INTRODUCTION

Retinoic acid (RA)¹ exerts profound effects on proliferation and differentiation, and has immunomodulatory properties. RA binds to nuclear receptors, the retinoic acid receptors, and retinoid X receptors (RAR,, and RXR,,) (1, 2, 3, 4, 5, 6), which belong to the superfamily of steroid and thyroid hormone receptors, including steroid hormone receptors, thyroid hormone receptors, vitamin D receptor and orphan receptors with as yet unknown ligands. These receptors are able to bind specific DNA sequences and act as ligand-inducible transcription factors. RARs, thyroid hormone receptors, and vitamin D receptor require the promiscuous partner RXR to form receptor heterodimers (6). These heterodimers interact with specific response elements (hormone response element) containing hexanucleotide receptor-binding half-sites which are organized as inverted repeats or direct repeats spaced by various numbers of nucleotides. A number of genes containing such hormone response elements in their regulatory regions have been identified (7).

Other mechanisms by which RARs can regulate gene expression include direct protein-protein interaction with other non-related transcription factors. RA can act as a negative regulator of AP-1-responsive genes by interaction of RARs with c-Jun, a member of the AP-1 transcription factor complex, resulting in prevention of AP-1 binding to its response element, as demonstrated in the case of the collagenase promoter (8). Another RA responsive gene is interleukin-2 (IL-2), which is down-regulated by RA in the Jurkat T cell leukemia line. Felli et al. (9) demonstrated that an octamer motif between positions 96 and 66 constitutes one target of the RAR-mediated inhibition of protein kinase C- and calcium-mediated activation of the IL-2 enhancer. This motif binds the inducible nuclear factor(s) OAP⁴⁰, consisting of the Oct-1 and Oct-2 trans-acting factors and an AP-1 complex. The AP-1 complex appeared to be the target of the inhibitory action of RAR on this sequence motif (9, 10).

Some genes do not contain regulatory sequences able to interact with RARs but can be regulated by RA, via events secondary to activation and induction of transcription factors by RA. There is evidence that retinoid-induced differentiation is accompanied in many systems by dramatic changes in the expression of transcription factors and cellular responses to growth factors and cytokines. RA-mediated induction of differentiation in embryonal carcinoma cells could be attributed to such changes. RA treatment of NTera-2 (NT-2) embryonal carcinoma cells has been shown to induce the transcription factor AP-2 (11). Another transcription factor complex induced by RA in this cell line is the NF-B complex, an inducible transcription factor initially identified as a heterodimer consisting of a 50-kDa subunit (p50; NFKB1) and a 65-kDa subunit (p65; RelA), which is released from its cytoplasmic inhibitor IB upon activation (12). Segars et al. (13) showed that induction of major histocompatibility class I genes by RA in NT-2-cells involves activation of RAR-RXR heterodimers and induction of NF-B p50-p65 (13). The RA-mediated activation of p50 and p65 which they observed is accompanied by an increase in p50 and p65 mRNA levels and is quite distinct from the previously documented activation of NF-B by a series of post-translational changes (12).

RA has been shown to regulate various genes which do not contain classical RA response elements (14), for instance, the chemotactic factor IL-8. RA has been shown to up-regulate IL-8 expression in some cell types, including fibroblasts (15), neuroblastoma cells (16), and a human ovarian carcinoma cell line (17). In the human acute promyelocytic leukemia cell line HL-60, RA itself did not up-regulate IL-8 mRNA but potentiated the 12-O-tetradecanoylphorbol-13-acetate induced stimulation of IL-8 transcripts (18). However, the exact mechanism by which RA can regulate expression of IL-8 is still unclear.

