Interleukin 6 induces the expression of vascular endothelial growth factor.

Angiogenesis, the formation of new blood vessels, is induced by various growth factors and cytokines that act either directly or indirectly. Vascular endothelial growth factor (VEGF) is a specific mitogen for vascular endothelial cells and therefore has a central role in physiological events of angiogenesis. Interleukin-6 (IL-6) expression on the other hand is elevated in tissues that undergo active angiogenesis but does not induce proliferation of endothelial cells. We demonstrate using Northern analysis that treatment of various cell lines with IL-6 for 6-48 h results in a significant induction of VEGF mRNA. The level of induction is comparable to the documented induction of VEGF mRNA by hypoxia or cobalt chloride, an activator of hypoxia-induced genes. In addition, it is demonstrated by transient transfection assays that the effect of IL-6 is mediated not only by DNA elements at the promoter region but also through specific motif(s) located in the 5′-untranslated region (5′-UTR) of VEGF mRNA. Our results imply that IL-6 may induce angiogenesis indirectly by inducing VEGF expression. It is also shown that the 5′-UTR is important for the expression of VEGF. The 5′-UTR of VEGF is exceptionally long (1038 base pairs) and very rich in G + C. This suggests that secondary structures in the 5′-UTR might be essential for VEGF expression through transcriptional and post-transcriptional control mechanisms.

Angiogenesis, the formation of new blood vessels, is induced by various growth factors and cytokines that act either directly or indirectly. Vascular endothelial growth factor (VEGF) is a specific mitogen for vascular endothelial cells and therefore has a central role in physiological events of angiogenesis. Interleukin-6 (IL-6) expression on the other hand is elevated in tissues that undergo active angiogenesis but does not induce proliferation of endothelial cells. We demonstrate using Northern analysis that treatment of various cell lines with IL-6 for 6 -48 h results in a significant induction of VEGF mRNA. The level of induction is comparable to the documented induction of VEGF mRNA by hypoxia or cobalt chloride, an activator of hypoxia-induced genes.
In addition, it is demonstrated by transient transfection assays that the effect of IL-6 is mediated not only by DNA elements at the promoter region but also through specific motif(s) located in the 5-untranslated region (5-UTR) of VEGF mRNA. Our results imply that IL-6 may induce angiogenesis indirectly by inducing VEGF expression. It is also shown that the 5-UTR is important for the expression of VEGF. The 5-UTR of VEGF is exceptionally long (1038 base pairs) and very rich in G ؉ C. This suggests that secondary structures in the 5-UTR might be essential for VEGF expression through transcriptional and post-transcriptional control mechanisms.
Angiogenesis, the formation of new blood vessels, is a multistep process in which many growth factors and cytokines have an essential role. Two types of angiogenesis promoting agents were characterized: 1) those that indirectly stimulate angiogenesis like tumor necrosis factor ␣ (TNF-␣) 1 and transforming growth factor ␤ (TGF-␤). 2) Those that act as direct angiogens such as acidic and basic fibroblast growth factor (aFGF and bFGF, respectively) or vascular endothelial growth factor (VEGF) (1)(2)(3)(4)(5)(6).
VEGF is a very potent angiogenic agent that acts as specific mitogen for vascular endothelial cells through specific cell surface receptors. VEGF is encoded by a single gene, however, 4 isoforms of 206, 189, 165, and 121 amino acids long are produced as a result of alternative splicing (7,8). The 165-amino acid long isoform is the most abundant. However, no differences in the biological activities of these isoforms have been reported although differences in their receptor binding abilities and their capability to interact with the ECM were demonstrated (9,10).
The promoter region of VEGF has been cloned, sequenced, and found to contain numerous putative binding sites for various transcription factors such as AP1, AP2, and SP1 (11). Interestingly, VEGF contains a very long (1038 bp) 5Ј-UTR that was also noted in a related cytokine, Platelet-derived growth factor (PDGF), and found to contain all the DNA binding motifs described above. This 5Ј-UTR is characterized by a high G ϩ C content upstream to the translation initiation site which is also characteristic of 5Ј-UTRs corresponding to PDGF, placenta growth factor and TGF-␤ (12)(13)(14).
