All-trans-retinoic acid inhibits Jun N-terminal kinase-dependent signaling pathways.

Retinoids, including retinol and retinoic acid derivatives, inhibit the growth of normal human bronchial epithelial (HBE) cells. The signaling pathways through which retinoids mediate this effect have not been defined. Normal HBE cell growth is stimulated by treatment with a variety of growth factors that increase mitogen-activated protein (MAP) activity. In this study, we examined MAP kinase-dependent pathways as potential targets of retinoid signaling and the role of MAP kinases in retinoid-induced c-fos gene regulation. All-trans-retinoic acid (t-RA) inhibited Jun N-terminal kinase (JNK) and, to a lesser extent, extracellular signal-regulated kinase activity in normal HBE cells. t-RA reduced c-fos mRNA and protein levels by decreasing c-fos gene transcription. The c-fos promoter was activated by co-transfection with a constitutively active JNK kinase (SEK)-1 and suppressed by a dominant negative JNK kinase kinase (MEKK)-1. Furthermore, c-fos expression was inhibited by agonists of retinoic acid receptors (RARs) or retinoid X receptors (RXRs), and suppression of c-fos promoter activity by t-RA was abrogated by treatment with antagonists of RAR-alpha or of all the RXRs. These findings provide the first evidence that t-RA inhibits JNK activity and demonstrate a potential role of JNK-dependent pathways in the suppression of c-fos expression by t-RA. Furthermore, c-fos expression was inhibited through activation of RAR- and RXR-dependent signaling pathways. In light of the growth activation induced by JNK/SEK-dependent pathways in a variety of cells, these data support further investigation into the role of JNK-dependent signaling in the growth-suppressive effects of retinoids.

Retinoids control normal tracheobronchial epithelial growth and differentiation. Rodents that are deprived of vitamin A develop squamous metaplasia in the tracheobronchial epithelium, and normal epithelial differentiation is restored by vitamin A supplementation (1,2). In tissue culture, human bronchial epithelial (HBE) 1 cells undergo squamous differentiation with a variety of agents, and all-trans-retinoic acid (t-RA) in-hibits this process (3)(4)(5)(6)(7). HBE cells treated with t-RA develop mucociliary features in collagen gels (3,8). Grown in monolayer cultures, retinol-treated HBE cells undergo growth arrest with no evidence of morphologic differentiation (9).
The signaling pathways that mediate the biologic effects of retinoids in normal HBE cells have not been defined. Potential candidates are receptor tyrosine kinase-dependent pathways, which stimulate the growth of normal HBE cells (9,18). Previous studies have shown that growth factor-and retinoid-dependent signaling pathways act antagonistically. t-RA inhibits expression of the epidermal growth factor (EGF) receptor in human trophoblast cells and reduces the level of one of its ligands, transforming growth factor-␣, in normal HBE cells and teratocarcinoma cells (19 -21). Overexpression of transforming growth factor-␣ in teratocarcinoma cells antagonizes the growth inhibitory effects of t-RA, and overexpression of the EGF receptor in HL-60 cells blocks the cytodifferentiating effects of t-RA (22,23). The EGF receptor activates signaling through mitogen-activated protein (MAP) kinase cascades, including the p42 and p44 (extracellular signal-regulated kinase; ERK) kinases, the c-Jun N-terminal kinases (JNK)/stress-activated protein (SAP) kinases, and the p38 kinase (reviewed in Refs. 24 and 25). Activation of these kinase pathways phosphorylates Elk-1, which is a component of the ternary complex that includes the serum response factor, SAP-1, and SAP-2, leading to increased Elk-1 DNA binding, ternary complex formation, and transcriptional activation (26). Through this mechanism, EGF treatment activates the Ets/serum response element motif in the c-fos promoter, increasing c-fos expression (27,28).
In this study, we examined MAP kinase-dependent pathways as potential targets of retinoid signaling and the role of MAP kinases in retinoid-induced c-fos gene regulation. t-RA inhibited JNK activity and reduced c-fos mRNA and protein levels by decreasing c-fos gene transcription. The c-fos promoter was activated by JNK-dependent pathway stimulation and inhibited by agonists of RARs or RXRs. These findings provide evidence that t-RA inhibited JNK activity and demonstrate a potential role of JNK-dependent pathways in the suppression of c-fos expression by t-RA. Furthermore, t-RA sup-pressed c-fos expression through activation of RAR-and RXRdependent signaling pathways.