IL-8 is a member of the C-X-C subfamily of chemokines (19, 20) and is produced by a variety of cell types, including monocytes/macrophages, fibroblasts, endothelial cells, synovial cells, keratinocytes, epithelial cells, and various tumor cells (21, 22, 23, 24). Classical inducers of IL-8 are inflammatory stimuli, such as IL-1, tumor necrosis factor (TNF) and also bacterial lipopolysaccharides, 12-O-tetradecanoylphorbol-13-acetate, viruses, and double-stranded RNA (25). IL-8 is regulated both at the transcriptional and post-transcriptional level. In the 3-flanking region, the IL-8 gene contains the repetitive ATTTA motif, which is responsible for destabilization of various cytokine mRNAs (26). Within the 5-flanking region representing the IL-8 promoter, the gene contains potential binding sites for the transcription factors AP-1, AP-2, AP-3, HSE, HNF-1, IRF-1, glucocorticoid receptor, NF-B, NF-IL-6, and the octamer factor (27). It has been suggested that the AP-1 and octamer binding motifs are dispensable for IL-8 gene activation, while the NF-B and NF-IL-6 binding sites appear to be sufficient (28). Recent studies now indicate that the region spanning the nucleotides 94 to 70 relative to the transcription start site of the IL-8 gene is essential for both induction and repression by certain stimuli (28, 29, 30, 31), mediated mainly by the transcription factor complexes NF-IL-6 and NF-B. NF-IL-6 was originally identified as an IL-1-induced transactivator of the human IL-6 gene and belongs to the CCAAT/enhancer-binding protein (C/EBP) family of transcription factors (32). Several members of the NF-B family, such as p50 (NFKB1), p65 (RelA), c-Rel, and p52 (NFKB2), have been shown to bind the NF-B motif of the IL-8 promoter. While p50, p65, and c-Rel are able to bind these sequences efficiently, NF-IL-6 binds to its own binding site very weakly. However, binding of NF-B to its site results in strong cooperative binding of NF-IL-6 to the adjacent site (29).

In this report we demonstrate that all-trans retinoic acid (ATRA) is able to activate the promoter of the IL-8 gene in a human melanoma cell line and shows synergistic effects when combined with the pro-inflammatory cytokine TNF- in a dose- and time-dependent manner. The synergistic induction of the IL-8 promoter by ATRA and TNF- requires an intact NF-B binding site and is coupled to changes in the composition of NF-B complexes bound to the cognate site of the IL-8 promoter. Moreover, we suggest that ATRA sensitizes the cells to treatment with TNF-, via removal and/or inactivation of an inhibitory factor(s), an event which allows enhanced binding of NF-B.

EXPERIMENTAL PROCEDURES

All-trans-retinoic acid was obtained from Sigma. 9-cis-Retinoic acid and 4-[2-(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)-1-propenyl]benzoic acid (TTNPB) were provided by J. Eliason (Nippon Roche Research Center, Japan). Retinoids were dissolved in dimethyl sulfoxide as a 10 mM stock solution and stored under light protection at 70 °C. Recombinant human TNF- (specific activity: 2 × 10⁷ units/mg) and recombinant human IFN-g (specific activity: 1 × 10⁶ units/mg) were purchased from Genzyme (Cambridge, MA). Recombinant human IL-1 was kindly supplied by P. Ramage (Sandoz Pharma AG, Basel, Switzerland).

The 5-flanking sequences of the IL-8 gene were generated by polymerase chain reaction from a human genomic DNA library using primers derived from published sequences (27). The full-length promoter of the IL-8 gene, NAP (nucleotides 1470 to +40), and the truncated promoter, NAPs (nucleotides 101 to +40), were cloned upstream of the firefly luciferase coding region in the pGL2-basic vector (Promega, Madison, WI). For stable transfection of the human melanoma cell line G-361 with the truncated IL-8 promoter construct NAPs, a neomycin resistance gene was introduced into the BamHI-SalI sites of the pGL2-basic vector. The promoter construct containing a mutated NF-B binding site was derived from the NAPs construct by polymerase chain reaction mutagenesis and subcloned into the pGL2-basic vector. The NF-B binding site (⁸⁰GGAATTTCCTC⁷⁰) was altered to GGccggTCCTC. All constructs were confirmed by sequencing using the dideoxy chain termination method (Sequenase Kit version 2.0; U. S. Biochemical Corp., Cleveland, OH).

The human melanoma cell line G-361 was purchased from ATCC (Rockville, MD). The cell line A3 was generated by transfection with the NAPs promoter-luciferase construct, and selection and subsequent growth in a medium containing 350 µg/ml of the antibiotic G418 (Geneticin; Life Technologies, Inc., Scotland). Both cell lines were cultivated in minimal essential medium supplemented with 5% fetal calf serum. All cells were incubated at 37 °C in a humidified 5% CO₂ atmosphere. Tests for mycoplasma contamination were negative. Transient transfections of G-361 cells using lipofectamine reagent (Life Technologies, Inc.) were performed as described previously (33).