VEGF expression was reported in normal tissues like lung, kidney, adrenal gland, liver, stomach, heart, and peritoneal macrophages and during normal physiological conditions such as cyclical corpus luteum formation, wound healing, and embryo development. Conversely, the expression of VEGF was also reported during abnormal physiological conditions such as tumor angiogenesis (2,(15)(16)(17)(18)(19).
The expression of VEGF promotes the formation of new blood vessels in a regulated mechanism that is sensitive to hypoxia and is not completely characterized. Recently, it was demonstrated that physiological conditions such as hypoxia (or treatment with cobalt, that mimics hypoxia) can induce VEGF expression (20,21). In addition, induced expression of VEGF was also noted in cells treated with interleukin-1␤ (IL-1␤), TGF-␤, PDGF-B, and 12-O-tetradecanoylphorbol-13-acetate (22)(23)(24)(25). Yet, the molecular mechanism governing the expression of VEGF are not characterized and it is likely that other cytokines might promote the expression of VEGF as well.
In this work, we have studied the effect of various cytokines on the expression of VEGF. We show that VEGF mRNA is induced by IL-6. This induction is also demonstrated by transient transfection assays in which the promoter region of VEGF including its exceptional 5Ј-UTR were connected to the reporter gene chloramphenicol acetyltransferase (CAT). We show that DNA element(s) located upstream to the transcription initiation site mediate in part the response to the IL-6. In addition, we demonstrate that the induction of VEGF is also mediated through specific DNA element(s) located at the 5Ј-UTR of the gene. The data suggests that the 5Ј-UTR elements cooperate synergistically with DNA elements located upstream to the transcription initiation site. It is also shown that 5ЈUTR of VEGF has an important role in the expression of VEGF. Fi-nally, we propose that IL-6 should be considered as an indirect inducer of angiogenesis that exerts its activity through the induction of VEGF.

MATERIALS AND METHODS
Cell Culture-The cell line A431 was received from ATCC (Rockville, MD). The cell lines C6 and L8 were kindly obtained from Drs. E. Keshet (The Hebrew University, Israel) and S. Salzberg (Bar-Ilan University, Israel), respectively. The cell lines were all grown in Dulbecco's modified Eagle's medium supplemented with gentamicin, glutamine, 10% fetal calf serum for A431 and C6 cells and 15% fetal calf serum for L8 cells.
Northern Blot Analysis-Cells were grown to 75% confluency and a typical set of experiments included untreated cells and cells treated with cytokines and cobalt at indicated times. R-IL-6 (80 ng/ml) was kindly obtained from Inter-Lab Ltd. . Following the incubation, the cells were harvested and RNA was prepared using Tri Reagent (MRC, Cincinnati, OH). Northern blot analyses were carried out as described previously (26). 32 P-Labeled VEGF and human interferon regulatory factor 1 probes were prepared as described previously (26,27), respectively. The blots were either exposed to x-ray film that were scanned and analyzed by Scanalytics densitometry system using RFLPscan computer analysis program (Billerica, MA) or analyzed by a Fuji PhosphorImager.
