J Biol Chem, Vol. 275, Issue 1, 642-650, January 7, 2000

Regulation of Vascular Endothelial Growth Factor Expression in Human Keratinocytes by Retinoids^*

Bárbara Vega Diaz, Marie-Cécile Lenoir, Annie Ladoux, Christian Frelin, Michel Démarchez, and Serge Michel^§

From Galderma R&D, 635 route des Lucioles, BP67, 06902 Sophia Antipolis cedex, Valbonne, France and the  Institut de Pharmacologie Moléculaire et Cellulaire CNRS UPR 411, 660 route des Lucioles, 06560 Valbonne, France

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Vascular endothelial growth factor (VEGF) is overexpressed in hyperproliferative diseases, such as psoriasis and cancers, which are characterized by increased angiogenesis. Experimentally, VEGF overexpression can be induced by the treatment of cell cultures and biological tissues with phorbol esters, such as 12-O-tetradecanoylphorbol-13-acetate (TPA). Using normal human keratinocytes in conventional cultures and skin grafted onto nude mice in vivo, we show that retinoids can inhibit TPA-mediated VEGF gene induction at the transcriptional level. Because retinoids are biologically active either by interacting with the nuclear retinoic acid receptors or by interfering with the activator protein 1 (AP1) transcription factor, we studied the effect of the retinoic acid derivative CD 2409, which exhibits strong anti-AP1 activity but does not bind to the known retinoic acid receptors in vitro. The results demonstrate that the inhibition of VEGF expression by retinoids only depends on their anti-AP1 activity and does not require gene transactivation via retinoic acid response elements. Because the VEGF promoter contains four potential AP1 binding sites, we used different promoter constructs to identify the functional site responsible for TPA induction and retinoid inhibition. This site turned out to be localized at position 621 of the 5' flanking region of the VEGF gene.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Vascular endothelial growth factor (VEGF)¹ was first described as a tumor cell derived factor that induced vascular hyperpermeability to plasma proteins (1). It was further characterized as an endothelial cell specific mitogen with the capacity to induce angiogenesis in a number of experimental in vivo models (2-4). VEGF is a secreted homodimeric glycoprotein of 40-45 kDa that selectively binds to two high affinity tyrosine kinase receptors on endothelial cells (5, 6). Four different human isoforms have been isolated to date, resulting from alternative splicing of VEGF mRNA (7). The two larger variants, VEGF ₁₈₉ and VEGF ₂₀₆, remain cell-associated, whereas the two smaller, forms VEGF₁₂₁ and VEGF ₁₆₅, are secreted (8). VEGF is the major angiogenic factor that regulates the growth of new capillaries from preexisting blood vessels, a process that involves the extravasion of plasma proteins, degradation of the extracellular matrix, and endothelial cell migration and proliferation, as well as capillary tube formation (9).

In normal human skin, VEGF is both expressed and secreted by epidermal keratinocytes. Neovascularisation, which occurs during wound healing, is associated with an enhanced expression of VEGF by migrating keratinocytes and with the up-regulation of VEGF receptors on dermal microvessels (10). VEGF expression is up-regulated in certain skin diseases involving vascular hyperproliferation, such as psoriasis (11), delayed-type skin hypersensitivity reactions, bullous diseases (12), and Kaposi's sarcoma (13). In cultured human keratinocytes, the expression of VEGF is increased by serum, transforming growth factor-1, tumor necrosis factor , keratinocyte growth factor (14), UVB, oxidants such as H₂O₂ (15), and hypoxia (16).

Retinoids consist of both natural and synthetic vitamin A derivatives, which are potent agents for the treatment of different skin disorders (17). They exert their biological effects via two families of nuclear receptors, which belong to the superfamily of steroid/thyroid hormone nuclear receptors. They comprise the retinoic acid receptors (RAR , , and ), which bind with both all-trans retinoic acid (RA) and 9-cis RA, and the retinoid X receptors (RXR , , and ), which only bind with 9-cis RA. The two classes of receptors are ligand-dependent transactivating factors that regulate gene expression by interacting with the promoter of target genes in the form of RXR/RXR homodimers or RAR/RXR heterodimers (18-20). They can also indirectly down-regulate the expression of certain genes, by antagonizing the effect of the AP1 transcription factor formed by heterodimers of proteins of the c-Jun and c-Fos family (21).