For evaluation of luciferase activity, cells were seeded at a density of 5 × 10⁴ cells/well of a 48-well plate and allowed to attach overnight. On the next day medium was aspirated and replaced by minimal essential medium supplemented with 1% fetal calf serum and the individual stimuli. After exposure to the stimuli for various time periods, cells were lysed and extracts were evaluated using luciferase assay reagent (Promega) in a scintillation counter.

For EMSA experiments the following double-stranded oligonucleotides were used: wild type (IL-8), 5-AATTCCAGTTGCAAATCGTGGAATTTCCTG-3; NF-B mutated (IL-8), 5- AATTCCAGTTGCAAATCGTGGCCGGTCCTG-3; nonspecific oligonucleotide (retinoic acid response element, RARE), 5-AATTCGATCCGCTAGCAAGGGTTCACCGAAAGTTCACTCGCATA-3. The oligonucleotides, when annealed with their complement, generated EcoRI-compatible ends and were labeled with Klenow polymerase using [³²P]dATP, and purified chromatographically.

Nuclear extracts were prepared according to Andrews and Faller (34) with modifications. 5 × 10⁶ cells were seeded in 12 ml of minimal essential medium supplemented with 1% fetal calf serum in a 100-mm tissue culture dish and allowed to attach overnight. On the next day, cells were treated with the individual stimuli for various time periods. Then medium was aspirated and cells were scraped and suspended in 1.5 ml of ice-cold phosphate-buffered saline. After 10 s centrifugation in a microcentrifuge, phosphate-buffered saline was aspirated completely. Cells were suspended in 400 µl of ice-cold buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.150 mM MgCl₂, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and allowed to swell on ice for 10 min. Then 25 µl of a 10% Nonidet P-40 solution were added and cells were lysed by vortexing for 10 s. After 30 s centrifugation in a microcentrifuge, the buffer was completely aspirated and the pelleted nuclei suspended in 25 µl of buffer C (20 mM HEPES, 20% glycerol, 0.420 M NaCl, 0.150 mM MgCl₂, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and incubated on ice with shaking for 15 min. After centrifugation for 2 min the extract was removed and diluted with 75 µl of buffer D (20 mM HEPES, 20% glycerol, 50 mM KCl, 0.5 mM EDTA, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride) and stored at 70 °C. For EMSA approximately 5 µg of nuclear extract was used for each binding reaction. Binding reactions were carried out at room temperature for 15 min in a 20-µl binding reaction mixture, containing 20 mM HEPES, 50 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 5% glycerol, 1 µg of poly(dI·dC) and 0.4 ng of labeled oligonucleotide. For competition assays, 100-fold molar excess of unlabeled oligonucleotide was added to the binding reaction 10 min before addition of the radiolabeled oligonucleotide. Were indicated, antibodies were added to the binding reaction before the radioactive probe and poly(dI·dC), and incubated for 25 min at room temperature. The unbound oligonucleotides and protein-DNA complexes were separated on a 5% polyacrylamide gel in 0.5 × TBE at a constant current of 10 mA. The gels were then dried at 80 °C for 1 h and exposed to Kodak X-Omat AR-5 at 70 °C.

Anti-p65, anti-p50, anti-p52, anti-c-Rel, anti-RelB, and anti-Sp-1 antibodies were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA).

One million cells were seeded into each well of a 6-well plate and allowed to attach overnight. The next day, medium was aspirated and replaced by minimal essential medium supplemented with 1% fetal calf serum and the individual stimuli. Total RNA was isolated by a single-step guanidinium isothiocyanate procedure (35). RNA was separated on a 1% agarose gel containing 1.3% formaldehyde. The gels were washed twice with 10 × SSC for 20 min, and RNA was then transferred to nylon membranes (Schleicher & Schuell, Dassel, Germany) by capillary transfer in 10 × SSC. After fixing the RNA to the membrane by baking at 80 °C for 2 h, hybridization was carried out in rapid hybridization buffer (Amersham, Bucks, UK) with ³²P-labeled probes for 12 h at 65 °C. After washing twice with 2 × SSC, 0.1% SDS for 5 min and twice with 0.1 × SSC, 0.1% SDS at 65 °C for 15 min, membranes were exposed to Kodak X-Omat AR-5 at 70 °C.

