Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Angiogenssis, the formation of new blood vessels, is a crucial process in tumor progression. Among many tumor-derived angiogenic factors that directly stimulate the endothelial motility and proliferation, vascular endothelial growth factor (VEGF) is the most potent peptide that acts as specific mitogen for vascular endothelial cells (1). Besides its capacity to induce angiogenesis (3) and to increase vascular permeability (1), VEGF can also stimulate production of tissue factor (4), collagenase (5), plasminogen activators, and their inhibitors (6) in these cells.

Considerable evidence lends support to the postulated role for VEGF as a prime regulator of tumor angiogenesis. High levels of VEGF are produced by various types of human cancer lines in vitro and in surgically resected tumors of the human gastrointestinal tract (7), ovary (8), brain (9), and kidney (10). Capillaries are clustered along VEGF-producing tumor cells, and tumor angiogenesis and subsequent tumor growth are inhibited in vivo by antibodies directed against VEGF (11) or soluble VEGF receptors (12). Furthermore, elevated serum levels of VEGF have been reported in patients with lung (13) and gynecologic cancers. Clinical significance of VEGF production by tumors has also been suggested by the fact that VEGF-producing tumors had significantly lower survival compared with the VEGF-negative tumors in squamous cell lung cancer (14) and that VEGF expression was independent of prognostic factors for survival (15). High pretreatment serum level of VEGF was associated with poor outcome in small-cell lung cancer (13). These properties have made the study of the VEGF gene expression relevant and significant for the control of unrestricted growth of tumors.

Expression level of the VEGF mRNA is tightly regulated by both transcriptional and post-transcriptional mechanisms. Hypoxia (16) as well as a variety of cytokines and growth factors, including epidermal growth factor (17), transforming growth factor  (TGF-) (18), TGF- (19), interleukin (IL)-1, and IL-6 (20) were shown to induce VEGF expression in several cell lines. IL-1 induces the VEGF gene expression at the Sp1-binding site, whereas TGF- induces it at the AP-2-binding site. Recent studies have demonstrated the intracellular signaling pathways and genetic elements involved in controlling its expression. The hypoxia-inducible factor 1 (HIF-1) and c-Src play a central role in hypoxia-mediated VEGF gene expression (21). p42/p44 MAP kinase activates VEGF gene expression through a G+C-rich region of the promoter (22).

Because the expression of VEGF has been implicated in tumor angiogenesis, pharmacologic intervention that affects the VEGF expression may influence tumor progression and prognosis of the cancer patients. Retinoids, the natural vitamin A derivatives (23, 24) and synthetic analogs (25), have long been known to regulate a broad range of biologic processes, including growth and differentiation in many normal and malignant cell types. In addition, recent clinical trials have suggested that retinoids may be effective for the treatment of human lung cancer (26) and hepatocellular carcinoma (27). We therefore anticipated that retinoids would negatively regulate the VEGF gene expression in tumor cells.

In this study, we examined the effects of at-RA on the VEGF expression in human bronchioloalveolar carcinoma NCI-H322 cells to explore the possibility that at-RA can serve as a novel therapeutic agent interfering with growth of lung cancer cells. Contrary to our expectation, the results of the present study indicate that retinoids significantly enhance the VEGF mRNA expression and protein production, and that Sp1 sites play an important role in mediating this effect. These results suggest that further consideration of the relevance and efficacy of the chemoprevention of lung cancer by at-RA should be warranted.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents

at-RA, genistein, and cycloheximide were purchased from Sigma Chemical Co. (St Louis, MO). Stock solutions of at-RA were prepared in ethanol preserved at -30°C. Phenol red-free RPMI 1640 medium, fetal bovine serum (FBS), penicillin, and streptomycin were purchased from GIBCO BRL, Gaithersburg, MD. A random primer kit was purchased from Boehringer Mannheim Corp., Indianapolis, IN [-³²P]dCTP (3,000 Ci/mmol) and [-³²P] ATP (6,000 Ci/mmol) were obtained from Amersham Co.(Arlington Heights, IL). Human VEGF enzyme-linked immunosorbent assay (ELISA) kit was obtained from Immuno Biological Laboratories (IBL) Co. (Fujioka, Gunma, Japan). Polyclonal antibodies for Sp1, Sp3, and Egr-1 were purchased from Santa Cruz Biotechnology, Santa Cruz, CA. Tfx-50 was obtained from Promega, Madison, WI.