Because retinoids are used to treat cancers (22) and skin diseases such as psoriasis (17) in which an overexpression of VEGF is involved, we have studied their effect on the expression of VEGF at the mRNA and protein level in cultured keratinocytes and human skin grafted onto the nude mouse. There are preliminary data from our laboratory (23) and others (24) indicating that natural and synthetic retinoids are able to down-regulate VEGF expression in cultured human keratinocytes; however, their mechanism of action has not been studied yet. In this paper we show, that it is the anti-AP1 activity of the retinoid molecules that is responsible for the inhibition of the VEGF expression, and the AP1 site in the human VEGF promoter responsible for this negative regulation has been identified.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Retinoids Employed-- The retinoids used were as follows: RA, CD 367 (4-(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-anthracen-2-yl)-benzoic acid), Am 580 (4-[(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-naphthalene-2-carbonyl)-amino]-benzoic acid), CD 2019 (6-[4-methoxy-3-(1-methyl-cyclohexyl)-phenyl]-naphthalene-2-carboxylic acid), CD 437 (6-[3-(1-adamantyl-4-hydroxy-phenyl]-naphthalene-2-carboxylic acid), CD 271 (adapalene), CD 2665 (4-[6-methoxyethoxymethoxy-7-(1-adamantyl)2-naphthyl]benzoic acid), and CD 2409 (4-[1-hydroxy-3-(5,5, 8,8-tetramethyl-5,6,7,8-tetrahydro-naphthalen-2-yl)-prop-2-ynyl]-benzoicacid). For references, see Table I.

Cell Culture Conditions-- Normal human keratinocyte (NHKs) were isolated from human skin obtained from plastic surgery. The cells were cultured by the method of Rheinwald and Green (25). They were propagated in serum-free keratinocyte basal medium (Clonetics, San Diego, CA) supplemented with 0.4% (v/v) bovine pituitary extract, 10 ng/ml epidermal growth factor, 5 µg/ml insulin, and 0.15 mM calcium. For all experiments, second passage keratinocytes were used. Subconfluent keratinocyte cultured in 60-mm dishes were incubated for 4 h in serum and growth factor-free keratinocyte basal medium either with or without retinoids. The latter were dissolved in Me₂SO at the desired concentrations. In some experiments, the cells were preincubated with retinoids for 16 h before the addition of 100 nM 12-O-tetradecanoylphorbol-13-acetate (TPA) (Sigma) for the last 8 h.

Human Skin Grafts-- Pathogen-free congenitally athymic nude mice, Swiss nu/nu (Iffa-Credo, Les Oncins, France), aged 5-7 weeks, were anesthetized with sodium pentobarbital (Nembutal). A graft site on the anterolateral back was prepared with 70% ethanol, after which a circular piece of skin (1 cm in diameter) was removed down to the panniculus carnosus. Human skin, obtained from plastic surgery after informed consent of the patients, was cut into 1-cm-diameter pieces and fitted into the prepared graft sites. To protect the human skin, grafts were first covered by a dermal equivalent and then protected by a surgical tape reinforced with an extensible bandage, which was changed twice a week over a 6-week period (26). Retinoids and TPA were simultaneously applied at the graft site for 6 h, and human skin was removed for RNA analysis.

RAR Binding Assay-- The assay was performed as described by Cavey et al. (27). Briefly, COS-7 cells were transfected with the different pSG-derived expression vectors encoding for human RARs using the polybrene technique (28). Cells were lysed, and the nuclei were recovered by centrifugation. For competition binding assays, nuclear extracts were incubated with [³H]CD 367 (2 nM) as the radioligand and various concentrations of the retinoid to be tested. Separation of free and bound ligand was performed by high-performance size exclusion chromatography. The dissociation constant (K_d value) for each retinoid was determined by nonlinear regression analysis using the Origin software (Microcalc Software Inc.).

RAR Transactivation Assay-- This assay was performed as described previously (29). Briefly, HeLa cells were cotransfected with 2 µg of expression vectors encoding for human RAR , RAR , or RAR  and with 5 µg of the TRE3-tk-chloramphenicol acetyltransferase reporter plasmid, which responds equally well to RAR , RAR , and RAR . The cells were grown for 24 h in the presence of different concentrations of the various retinoids. Chloramphenicol acetyltransferase activity was determined in lysates by enzyme-linked immunosorbent assay (ELISA) (Roche Molecular Biochemicals). The retinoid concentrations that produced half maximal activation (AC₅₀) were determined from dose response curves, using the Origin software (Microcalc Software Inc.)

AP1 Transrepression Assay-- HeLa cells were transfected with a construct containing the collagenase promoter from position 73 to + 63 (30) cloned upstream of the reporter gene encoding chloramphenicol acetyltransferase. Transfected cells were treated with retinoids at 1 µM for 5 h, and then 10 nM TPA was added for a further 16 h. The amount of chloramphenicol acetyltransferase in cell lysates was determined by ELISA.