IL-8 was detected using a 350-base pair cDNA fragment containing the IL-8 coding region. IB was detected using a 0.9-kilobase fragment of the porcine IB cDNA (36). p65 was detected using a 1.8-kilobase HindIII-XbaI fragment of the human p65 cDNA (CMV 4T p65) kindly provided by W. C. Greene (Gladstone Institute, San Francisco). Glyceraldehyde-3-phosphate dehydrogenase as detected by a EcoRI-HindIII fragment of the rat glyceraldehyde-3-phosphate dehydrogenase cDNA from P. Amstad (ISREC, Lausanne, Switzerland).

RESULTS

The cell line A3 was derived from the human melanoma cell line G-361 by stable transfection with a reporter gene construct, representing a truncated IL-8 promoter fused to the firefly luciferase coding region. The IL-8 promoter construct, designated NAPs, contains nucleotides 101 to +40 relative to the transcription start of the IL-8 gene, and thus contains the intact binding sites for the transcription factors NF-IL-6 and NF-B, but lacks the AP-1 binding site (33).

We investigated the effects of the retinoids ATRA, 9-cis-retinoic acid, and the arotinoid TTNPB on expression of the reporter gene. Stimulation of A3 cells for 24 h with either 5 µM ATRA or 9-cis-RA induced 5-7-fold enhanced luciferase expression, whereas the RAR-specific arotinoid TTNPB (5 µM) had less effect, activating the IL-8 promoter by only 1.5-2-fold. The effects of pro-inflammatory cytokines were also studied in this cell line. While stimulation with 1 ng/ml IL-1 for 24 h induced a 4-7-fold enhanced luciferase expression, treatment with 1 ng/ml TNF- for the same time period caused a 50-fold activation of reporter gene expression. Stimulation of cells with IFN- (250 units/ml) did not induce any significant promoter activity (Table I).

Table I. Activation of the IL-8 promoter by retinoids and inflammatory cytokines in A3 cells A3 cells were treated with various retinoids and inflammatory cytokines for 24 h. Luciferase values were obtained from cell lysates as described under ``Experimental Procedures.'' Relative luciferase activities compared to untreated control cells from one representative experiment from at least four individual experiments are shown. Stimulus Relative luciferase activity All-trans-retinoic acid (5 µM) 6.4  ± 0.8 9-cis-Retinoic acid (5 µM) 5.7  ± 0.2 TTNPB (5 µM) 1.7  ± 0.2 IL-1 (1 ng/ml) 5.0  ± 1.2 TNF- (1 ng/ml) 52.0  ± 5.4 IFN- (250 units/ml) 1.0  ± 0.2

ATRA and TNF- Show Dose-dependent Synergy in Their Activation of the IL-8 Promoter

To investigate IL-8 promoter activation by ATRA, either alone or in combination with TNF-, A3 cells were exposed to various doses of both stimuli for 24 h. Treatment of A3 cells with 0.1-5 µM ATRA enhanced promoter activity 2-3-fold at 1 µM and 5-7-fold at 5 µM. Exposure of A3 cells to 0.1-10 ng/ml TNF- resulted in enhanced luciferase expression, even at the lowest concentration of 0.1 ng/ml (8-10-fold), which increased markedly at concentrations of 1 and 10 ng/ml TNF- (50- and 100-fold, respectively). Combination of 0.1-10 ng/ml TNF- with 1 or 5 µM ATRA resulted in a synergistic promoter activation, with a more than 300-fold enhanced expression of the reporter gene after stimulation with 10 ng/ml TNF- and 5 µM ATRA (Fig. 1).

ATRA and TNF- show dose-dependent synergy in their activation of the IL-8 promoter.

ATRA and TNF- Activate the IL-8 Promoter in a Time-dependent Manner

A time course was performed to further evaluate the synergistic IL-8 promoter activation by ATRA and TNF-. A3 cells were stimulated with 5 µM ATRA alone or in combination with 10 ng/ml TNF- for various lengths of time. Treatment of A3 cells with ATRA for 4 h had no effect on reporter gene expression, whereas simultaneous stimulation with TNF- and ATRA for the same time period resulted in significantly enhanced luciferase expression compared to cells treated with TNF- only. This effect became more obvious after 8 h treatment, where no enhanced luciferase expression was observed by stimulation with ATRA alone, but a pronounced synergistic enhancement of promoter activity was seen with the combination of ATRA and TNF- (35-fold) compared to cells stimulated with TNF- alone (12-fold). After 16 h, a small increase in luciferase expression was observed in cells treated with ATRA alone (2-3-fold). However, at this time point, a greater than 200-fold, synergistic activation of the IL-8 promoter was induced by combined stimulation with ATRA and TNF-. Between 16 and 24 h a continuous increase in the activation of the IL-8 promoter by all treatments was observed (Fig. 2).

