Inhibition of p90 ribosomal S6 kinase-mediated CCAAT/enhancer-binding protein beta activation and cyclooxygenase-2 expression by salicylate.

We have previously shown that salicylate at a pharmacological concentration suppresses CCAAT/enhancer-binding protein beta (C/EBPbeta) binding, thereby reducing cyclooxygenase-2 (COX-2) and inducible nitric-oxide synthase expression (Saunders, M. A., Sansores-Garcia, L., Gilroy, D. W., and Wu, K. K. (2001) J. Biol. Chem. 276, 18897-18904; Cieslik, K., Zhu, Y., and Wu, K. K. (2002) J. Biol. Chem. 277, 49304-49310). We postulated that salicylate targets a kinase that phosphorylates and activates C/EBPbeta. Here we report the identification of p90 ribosomal S6 kinase (RSK) as a target of salicylate. Salicylate inhibited RSK in vivo and blocked the activity of RSK2 purified from cells stimulated by phorbol 12-myristate 13-acetate (PMA). Mutation of the RSK-phosphorylation site (T266A) of C/EBPbeta abrogated PMA-stimulated C/EBPbeta binding activity. RSK activation was required for PMA-induced COX-2 transcriptional activation. Salicylate also inhibited Ras and extracellular signal-regulated kinase (ERK) activation induced by PMA. We conclude that salicylate inhibits C/EBPbeta-mediated COX-2 transcriptional activation by blocking RSK activity and Ras signaling pathway.

Salicylic acid (SA) 1 is a natural compound produced by diverse plants as a signaling molecule to defend against environmental insults (1,2). Aspirin is synthesized from SA and is deacetylated to generate SA in human-circulating blood (3,4), and both aspirin and SA exert anti-inflammatory and anti-neoplastic actions (5,6). Although the mechanisms by which SA controls various pathophysiological processes are not entirely clear, recent studies have shown that SA controls the activation of pro-inflammatory and pro-proliferative transcrip-tional activators including nuclear factor-B (NF-B), AP-1, cyclic AMP response element-binding protein (CREB), c-Myc, and CCAAT/enhancer-binding protein ␤ (C/EBP␤) (7)(8)(9)(10)(11). The effect of SA on transcriptional activation appears to be concentration-dependent. Its inhibition of NF-B, AP-1, and CREB requires suprapharmacological concentrations generally at 10 -20 mM (7)(8)(9), whereas its inhibition of C/EBP␤ occurs at pharmacological concentrations at 10 Ϫ6 -10 Ϫ4 M (11). The mechanisms by which SA exerts differential concentration-dependent inhibition of transcriptional activators remain to be fully elucidated. It has been shown that SA at suprapharmacological concentrations inhibits IB kinase ␤, thereby blocking IB phosphorylation and dissociation from NF-B (12). As SA at suprapharmacological concentrations has been shown to inhibit a large number of kinases (13), the action on IB kinase ␤ may not be specific. Our previous work has shown that SA at therapeutic concentrations selectively inhibits C/EBP␤ binding to its cognate DNA binding site and does not perturb NF-B activation, suggesting a different mode of action (11). However, the mechanism by which it blocks C/EBP␤ activation remains to be elucidated.