RNA Isolation-- Total RNA was isolated from cultured keratinocytes or reconstructed epidermis using the Trizol method (Life Technologies, Inc.) according to the manufacturer's procedure and stored at 80 °C until use. Total RNA from epidermal skin grafts was isolated as described by Chomczynski and Sacchi (31).

RT-PCR and Semiquantitative PCR-- The oligonucleotide primers for PCR were synthesized by Life Technologies, Inc. The sequences were GAPDH sense (5'-AATCCCATCACCATCTTCCA-3') and antisense (5'-GTCATCATATTTGGCAGGTT-3') oligonucleotide and CRABPII sense (5'-GCCACCATGCCCAACTTCT-3') and antisense (5'-GGCCACTCACTCTCGGACGTA-3') oligonucleotide. The amplification products were predicted to be 558 base pairs for GAPDH and 427 base pairs for CRABPII. VEGF primers sequences were as follows: sense oligonucleotide, 5'-CCATGAACTTTCTGCTGTCTT-3'; antisense oligonucleotide, 5'-ATCGCATCAGGGGCACACAG-3'. The VEGF primers were chosen in exons 1 and 3, resulting in a 249-base pair PCR product irrespective of the splice form produced.

RT-PCR was carried out using 5 µg of total RNA extracted from cultured cells and skin grafts. After denaturation in diethylpyrocarbonate-treated water for 10 min at 70 °C, RNA was reverse-transcribed into first strand cDNA using SuperScriptII RNase H-reverse transcriptase (10 units/reaction, Life Technologies, Inc.) and 0.5 µg of oligo(dT) as primer, at 42 °C for 50 min in a total volume of 20 µl in a buffer containing 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl₂, 1 mM dNTP, 10 mM dithiothreitol, and 20 units RNasin. Reverse transcriptase was inactivated at 70 °C for 15 min, and the RNA template was digested by RNase H at 37 °C for 20 min. Each experiment included samples containing no reverse transcriptase (negative controls) to exclude amplification from contaminating genomic DNA. Semiquantitative RT-PCR amplification was performed with a PTC 225 thermal cycler (MJ Research), following a 1-min period of denaturation at 94 °C, under the following conditions: denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, for a total of 30 cycles. The assay mixture contained 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 1.5 mM MgCl₂, 0.1 µM of oligonucleotide primers, dNTPs (100 µM of dATP, dGTP, dTTP, 10 µM dCTP), 0.5 µCi of [³²P]dCTP, 0.5 units of Taq DNA polymerase, and 5 µl of 100-fold diluted cDNA mixture. The final product was extended for 3 min at 72 °C. In each experiment, RT positive controls (templates containing cDNA encoding for VEGF) and negative control (without DNA) were included. The PCR products were then electrophoresed on 6% (w/v) acrylamide gels. Radioactivity in each band was quantified by the storage phosphorimaging technique. The screens were scanned using a Fuji BAS 2000. The signal was quantified in photostimulating luminescence units using the Tina image analysis software. Results were expressed for each sample as band intensity relative to that of GAPDH. An optimum number of PCR cycles was determined in the region of exponential amplification. 10-Fold logarithmic dilutions of the cDNA mixture were used to verify the linear correlation between the intensity of the radioactive signal and the initial amount of cDNA.

Northern Blots-- 10 µg of total RNA were separated by denaturing electrophoresis on 1.2% agarose formaldehyde gels and transferred to Nytran membranes (Schleicher and Schuell) prior to hybridization with selected probes. The probe used for human VEGF was the coding sequence of VEGF₁₆₅ subcloned into the BamHI site of the pBluescript SK() plasmid. The GAPDH probe was a gift of Dr. F. Moreau-Gachelin (Paris, France). cDNA probes were labeled using [³²P]dCTP and the Prime-a-Gene labeling system (Promega). Radioactivity in each band was quantified according to the method described above for PCR products. The VEGF mRNA levels were normalized to GAPDH mRNA levels to compensate for loading errors.

Human VEGF ELISA-- 96-well plates coated with anti-human VEGF monoclonal antibody were purchased from R&D Systems. Keratinocyte culture supernatants were added to the wells, and VEGF was bound by the immobilized antibody. After extensive washing, a peroxidase-linked polyclonal antibody recognizing VEGF₁₂₁ and VEGF₁₆₅ was added to the wells; after washing, a substrate solution was added, and the plates were incubated for 5 min at room temperature. Absorbance was measured at 620 nm with an ELISA plate reader (SLT Lab Instruments, 340 ATC).