EXPERIMENTAL PROCEDURES
Cells and Culture Conditions-Normal HBE cells were cultured from bronchial mucosal biopsy samples taken from fresh surgical specimens as monolayer cultures on standard plasticware in keratinocyte serumfree medium (Life Technologies, Inc., Gaithersburg, MD) containing bovine pituitary extract and EGF as described previously (29). The MEK-1 inhibitor PD98059 was purchased from Calbiochem. For experiments involving RNA or protein preparation, cells were seeded on 10-cm plates at a density of 10 5 /plate. For transient transfection assays, cells were seeded on 6-well plates at a density of 2 ϫ 10 5 /well. Retinoid treatment was begun the day after cell seeding.
Western Blot Analysis-Whole cell lysates were prepared in MEGA-RIPA buffer as described previously (29). Protein lysate (50 g) was separated by electrophoresis on a SDS-7.5% polyacrylamide gel, transferred onto a BA-S-83-reinforced nitrocellulose membrane (Schleicher & Schuell, Inc., Keene, NH), and immunoblotted overnight at 4°C with a primary monoclonal antibody to c-fos, JNK-1, or ERK-1 and ERK-2 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Binding was detected by using the ECL kit (Amersham, Inc.) according to the manufacturer's directions.
Reporter Plasmids and Expression Vectors-The human c-fos promoter from Ϫ327 to ϩ40 (a PstI-PstI genomic fragment which was a gift from Dr. Jaideep Chaudhary, University of California-San Francisco) was subcloned in sense (fos-LUC) and antisense (sof-LUC) orientation into the pGL3-basic luciferase reporter plasmid (Promega, Inc.). The constitutively active mutant cDNAs for SEK-1 (ED mutant, under the control of the elongation factor promoter) and MEK-1 (R4F mutant, under the control of the CMV promoter) were gifts from Dr. John M. Kyriakis (Harvard Medical School, Boston, MA) and Dr. Natalie G. Ahn (University of Colorado, Boulder, CO), respectively (34,35). The dominant negative mutant cDNA for MEKK-1 under the control of the SV40 promoter was a gift from Dr. F. X. Claret (University of California, San Diego) (36). SV40-driven expression vectors were purchased; these contained the DNA-binding domain of GAL-4 alone (GAL-DBD) or fused to the transcription activation domain of Elk-1 (Elk-GAL-DBD) or c-jun (JUN-GAL-DBD) (Stratagene, Inc., La Jolla, CA). GAL-UAS-LUC is a luciferase reporter plasmid containing five repeats of the GAL4 binding element (Stratagene).
Transient Transfection Assays-The day after seeding, the cells were transfected with luciferase reporter plasmid (2 g/well) for 6 h using LipofectAMINE (Life Technologies, Inc.). The transfectant solution was removed and the cells were treated with retinoids for 24 h. The cells were subjected to luciferase assays as described previously (29). Luciferase activities were expressed as the means and standard deviations of five identical wells and were normalized to cell number (2 ϫ 10 5 cells/well).
Nuclear Run-on Analysis-Cells were treated with media alone or t-RA (10 Ϫ6 M) for 6 h. Cells were lysed in hypotonic buffer (0.8% Nonidet P-40, 10 mM Hepes pH 7.9, 10 mM KCl, 0.75 M spermidine, 0.15 M spermine, 0.1 mM EDTA, pH 8.0, 0.1 mM EGTA, 10 mM dithiothreitol) using a Dounce homogenizer on ice, and centrifugation at 1,500 rpm. The pellets (nuclei) were resuspended thoroughly in transcription buffer (150 mM KCl, 5 mM MgCl 2 , 1 mM MnCl, 20 mM Hepes pH 7.9, 10% glycerol). Radioactive RNA was synthesized from the nuclei by adding the nucleotide mixture (transcription buffer containing 5 mM dithiothreitol, 1 mM ATP, 1 mM GTP, 1 mM CTP, 250 Ci of [␣-32 P]UTP) and incubating at 25°C for 30 min. The nuclei were centrifuged at 1,500 rpm for 6 min, resuspended in DNase I buffer (2 mM Tris, pH 7.5, 5 mM MgCl 2 ), and treated with 5 l of DNase I at room temperature for 10 min to digest genomic DNA. Nuclear protein was digested by the addition of 5 l of proteinase K, 500 l of 1% SDS, 1 ϫ TE, 10 l of 0.5 M EDTA, and 240 l of 1 ϫ TE and incubated at 40°C for 30 min. RNA was prepared from the nuclear lysates by phenol/chloroform extraction, NH 3 OAc/EtOH precipitation overnight at Ϫ20°C, centrifugation at 14,000 rpm for 10 min at 4°C, resuspension of the pellet in DNase I buffer, and DNase I digestion as described above. Proteinase K treatment was repeated followed by phenol/chloroform extraction, NH 3 OAc/ EtOH precipitation. The RNA pellet was resuspended in water, and incorporation of 32 P was quantitated with a scintillation counter. The c-fos and glyceraldehyde-3-phosphate dehydrogenase cDNAs and the c-fos expression vector with no insert (negative control) were immobilized on nylon membranes (5 g/mm 2 ) by slot-blot and prehybridized (5 ϫ SSPE, 5 ϫ Denhardt's solution, 0.5% SDS, 50% deionized formamide, 200 mg/ml tRNA, and 500 mg/ml single-stranded DNA) at 42°C for 2 h. The RNA was boiled for 3 min, chilled on ice, added to hybridization solution (the same as the prehybridization solution) in sufficient quantity to cover the slot-blot membrane at a concentration of 6.6 ϫ 10 7 cpm/ml, and incubated at 42°C for 48 h. The membranes were washed to a stringency of 0.1 ϫ SSC, 1% SDS at 65°C and autoradiographed.