Construction of VEGF-CAT Constructs-The bacteriophage genomic clone (4), originally generated from human lung fibroblasts that contains the first two exons of VEGF and the 3.4-kb sequence upstream to the translation initiation site, was kindly obtained from Dr. J. Abraham (California Biotechnology Inc.) (11). A 5.5-kb EcoRI fragment from 4 was cloned into the EcoRI site of pUC18 (pUC18-5.5). This recombinant pUC18 plasmid was digested with EcoRI and NarI and a 3.3-kb segment corresponding to DNA sequences 83 bp upstream to the first ATG codon of VEGF was cloned into the same restriction sites in pUC19 thus generating pUC19-3.3. This plasmid was digested with NarI and HindIII and the resulting fragment (3.3 kb) was cloned into the promoterless CAT construct pBLCAT3 (28) that was digested with AccI and HindIII, thus generating p3.4CATÉNrX. To generate internal deletions p3.4CATÉNrX was digested with NheI, thus removing 695 bp. The digested plasmid was self-ligated thus generating p3.4CATÉNhe. Similarly p3.4CATÉNrX was digested with NheI and XhoI, thus deleting 989 bp upstream to the translation initiation site, that was then religated to itself following fill-in reaction with Klenow fragment generating the reporter plasmid p3.4CATÉNX. All 3 CAT constructs were digested with PstI, self-ligated, and as a result a 1571-bp fragment was deleted from the 5Ј region of each clone and the equivalent plasmid were termed p1.8CATÉNrX, p1.8CATÉNhe, and p1.8CATÉNX. Polymerase chain reaction was employed to construct CAT reporter plasmids that contain the missing 83 bp located 5Ј to the first ATG codon. The 3Ј primer corresponding to the first 15 bp located upstream to the first ATG to which a XhoI site was engineered (L-4, 5Ј-CATATGCTCGAGGGTTTCGGAGGCCCG-3Ј) was amplified with the 5Ј primer corresponding to the native PstI site located Ϫ1829 bp upstream to the translation initiation site (L-3, 5Ј-TCACTTTGAT-GTCTGCAGGC-3Ј) using pUC18-5.5 as a template and the resultant 1829-bp fragment was digested with PstI and XhoI and was replaced with the corresponding fragment in plasmid p1.8CATÉNrX generating p1.8CAT reporter plasmid. To generate a reporter plasmid that contains the available promoter region of VEGF and the 5Ј-UTR (from position Ϫ3400 up to ϩ1), p3.4CATÉNrX was digested with PstI and the resultant 1751-bp fragment was cloned into the PstI site of p1.8CAT generating p3.4CAT. The various CAT constructs are schematically illustrated in Fig. 4, A and B, and their features are described under "Results." DNA Transfections and CAT Assays-L8 cells were transfected by the calcium phosphate-DNA coprecipitation method as described before (29). Cells were transfected with 5 g of various plasmid DNA and 15 g of pUC19 as carrier DNA. For IL-6 treatments, the cells were washed twice with phosphate-buffered saline 24 h following transfection, and then treated with human IL-6 at indicated concentration and harvested 48 h later. For cobalt chloride treatment, the cells were washed twice with phosphate-buffered saline as above and cobalt chlo-ride was added to a final concentration of 550 M for 6 h from a 100-fold stock solution prepared in the growth medium. The cells were then washed again with phosphate-buffered saline and harvested 42 h later. CAT assays were performed and normalized for protein concentration and transfection efficiencies as described previously (26). In principal, transfection efficiency was monitored by ␤-galactosidase activity driven by the Rous sarcoma virus long terminal repeats that were cotransfected in each experiment. Each set of transfection experiments was repeated at least three times generating similar results.

Effect of Various Cytokines on the Expression of VEGF-It is
well documented that various cytokines have a role in biological processes such as: inflammation, wound healing, and tumor progression, in which angiogenesis takes part. To investigate if cytokines can affect angiogenesis by influencing the expression of VEGF mRNA, we have tested the effect of IL-6, TNF-␣, and IFN-␤ on the expression of VEGF in A431 cells (human epidermoid carcinoma) that constitutively express low levels of VEGF (30). Fig. 1 shows that exposures of A431 cells to IFN-␤ for 5 h and to TNF-␣ for 20 h resulted in induced VEGF mRNA levels. In addition, treatment of cells with IL-6 for 5 h resulted in a moderate induction of VEGF mRNA level that was further induced following 24 h of treatment. This elevated level was also apparent after 48 h of treatment (data not shown). As a positive control, the cells were treated for 5 h with cobalt, which mimics anaerobic conditions (21,31). As expected, a significant induction in VEGF mRNA was observed following cobalt treatment (Fig. 1). To show that the observed induction of VEGF is specific, the same membranes were also probed with labeled DNA fragment derived from interferon regulatory factor-1, a transcriptional activator that mediates interferon responsiveness. As expected (26), interferon regulatory factor-1 was induced mainly by IFN-␤ and to a lesser extent by IL-6 ( Fig. 1). Treatment of cells with either TNF-␣ or cobalt had no effect on the expression of interferon regulatory factor-1. The lower panels show relative amounts of 28 S rRNA.