C/EBP␤ mediates transcription of several pro-inflammatory cytokines and mediators (for a review see Ref. 14). Work from our and other laboratories has shown that C/EBP␤ binding to its cognate promoter motif is essential for cyclooxygenase-2 (COX-2) transcriptional activation induced by phorbol 12-myristate 13-acetate (PMA), interleukin-1␤, and lipopolysaccharide (15)(16)(17). Sodium salicylate at 10 Ϫ5 M selectively inhibited C/EBP␤ binding to COX-2 promoter, thereby suppressing COX-2 expression (11). It is unclear how SA suppresses C/EBP␤ binding. C/EBP␤ is a member of the C/EBP family of basic leucine zipper transcription factors (for a review see Ref. 18). It forms homodimers as well as heterodimers with other C/EBP isoforms, and the dimers bind to a specific DNA sequence at the promoter region of the target genes. C/EBP␤ contains an intramolecular autoinhibitory element that hinders its binding to DNA, and phosphorylation at several discrete threonine or serine residues enables C/EBP␤ to bind DNA presumably by releasing the autoinhibitory element (19,20). It has been reported in in vitro experiments that phosphorylation of C/EBP␤ by extracellular signal-regulated kinase (ERK) pathway, protein kinase A, p90 ribosomal S6 kinase (RSK), or CaM-dependent kinase IV is associated with increased C/EBP␤ binding activity (21)(22)(23)(24). The residues phosphorylated by these kinases have been identified to be corresponding to Thr 235 (ERK), Thr 266 (RSK), Ser 288 (protein kinase A), and Ser 325 (CaM-dependent kinase IV) in human C/EBP␤. Little is known about the kinases that activate C/EBP␤ binding activity in response to stimulation by PMA and pro-inflammatory media-tors in vivo, nor is it known whether SA reduces C/EBP␤ binding activity by suppressing any of the kinases. In this study, we tested the hypothesis that PMA stimulates C/EBP␤mediated COX-2 transcriptional activation by inducing one of the kinases that phosphorylate C/EBP␤ and SA blocks the activity of this kinase. To test this hypothesis, we mutated each of the four phosphorylation residues, constructed them into a FLAG expression vector, and determined the binding activity of wild-type (WT) and mutant C/EBP␤ to an authentic C/EBP␤ enhancer element by streptavidin-agarose pull-down and electrophoretic mobility shift assays. The results show that mutation of the RSK phosphorylation residue (T266A) abrogated PMA-stimulated C/EBP␤ binding. Overexpression of a dominant negative (DN) mutant of RSK2 suppressed PMA-induced C/EBP␤ binding and COX-2 transcriptional activation. PMA increased RSK activity, which was inhibited by sodium salicylate in a concentration-dependent manner. The results suggest that SA directly inhibits RSK activity.

EXPERIMENTAL PROCEDURES
Materials-Lipofectin, minimum Eagle's medium, Dulbecco's modified Eagle's medium high glucose medium, Opti-MEM I medium and antibiotics were obtained from Invitrogen. Rabbit polyclonal COX-2 antibodies were obtained from Cayman (Ann Arbor, MI). Antibodies against ERK, phosphorylated ERK, and HA were obtained from Cell Signaling Technology (Beverly, MA). C/EBP␤ antibodies and Protein A/G Plus-agarose were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Enhanced chemiluminescence solution and BCA reagent for protein assay were obtained from Pierce. SDS-PAGE-ready gels (acrylamide:bis 29:1) were obtained from Bio-Rad. Fetal bovine serum, anti-FLAG antibody, PD98059, PMA, and streptavidin immobilized on 4% beaded agarose were obtained from Sigma. Luciferase assay system and T4 polynucleotide kinase were obtained from Promega (Madison, WI). Constitutively active (CA) Ras, DN Ras, pUSE vectors, Ras activation assay kit, and S6 kinase assay kit were purchased from Upstate Biotechnology (Charlottesville, VA). [␥-32 P]ATP was obtained from Amersham Biosciences. Biotinylated oligonucleotides were obtained from Integrated DNA Technologies (Coralville, IA), and non-modified oligonucleotides were obtained from Genosys (Woodlands, TX). GF109230X (bisindolylmaleimide I), rapamycin, AG490, SB203580, and LY294002 were obtained from Calbiochem.
Cell Culture-Human foreskin fibroblasts (H68) obtained from ATCC (Manassas, VA) were cultured in high glucose Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and penicillin/ streptomycin as previously described (11). Cells were washed and cultured in serum-free medium for 24 h before experiment. After washing, they were incubated in fresh serum-free medium in the presence or absence of PMA for 4 h in most of the experiments unless otherwise specified. An identical procedure was applied to human embryonic kidney epithelial cells 293 (ATCC), which were cultured in minimum Eagle's medium.
Plasmids-Human COX-2 promoter fragment (Ϫ891/ϩ9) subcloned into pGL3 luciferase expression vector was prepared as previously described (25). HA-tagged WT RSK2 and K451A mutant constructed into pKH3 as well as pKH3 empty vector were kindly provided by Dr. Jeffrey A. Smith (University of Virginia). The K451A mutation at the C-terminal kinase active site diminished the C-terminal kinase activity. The mutant was reported to retain a basal S6 peptide phosphorylation activity (26) and function as a DN mutant against RSK activation. DN-H-Ras (S17N) and CA-H-Ras (Q61L) constructed into pUSE were purchased from Upstate Biotechnology. Human C/EBP␤ mutants, T235A, T266A, S288A, and S325A, were prepared by PCR-based sitedirected mutagenesis as described previously (27). FLAG-tagged WT and mutant C/EBP␤ were constructed in pCMV-Tag2B. WT C/EBP␤ was a gift of Dr. Philip Auron (Harvard Medical School).