RESULTS
We examined the effect of t-RA treatment on MAP kinase activity in normal HBE cells. ERK and JNK activity were detected in normal HBE cells, and t-RA inhibited the activity of JNK and, to a lesser extent, ERK (Fig. 1A). Densitometric analysis revealed that at 24 h of t-RA treatment, JNK and ERK activities were 11 and 71%, respectively, of untreated cells. Transient co-transfection experiments were performed using reporter plasmids containing GAL4-response elements (GAL-UAS-LUC) and expression vectors containing the GAL4-DBD fused with the transcriptional activating domain of Elk-1 (Elk-GAL-DBD) or c-jun (Jun-GAL-DBD). GAL-UAS-LUC was activated by Elk-GAL-DBD and Jun-GAL-DBD, and t-RA inhibited 40% of the activity induced by Jun-GAL-DBD but not Elk-GAL-DBD (Fig. 1B). These findings demonstrate a prominent inhibition of JNK activity by t-RA treatment.
ERK-and JNK-dependent signaling pathways activate c-fos expression (27,28). Therefore, we examined changes in c-fos levels with t-RA treatment. c-Fos protein levels were inhibited by t-RA in normal HBE cells ( Fig. 2A). Moreover, t-RA decreased c-fos mRNA levels both time and dose dependently (Fig. 2, B and C). The effect of t-RA on c-fos gene transcription was examined. Nuclear run-on analysis revealed that t-RA decreased c-fos gene transcription (Fig. 3A). Transient trans-fection assays with reporter plasmids containing the c-fos promoter fragment from Ϫ327 to ϩ40 in sense (fos-LUC) or antisense (sof-LUC) orientation demonstrated that t-RA suppressed fos-LUC activity, and this effect was dependent on t-RA dose (Fig. 3B). These findings reveal that t-RA suppressed c-fos gene transcription through proximal c-fos promoter elements.
The role of MAP kinase-dependent pathways in the regulation of c-fos promoter activity was examined. Transient cotransfection assays with fos-LUC and expression vectors containing constitutively active mutant MEK-1 or JNK kinase (SEK)-1 revealed that fos-LUC was minimally activated by MEK-1 and to a greater extent by SEK-1 (Fig. 4A). Treatment of normal HBE cells with the MEK-1 inhibitor PD98059 suppressed ERK activity but had no measurable effect on fos-LUC activity (data not shown). Co-transfection of an expression vector containing a dominant negative mutant JNK kinase kinase (MEKK)-1 inhibited fos-LUC activity (Fig. 4B). Assays of JNK and ERK activities in normal HBE cells transiently transfected with the dominant negative mutant MEKK-1 expression vector revealed selective inhibition of JNK activity (Fig. 4C). These findings support a role for JNK-dependent pathways in the regulation of c-fos promoter activity.
Retinoids induce signaling events through nuclear receptor activation. The potential roles of RAR-and RXR-dependent pathways were investigated in the suppression of c-fos expression. Normal HBE cells express all RAR and RXR gene family members (29). Treatment with RAR-(TTNPB) or RXR (SR11235)-selective agonists inhibited c-fos mRNA levels and activation of the c-fos promoter (Fig. 5, A and B), suggesting that activation of RAR-and RXR-dependent pathways inhibited c-fos expression through suppression of proximal c-fos promoter elements. Furthermore, treatment with antagonists of RAR-␣ (LG100629) or of all the RXRs (LG100754) partially abrogated the inhibition of c-fos promoter activity by t-RA (Fig.  5C), providing evidence that t-RA inhibited c-fos promoter activity through RAR-and RXR-dependent signaling pathways. DISCUSSION In this study, we examined the effect of t-RA treatment on MAP kinase-dependent signaling pathways and the role of these pathways in t-RA-induced c-fos gene regulation. Our findings provide the first evidence that t-RA inhibits JNK and, to a lesser extent, ERK activity. Furthermore, we showed that t-RA suppressed c-fos mRNA and protein levels through a reduction in c-fos gene transcription. Previous studies have shown that t-RA inhibits c-fos gene transcription (37,38), but the signaling pathways through which t-RA mediates this effect have not been examined. The c-fos promoter is activated by both ERK-and JNK-dependent pathways through activation of the ternary complex factor (27,28). Consistent with these findings, we observed activation of the c-fos promoter by both MEKand JNK-dependent pathways. Furthermore, inhibition of JNKdependent pathways by transfection of a dominant negative mutant JNK kinase kinase (MEKK)-1 was sufficient to suppress c-fos promoter activity. These findings support the hypothesis that JNK inhibition contributes to retinoid-induced suppression of c-fos expression. While the suppression of JNK activity was more profound, ERK inhibition may also contribute to the inhibition in c-fos expression in t-RA-treated normal HBE cells.
Retinoid actions are mediated through nuclear receptor activation. RAR-and RXR-selective retinoids have distinct bio-logic effects on cells in tissue culture (39), suggesting that they activate different signaling pathways. Consistent with this notion, RAR-and RXR-selective retinoids can regulate the expression of distinct target genes or have opposing effects on the same target gene (40,41). Interestingly, both RAR-and RXR- selective agonists inhibited c-fos expression in this study, suggesting that RAR-and RXR-dependent pathways share certain signaling events. Supporting this possibility, both RAR-and RXR-selective agonists inhibit the squamous differentiation of HBE cells (42). This appears to be at least partly the result of the suppressive effects of these retinoids on the AP-1 transcription factor, which binds to the AP-1 sites in the promoters of a variety of genes required for the induction of HBE squamous differentiation, activating their expression (42). Furthermore, we found that an RAR-␣ antagonist abrogated the effect of t-RA on the c-fos promoter, supporting a role for RAR-␣ in the inhibition of c-fos expression. In addition to reducing c-fos expression, RAR-␣ also activates the expression of transforming growth factor-␤ and insulin-like growth factor-binding pro-tein-3 in normal HBE cells (43). Unexpectedly, an RXR antagonist that blocks activation of RXR homodimers abrogated the inhibition of c-fos promoter activity by t-RA. t-RA does not transcriptionally activate RXRs (reviewed in Ref. 10). A potential explanation for this finding is that t-RA can undergo stereoisomerization intracellularly to 9-cis-retinoic acid (44) and thus activate RXR-dependent signaling, which the RXR antagonist would block.
Mechanisms by which retinoid receptors suppress c-fos expression have not been elucidated. Retinoid receptors modulate gene expression directly by binding to response elements within gene promoters and indirectly by binding to other transcription factors. The c-fos promoter from Ϫ327 to ϩ40 (45) contains a variety of response elements that regulate its activ- ity, including binding sites for cAMP response element-binding protein, STAT factors, YY-1, AP-1, and the ternary complex, but there are no consensus retinoid response elements (AG-GTCA repeats separated by 1, 2, or 5 nucleotides), suggesting that retinoid receptors do not bind directly to this portion of the c-fos promoter. We demonstrated a potential role for JNK-dependent pathways in the suppression of c-fos expression by t-RA, supporting an indirect interaction of retinoid receptors with the c-fos promoter through JNK-dependent signaling pathways. Interactions between retinoid receptors and MAP kinase pathways have been not described. The estrogen receptor is phosphorylated through MAP kinase-dependent pathways, enhancing estrogen receptor transcriptional activity (46), which provides a precedent for investigating interactions between retinoid receptors and components of MAP kinase signaling pathways.
Normal HBE cells secrete a variety of peptide growth factors that stimulate growth through autocrine and paracrine mechanisms (9,18). It is possible that the growth of normal HBE cells is dependent upon MAP kinase activation by endogenous and exogenous growth factors. This notion is supported by the observation that withdrawal of EGF or bovine pituitary extract, which contains a number of peptide growth factors, from the media attenuates the growth of normal HBE cells (9). We and others previously showed that treatment with t-RA inhibits normal HBE cell growth, and interruption of an autocrine growth stimulus contributed to the growth inhibition induced by t-RA (9,29). Inhibition of JNK or other components of growth factor-dependent signaling pathways may be an important mechanism by which retinoids inhibit the growth of normal HBE cells, and future work will examine this possibility.