It was demonstrated by Keshet and co-workers (32) that IL-6 is expressed during the angiogenesis that accompanies folliculogenesis of the maternal decidua and during early postimplantation development. Therefore, we have focused on the effect of IL-6 on VEGF expression. Since A431 cells are not very suitable for transfection, the ability of IL-6 to induce VEGF mRNA was tested in two rat-derived cell lines, L8 cells (skeletal muscle myoblasts) and C6 cells (Glioma cells) in which VEGF is hypoxia-inducible (20). The results shown in Fig. 2 clearly demonstrate that VEGF mRNA expression in the two cell lines was augmented by both IL-6 and cobalt. Thus, the effect of IL-6 on VEGF expression is not limited to A431 cells.
IL-6 Enhances the Activity of a CAT Reporter Gene Driven by VEGF Promoter Region-To examine the role of specific regulatory elements within the VEGF promoter region in mediating the response to IL-6 and cobalt, a genomic DNA segment was connected to the reporter gene CAT. This ϳ3.4-kb genomic fragment also contains the 5Ј-UTR up to the translation initiation site (see illustration of construct p3.4CAT in Fig. 4A, for details see "Materials and Methods"). A promoter region is defined as regulatory DNA sequences located upstream to the transcription initiation site. Since VEGF has an exceptionally lengthy 5Ј-UTR that also contains putative binding sites for various transcription factors it was included in the initial CAT construct (pCAT3.4). The reporter construct p3.4CAT was introduced into L8 cells and the cells were treated with increasing amounts of IL-6. A 3-fold increase in CAT activity was detected in cobalt treated cells (Fig. 3). A similar potentiation of CAT activity (3-fold) was also observed in L8 cells treated with 80 ng/ml IL-6 and to a lesser extent with 40 or 160 ng/ml as compared with the CAT activity of untreated cells (Fig. 3). Similar results were also obtained with C6 cells and U937 cells (human promonocytic cells) (data not shown).
The 5Ј-UTR of VEGF Is Essential for Its Responsiveness to IL-6 and Cobalt-To determine which part of the promoter region of the 5Ј-UTR of VEGF confers responsiveness to either IL-6 or cobalt, a series of deletions were generated in the 3.4-kb segment. The various deletions made in the promoter region or in the 5Ј-UTR are illustrated in Fig. 4, A and B. Using available restriction sites, we have made three types of deletions in the 5Ј-UTR region. The first construct, p3.4CATÉNrX, contains a 83-bp deletion of an NarI-XhoI fragment located upstream to the translation initiation site that is G ϩ C-rich (83%). In the second construct (p3.4CATÉNhe), a 695-bp fragment confined by NheI restriction sites located in the 5Ј-UTR 49 bp downstream to the transcription initiation site was deleted. In the third construct, p3.4CATÉNX, most of the 5Ј-UTR except for 49 bp adjacent to the transcription initiation site was removed (Fig. 4A). An EcoRI-PstI fragment (1571 bp) was deleted from all these CAT constructs thus leaving 791 bp upstream to the transcription initiation site (see 1.8CAT plasmids illustrated in Fig. 4B). All 8 constructs were transfected into L8 cells that were either treated with IL-6 (80 ng/ml) or cobalt or not treated.
We first wanted to test the effect of the 5Ј-UTR on the promoter region. For that purpose, we have compared the CAT activity of the full-length construct, p3.4CAT, with that of the other 3.4-kb constructs containing deletions in that region. The results of these transfection experiments are shown in Fig. 4A. As expected, both IL-6 and cobalt promoted CAT activity of the full-length promoter that also contains 5Ј-UTR (construct p3.4CAT) as compared to CAT activity of untreated cells. However, in cells transfected with the construct p3.4CATÉNX, in which most of the 5Ј-UTR was deleted (except for 49 bp downstream to transcription start site), no significant change in CAT activity in response to IL-6 or cobalt was detected. Interestingly, this construct (p3.4CATÉNX) still contains "a classical promoter region" which is missing the 5Ј-UTR. However, the CAT activity derived by this construct in untreated cells was reduced by 50% when compared to that of cells transfected with the full-length construct. We have also studied the effect of shorter internal deletions in the 5Ј-UTR on the ability to promote CAT activity in response to IL-6 and cobalt. Deletion of only 83 bp in the 5Ј-UTR near the translation start site (p3.4CATÉNrX) resulted in a significant reduction in CAT activity of the untreated transfected cells (at least by 30%) and in a weak response to IL-6, although the response to cobalt was still sustained. A reporter construct containing an internal deletion of 695 bp out of the 1038 bp of the 5Ј-UTR (p3.4CATÉNhe) was also tested. When introduced into the cells, the level of CAT activity as compared to the full-length construct was also reduced by 25% but no significant response to either cobalt or IL-6 was noted. These results imply that the 5Ј-UTR potentiates the expression of the VEGF promoter and that response elements to either IL-6 or cobalt may be located within that region. resulted in a modest response to cobalt treatment and a weak response to IL-6. The remaining constructs, p1.8CATÉNhe and p1.8CATÉNX, promoted CAT activities in untreated transfected cells that were 4-fold lower as compared to that of cells transfected with the full-length 3.4-kb construct. No significant response to IL-6 was observed with these constructs and the response to cobalt was residual. These results indicate that the 1751-bp deleted segment contains DNA motifs that work cooperatively with DNA or RNA elements located at the 5Ј-UTR and confirm the fact that this region is essential for expression of VEGF. DISCUSSION Formation of new blood vessels, angiogenesis, is mediated through the activation of parental vessels by specific inducers that either act directly or indirectly. VEGF is a direct angiogenic agent and its role in promoting angiogenesis during normal and abnormal physiological conditions is well documented (1). This suggests that its expression is subjected to complex regulatory mechanisms. Several stimulators of VEGF expression have been reported including hypoxia and cobalt chloride (20,21,31,33), and various cytokines and growth factors such as IL-1␤, TGF-␤, and PDGF-B (22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)35).