Transfection and Promoter Activity Analysis-Cells at 60% confluence in a 35-mm dish were transfected with 2 g of COX-2 luciferase expression vector using Lipofectin according to a procedure previously described (28). 18 h after transfection, cells were incubated in serum-free medium for 24 h followed by stimulation with PMA for 4 h. Cells were harvested, and luciferase activity was measured in a Turner TD-20/20 luminometer. Each experiment was repeated at least three times.
Western Blot Analysis-Western blotting was performed by a procedure described previously (28). 20 g of proteins was loaded to each lane, separated by 4 -15% gradient SDS-polyacrylamide gel electrophoresis, and transferred to a nitrocellulose membrane. Proteins were detected by enhanced chemiluminescence method using specific antibodies.
Streptavidin-Agarose Pull-down Assay-Binding of nuclear extract proteins to biotinylated probes were assayed by streptavidin pull-down as previously described (29). 400 g of nuclear extract proteins in 600 l of PBSI buffer were incubated with a mixture of 4 g of double-stranded biotinylated 21-bp oligonucleotides containing a human COX-2 promoter C/EBP enhancer element sequence (Ϫ132/Ϫ124, shown underlined), 5Ј-GCTTACGCAATTTTTTTAAGG-3Ј, and 40 l of 4% beadedagarose conjugated with streptavidin for 2 h on a rocking platform at room temperature. The beads were collected by centrifugation at 550 ϫ g for 1 min, washed three times with PBSI, and resuspended in 40 l of Laemmli sample buffer. Nuclear proteins bound to the beads were dissociated by incubating the mixture at 95°C for 5 min and analyzed by Western blotting.
Electrophoretic Mobility Shift Assay (EMSA)-21-bp oligonucleotides containing C/EBP COX-2 promoter enhancer element as above were end-labeled with [␥-32 P]ATP using T4 kinase. EMSA was performed by incubating 5 g of nuclear extract proteins with 20 ng of 32 P-labeled probe in binding buffer (4% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA, 50 mM NaCl, 10 mM Tris, 0.5 mM DTT, and 0.02 mM PMSF) at room temperature for 20 min. The mixture was applied to a 4% polyacrylamide gel and electrophoresed at 175 V for 2 h. DNA-protein complexes were detected by autoradiography.
RSK Assay-RSK activity was determined using an assay kit from Upstate Biotechnology. The test sample was incubated in assay dilution buffer (20 mM MOPS, 25 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM sodium orthovanadate, and 1 mM DTT) containing inhibitor mixture (20 M PKC inhibitor peptide, 2 M protein kinase A inhibitor peptide, and 20 M R24571), magnesium/ATP mixture (75 mM MgCl 2 , 500 M ATP), and S6 substrate peptide (50 M AKRRRLSSLRA) at 30°C. 15 Ci of [␥-32 P]ATP was added and incubated for 20 min. The mixture was spotted on P81 paper and washed with 0.75% phosphoric acid, and radioactive peptides were measured in a scintillation counter (LSC Beckman).
Ras Activation Assay-Ras activation was assayed using a kit from Upstate Biotechnology. Cells were stimulated with PMA for 15 min and lysed with Mg 2ϩ lysis/wash buffer (25 mM Hepes, 150 mM NaCl, 1% Nonidet P-40, 10 mM MgCl 2 , 1 mM EDTA, and 2% glycerol), and cell extract was incubated with Raf-1-conjugated agarose beads for 1 h at 4°C. Agarose beads were pelleted by centrifugation, washed with Mg 2ϩ lysis/wash buffer provided by Upstate Biotechnology, and resuspended in 2ϫ Laemmli reducing sample buffer. Samples were analyzed by Western blotting using an anti-Ras antibody.

T266A C/EBP␤ Mutant Had a Reduced Binding Activity-To
determine whether the C/EBP␤ phosphorylation residues (Thr 235 , Thr 266 , Ser 288 , and Ser 325 ) are involved in PMA-induced C/EBP␤ binding to COX-2 promoter, we mutated each residue individually to alanine, constructed it into a FLAG expression vector, and expressed it in human fibroblasts by transient transfection. Each mutant expressed a similar amount of C/EBP␤ proteins as wild type detected with a C/EBP␤ antibody (Fig. 1A). Major bands representing fulllength and liver-enriched transcription active protein were de-tected in WT-and mutant-transfected cells (Fig. 1A). To evaluate the influence of C/EBP␤ mutation on its binding activity, we treated human fibroblasts that had been transfected with FLAG-tagged WT or mutants with PMA (100 nM) for 4 h and nuclear extracts prepared from the treated cells were incubated with a biotinylated C/EBP probe and streptavidin-agarose beads as described under "Experimental Procedures." FLAG-C/EBP␤s bound to the C/EBP probe were pulled down by centrifugation, and FLAG-WT and FLAG mutant were resolved by Western blot analysis using a FLAG antibody. FLAG-C/EBP␤ was not detected in vector control, and in the absence of PMA stimulation, only trace of FLAG-C/EBP␤ was detected in the complex (Fig. 1B). PMA induced a large increase in FLAG-C/EBP␤ binding (Fig. 1B). When compared with the WT, the T266A C/EBP␤ level in the complex was reduced by ϳ50%, whereas the T235A, S288A, or S325A level was not significantly changed (Fig. 1B). To corroborate the binding data from the streptavidin-agarose pull-down assay, we analyzed C/EBP␤ binding by EMSA. We have previously shown by EMSA and supershift assays with specific antibodies for each isoform of C/EBP that PMA selectively stimulated complex formation between C/EBP␤ and a 32 P-labeled probe containing the authentic C/EBP enhancer element sequence (Ϫ132 to Ϫ124) of COX-2 promoter (11). The complex was abolished by 50-fold excess of unlabeled probes or a mutated probe (11). We repeated the experiments, and the results were similar (data not shown). We next transfected cells with FLAG-WT, FLAG mutants, or FLAG vector control and analyzed protein-DNA complex by gel retardation. The retarded band of T266A was diminished when compared with that of WT and vector control (Fig. 2), suggesting that T266A may be a dominant negative mutant that inhibits the binding of native C/EBP␤ to the probe. Taken together, these results indicate that Thr 266 is required for PMA-induced C/EBP␤ binding.
RSK Activation Was Required for COX-2 Transcriptional Activation-Because Thr 266 is a target of RSK phosphorylation, we reasoned that PMA may induce C/EBP␤ binding and C/EBP␤dependent COX-2 transactivation by activating RSK. Therefore, we evaluated the effect of PMA on RSK activity. Cells transfected with wild-type HA-RSK2 (RSK-WT), mutant HA-RSK2 (RSK-DN), or vector were treated with or without PMA for 15 min. HA-RSK2 was pulled down, and their activities were measured. RSK activity in cells transfected with RSK-WT was increased by PMA by ϳ2-fold (Fig. 3). Consistent with a previous report (26), RSK-DN retained a basal activity but failed to respond to PMA stimulation (Fig. 3). We next assessed the influence of RSK-DN on COX-2 promoter activity. Cells were transfected with RSK-WT, RSK-DN, or vector control, and COX-2 promoter activity was determined by co-transfection of cells with a luciferase expression vector containing a core COX-2 promoter. RSK-WT augmented PMA-induced COX-2 promoter activity, whereas RSK-DN abrogated COX-2 promoter activity stimulated by PMA (Fig. 4A). Corresponding to the promoter data, RSK-WT enhanced PMA-induced C/EBP␤ binding, whereas RSK-DN completely suppressed it (Fig. 4B).
Sodium Salicylate Inhibited RSK Activation-It has been shown that sodium salicylate at pharmacological concentrations inhibits COX-2 transcriptional activation induced by PMA and other pro-inflammatory mediators by blocking C/EBP␤ binding to the C/EBP enhancer element at Ϫ132/Ϫ124 of human COX-2 promoter (11). To determine whether salicylate may target RSK, we treated HA-RSK2-transfected human fibroblasts with sodium salicylate at increasing concentrations for 30 min followed by PMA for 15 min. RSK2 was immunoprecipitated using a HA antibody, and RSK activity of the precipitated RSK2 was assayed. Sodium salicylate inhibited PMA-stimulated RSK activity in a concentration-dependent fashion, and maximal inhibition was noted at 10 Ϫ5 M (Fig. 5A). To provide direct evidence for inhibition of RSK by SA, we pretreated HA-RSK2-transfected cells with or without PMA for 15 min and isolated RSK2 with a HA antibody. After extensive washing, HA-RSK2 was incubated with sodium salicylate (10 Ϫ5 M) for 30 min and RSK activity was measured. PMAstimulated RSK activity was suppressed by SA to an extent comparable with that in vivo (Fig. 5B). These results suggest that SA at pharmacological concentrations inhibit RSK activity stimulated by PMA.
Sodium Salicylate Suppressed PMA-induced Ras Signaling Pathway-To gain insight into the signaling pathway through which PMA induces RSK-mediated C/EBP␤ binding and COX-2 transcriptional activation, we treated fibroblasts with inhibitors of major signaling molecules before the addition of PMA. PMA-induced COX-2 protein expression was suppressed by PKC and MEK-1 inhibitors but not phosphatidylinositol-3 kinase inhibitor (Fig. 6A). These results suggest that PMAinduced COX-2 expression is mediated via the Ras signaling pathway. The effects of DN-Ras and CA-Ras on COX-2 promoter activity were evaluated. Compared with vector control, CA-Ras augmented basal COX-2 promoter activity without enhancing PMA-induced activity, whereas DN-Ras reduced the basal activity as well as abrogated PMA-induced increase in COX-2 promoter activity (Fig. 6B). These results suggest that Ras plays an essential role in regulating COX-2 transcriptional activation. To determine whether salicylate inhibits Ras activation, we measured Ras activity in PMA-treated cells in the presence or absence of sodium salicylate (10 Ϫ5 M). The Ras activity was undetected at basal state and was highly increased after PMA stimulation (Fig. 7A). PMA-stimulated Ras activation was abrogated by salicylate, whereas salicylate had no effect on the basal activity (Fig. 7A). In accord with Ras activation, ERK1/2 were activated by PMA, which was suppressed by sodium salicylate (Fig. 7B). To analyze the possible involvement of a feedback activation of Ras by RSK, we transfected cells with RSK-WT, RSK-DN, and vector control and treated the transfected cells with or without PMA for 15 min. Ras activity in the cell lysates was determined. Neither RSK-WT nor RSK-DN had a significant effect on basal or PMA-stimulated Ras activity when compared with vector control (Fig. 8). DISCUSSION Results from this study indicate that salicylate inhibits PMA-induced COX-2 transcriptional activation by blocking RSK activation. Several lines of evidence support this conclusion. First, PMA is capable of activating RSK, which is inhib- FIG. 3. RSK activity in cell lysates prepared from fibroblasts transfected with HA-tagged RSK2 expression vectors. Cells were transfected with wild-type HA-RSK2 (RSK-WT), RSK-DN, or vector control followed by stimulation with PMA for 15 min. HA-RSKs were immunoprecipitated, and RSK activity was assayed using a synthetic S6 peptide. Each bar represents mean Ϯ S.E. of three separate experiments.

FIG. 4. Suppression of COX-2 promoter activity by RSK-DN.
A, fibroblasts were co-transfected with human COX-2 promoter construct and WT, DN, or control RSK vector. The transfected cells were treated with or without PMA for 4 h, and the expressed luciferase activity was measured. Each bar denotes mean Ϯ S.E. of three separate experiments. B, nuclear extracts from transfected cells treated with or without PMA were analyzed for binding to [ 32 P]C/EBP probe by EMSA. This graph is representative of three experiments, and the mean densitometry results are shown in the lower panel.
ited by a DN-RSK2 mutant. Concordantly, this dominant negative mutant inhibits PMA-induced C/EBP␤ binding and COX-2 promoter activity. Second, mutation of the RSK phosphorylation site of C/EBP␤ results in a poor activation and binding activity in response to PMA stimulation. Third, salicylate inhibits PMA-induced RSK activation in a concentrationdependent manner in accord with the inhibition of COX-2 protein expression and promoter activity. It was previously reported that sodium salicylate inhibited unstimulated RSK2 only at very high concentrations (20 mM) but was capable of abolishing PMA-stimulated RSK2 at 5 mM (9). Taken together, these results suggest that SA at therapeutic concentrations exerts direct inhibitory action on activated RSK molecules. ERK1/2 activate RSK by docking to the C-terminal region of RSK and phosphorylate several residues at the C-terminal kinase domain and the linker region, thereby activating the N-terminal kinase (30). RSK has been shown to complex with ERK in several cell types (30 -32). This raised the possibility that SA may inhibit RSK via inactivation of ERK1/2. However, this is unlikely because ERK1/2 were not reported to be complexed with purified RSK2 (9). A recent study has shown that the complex formation is dynamically regulated. ERK forms complexes with RSK in unstimulated cells but dissociates from RSK following cell stimulation with PMA and other agonists (33). Furthermore, our mutant data indicate that ERK1/2 are not directly involved in C/EBP␤ phosphorylation because mutation of ERK1/2 phosphorylation residue (Thr 235 ) did not alter C/EBP␤ binding.