Transfection Plasmids and VEGF Transactivation Assay-- NHK cells were transiently transfected using the polybrene procedure (28) with 5 µg of different constructions containing VEGF promoter fragments cloned into the pGl2-basic luciferase reporter plasmid (32) kindly provided, with the permission of Dr J. Abraham (Scios Nova Inc., Sunnyvale, CA), by Drs. A. Damert and W. Risau (Max-Plank-Institut für Physiologische und Klinische Forschung, Bad Nauheim, Germany). After 6 h of incubation in the presence of polybrene (30 µg/ml) and plasmid DNA, the keratinocytes were shocked with 30% Me₂SO for 5 min, washed twice with phosphate-buffered saline, and refed with culture medium. Cells were also transfected with 5 µg of a reporter plasmid containing three copies of the synthetic oligonucleotide (5'-GGCAAAGTGAGTGACCTGCTTT-3') derived from position 614 to 635 of the VEGF promoter, cloned upstream of the Herpes virus thymidine kinase promoter in the TK-Luc+ (HSB) vector, a kind gift of Dr. P. Balaguer (Pathologie des Récepteurs Nucléaires, INSERM U 439, Montpellier, France).

A 0.7 kb VEGF promoter fragment containing the putative 621 AP1 binding site was prepared from the full-length VEGF promoter using the NheI restriction enzyme. The fragment was subcloned into the corresponding restriction site of the pGl2 basic vector (Promega) and subjected to site-directed mutagenesis according to the manufacturer's procedure (QuickChange^TM site-directed mutagenesis kit, Stratagene). The 621 AP1 binding site displaying the nucleotide sequence TGAGTGA was mutated to give TTAGTTA, a sequence inactive for AP1 binding (33). After verification by sequencing, the 0.7-kilobase fragment was ligated in the correct orientation into the NheI-digested VEGF promoter. The mutated promoter thus obtained was subcloned into the pGl2-basic luciferase reporter plasmid.

Transfected NHKs were treated with 100 nM CD 2409 or 100 nM dexamethasone for 16 h, than 100 nM TPA was added for an additional 8 h. Luciferase activity was determined using the Luclite kit (Packard) and the Microbeta Trilux (Wallac EG&G) luminescence counter.

Electrophoresis Gel Shift Assay-- Nuclear extracts were prepared from NHK cells according to the method of Dignam et al. (34). Cells were lysed in 10 mM Hepes, pH 7.9, containing 1.5 mM MgCl₂, 10 mM KCl and 0.5 mM dithiothreitol. After centrifugation at 15,000 rpm for 15 min, the nuclear pellet was suspended in 20 mM Hepes, pH 7.9, containing 25% (v/v) glycerol, 0.42 M NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, and 0.5 mM dithiothreitol. After 30 min of agitation, the nuclear suspension was centrifuged again, and the supernatant was dialyzed against 20 mM Hepes, pH 7.9, 20% (v/v) glycerol, 0.1 M KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride. The nuclear extracts were incubated in loading buffer (20 mM Hepes, pH 7.9, 30 mM KCl, 20% glycerol, 0.1% Nonidet P-40, 0.2 mM EDTA, 4 mM MgCl₂, 4 mM spermidine, 100 µg/ml each poly(dI-dC), and salmon sperm DNA) with the oligonucleotides that were previously labeled with ³²P-ATP with T₄ polynucleotide kinase. The sequence of the AP1 consensus oligonucleotide derived from positions 621 of the human VEGF promoter was 5'-AGGGGCAAAGTGAGTGACCTGCTT-3'. The sequence of the AP1 consensus oligonucleotide derived from the collagenase promoter was 5'-CGCTTGATGAGTCAGCCGGAA-3'. The sequence of the AP2 consensus oligonucleotide was 5'-GATCGAACTGACCGCCCGCGGCCCGT-3'. The mixture was incubated for 30 min at 4 °C and subjected to 5% polyacrylamide gel electrophoresis. Following migration, the gel was analyzed by the storage phosphorimaging technique using a Fuji BAS 2000 screen.

Statistical Analysis-- The results given in the form of histograms are the average (±S.E.) obtained from three independent experiments, each of which provided two to five samples for the same experimental condition. They were analyzed using the two-sided Student's t test.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Regulation of the VEGF mRNA Baseline Level in Cultured NHKs by Retinoids-- In a first series of experiments, the effect of different RAR subtype selective agonists and of an RAR antagonist on the basal expression of VEGF mRNA was determined in cultured human keratinocytes. The concentrations of retinoids were chosen according to their binding affinities for the different RARs (Table I). Time course experiments performed with RA showed that the down-regulation of VEGF mRNA is at its maximum after 4 h (Fig. 1A).