We have studied the effect of IL-6 on VEGF expression. IL-6 is a multi-functional cytokine that is produced by many cells and has pleiotropic effects. IL-6 acts on a wide range of tissues and cell lines and promotes cell growth and differentiation on the one hand and growth arrest on the other hand (36 -38). Previously, it was demonstrated that the in vivo expression of IL-6 accompanies vascularization in the reproductive tissues (32). Moreover, IL-6 expression has been noted during wound healing and tumor growth (38,39). It was suggested that IL-6 may induce angiogenesis and may also induce motility of cells such as endothelial cells (32,38,40). We provide evidence indicating that IL-6 may promote angiogenesis through the induction of VEGF expression. This induced expression is mediated by specific DNA motifs located on the putative promoter FIG. 4. The induced CAT activity in response to treatments with IL-6 and cobalt is mediated by DNA motifs located at the promoter region and at the 5-UTR of VEGF. A, the promoter region of VEGF including the 5Ј-UTR (ϳ3.4-kb segment) was connected to CAT reporter gene, construct p3.4CAT. In addition, three internal deletions were generated in the 5Ј-UTR of the p3.4CAT construct that resulted in the formation of p3.4CATÉNrX (a deletion of 83 bp), p3.4CATÉNhe (a deletion of 695 bp), and p3.4CATÉNX (a deletion of 989 bp). All 4 constructs are illustrated. Each construct was transfected into L8 cells that were treated with IL-6 or cobalt chloride (Co) or not treated (Ϫ) and CAT assays were performed as described under "Materials and Methods." B, a deletion of 1571 bp at the 5Ј end of all p3.4CAT constructs was generated resulting in p1.8CAT constructs illustrated in the figure. The various constructs were transfected into L8 cells as described in panel A. In each set of experiments (panels A or B) the CAT activity was calculated in relative units and the results are shown next to the corresponding CAT spots. Abbreviations in the figure represent restriction sites (RI), EcoRI;, N, NarI; Nhe, NheI; X, XhoI. Arrow, transcription start site; numbers, relative distance from the translation start site. Statistical significance was determined by the t test with a probability value (p) Ͻ 0.1. region of VEGF as well as by specific elements located in the 5Ј-UTR. IL-6 exerts its biological effects through association with specific cell surface receptors resulting in the activation of specific transcription factors that interact with two types of cis-acting DNA control elements mediating IL-6 response. Type I elements have a consensus sequence agTgNNGYAA which serves as binding sites for NF-IL-6, and type II IL-6 response elements have a consensus sequence CTGGGA that binds IL-6-RE binding protein (41)(42)(43). Interestingly, two type I response elements are located at DNA sequences flanking the translation initiation site of VEGF. The first is located between positions Ϫ796 and Ϫ804 (GTGCTGGAA) within the 5Ј-UTR and the second is located between positions Ϫ2313 and Ϫ2320 (TGAGGGAA) upstream to the transcription initiation site. Our results provide evidence that both elements might be functional. When DNA sequences between positions Ϫ1829 and Ϫ3400 were deleted, the response to IL-6 was decreased by 2-fold in comparison to the response of the full-length promoter plus 5Ј-UTR. This suggests that part of the IL-6-REs are confined to the deleted region (Fig. 4). In addition, internal deletions within the 5Ј-UTR abrogated the effect of IL-6, indicating that a possible functional IL-6-RE is also contained within that region. These findings indicate that the putative IL-6-REs at the promoter region of VEGF and its 5Ј-UTR are not only functional but also operate in a cooperative manner. Our results, however, do not enable us to determine whether motifs in the 5Ј-UTR regulate the expression of VEGF at the transcriptional or post-transcriptional level.
Recently, it was reported that IL-6-like VEGF is induced in response to hypoxia (44). Thus, our findings together with these recent results suggest that induction of IL-6 by hypoxia may promote the expression of VEGF that eventually leads to angiogenesis. The involvement of IL-6 in angiogenesis is apparently not limited to the hypoxia conditions since it is also produced during wound healing (39) and ovulation (32). Therefore, IL-6 should be considered as an indirect angiogenic factor. Interestingly, ␣ 2 -macroglobulin is an acute-phase protein that is also induced by IL-6 (42,43). It was reported that VEGF is inactivated by binding to ␣ 2 -macroglobulin (45). Since IL-6 induces ␣ 2 -macroglobulin it may eventually lead to the inactivation of circulating VEGF.
We also demonstrate that other cytokines such as IFN-␤ and TNF-␣ can also induce the expression of VEGF mRNA. The involvement of IFN-␤ in angiogenic processes have not been described before. However, IFN-␤ is expressed during inflammation, rheumatoid arthritis, and wound healing. Thus, it is probable that expression of IFN-␤ in response to these disorders might be one of the signals that triggers the angiogenic process through the induction of VEGF expression. Direct proof for such regulatory mechanisms awaits future studies. On the other hand, TNF-␣ has been implicated as an indirect angiogenic factor (46). Our data suggests as published very recently, that it acts through the induction of the expression of the direct angiogenic agent, VEGF (47).