RSK family proteins comprise three closely related members, RSK 1, 2, and 3, and several distantly related members (for a review see Ref. 34). RSK1 and RSK2 (RSK1/2) are highly homologous and share common functional properties (33). Both RSK1/2 are inhibited by DN-RSK2 (31). By contrast, RSK3 and other members are regulated differently and have different functional properties (30). Thus, it is likely that RSK1 and RSK2 are involved in PMA-induced C/EBP␤ activation and are targets of SA. It is unclear how salicylate inhibits RSK1/2 activity. RSK1/2 contain two kinase domains, a C-terminal kinase domain, which is activated by ERK1/2, and an N-terminal kinase domain, which is activated by C-terminal kinase via autophosphorylation and fully activated by phosphoinositidedependent kinase-1 (34). Activated N-terminal kinase phosphorylates a number of exogenous substrates including C/EBP␤. The structural basis for the activation of N-terminal kinase is not entirely clear, but a recent report on the structure of MSK-1, a distant relative of RSK1/2, provides important insight (35). Structural analysis of MSK-1 has shown that the ATP binding site of the kinase domain is blocked by a ␤-sheet, implying that activation of MSK-1 as well as its related proteins involves the displacement of this autoinhibitory ␤-sheet to allow for ATP binding to its specific pocket. It is interesting to note that salicylate inhibits ATPase activity of immunoglobulin heavy chain-binding protein (BiP) (36). This raises the possibility that salicylate may block ATP binding site of the N-terminal kinase domain. Further studies are needed to investigate this possible mechanism.
Because RSK1/2 are activated by ERK1/2 and phosphoinositide-dependent kinase-1, we determined whether PMAinduced COX-2 expression requires Ras and phosphatidylinositol 3-kinase signaling pathways and whether SA might have an additional effect on blocking the signaling molecules. Our re-sults show that Ras and ERK1/2 are activated by PMA and are essential for PMA-induced COX-2 expression, which are inhibited by SA. Because RSK1/2 has been shown to phosphorylate SOS (37), an upstream signaling molecule of Ras, and thus may regulate Ras activation by a feedback mechanism, we tested the hypothesis that SA indirectly inhibits Ras and ERK1/2 activation via RSK1/2 suppression. Our data did not support this hypothesis, because Ras activation by PMA was not blocked by dominant negative mutant of RSK2. It is possible that SA may act on a signaling molecule upstream of Ras. One possible candidate is PKC, which is essential for PMA-induced COX-2 expression and is known to activate the Ras signaling pathway. PMA is known to activate multiple isoforms of classic and novel PKC isoforms. In human embryonic kidney 293 cells, PMA activates novel PKCs including PKC , PKC ␦ , and PKC (38). However, it has been shown that salicylate does not inhibit the activity of any of the PMA-activated isoforms (38). Thus, PKC may not be the upstream target. The mechanism by which SA inhibits PMA-induced Ras and ERK1/2 activation remains to be elucidated. Nevertheless, our results indicate that SA inhibits C/EBP␤-mediated COX-2 expression by suppressing RSK1/2 activity via direct and indirect pharmacological mechanisms.
In summary, our results have shown that SA inhibits RSK activation, thereby reducing RSK-induced C/EBP␤ binding and transcriptional activation of COX-2. Our results further show that SA blocks Ras and ERK activation, suggesting that salicylate may inhibit other kinases that share active site configuration with RSK1/2. As RSK-related kinases activate a large number of transcription factors including CREB, IB, Fos, CREB-binding protein, and p300 coactivator (39 -41), salicylate may control the expressions of diverse genes important in inflammation and cell proliferation. Therefore, RSK and related kinases may be potentially valuable targets for drug discovery.