Cell Culture

NCI-H322 cells derived from human bronchioloalveolar carcinoma cells were obtained from American Type Culture Collection (ATCC, Rockville, MD). These cells were grown at 37°C in 5% CO₂ atmosphere in phenol red-free RPMI 1640 medium (GIBCO BRL) supplemented with 10% fetal bovine serum and 1% penicillin/ streptomycin (GIBCO BRL).

Northern Blot Analysis

Total RNA was extracted from NCI-H322 cells by using ISOGEN (Nippon Gene, Toyama, Japan). Total RNA was electrophoresed in 1.2% agarose/formaldehyde gels containing 25 mmol/liter MOPS buffer (pH 7.8), 1 mmol/liter ethylenediamine tetraacetic acid (EDTA), and 2.2 mol/liter formaldehyde. Northern blot analysis was performed using 20 µg of RNA sample. RNA was transferred to nylon membrane (Hybond-N+; Amersham) as described by the manufacturer. After being UV-crosslinked, blots were stained with 0.04% methylene blue to verify the relative quality and quantity of the RNA. Filters were prehybridized for 6 h at 42°C with 40% formamide, 5× saline sodium citrate (SSC), 5× Denhardt's solution (0.1% Ficoll, 0.1% bovine serum albumin [BSA], and 0.1% polyvinylpyrrolidone), 1% sodium dodecyl sulfate (SDS) and 0.01 mg/ml denatured salmon sperm DNA. Hybridization was performed at 42°C for 12 h using probes included 642-bp fragment of human VEGF cDNA sequence. The probe was radiolabeled with [-³²P]dCTP (Amersham) using a random primer DNA labeling kit (Boehringer Mannheim). Filters were washed with 0.1% SDS containing 2× SSC at 42°C, and exposed to XAR film (Kodak, Rochester, NY) at -80°C for 72 h. Developed films were scanned and analyzed by a computer program (NIH Image 1.49) to measure the relative intensity of each band.

Measurement of VEGF Protein

The amount of VEGF protein was measured in the conditioned medium of 1 µM at-RA-treated NCI-H322 cells with sandwich ELISA analysis (IBL). The samples were sandwiched between a monoclonal antibody against human recombinant VEGF. A second polyclonal antibody against VEGF was conjugated to horseradish peroxidase and then added to the mixture. Color develops by addition of hydrogen peroxide and chromogen tetramethylbenzidine, and the intensity was measured at 450 nm. The amount of VEGF protein production was revised by the amount of total protein of each dish.

Plasmid Constructions

We obtained the mouse VEGF 1,207/pGL2 plasmid from Dr. D'Amore (28). A plasmid 450 LUC was made by subcloning the MluI insert from 1,207 LUC into corresponding site of pGL2. Other plasmids were prepared by polymerase chain reaction (PCR) from plasmid -1,207 LUC using the reverse primer (nt +372) with a BglII site (underlined), 5'-CCCAGATCTTCTCTCTGA CCGGTCTCT-3'. Sequences for PCR upstream primers were: 927 LUC, 5'-CACAGTGCATACGTGGGTTTCCACAGG-3'; 100 LUC, 5'-GGTGCCTGGCTCCACCAGACCGTCCCC-3'; 81 LUC, 5'-CCGTCCCCGGGGCGGGTCTGGGCGGGGCTT-3'; 81 (Sp1 µ) LUC, 5'-CCGTCCCCGGTTCGGGTCTGGGCTT GGCTT-3'; and 1 LUC, 5'-AAGCGCAGAGGCTTGGGGCA GCCGAGC-3'.

The resulting PCR products were subcloned into the pCR2.1 (Invitrogen, Carlsbad, CA), and then subcloned into the KpnI/BglII sites of the promoterless luciferase reporter gene vector, pGL2 (Promega).