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Inhibition of VEGF mRNA level in cultured NHKs by retinoids. A, cells were incubated with RA at 1 µM from 0 to 24 h. Total RNA was extracted at various time points, and the VEGF mRNA level was determined relative to that of GAPDH by Northern blot analysis. B, cells were incubated for 4 h with Am 580, CD 367, or CD 437 at 10 nM; CD 2019 or CD 271 at 100 nM; and CD 2665 at 1 µM. CD437 (10 nM) and CD 2665 (1 µM) were applied together. Total RNA was extracted, and the VEGF mRNA level was determined, relative to that of GAPDH, by semiquantitative RT-PCR. NS, nonsignificant; *, p < 0.1; **, p < 0.05.

Fig. 1B shows that the baseline level of VEGF mRNA is significantly reduced with the potent RAR pan-agonist CD 367, the selective RAR agonist Am 580, the selective RAR, agonist CD 271, and the selective RAR  agonist CD 437. The RAR, antagonist CD 2665 had no effect on the VEGF mRNA transcription and did not inhibit the effect of the RAR agonist CD 437.

Regulation of TPA-induced VEGF mRNA and Protein Levels in Cultured NHKs by Retinoids-- In a second series of experiments, the time course of VEGF expression during treatment of NHKs with 100 nM TPA was determined. The induction of VEGF mRNA was at a maximum after 8 h and then decreased slowly (Fig. 2A).

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Inhibition of TPA-induced VEGF mRNA level in cultured NHKs by retinoids. A, keratinocytes in serum-free medium were treated for different periods of time with 100 nM TPA. B, cells were treated with the indicated retinoid agonists or antagonist for 16 h, and 100 nM TPA was added for the last 8 h. Total RNA was extracted at various time points from the experiments described in A and B, and VEGF mRNA level was determined relative to that of GAPDH by semiquantitative RT-PCR. NS, nonsignificant; *, p < 0.1; **, p < 0.05.

Subsequently, retinoids detailed in Table I were tested for their inhibitory effect on TPA-induced VEGF mRNA expression. All of the RAR agonists that displayed anti-AP1 activity (compare with Table I) inhibited VEGF mRNA induction regardless of their RAR subtype selectivity, as shown by Northen blot and RT-PCR analysis (Fig. 2B). The RAR, antagonist CD 2665 did not demonstrate any anti-AP1 activity, as it did not affect the level of VEGF mRNA. Because all molecules that diminished the VEGF mRNA level were active in the AP1 transrepression assay (Table I), we suggested that the VEGF inhibition by retinoids is related to their ability to antagonize the AP1 factor.

In order to prove this hypothesis we used CD 2409, a selective anti-AP1 retinoid displaying no in vitro affinity for the three RAR subtypes (Table I) and a weak transcriptional activity via an RARE (Table II). As shown in Fig. 3A, CD 2409 inhibited the binding of the AP1 nuclear protein complex to the AP1 consensus oligonucleotide sequence derived from the collagenase promoter. As shown in Fig. 3B, CD 2409 did not inhibit the binding of the AP2 protein to the AP2 consensus oligonucleotide sequence.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Electrophoretic mobility shift assay using 32P-labeled AP1 oligonucleotide containing the AP1 site derived from the human collagenase promoter or a consensus AP2 site. Nuclear extracts were prepared from control cells (lane 2) and from keratinocytes treated with 100 nM CD 2409 or 100 nM TPA (lanes 3 and 4), or 100 nM CD 2409 and 100 nM TPA (lane 5). A, the extracts were incubated with the 32P-labeled AP1 site derived from the collagenase promoter (lanes 2-5) and with a 100-fold excess of the cold AP1 oligonucleotide (lane 6). The 32P-labeled AP1 oligonucleotide without nuclear extract is shown as a control in lane 1. The location of the AP1 complex is shown by an arrow. B, the extracts were incubated with the 32P-labeled consensus AP2 site (lanes 2-5) and with a 100-fold excess of the cold AP2 oligonucleotide (lane 6). The 32P-labeled AP2 oligonucleotide without nuclear extract is shown as a control in lane 1. The location of the AP2 complex is shown by an arrow. P, free probe.