ATRA and TNF- activate the IL-8 promoter in a time-dependent manner.

IL-8 mRNA Is Up-regulated by ATRA and TNF- in A3 Cells

Northern analysis was performed to investigate whether ATRA and TNF- are able to increase levels of IL-8 mRNA in A3 cells. Cells were treated with 5 µM ATRA, 10 ng/ml TNF-, or a combination of ATRA and TNF- for various lengths of time before isolation of RNA. Hybridization with ³²P-labeled IL-8 cDNA demonstrated that treatment of cells with ATRA induced only slight expression of IL-8 specific transcripts. TNF- induced expression of IL-8 transcripts after 2 h, increasing up to 24 h. Simultaneous treatment of cells with ATRA and TNF- resulted in a clear synergistic induction of IL-8 transcripts, particularly marked at 16 and 24 h (Fig. 3).

IL-8 mRNA is up-regulated by ATRA and TNF- and synergistically by ATRA and TNF- in A3 cells.

The NF-B Binding Site of the IL-8 Promoter Is Essential for Synergistic Activation by ATRA and TNF-

Since A3 cells are not themselves suitable for evaluation of the binding sites in the IL-8 promoter essential for activation by ATRA and TNF-, the parental cell line G-361 was used to study the effects of ATRA and TNF- on various IL-8 promoter constructs. G-361 cells were transiently transfected with reporter gene constructs, containing either the ``full-length'' promoter (1470 to +40 relative to transcription start; NAP) or the truncated promoter (101 to +40 relative to transcription start; NAPs). 28 h post-transfection, cells were exposed to 5 µM ATRA, 10 ng/ml TNF- or ATRA combined with TNF- for 24 h. Stimulation of cells transfected with the full-length promoter construct resulted in 8-10-fold ATRA-mediated luciferase expression, which was higher than the effect of ATRA on the truncated IL-8 promoter construct (2-3-fold over control). However, treatment of G-361 cells with TNF- resulted in a pronounced activation of luciferase expression in both cases (20-fold for NAP, 30-40-fold for NAPs) and a synergistic activation when TNF- was combined with ATRA (120-150-fold) (Fig. 4).

The NF-B binding site is essential for synergistic activation of the IL-8 promoter by ATRA and TNF-.

To determine whether the NF-B binding site of the IL-8 promoter is essential for this synergistic activation, a mutation was introduced into the NF-B binding site of the truncated IL-8 promoter construct NAPs. Transfection of G-361 cells with this promoter construct followed by stimulation with ATRA slightly enhanced luciferase expression, while TNF- failed to induce significant promoter activity. However, the mutation of the NF-B binding site abolished the synergy between ATRA and TNF- (Fig. 4).

ATRA Does Not Affect Expression of IB Transcripts or Enhance TNF- Mediated Up-regulation of p65 mRNA

To determine whether the expression of IB, the cytoplasmic inhibitor of NF-B, can be negatively regulated by ATRA, Northern blot analysis was performed. Treatment with TNF- clearly induced expression of IB mRNA, and these enhanced levels were not affected by co-treatment with ATRA. Stimulation with ATRA alone had no effect on IB mRNA levels (Fig. 5). The synergistic effect of ATRA and TNF- on IL-8 gene activation is therefore independent of IB expression in this cell line.

ATRA does not affect expression of IB transcripts or enhance TNF- mediated up-regulation of p65 mRNA.

Blots were rehybridized with a p65-specific probe to determine whether p65 induction contributes to the synergy between TNF- and ATRA. While ATRA or TNF- alone up-regulated p65 mRNA levels, ATRA and TNF- together did not induce synergistic expression of p65 transcripts (Fig. 5).

NF-B Binding Activity Is Induced by ATRA and TNF- in A3 Cells

To investigate the nuclear trans-acting factors induced by ATRA and TNF-, EMSAs were performed. Nuclear extracts, prepared from cells stimulated with TNF- or a combination of TNF- with ATRA for various lengths of time, were incubated with a ³²P-labeled oligonucleotide representing the NF-IL-6 and NF-B binding sites of the IL-8 wild type promoter.

Fig. 6a shows that nuclear extracts prepared from cells incubated with 10 ng/ml TNF- for 4, 8, 16, and 24 h contain multiple factors (complex B-D) which bind to the IL-8 wild type promoter sequences. Simultaneous treatment of A3 cells with ATRA and TNF- for 4 and 8 h produced the same pattern as cells stimulated with TNF- alone. In contrast, simultaneous exposure of cells to ATRA and TNF- for 16 and 24 h induced formation of a single, faster migrating complex (complex A). The same complex was also apparent when cells were preexposed to ATRA for 16 h and additionally stimulated with TNF- for 3 h (Fig. 6a). This complex bound specifically to the NF-B binding site of the IL-8 promoter, since competition assays performed with 100-fold molar excess of unlabeled wild type oligonucleotide completely abolished this binding activity. In contrast, binding activity remained when competition was performed with an oligonucleotide containing a mutated NF-B but intact NF-IL-6 binding site (Fig. 6b). Additionally, supershift experiments were performed using antibodies directed against the various NF-B proteins. These antibodies have been previously shown to react with their respective target proteins (37). Anti-p50 partially diminished formation of this complex, while anti-p65 almost completely blocked formation of complex A and induced a clear supershifted band. Addition of anti-p52, anti-c-Rel, anti-RelB, or an irrelevant antibody (anti-Sp-1) had no effect on formation of complex A (Fig. 6c).

NF-B binding activity is induced by ATRA and TNF- in A3 cells. EMSAs using nuclear extracts from A3 cells after stimulation with TNF- (10 ng/ml) or a combination of ATRA (5 µM) and TNF- were performed. EMSA was performed with a ³²P-labeled oligonucleotide representing the NF-IL-6 and NF-B site of the IL-8 promoter. a, cells were treated for 4, 8, 16, and 24 h with 10 ng/ml TNF- (lanes 2-5), for 4, 8, 16, and 24 h with ATRA and TNF- (lanes 6-9), and for 16 h with ATRA followed by 3 h stimulation with TNF- (lane 10) before preparation of nuclear extracts. Unstimulated control is shown in lane 1. b, competition experiments were performed with nuclear extracts prepared from A3 cells stimulated for 16 h with 5 µM ATRA followed by a 3-h exposure to TNF-. 100-fold molar excess of unlabeled oligonucleotide was added to the reaction mixture 10 min before incubation with the radiolabeled wild type IL-8 promoter oligonucleotide. Results shown include control extract (lane 1), stimulated cells (lane 2), competition with the wild type IL-8 promoter oligonucleotide (lane 3), competition with an IL-8 promoter oligonucleotide containing a disrupted NF-B binding site (Bm) (lane 4), and competition with a nonspecific oligonucleotide, representing the RA response element of the RAR promoter (lane 5). Sequences of each oligonucleotide are shown under ``Experimental Procedures.'' c, supershift experiment performed with nuclear extracts prepared from A3 cells stimulated for 16 h with 5 µM ATRA followed by a 3-h exposure to TNF-. Results shown include nuclear extracts from stimulated cells with no antibody added (lane 1), 2 µl of anti-p50 (lane 2), 2 µl of anti-p65 (lane 3), 2 µl of anti-p52 (lane 4), 2 µl of anti-c-Rel (lane 5), 2 µl of anti-RelB (lane 6), and 2 µl of anti-Sp-1 (lane 7). d, competition experiments were performed with nuclear extracts prepared from A3 cells stimulated for 3 h with TNF-. 100-fold molar excess of unlabeled oligonucleotide was added to the reaction mixture 10 min before incubation with the radiolabeled wild type IL-8 promoter oligonucleotide. Results shown are control extract (lane 1), stimulated cells (lane 2), competition with the wild type IL-8 promoter oligonucleotide (lane 3), competition with an IL-8 promoter oligonucleotide containing a disrupted NF-B binding site (Bm) (lane 4), and competition with a nonspecific oligonucleotide, representing the RA response element of the RAR promoter (lane 5). e, supershift experiment performed with nuclear extracts prepared from A3 cells stimulated for 3 h with TNF-. Results shown are nuclear extracts from stimulated cells with no antibody added (lane 1), 2 µl of anti-p50 (lane 2), 2 µl of anti-p65 (lane 3), 2 µl of anti-p52 (lane 4), 2 µl of anti-c-Rel (lane 5), 2 µl of anti-RelB (lane 6), and 2 µl of anti-Sp-1 (lane 7).

The slower migrating complexes (complexes C and D), which were induced by TNF-, disappeared upon 16 h treatment with ATRA. Competition experiments with extracts from cells stimulated with TNF- for 3 h, using excess unlabeled wild type oligonucleotide, abolished binding of complexes C and D, while competition with the oligonucleotide containing the intact NF-IL-6 but mutated NF-B binding site did not affect formation of these complexes (Fig. 6d). To determine whether these complexes also consist of NF-B proteins, supershift experiments were performed. Complex D formation was partially reduced by anti-p65 antibody, while formation of complex C was prevented by both anti-p50 or anti-p65. Complex B formation was only partially reduced by these antibodies. Addition of anti-p50, anti-p52, anti-c-Rel, anti-RelB, or the irrelevant anti-Sp-1 antibody did not alter binding activities (Fig. 6e). These results indicate that p65 participates in the formation of both complexes C and D, which are released from the NF-B binding site by ATRA.

DISCUSSION

We have established the cell line A3, which was derived from the human melanoma cell line G-361 by stable transfection with a reporter gene construct containing a portion of the IL-8 promoter fused to the firefly luciferase coding region, and show that the promoter can be activated by ATRA. Combination of ATRA with TNF- induced synergistic activation of the IL-8 promoter in a dose- and time-dependent manner. These results are also reflected by IL-8 mRNA levels, which are synergistically induced by ATRA and TNF- in this cell line. Using transient transfections of the parental cell line G-361 with various promoter constructs, we established that the synergistic activation of the IL-8 promoter requires an intact IL-8 NF-B binding site. Furthermore, EMSA experiments performed with nuclear extracts from A3 cells demonstrated that stimulation with ATRA and TNF- resulted in changes in the composition of NF-B complexes bound to the IL-8 promoter and enhanced binding of a p50 and p65 containing NF-B complex compared to cells treated with TNF- alone.

Several studies have demonstrated that transcriptional activation of the IL-8 gene by inflammatory cytokines, such as IL-1 and TNF, is mediated through sequences located between nucleotides 94 and 71 of the IL-8 promoter (38, 39, 40, 41). These sequences represent binding sites for members of the NF-B and NF-IL-6 transcription factor families. It has been shown that binding of NF-B factors is essential for activation of IL-8 gene transcription (38, 42), whereas maximal induction of promoter activity is obtained when either an intact AP-1 or NF-IL-6 site is also present, to enable cooperation of NF-B with one of these transcription factors (28, 40). Kunsch et al. (42) demonstrated that maximal transcriptional activation of IL-8 gene expression by phorbol 12-myristate 13-acetate in the human Jurkat T cell leukemia line requires intact binding sites for both NF-B and NF-IL-6 proteins. This cooperation was further confirmed by co-transfection of NF-B p65 and NF-IL-6 expression vectors with an IL-8 reporter gene construct, which resulted in synergistic promoter activation, while co-transfection of NF-IL-6 and p50 expression vectors did not increase reporter gene expression over NF-IL-6 alone (28). Moreover they demonstrated that the NF-B element of the IL-8 promoter contains a site which exclusively binds p65, whereas p50/p50 homodimers or p50/p65 heterodimers are not able to bind to these DNA sequences (42). Stein and Baldwin (29) showed that the NF-B element of the IL-8 promoter does not exclude binding of p50/p50 homodimers and p50/p65 heterodimers and demonstrated that p50 expressed in COS cells has binding properties different from bacterially expressed p50, which itself does not bind the IL-8 NF-B element.

In addition to activation of the IL-8 promoter through these binding sites, repressive effects on IL-8 gene transcription have also been shown to be mediated via these DNA sequences. Stein and Baldwin (29) demonstrated that co-transfection of an IL-8 NF-B promoter-CAT construct and increasing amounts of a C/EBP-/NF-IL-6 expression vector led to a reduction of reporter gene expression at high amounts of expression vector. Another repressive effect, mediated through the NF-B binding site of the IL-8 promoter, was reported by Oliveira et al. (30). Simultaneous treatment of human diploid FS-4 fibroblasts with interferon- and TNF- resulted in a reduction of TNF- induced promoter activity. While interferon- did not influence induction or activity of NF-B binding, EMSA demonstrated that an additional complex, binding to the NF-B element of the IL-8 promoter, occurred after simultaneous treatment with TNF- and interferon-. Further competition and supershift assays showed that this complex might be composed of p65 and other NF-B proteins, such as p50, and members of the C/EBP-/NF-IL-6 family.

To investigate the binding activities of nuclear extracts prepared from A3 cells, we performed EMSAs using an oligonucleotide representing the intact NF-IL-6 and NF-B binding sites. Either simultaneous treatment of A3 cells with ATRA and TNF- for 16 h or pretreatment with ATRA for 16 h and additional stimulation with TNF- for 3 h resulted in the formation of a single complex (complex A), which bound specifically to the NF-B binding site. This complex contains p65 and p50, as demonstrated by supershift experiments. This enhanced binding of NF-B to the IL-8 promoter sequence after stimulation with ATRA and TNF- is consistent with the synergy observed for IL-8 promoter activation and expression of IL-8 transcripts. However, we also observed that exposure of cells to ATRA for more than 16 h resulted in the loss of two binding activities, designated complexes C and D, which probably represent inhibitory activities. These complexes were present in extracts from unstimulated cells and were induced by treatment with TNF- alone for 3 h. The exact nature of these protein complexes cannot be fully established from our experiments, but we observed that both complexes bind specifically to the NF-B binding site of the IL-8 promoter, since competition with an oligonucleotide containing a mutated NF-B but intact NF-IL-6 binding site did not prevent binding of complexes C and D. Supershift experiments demonstrated that complex D is reduced after addition of anti-p65, whereas complex C is abolished after addition of both anti-p50 and anti-p65. Antibodies directed against the other members of the NF-B family did not induce any changes in complex formation at all. Stein and Baldwin (29) demonstrated that nuclear extracts from TNF--stimulated HeLa cells contain three complexes binding to the IL-8 promoter sequences from nucleotides 97 to 69 and showed that the slowest migrating complex consists of p65 and C/EBP-/NF-IL-6. They propose that NF-IL-6, when present in excess over NF-B, can interact with NF-B, resulting in inhibition of NF-B-dependent gene transcription (29). Both complexes (complexes C and D) which were observed in TNF- stimulated A3 cells may represent binding activities similar to those seen in HeLa cells. Thus p65 and another factor, distinct from known members of the NF-B family of transcription factors, may participate in formation of these potential inhibitory activities.

Another regulatory mechanism mediated by the NF-B binding site is the repressive effect of glucocorticoids. Dexamethasone has been shown to inhibit IL-1-induced IL-8 gene expression in the human glioblastoma cell line T98G and diminish formation of NF-B complexes composed of p50 and p65. It has been suggested that this effect is mediated directly by the glucocorticoid receptor, via protein-protein interaction with NF-B (31). The ability of the glucocorticoid receptor to associate physically with the p65 subunit of NF-B was previously demonstrated (43, 44). Recent studies performed on the immunosuppressive effect of glucocorticoids have demonstrated a novel role in their interaction with NF-B. While NF-B activation by TNF- involves degradation of the cytoplasmic inhibitor IB, one essential function of the glucocorticoid dexamethasone has been identified as the transcriptional activation of the IB gene, resulting in an increased rate of IB protein synthesis. In the presence of dexamethasone, NF-B associates with newly synthesized IB, resulting in a reduced rate of NF-B translocation to the nucleus (45, 46). To determine whether ATRA induces the opposite effect in A3 cells by inhibition of IB expression, Northern blot analysis was performed. We did not observe an effect of ATRA on TNF- induced IB mRNA expression, suggesting that ATRA does not influence the regulation of IB. The levels of p65 transcripts induced by TNF- in these cells were not further enhanced by additional treatment with ATRA, excluding the possibility that the synergy between ATRA and TNF- is mediated via enhanced expression of p65.

We conclude that ATRA is able to cause transactivation of the IL-8 gene, which does not contain classical RA response elements in its regulatory regions. In addition we demonstrate that ATRA is able to cooperate with the inflammatory cytokine TNF- to induce a synergistic transactivation of the IL-8 gene, and that this effect is mediated by the transcription factor complex NF-B. Moreover we suggest that the synergistic activation of the IL-8 promoter by ATRA and TNF- could be a two-step process, involving the ATRA-mediated release of potential inhibitory factors from the NF-B binding site and enhancement of TNF--induced p50 and p65 binding to their cognate site. The observed effect of ATRA and TNF- on activation of NF-B demonstrates another level of retinoid-mediated gene regulation which probably plays a role in the regulation of genes other than IL-8. It is therefore of importance to determine whether similar effects occur in the regulation of other RA responsive cytokines, and whether the observed effects are cell type specific.