We obtained from Dr. Abraham JA (Scinos Inc., CA) the -1,180LUC which contains DNA fragment from 1,180 to +338 of the human VEGF gene fused to luciferase reporter plasmid. Plasmids -480LUC and 89 LUC were made by subcloning the BglII and SmaI insert from 1,180 LUC into corresponding site of pGL3 (Promega).

Transient Transfection, Luciferase Assay, and Preparation of Cell Lysates

For transient transfection, NCI-H322 cells were seeded at 5 × 10⁴ cells per dish. These cells were transfected with 1 µg of reporter plasmid by the lipofection method using Tfx-50 (Promega). After 24 h, transfected cells were washed twice with phosphate-buffered saline (PBS) and stimulated in the absence (vehicle) or presence of 1 µM at-RA. After 72 h incubation, cells were harvested for luciferase assays. Cells were washed with PBS two times and lysated in 120 µl of Cell Cultured Lysis Regent (CCLR; Promega). The cell lysate was scraped and centrifuged to remove cell debris. Luciferase activity was measured with a Berthold Lumat LB9501 luminometer. Moreover, cotransfection with 1 µg of reporter plasmid and 1 µg of expression plasmid for retinoic acid receptor  (RAR), which was generous gift from Dr. K. Umezono (Kyoto University, Kyoto, Japan), was performed by the same methods. Each transfection was repeated and the mean ± the standard error (S.E.) of the mean was plotted.

Nuclear Extracts

Nuclear extracts were prepared from NCI-H322 control cells or cells treated with 1 µM at-RA for 72 h in RPMI 1640 medium without fetal bovine serum. Briefly, confluent cells were washed two times with PBS, scraped, and collected with PBS. After that, 5 ml of ice-cold Buffer A (50 mM Hepes-KOH [pH 7.8], 420 mM KCl, 0.1 mM EDTA [pH 8.0], 5 mM MgCl₂, 20% glycerol, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1 mM dithiothreitol [DTT], 2 µg/ml of aprotinin, leupeptin, and pepstatin, 1 mM sodium orthovanadate) was added and incubated on ice for 5 min. After centrifugation at 15,000 rpm for 5 min, nuclei were downs homogenized. Nuclei were pelleted by being centrifuged at 15,000 rpm for 5 min again and resuspended in 300 µl of nuclear extraction buffer C (50 mM Hepes-KOH [pH 7.8], 420 mM KCl, 0.1 mM EDTA [pH 8.0], 5 mM MgCl₂, 20% glycerol, 1 mM PMSF, 1 mM DTT, 2 µg/ml of aprotinin, leupeptin, and pepstatin, 1 mM sodium orthovanadate). After incubation and rocking at 4°C, the lysates were cleared of debris by centrifugation.

Oligonucleotide and Electrophoretic Mobility Shift Assays

The sequences of the oligonucleotides used as probes or competitors in electrophoretic mobility shift assays (EMSAs) were as follows (Sp1-binding sites are underlined, and mutated bases are indicated by boldfaced letters): VEGF -81/52, 5'-CCG TCC CCG GGG CGG GTC TGG GCG GGG CTT-3'; VEGF -81/52 µ, 5'-CCG TCC CCG GTT CGG GTC TGG GCT TGG CTT-3'; Egr-1 consensus, 5'-CGCCCTCGCCCCCGCGCCGGG-3'; Sp1 consensus, 5'-ATTCGATCGGGGCGGGGCGAGC-3'; and AP-2 consensus, 5'-GATCGAACTGACCGCCCGCGGCCCGT-3'.

All probes were 5' end labeled with T4 polynucleotide kinase and [-³²P]ATP. Binding reactions were performed for 15 min on ice with 10 µg nuclear extracts in 10 µl Buffer C, and ~ 10,000 cpm of ³²P-labeled oligonucleotides. Polyacrylamide gel (6%) was used for EMSA. For the competition experiments, 0.5 ng labeled oligonucleotides was mixed with 50 ng unlabeled competitor oligonucleotides. For supershift assay, nuclear extracts were preincubated with 2 µl of anti-Egr-1, Sp1, Sp3 polyclonal antibodies (Santa Cruz Biotechnology) in the binding buffer for 1 h at 4°C before initiation of the binding reaction. ³²P-labeled probes were then added, the incubation was continued for 20 min before electrophoretic separation on 5% polyacrylamide gel at 100 V, and gels were dried and exposed to Kodak XAR-5 film.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