CD 2409 inhibited the basal expression level of VEGF mRNA (Fig. 4A) to a level similar to RA (see Fig. 1A), and its effect was maximum after 4 h of treatment. CD 2409 also inhibited the TPA-induced VEGF mRNA level, as did dexamethasone, a well known anti-AP1 compound (35) (Fig. 4B). This inhibition was dose-dependent and was maximum at 100 nM (result not shown). In addition, CD 2409 inhibited the basal and TPA-induced secretion of VEGF₁₂₁ and VEGF₁₆₅ as determined by ELISA (Fig. 4C). The discrepancy between the inhibition of basal VEGF expression at the mRNA (Fig. 4B) and protein level (Fig. 4C) can be explained by the fact that the inhibitory effect of retinoids on mRNA expression is maximum after 4 h and then diminishes (Fig. 1B), whereas its manifestation at the protein level needs more time. In this particular experiment, the incubation time was 24 h.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Inhibition by CD 2409 of basal or TPA-induced VEGF mRNA level in cultured NHKs. A, keratinocytes in serum-free medium were incubated with CD 2409 at 0.1 µM from 0 to 24 h. Total RNA was extracted at various time points, and the VEGF mRNA level was determined relative to that of GAPDH by Northern blot analysis. B, keratinocytes in serum-free medium were treated for 16 h with the anti-AP1 selective retinoid CD 2409 (100 nM) or a 1 µM concentration of the reference compound dexamethasone. TPA was added for the last 8 h. Total RNA was prepared, and VEGF mRNA expression was determined, relative to that of GAPDH, by Northern blot analysis. C, secretion of VEGF polypeptides by cultured keratinocytes was determined using an ELISA. NS, nonsignificant; *, p < 0.1; **, p < 0.05; ***, p < 0.01.

Regulation by CD 2409 of VEGF and CRABPII mRNA Levels in Human Skin Grafts-- The effect of CD 2409 on VEGF mRNA was also evaluated in vivo in human skin grafted onto the nude mouse. Grafted human skin preserves most of its original characteristics for the life span of the graft (26).

The VEGF mRNA level was markedly increased 6 h after topical treatment with 0.01 and 0.003% TPA, and it returned to control levels after 24 h (Fig. 5A). In the subsequent experiments, TPA was used at 0.01%, and the VEGF mRNA levels were analyzed 6 h after treatment.

View larger version (34K): [in this window] [in a new window]   Fig. 5.   Effect of the anti-AP1 selective retinoid CD 2409 on TPA-induced VEGF and CRABPII mRNA expression in human skin grafts. A, human skin grafts were treated topically with the indicated TPA concentrations. Total RNA was prepared after 6 and 24 h of treatment, and the level of VEGF mRNA was determined, relative to that of GAPDH, by semiquantitative RT-PCR. B, grafts were treated topically with the reference compound dexamethasone (0.05% (w/v)) or CD 2409 (0.01 or 0.1% (w/v)). Both compounds were also applied in combination with TPA at 0.01% (w/v). After 6 h, total RNA was prepared, and the VEGF (B) or CRABPII (C) mRNA was determined, relative to that of GAPDH, by semiquantitative RT-PCR. NS, nonsignificant; *, p < 0.1; **, p < 0.05; ***, p < 0.01.

CD 2409 alone displayed no significant effect on the basal level of VEGF mRNA at concentrations of either 0.01 or 0.1%. However, the TPA-induced VEGF mRNA expression was inhibited when skin grafts were simultaneously treated with 0.1% CD 2409 and TPA. Dexamethasone, the anti-AP1 reference compound, also displayed an inhibitory effect (Fig. 5B).

According to the in vitro data shown in Table II, CD 2409 has weak RARE transactivating potential. To confirm the missing RAR agonistic activity, the effect of CD 2409 on the expression of CRABPII mRNA, a well characterized marker of retinoid activity (36), was tested in vivo. As expected from its lack of RARE activity, CD 2409 failed to affect CRABPII mRNA level in TPA-treated human skin grafts (Fig. 5C).

Effect of the Selective Anti-AP1 Retinoid on the Transactivation of the VEGF Promoter-- NHKs were transfected with constructs containing the luciferase reporter gene linked either to the full-length VEGF promoter or to the two different promoter fragments shown in Fig. 6B. Transfected cells were treated with the selective anti-AP1 retinoid CD 2409 for 16 h, after which time, TPA was added for the last 8 h. As shown in Fig. 6 A, a, TPA treatment stimulated transactivation mediated by the full-length promoter, which contains four potential VEGF AP1 sites. CD 2409 inhibited this stimulation by 80%. When the promoter construct b of Fig. 6B containing a deletion of the potential AP1 site at 621 was used, transactivation was no longer stimulated by TPA treatment, and consequently CD 2409 had no effect (Fig. 6 A, b). This suggests that the AP1 binding site at position 621 is the functional one. NHKs were also transfected with the reporter construct c containing only the AP1 site at position 621. As shown in Fig. 6A, c, TPA indeed increased the transactivation level. The TPA-induced transcription of the reporter gene was reduced to the basal level by CD 2409 (Fig. 6A, c).

