimulation by hypoxia. Deletion of this element virtually abolished the transcriptional stimulation by hypoxia (Fig. 4B).
To test whether the regulatory sequences located within the 430-bp element are sufficient to mediate the hypoxic response in a heterologous promoter context, we inserted fragments of different sizes including the putative HIF binding site 2.8 kb upstream of the SV40 promoter driving the luciferase reporter gene (pGL2prom, Promega). Transient transfection experiments in Hep3B cells and subsequent exposure to hypoxia for 36 h revealed that fragments of 500 (data not shown), 264, 100, 50, and 40 bp were sufficient to mediate transcriptional activation by hypoxia, whereas a 30-bp fragment failed to do so (Fig. 5, A and B). Similar results were obtained in HeLa cells or when the fragments were introduced in a position next to the SV40 promoter (data not shown). It was previously shown (35) that mutations of residues 2-4 of the human EPO HIF consensus (ACGTGCT to AAAAGCT) lead to impaired hypoxia response when tested in a 50-bp sequence introduced in a SV40 promoter-CAT reporter gene construct. In a similar approach, we mutated the Flt-1 consensus ACGTGGA to AAAAGGA within the 50-bp wild-type Flt-1 fragment cloned upstream of the SV40 promoter. When transiently transfected in Hep3B (Fig. 5) or HeLa cells (data not shown), a significant decrease in terms of hypoxia inducibility between the wild-type and the mutated sequence could be observed (from 2.4-to 1.2-fold). We conclude that a 40-nucleotide sequence present within the Flt-1 promoter (Ϫ976 to Ϫ937) is sufficient to mediate transcriptional activation by hypoxia when tested in a heterologous promoter context. Most significantly, this stimulation was nearly abolished when mutations in positions 2-4 of the putative HIF binding site were introduced. These findings suggest that HIF-1, a transcription factor previously shown to mediate hypoxic stimulation of the EPO and VEGF genes, is also required for the regulation of Flt-1 expression.
To determine whether the hypoxia inducibility conferred by the 40-bp Flt-1 sequence is of a magnitude comparable with that conferred by other HIE sequences, we introduced the previously described regulatory sequences of the EPO (50 bp) (35) and VEGF (47 bp) (11) genes in remote enhancer positions on luciferase reporter gene constructs. As shown in Fig. 6, transient transfection experiments with HeLa cells revealed a 2.9-fold stimulation by the Flt-1 HIE, a 3.2-fold stimulation by the EPO-derived HIE, and a 2.3-fold stimulation by the VEGF sequences. The SV40 promoter led to only 1.4-fold activation under the same conditions. Similar levels of hypoxic induction have been previously obtained when the EPO or VEGF sequence was tested in conjunction with other reporter genes or other cell types (11,(37)(38)(39)(40). DISCUSSION Our results indicate that the genes for two VEGF receptors, Flt-1 and Flk-1/KDR, are differentially regulated by hypoxia in cultured endothelial cells. To monitor gene expression, we use real time PCR technology. This novel approach allows very

FIG. 1. Quantitative analysis of Flt-1 and Flk-1/KDR expression in HUVE cells by real time RT-PCR.
A, analysis of total RNA (100 ng) isolated from HUVE cells incubated in normoxic (5% CO 2 , 37°C) or hypoxic conditions (5% CO 2 , 0% O 2 , 95% N 2 ) for the indicated length of time. Standard curves for Flt-1 and ␤-actin were generated by serial dilutions of total RNA isolated from HUVE cells in a separate experiment (data not shown). The relative Flt-1 expression levels were calculated dividing the Flt-1 levels by the ␤-actin levels measured in the same RNA preparation. Linear regression of the ␤-actin standard curve was used to normalize the results for the Flt-1 signals to 100 ng of total RNA (27). Data shown are from duplicate analyses of two independent experiments. B, analysis of total RNA (100 ng) from HUVE cells for Flk-1/KDR expression. A standard curve for Flk-1/KDR was generated by serial dilution of total RNA isolated from HUVE cells in a separate experiment (data not shown). Data analysis was performed as outlined in A.