at-RA Increases VEGF mRNA Levels in Cultured Bronchioloalveolar Carcinoma Cell

As a first step toward understanding retinoid action in bronchioloalveolar carcinoma NCI-H322 cells, we measured the steady state mRNA levels of all RAR receptors and RXR. All of three retinoic acid receptors, RAR, RAR, and RAR, and their heterodimer partner, RXR, were detected by RT-PCR analysis (data not shown). Thus, this system allowed us to examine the effects of at-RA on VEGF expression.

To investigate the effects of retinoids on VEGF expression, subconfluent NCI-H322 cells were stimulated with 1 µM at-RA in ethanol for 24, 48, or 72 h. Effects of at-RA on VEGF mRNA levels in the NCI-H322 cells were comparable between the confluent cultures and subconfluent cultures (data not shown). NCI-H322 cells stimulated with ethanol alone consistently had no effect on VEGF expression, but 1 µM at-RA gradually increased VEGF expression, which resulted in up to 5-fold increase at 72 h after stimulation (Figure 1A). Moreover, subconfluent NCI-H322 cells were stimulated with at-RA (0.1, 1, 2.5, 5, or 10 µM) for 72 h. Whereas NCI-H322 cells stimulated with vehicle alone had no effect on VEGF expression, at-RA increased VEGF expression in a dose-dependent manner (Figure 1B).

View larger version (46K): [in this window] [in a new window]   Figure 1.   Effects of at-RA on VEGF mRNA levels in NCI-H322 cells. (A) Northern blot analysis. Total cellular RNA (20 µg) was prepared from subconfluent NCI-H322 cells that were cultured in the absence (vehicle) or presence of 1 µM at-RA for indicated times, and analyzed by Northern blotting for VEGF mRNA. -Actin indicates that comparable amount of total RNA actually blotted onto a membrane (upper panel). Densitometric analysis of the Northern blots (lower panel). VEGF mRNA levels yielded by scanning the autoradiographs were normalized to the -actin signal. The results are indicated as values relative to VEGF mRNA levels in the control cells. They are expressed as the mean ± SEM. (B and C) Northern blot analysis indicating the dose-dependent effect of at-RA on the VEGF gene expression in NCI-H322 cells (B) and A549 cells (C). Subconfluent cells were incubated in serum-free medium for 24 h before testing. After cells were treated for 72 h with increasing doses of at-RA, ranging from 0-10 µM, total RNA (20 µg/lane) was extracted and analyzed by Northern blots. The membranes were hybridized with the VEGF and -actin cDNA probes (upper panel). Densitometric analysis of the Northern blots. Quantification of VEGF mRNA levels was performed as described above (lower panel).

Next, to determine whether an increase in VEGF expression by at-RA is not specific to NCI-H322 cells, we examined the effects of at-RA on the A549 cells, human lung adenocarcinoma cells (Figure 1C). Results showed that the induction of VEGF mRNA by at-RA was comparable between A549 cells and NCI-H322 cells.

at-RA Increases VEGF Protein Production

We then examined the VEGF protein production in response to at-RA in NCI-H322 cells. Subconfluent NCI-H322 cells grown in serum-free medium for 24 h were stimulated with at-RA for 72 h and the supernatants were analyzed by ELISA for VEGF protein production. As shown in Figure 2, VEGF protein levels were increased in proportion to at-RA concentration.

View larger version (9K): [in this window] [in a new window]   Figure 2.   at-RA-induced VEGF protein production in NCI-H322 cells. Subconfluent NCI-H322 cells were stimulated with at-RA, ranging from 0-10 µM, for 72 h and the supernatants were analyzed by enzyme-linked immunosorbent assay for VEGF protein production. VEGF protein production was revised by the amount of the total protein contained in subconfluent NCI-H322 cells. The results are arbitrarily indicated as values relative to VEGF protein levels in the control cells and are the mean ± SEM of three separate experiments in duplicate. *P < 0.05 compared with control cells (n = 6).

Increased VEGF Expression Is Regulated at Transcriptional Level

To determine whether the expression of at-RA-induced VEGF mRNA is regulated at the transcriptional levels, we measured the half-life of VEGF mRNA by performing a standard mRNA decay assay using actinomycin D, which prevented the transcription of the genes. Results showed that the half time of the decrease in VEGF mRNA levels was 2.6 h versus 2.4 h (vehicle versus at-RA), indicating that at-RA did not significantly affect the stability of VEGF transcripts.

We then performed transient transfection of VEGF promoter/luciferase reporter gene into NCI-H322 cells. We chose 1 µM of at-RA for following experiments because previous studies showed that the serum concentration of at-RA up to 3 µmol/liter had no toxicity when rats were dosed orally with at-RA (29, 30) and maximal differentiation of human promyelocytic leukemia cell line (HL-60) occurred at 1 µM at-RA (23, 27). As shown in Figure 3, luciferase activity of -1207Luc, which contains mouse VEGF promoter sequence spanning from -1207 to +372, was significantly increased in response to 1 µM at-RA. Progressive deletion of the 5'-flanking sequence from -1207 to -100 did not abrogate the effects of at-RA on VEGF promoter. However, further deletion to 1 eliminated the ability of at-RA to enhance transcription, thus suggesting that sequence important for positive regulation of VEGF promoter activity by at-RA is located within the region between 100 and -1. These results indicate that at-RA increases VEGF expression at the transcriptional level and the promoter region downstream of -100 bp contains at-RA response element.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Progressive 5'-deletion analysis of at-RA response region. Deletion analysis of retinoic acid responsive element within the VEGF promoter in NCI-H322 cells. NCI-H322 cells were transiently transfected with the indicated constructs. Twenty-four hours after transfection, cells were cultured in the absence (vehicle) or presence of 1 µM at-RA for 72 h. Luciferase activities are expressed relative to that of -1 Luc in the vehicle-treated cells, which is set at 1.0. Values are mean ± SEM. *P < 0.05 compared with control cells (n = 6).

Identification of the Nuclear Factors Binding to VEGF Promoter Downstream of -100

The observations described above led us to determine the sequence elements that can function as potential binding sites for the nuclear factors that are involved in at-RA- mediated VEGF expression. Searching for the putative transcription binding sequence between -100 and -1 revealed that G+C-rich sequence spanning between -81 and -52 matches very closely the consensus sequence of Sp1-binding site. To test the ability of this sequence to interact with nuclear proteins, we performed EMSA with nuclear extracts from NCI-H322 cells and the ³²P-labeled double-stranded oligonucleotides containing the sequence between 81 and 52. As shown in Figure 4, this probe gave rise to three major DNA-protein complexes, designated as C1, C2, and C3. Nuclear extracts from NCI-H322 cells stimulated with at-RA showed the comparable intensity with the nuclear extract from unstimulated cells. All three shifted complexes proved to be sequence-specific because the formation of these complexes was inhibited by excess amount of unlabeled consensus Sp1-binding sequence but not by Egr-1-binding sequence and AP-2-binding sequence. To directly verify that the shifted complexes C1, C2, and C3 contain Sp1 or Sp1-related proteins, we performed EMSA in the presence of Sp1- or Sp3-specific antisera. Addition of the Sp1 antibody resulted in a supershift of a complex C1, indicating that Sp1 is a principal DNA-binding component of this complex. The Sp3 antibody supershifted the complexes C2 and C3. We also tested the effects of Egr-1 antibody on the complex formation. Addition of Egr-1 antibody had no effect on the formation of complexes. These results provided evidence that Sp1 and Sp3, but not Egr-1, bind to the VEGF 81/52 probes, and the binding activity of Sp1 and Sp3 to these G+C-rich sequences was not affected by at-RA treatment.

View larger version (72K): [in this window] [in a new window]   Figure 4.   Electrophoretic mobility shift assays (EMSA) of the retinoic acid response element. An EMSA was performed using a 32P- labeled probe containing the VEGF -81/52 sequence: 5'-CCGTCCCCGGGGCGGGTCTGGGCGGGGCTT-3' and nuclear extracts from NCI-H322 cells treated in the absence (vehicle) or presence of 1 µM at-RA for 30 min. A hundred molar excess of unlabeled oligonucleotides was added as competitors. Self and µ indicate -81/52 and -81/52 µ, respectively. Sp1, Egr-1, and AP-2 indicate Sp1-, Egr-1-, and AP-2-consensus binding sequence, respectively. DNA:protein complexes are indicated by C1, C2, and C3. For supershift assay, nuclear extracts were preincubated with 1 to 2 µl of anti-Egr-1, Sp1, Sp3 polyclonal antibodies. Supershifted complex is indicated by SS.

Mapping of at-RA Response Element to Sp1/Sp3-binding Site

To determine whether Sp1/Sp3 sites are required for at-RA- induced VEGF expression, we made a mutation construct, 81 (Sp1 µ) Luc, in which two Sp1/Sp3 sites located at -73 and -62 were mutated in the context of -81 Luc (Figure 5A). As shown in Figure 5B, although 81 Luc showed the significant induction in response to at-RA, the construct -81 (Sp1 µ) Luc was not responsive to at-RA. Inability of this construct to respond to at-RA was not due to the mutation of the sequence essential for the basal promoter activity, because promoter activity of -81 (Sp1 µ) Luc was significantly higher than that of the promoterless construct pGL2.

View larger version (22K): [in this window] [in a new window]   Figure 5.   Site-specific mutational analysis of the Sp1 site within the mouse VEGF promoter. (A) Sequence of the Sp1 site within the promoter region between 73 and -55 was boxed. Base substitution introduced into Sp1 site was underlined. (B) Effects of mutation in Sp1 site on at-RA-induced VEGF promoter activity. NCI-H322 cells were transiently transfected with either wild type -81 Luc or -81 (Sp1 µ) Luc. Twenty-four hours after transfection, cells were cultured in the absence (vehicle) or presence of 1 µM at-RA for 72 h. Luciferase activities are expressed relative to that of promoterless pGL2 in the vehicle-treated cells, which is set at 1.0. Values are mean ± SEM. *P < 0.05 compared with control cells (n = 6). (C) Effects of mutation in Sp1 site on RAR-induced VEGF promoter activity. NCI-H322 cells were transfected with either wild-type -81 Luc or -81 (Sp1 µ) Luc together with RAR or control vector pRc/CMV for 48 h. Luciferase activities are expressed relative to that of promoterless pGL2 cotransfected with pRc/CMV, which is set at 1.0. Values are mean ± SEM. *P < 0.05 compared with control cells (n = 6).

To verify that an induction of VEGF promoter activity by at-RA is mediated by the activation of RAR, a mammalian expression vector for RAR, which expressed RAR under the control of cytomegalovirus promoter, was cotransfected with either -81 Luc or -81 (Sp1 µ) Luc into NCI-H322 cells. As shown in Figure 5C, ligand-activated RAR rather than unstimulated RAR increased the transcription of the -81 Luc in an Sp1 site-dependent manner. These results are in complete agreement with the results obtained by at-RA treatment shown in Figure 5B.

To test whether at-RA-mediated VEGF expression is observed across the species of the VEGF gene, we used the human VEGF promoter/reporter constructs in the transient transfection assays. As shown in Figure 6B, the human VEGF promoter constructs, 1180 LUC, 480 LUC, and -89 LUC were significantly responsive to at-RA either with or without RAR expression vector. As in the case with mouse VEGF promoter, human VEGF promoter was highly activated by at-RA in the presence of RAR expression vector. Moreover, at-RA increased the human VEGF promoter -89Luc in an Sp1 site-dependent manner (Figure 6B).

View larger version (24K): [in this window] [in a new window]   Figure 6.   Luciferase assays of the human VEGF promoter. (A) Sequence of the Sp1 site within the human VEGF promoter region between 89 and -67. Base substitution introduced into Sp1 site was underlined. (B) Effects of mutation in Sp1 site on at-RA-induced VEGF promoter activity. NCI-H322 cells were transiently transfected with -1180 Luc, -480 Luc, -89 Luc, or -89 (Sp1 µ) Luc, along with either RAR expression vector (PAR/CMX) or control vector (pRc/CMV). Twenty-four hours after transfection, cells were cultured in the absence (vehicle) or presence of 1 µM at-RA for 72 h. Luciferase activities are expressed relative to that of promoterless pGL3 in the vehicle-treated cells, which is set at 1.0. Values are mean ± SEM. *P < 0.05 compared with control cells (n = 6).

Effects of Protein Synthesis Inhibitor on at-RA-Induced VEGF Expression

We attempted to investigate whether de novo protein synthesis was required for at-RA-induced VEGF expression. At 1 h before the stimulation of NCI-H322 cells with 1 µM at-RA, we added 40 µg/ml of cycloheximide and harvested the cells after 48 h. As shown in Figure 7A, the increase in VEGF mRNA levels by at-RA was completely blocked in the presence of cycloheximide, suggesting that de novo protein synthesis is required for at-RA-induced VEGF mRNA expression.

View larger version (33K): [in this window] [in a new window]   Figure 7.   Effects of cycloheximide and genistein on at-RA-induced VEGF expression. Northern blot analysis of NCI-H322 cells. Subconfluent NCI-H322 cells were incubated in serum-free medium for 24 h before testing. Cells were treated with either 40 µg/ml of cycloheximide (A) or 100 µmol/liter of genistein (C), at 1 h before the treatment with 1 µM at-RA for 48 h. Total cellular RNA (20 µg) was prepared from NCI-H322 cells and the membrane was hybridized with VEGF cDNA. -Actin indicates that comparable amount of total RNA actually blotted onto a membrane. Panels B and D indicate densitometric analysis of the Northern blots of A and C, respectively. VEGF mRNA levels yielded by scanning the autoradiographs were normalized to the -actin signal. The results are indicated as values relative to VEGF mRNA levels in the control cells. They are expressed as the mean ± SEM.

Effects of Tyrosine Kinase Inhibitors on at-RA-Induced VEGF Expression

Because the activation of the tyrosine kinases has been implicated in the at-RA-induced gene expression, we assessed the effects of genistein, a nonselective tyrosine kinase inhibitor, on the induction of the VEGF mRNA expression by at-RA. We added 100 µM of genistein in subconfluent NCI-H322 cells at 1 h before the treatment with at-RA. at-RA-induced VEGF expression was completely blocked by genistein (Figure 6C), suggesting that genistein-sensitive tyrosine kinases play an important role in at-RA-induced VEGF expression.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Retinoids have been implicated in the control of growth and differentiation of a variety of tumors and transformed cells by directly regulating growth factor expression or by altering the response and sensitivity of cells to growth factors. A number of studies have shown that retinoids exhibit antimitogenic activity at least partly through the activation of p21 (31) or by enhanced degradation of cyclin D1 (32). Although the precise molecular mechanism underlying the hypoproliferative activity has not yet been fully understood, the ability of retinoids to inhibit cellular proliferation resulted in their clinical use in the management of selected human cancers, as well as hyperproliferative disorders of the epidermis (e.g., psoriasis) (33). These considerations led us to hypothesize that at-RA may block tumor progression by inhibiting the expression of VEGF, a critical regulator of the tumor angiogeneis. However, the results of the present study indicated that at-RA induced the VEGF gene expression in two distinct lung cancer cells, NCI-H322 cells and A549 cells.

The finding that at-RA increased the expression of mitogenic growth factor VEGF was not expected but not surprising because it is well accepted that at-RA, in some circumstances, stimulates the synthesis of growth factors (34). Induced expression of VEGF was not associated with increased mitogenesis because no significant change was observed in thymidine incorporation rate in response to at-RA at the concentration of < 1 µM at-RA (data not shown). Failure of NCI-H322 cells to respond to VEGF is probably due to the absence of kinase domain-containing receptor, which is known to mediate the mitogenic effects of VEGF. In this regard, our finding of at-RA-induced VEGF production appears to be compatible with the well-accepted antiproliferative role of at-RA. However, the increased VEGF production by at-RA may lead to the stimulation of migration and proliferation of endothelial cells through a paracrine control.

Interestingly, we showed that at-RA activated VEGF transcription through Sp1-binding sites located at -73 and -62 of the VEGF promoter, as shown in Figure 5. Moreover, cotransfection of RAR expression vector enhanced the effects of at-RA on the wild type but not Sp1 site-mutated VEGF promoter, indicating that ligand-activated RAR increased VEGF transcription in an Sp1 site-dependent manner. It is not likely that RAR transactivates promoter through direct binding to Sp1 binding site because our gel mobility shift assays showed that shifted complex is competed by consensus Sp1 sequence but not by consensus RARE sequence (data not shown). There is a precedent report indicating that ligand-activated RAR may regulate the gene expression through modulating the function of other transcription factors in addition to direct binding to DNA. This process, termed cross-coupling, has been described as the mechanism underlying the inhibition of AP-1-dependent transcription by at-RA (35). In this process, RAR does not bind to DNA but rather appears to interact with c-jun and likely with additional proteins as well. In this regard, it is possible that transactivating function of Sp1 was increased by direct/indirect interaction with RAR. Such a model is supported by recent reports in which at-RA induces the binding activity of the Sp1 to its cognate binding sequence within the urokinase-type plasminogen activator (uPA) and tissue-type plasminogen activator (tPA) genes (36, 37). Those studies have demonstrated the physical interaction between RAR and Sp1. However, in contrast to their data, no increase in the Sp1 binding activity was observed in response to at-RA in NCI-H322 cells.

It should be noted that at-RA slowly induces VEGF mRNA expression; induction of VEGF is not detected until 24 h after at-RA treatment. This raised the possibility that at-RA induced the expression of a select subset of genes through RAR-RXR heterodimer and these gene products were able to increase Sp1-dependent transcription. In support of this hypothesis, results of our experiments using cycloheximide and genistein indicate that at-RA-induced VEGF expression requires new protein synthesis and tyrosine kinase activation. One of the candidate genes that are primarily activated by RAR is a signal transducer and activator of transcription 1 (STAT1) as described by Shang and colleagues (38). They demonstrated that at-RA induced STAT1 expression and its phosphorylation in MCF-7 cells independent of the activation of JAK1, JAK2, and Tyk2, which are well-known kinases responsible for STAT1 phosphorylation. In our system, we observed that STAT1 expression and its phosphorylation were not altered in at-RA-treated NCI-H322 cells (data not shown). However, the identification of the signaling molecules containing the tyrosine resides should be important to fully understand the mechanisms by which at-RA regulates the VEGF gene expression.

Weninger and coworkers indicated that retinoids downregulated VEGF production by normal human keratinocytes (39). But they also showed that at-RA increased the VEGF protein production of the supernatants in A431 (the epidermoid cancer cell line). Moreover, Diaz and associates indicated that retinoids downregulated the VEGF expression in human keratinocytes (40). The molecular mechanisms underlying the cell-type-dependent regulation of the VEGF expression by at-RA remain to be determined. It is intriguing to speculate that the difference in the expression levels of tyrosine kinases or their substrates may partly account for the different effects of at-RA between cancer cells and normal cells.

In conclusion, we have examined the ability of at-RA to regulate the expression of VEGF in NCI-H322 bronchioloalveolar carcinoma cells. Our studies demonstrate that at-RA induces VEGF gene expression by increasing Sp1-dependent transcription through a pathway involving de novo protein synthesis and tyrosine phosphorylation (Figure 8). As VEGF plays a critical role in tumor angiogenesis, our data underscore the importance of the careful evaluation of effects of at-RA on tumor progression in vivo.

View larger version (15K): [in this window] [in a new window]   Figure 8.   Model of the at-RA-induced VEGF expression. Upon stimulating the cells with at-RA, ligand-activated RAR induces the transcription of the unidentified gene encoding signaling molecule, which is capable of enhancing the transactivating function of the nuclear factors binding to the Sp1 sites. As a consequence, the transcription of the VEGF gene is increased, and the yielded VEGF promotes tumor angiogenesis. Genistein- sensitive tyrosine kinases may play a role at certain step in this process.