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Inhibition of the TPA-induced transactivation of the human VEGF promoter in NHKs by the anti-AP1 selective retinoid CD 2409. A, NHKs were transfected with plasmids a, b, and c (column groups a, b, and c) (as shown in B, below) containing different VEGF promoter fragments cloned upstream of the luciferase reporter gene. Transfected cells were treated with 100 nM CD 2409 for 16 h before addition of 100 nM TPA for 8 h. Luciferase activity was determined by luminescence measurement. NS, nonsignificant; *, p < 0.1; **, p < 0.05. B, diagram of human VEGF promoter-luciferase reporter gene constructs. Circles indicate the four potential AP1 sites within the upstream sequence of the human VEGF promoter.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Effect of the inactivation of the 621 AP1 binding site by site-directed mutagenesis on the transactivation of the VEGF promoter induced by TPA. A, NHKs were transfected with plasmids a and b (column groups a and b) (as shown in B, below) containing a VEGF promoter fragment cloned upstream of the luciferase gene and displaying a wild type or mutated 621 AP1 binding site, respectively. Transfected cells were treated with 100 nM CD 2409 for 16 h before addition of 100 nM TPA for 8 h. Luciferase activity was determined by luminescence measurement. NS, nonsignificant; *, p < 0.1; **, p < 0.05. B, diagram of human VEGF promoter-luciferase reporter gene constructs. Circles indicate the four potential AP1 sites within the upstream sequence of the human VEGF promoter. m, mutated site.

To gain further insight into the role of the AP1 binding site at position 621 for the inhibitory effect of CD 2409 on the TPA-induced transactivation of the VEGF promoter, we obtained by site-directed mutagenesis a promoter construct containing two mutations inactivating only the 621 site. This construct, as well as the unmodified control construct, is depicted in Fig. 7B. As shown in Fig. 7A, a, treatment with TPA of NHKs tranfected with the control luciferase vector containing the unmodified 621 site induced transactivation of the VEGF promoter, and CD 2409 inhibited this stimulatory effect. In contrast, the stimulatory effect of TPA on the transactivation of the VEGF promoter was completely abolished when NHKs were transfected with the construct containing the mutated 621 AP1 binding site (Fig. 7A, b). This result further supports the idea that among the four putative AP1 binding sites of the VEGF promoter the 621 site is the functional one.

Effect of CD 2409 on the Interaction of the AP1 Sequence of Human VEGF Promoter with NHK Nuclear Proteins-- To characterize the interaction of CD 2409 with AP1 binding to the 621 site, gel shift experiments were performed. Nuclear extracts were prepared from NHKs grown in the presence or absence of CD 2409 and TPA. As shown in Fig. 8, TPA increased AP1 activity by about 4-fold in treated cultures (lane 4) when compared with untreated cultures (lane 2). CD 2409 alone, at 100 nM, did not have any effect on the basal level of AP1 activity (lane 3). However, it inhibited the interaction, induced by TPA, of AP1 with its binding site (lane 5). Fig. 8, lane 6, displays the result of a competition experiment with a 100-fold excess of cold AP1 oligonucleotide, and lane 7 shows that an unrelated SP1 oligonucleotide has no effect.

View larger version (80K): [in this window] [in a new window]   Fig. 8.   Electrophoretic mobility shift assay using a 32P-labeled oligonucleotide containing the AP1 binding site derived from positions 614 to 635 of the human VEGF promoter. The 32P-labeled AP1 oligonucleotide was mixed with NHK nuclear extracts. Nuclear extracts were prepared from untreated cells (lane 2) or NHKs treated with either 100 nM CD 2409 (lane 3), 100 nM TPA (lane 4), or 100 nM TPA and 100 nM CD 2409 (lane 5). As controls, competition of the TPA-induced interaction of AP1 with a 100-fold molar excess of AP1 unlabeled oligonucleotide is shown in lane 6, and the effect of a 100-fold molar excess of SP1 unlabeled oligonucleotide in lane 7. P represents free probe. The location of the AP1 complex is shown by an arrow.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The effect of retinoids on the expression of VEGF was studied at the mRNA and protein level. Results obtained with cultured normal human keratinocytes show that (i) retinoic acid and synthetic retinoids inhibit basal expression of VEGF mRNA, (ii) retinoids also block TPA-induced VEGF expression, and (iii) this inhibitory effect does not depend on the RAR subtype selectivity of the compound.