accurate and reproducible quantification of gene expression and, unlike other quantitative PCR methods, it does not require post-PCR sample handling, thus preventing product contamination (26,27). While the expression of Flt-1 mRNA is up-regulated in HUVE cells, Flk-1/KDR expression remained unchanged or moderately down-regulated. Although the magnitude of the hypoxic stimulation of Flt-1 expression exhibited some variability between different preparations and donors, the effect was highly reproducible and also persisted over the entire period of analysis. Consistent with these findings, we identified by deletion analysis a 430-bp fragment in the promoter of Flt-1 necessary for mediating the hypoxic response. Further analysis of sequences involved in such response led to the identification of a heptamer element highly homologous to FIG. 2. Comparison of murine and human KDR/Flk-1 and Flt-1 promoter sequences. A, a 2.0-kb mouse genomic Pst-1 fragment in the pSK Flk-1-(Ϫ1824/ϩ148) vector was sequenced on both strands, as described under "Experimental Procedures." Mouse and human (28) sequences are aligned using a program developed at Genentech. The transcription start site is marked by an asterisk. B, both strands of a 3.4-kb mouse genomic NcoI fragment in pSK Flt-1 (Ϫ3181/ϩ276) were sequenced. The alignment with the human sequence (29) was done as described in A. The Flt-1 hypoxia-inducible element is shown as a bar; nucleotides within the putative HIF consensus binding site are marked with filled circles. The TATA box sequences are marked with a bar, and the transcription start site is indicated by an asterisk. the HIF binding site in the promoter of the Flt-1 but not the Flk-1/KDR gene. Binding sites for HIF have been found in other hypoxia-inducible genes such as EPO (13), VEGF, inducible nitric-oxide synthetase (41), glucose transporter (GLUT1) (42), and the glycolytic enzymes aldolase A, enolase 1, lactate dehydrogenase A, and phosphoglycerate kinase 1 (14,15,43) (see Table II). Similarly to the EPO and VEGF HIE sequences (11,35), introduction of mutations in nucleotides 2-4 within the putative Flt-1 HIF consensus binding site eliminated its ability to activate transcription in response to hypoxia. This element, albeit necessary, is not sufficient for hypoxia response element activity. We have shown that a fragment of 40 nucleotides, which includes the HIF binding site, mediates hypoxic responses, whereas a 30-bp fragment failed to do so. On the basis of these observations, we conclude that the flanking sequences to the HIF binding site also contribute to the hypoxic response.
Previous studies have shown that hypoxic up-regulation of the VEGF gene is of major pathophysiological significance (7)(8)(9). Using different cell lines for the analysis of VEGF promoters from different species, fragments ranging in size from 47 to 28 bp were found to mediate hypoxic stimulation of various reporter genes tested (11,38,40). Furthermore, sequences that mediate increased RNA stability were identified in the 3Јuntranslated region of the VEGF mRNA (10,44,45). Moreover, the von Hippel-Lindau tumor suppressor gene has been recently shown to play an important inhibitory role in the transduction of signals generated by changes in oxygen, possibly by post-transcriptional mechanisms (46). It is possible that such additional post-transcriptional events also participate in the regulation of Flt-1 expression.
In contrast to the transcriptional up-regulation of VEGF and Flt-1, the Flk-1/KDR message in endothelial cells exposed to hypoxia appears not to be increased. This suggests the presence of additional regulatory components leading to the upregulation observed in several in vivo models. Interestingly, recent studies suggest that Flk-1/KDR up-regulation in response to hypoxia is largely indirect and requires a so far unidentified paracrine mediator released by ischemic tissues (24). Such a soluble factor, which is not found in the supernatant of endothelial cells, may explain, at least in part, the discrepancies between in vivo and in vitro findings. Furthermore, Waltenberger et al. have suggested that hypoxia results in increased stability of the Flk-1/KDR mRNA (23). The relative contribution of paracrine factors and post-transcriptional events in the hypoxic induction of Flk-1/KDR remains to be determined. Our findings on the differential transcriptional regulation of the VEGF receptors in response to changes in oxygen levels further emphasize their different roles in mediating VEGF biologic functions. Gene knockout studies have  shown that both receptors are essential for the development of the embryonic vasculature, since mouse embryos null for either receptor died in utero between day 8.5 and day 9.5. Based on these and other studies, it can be deduced that Flk-1/KDR is required for the mitogenic effects of VEGF on endothelial cells and is critically involved in the regulation of angiogenesis, both FIG. 4. Functional analysis of the mouse Flt-1 promoter. A, schematic representation of reporter gene constructs. Exons are indicated by thick lines. The -fold induction by hypoxia was calculated as described in B. B, HeLa cells were incubated under normoxic or hypoxic conditions for 24 h. Luciferase signals were normalized for transfection efficiency to the signals obtained from the co-transfected Renilla Luciferase construct driven by the SV40 promoter. Data shown are from analysis of three independent experiments. The -fold induction in luciferase activity was calculated by dividing the relative luciferase activity levels obtained from cells incubated in hypoxic conditions by the levels obtained when cells were exposed to normoxic conditions for the same length of time. Relative luciferase activities were calculated by dividing the signals from the Flt-1 promoter-luciferase constructs by the signals obtained from the Renilla Luciferase driven by the SV40 promoter. Reporter and reference reporter gene constructs were transiently co-transfected in HeLa cells as described under "Experimental Procedures." Similar results were obtained with Hep3B cells (data not shown).
in the developing and in the adult animal (47). In contrast, Flt-1 appears to be primarily involved in endothelial cell morphogenesis, at least during embryonic development (48). The role of Flt-1 in the endothelium of the adult animal is less clearly defined. However, the finding that the Flt-1 mRNA is expressed in both proliferating and quiescent endothelial cells suggests a role for this receptor in the maintenance of endothelial cells (49).
Up-regulation of the Flk-1/KDR gene may be required for endothelial cells to acquire a proliferative phenotype and respond to VEGF. That hypoxia up-regulates this receptor largely by indirect mechanisms may reflect the need for a more complex regulation than that of Flt-1, being critically depend-ent on the integration of signals generated from the ischemic tissues and, possibly, also on changes in mRNA stability.
Regulation of VEGF target genes in the vascular endothelium may represent the integration of signals generated by both receptors in homo-or heterodimeric forms. Variations in the levels of one receptor species versus the other may lead to changes in the expression pattern of VEGF target genes. It will be interesting to investigate changes of VEGF target gene expression in endothelial cells in response to hypoxia.
Interestingly, very recent studies indicate that some effects of VEGF outside the vascular endothelium are mediated by the Flt-1 receptor. For example, it has been shown that the migration of monocytes/macrophages in response to VEGF is medi- FIG. 5. A 40-bp Flt-1 promoter fragment is sufficient to mediate transcriptional activation by hypoxia. A, schematic representation of luciferase reporter gene constructs. Putative transcription factor binding sites are indicated. B, Hep3B cells were transfected by electroporation and incubated for 48 h under normoxic or hypoxic conditions, as described. The -fold induction by hypoxia, shown in A, was calculated by dividing the relative luciferase activity levels obtained from cells incubated in hypoxia by the levels obtained when cells were exposed to normoxic conditions for the same length of time. Relative luciferase activities were obtained from Hep3B cells transiently transfected with the indicated luciferase fusion constructs. Flt-1 promoter fragments were present in the "enhancer" position (SalI site) of the pGL2prom vector (see "Experimental Procedures"). Data shown are means Ϯ S.E. of three independent experiments. Relative luciferase activity was calculated by normalizing the luciferase signals to the signals obtained from the co-transfected Renilla luciferase reference gene driven by the SV40 promoter. Similar results were obtained with HeLa cells (data not shown). ated by Flt-1, not by KDR (50,51). In this context, hypoxic up-regulation of Flt-1 may play an especially important role in amplifying the chemotactic effects of VEGF for mononuclear cells during tumorigenesis or in the course of wound healing. Furthermore, the role of VEGF in tumor biology has been broadened by the unexpected finding that this factor has the ability to interfere with the maturation of host professional antigen-presenting cells such as dendritic cells and thus may impair an immune response to cancer cells (52). RT-PCR analysis of dendritic cells revealed that Flt-1 mRNA but not Flk-1/ KDR mRNA is expressed in such cells.
In conclusion, further studies are required to fully appreciate the biological significance of the differential transcriptional regulation of the VEGF receptors by hypoxia, in particular the direct up-regulation of Flt-1. However, the high degrees of evolutionary conservation in the promoter region of the Flt-1 gene indicate that the processes involved in the transcriptional regulation of this gene are highly crucial.
FIG. 6. Comparison of hypoxia inducibility conferred by HIEs derived from VEGF, Flt-1, and EPO genes. HIEs from the human VEGF and EPO and the mouse Flt-1 genes were introduced in the "enhancer" position in the pGL2prom vector. HeLa cells were transfected by lipofection and incubated for 30 h under normoxic or hypoxic conditions. Relative luciferase activities were calculated by normalizing luciferase signals to the signals obtained from the cotransfected Renilla luciferase reference gene, driven by the SV40 promoter. The levels of transcriptional stimulation by hypoxia are calculated dividing relative luciferase activities obtained from cells grown under hypoxic conditions by the relative luciferase activities obtained from cells grown under normoxic conditions. Data shown reflect the means Ϯ S.E. of three independent experiments.