The 5Ј-UTR of VEGF is exceptionally long (1038 bp) and contains several putative DNA motifs for SP1, AP1, AP2, as originally noted by Tischer and colleagues (11), and also the putative IL-6-RE type I element. This region is rich in G ϩ C content. A similar structure was also reported for PDGF-B. It was demonstrated that the PDGF 5Ј-UTR acts as translational inhibitor and upon deletion translation efficiency increases (13,48). Furthermore, during megakaryocytic differentiation, in which PDGF-B is produced, translation repression is removed by cis-acting elements (48). Similarly, the 5Ј-UTR of placenta growth factor, albeit shorter (less then 300 bp), is composed of 73% G ϩ C base pairs with a short open reading frame that has an inhibitory effect on translation (12). Our results also demonstrate a possible role for the 5Ј-UTR in VEGF expression. However, unlike the 5Ј-UTRs of PDGF and placenta growth factor that serve as translation inhibitors, VEGF 5Ј-UTR enhances promoter activity as determined by transient assays. A deletion of 83 bp located in the vicinity of the translation start site resulted in a significant decrease in CAT activity (30 -40%), suggesting that this region has a role in VEGF expression. Since this region is characterized by a highly G ϩ C content (83%) it is most probable that it has a role in maintaining proper secondary structure of the 5Ј-UTR which is crucial for proper translation. A possible role for secondary structure was also suggested for PDGF 5Ј-UTR in inhibiting scanning mechanisms of ribosomes. It was proposed that such structured 5Ј-UTRs might serve as internal ribosome entry sites (48). Since the 5Ј-UTR of VEGF can theoretically form such structures, it is most probable that it is also subjected to translational regulation. DNA motifs that confer response to hypoxic treatment enable the binding of specific hypoxia induced factor (21). Similar DNA motifs were also observed for VEGF, however, only the motif located at the 3Ј-UTR was functional while the two homologous motifs at the promoter region of VEGF (Ϫ2856 to Ϫ2847 and Ϫ1958 to Ϫ1949) were not essential (49). It was found that a putative new hypoxia element was contained within a 100-bp region between locations Ϫ1882 to Ϫ1782. Our results support this finding and enables the margins of the hypoxia element to be further narrowed to positions Ϫ1782 to Ϫ1829 (construct p1.8CAT). The magnitude of increased CAT activity for the 3.4CAT construct following treatment with cobalt and for the 1.8CAT construct was similar (ϳ2-fold), implying that the hypoxia (cobalt) response element is indeed located downstream to position Ϫ1829. When most of 5Ј-UTR was deleted, the response to cobalt was not observed (Fig. 4). This is not in agreement with the reported results of Minchenko et al. (49) that also used constructs lacking the 5Ј-UTR to which luciferase was connected as the reporter gene. They demonstrated enhanced luciferase activity in response to cobalt in HeLa cells, and calculated their results as ratio between the activity of the reporter gene in treated cells and untreated cells and not in absolute light units. Therefore, it is likely that the extremely sensitive luciferase assay enabled the detection of variations even at very weak promoter activities. However, their results do not contradict ours showing that the 5Ј-UTR is essential for VEGF expression. Differences in cell lines should be ruled out since our data were also reproducible in HeLa cells (data not shown).
In conclusion, we provide evidence showing that the role of IL-6 in angiogenesis is mediated through the induction of VEGF, a potent angiogenic agent. We show that putative DNA motifs located at the promoter region as well as in the 5Ј-UTR are necessary for the responsiveness to IL-6. In addition, it is demonstrated that 5Ј-UTR enhances the basal promoter activity of VEGF as determined by transient transfection assays. These results indicate that the secondary structure of 5Ј-UTR may be important for an efficient expression of VEGF and suggest that transcription and post-transcription control mechanisms are involved.