Because we were not able to find, by computer search, an RARE consensus sequence in the VEGF promoter (7), a possible interpretation for the above results is that the inhibitory action of retinoids on basal and TPA-induced VEGF expression is not mediated by the direct interaction of RARs with their RAREs, but rather by another mechanism, such as the inhibition of AP1 activity (37). Comparing the inhibitory effect of the retinoids studied and their AP1 transrepression activity (Table I) reveals that there is indeed such a relationship. For example, the RAR/ antagonist CD 2665, which does not affect AP1 activity, does not inhibit VEGF expression either. To further prove that the effect of retinoids is due to their anti-AP1 properties, we used the selective anti-AP1 retinoid CD 2409. This compound is characterized by its weak transactivation via the RARE pathway and by its strong anti-AP1 potential, as exemplified by (i) transrepression of the collagenase promoter and (ii) gel shift analysis using the consensus AP1 site of the collagenase promoter. Up to now, several synthetic retinoids with anti-AP1 selectivity have been described in the literature using the same criteria (38-40).

CD 2409 was able to inhibit the basal as well as the TPA-induced VEGF mRNA expression in cultured keratinocytes. Its action was similar to dexamethasone, a reference AP1 inhibitor (35). CD 2409 was also able to block the basal and TPA-induced secretion of VEGF₁₂₁ and VEGF₁₆₅ polypeptides. The results obtained with cultured keratinocytes could be confirmed in vivo using human skin grafted onto the nude mouse.

The VEGF promoter contains four motifs homologous to the AP1 binding site, 5'-TGANT(C/A)NN-3' (41) at positions 621, 1527, 2265, and 2930 (7). Transactivation analysis with different promoter constructs revealed that in cultured keratinocytes, a unique site located at position 621 of the 5' flanking region of the VEGF gene is functional and responsible for the inhibitory effect of retinoids. The importance of this site was confirmed by gel shift analysis and by site-directed mutagenesis, in which inactivation of this site completely abolished the activation by TPA of the transactivation of the VEGF promoter. It is worth mentioning that the 621 AP1 binding site is located in the 5' untranslated region of the VEGF gene, a region containing DNA binding motifs important for the regulation of the expression of VEGF by platelet-derived growth factor (41) or interleukin 6 (42).

Altogether, our data strongly support the hypothesis that VEGF belongs to the family of genes the expression of which is down-regulated by retinoids via the inhibition of the AP1 pathway. This family includes the matrix metalloproteinases (collagenases and stromelysin) (32, 43, 44), growth factors (transforming growth factor ) (45), interleukin 1 (46, 47), and skin-specific genes such as keratin 5 (48) and involucrin (49).

A detailed study has shown that after phorbol ester stimulation of keratinocytes, three members of the Jun and Fos family, namely Fra-1, JunB, and Jun-D, bind to the involucrin promoter (49). Future studies should determine whether the same AP1 family members are involved in the regulation of the VEGF promoter.

There are other nuclear factors known to regulate VEGF expression. For example, in A431 cells, it is AP2 that mediates the effect of transforming growth factor  on VEGF transcription (50). The transcription factor HIF-1, induced by hypoxia, is a major regulatory factor for controlling the VEGF level in several cell types (51-54). How retinoids modulate these transcription factors, however, is not very well known. As far as AP2 is concerned, an elevation of its mRNA level in NT2 cells after retinoic acid treatment has been described (55).

The negative regulation of AP1 activity by retinoids is not completely understood. The formation of a protein complex between Jun/Fos and RAR/RXR (47) and the competition between RAR/RXR and AP1 for the integrator co-factor CBP/P300 (56) have been proposed as explanations. Retinoids may also down-regulate AP1 by activation of mitogen-activated protein kinase phosphatase-1 and -2, which specifically dephosphorylate c-Jun N-terminal protein kinases, a group of mitogen-activated protein kinases involved in c-Jun phosphorylation and activation (57).

VEGF is overexpressed in skin disorders such as psoriasis (11) and is also important for tumor growth and metastasis (58), where it stimulates neovascularization. Retinoids known for their antiangiogenic properties (59) are currently used for the treatment of metastatic cancers (22) and for psoriasis (17). It remains to be tested whether an anti-AP1 selective retinoid such as CD 2409 can be used as a therapeutic agent in the treatment of these diseases, in which an overexpression of VEGF is involved.

	ACKNOWLEDGEMENTS

We thank Prof. U. Reichert for helpful discussions. We also thank P. Rossio G. Terradura and A. Jomard for their help in the skin grafting technique.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ To whom correspondence should be addressed. Tel.: 33-4-93-95-70-70; Fax: 33-4-93-95-70-71.

	ABBREVIATIONS

The abbreviations used are: VEGF, vascular endothelial growth factor; AP, activator protein; ELISA, enzyme-linked immunosorbent assay; NHK, normal human keratinocyte; RA, all-trans retinoic acid; RAR, retinoic acid receptor; RXR, retinoid X receptor; RARE, retinoic acid response element; TPA, 12-O-tetradecanoylphorbol-13-acetate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PCR, polymerase chain reaction; RT, reverse transcription.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES