Regulation of c-Fos Gene Expression by NF-κB: A p65 Homodimer Binding Site in Mouse Embryonic Fibroblasts but Not Human HEK293 Cells

The immediate early gene c-Fos is reported to be regulated by Elk-1 and cAMP response element-binding protein (CREB), but whether nuclear factor (NF)-κB is also required for controlling c-Fos expression is unclear. In this study, we determined how NF-κB’s coordination with Elk/serum response factor (SRF) regulates c-fos transcription. We report that PMA strongly induced c-Fos expression, but tumor necrosis factor (TNF)-α did not. In mouse embryonic fibroblasts, the PMA induction of c-Fos was suppressed by a deficiency in IKKα, IKKβ, IKKγ, or p65. By contrast, in human embryonic kidney 293 cells, PMA induced c-Fos independently of p65. In accordance with these results, we identified an NF-κB binding site in the mouse but not human c-fos promoter. Under PMA stimulation, IKKα/β mediated p65 phosphorylation and the binding of the p65 homodimer to the NF-κB site in the mouse c-fos promoter. Furthermore, our studies demonstrated independent but coordinated functions of the IKKα/β-p65 and extracellular signal-regulated kinase (ERK)-Elk-1 pathways in the PMA induction of c-Fos. Collectively, these results reveal the distinct requirement of NF-κB for mouse and human c-fos regulation. Binding of the p65 homodimer to the κB site was indispensable for mouse c-fos expression, whereas the κB binding site was not present in the human c-fos promoter. Because of an inability to evoke sufficient ERK activation and Elk-1 phosphorylation, TNF-α induces c-Fos more weakly than PMA does in both mouse and human cells.


Introduction
The immediate early gene c-Fos is a protooncogene that can dimerize with c-Jun to form an activator protein (AP)-1 transcription complex [1]. c-Fos plays crucial roles in the molecular mechanisms underlying a variety of cellular processes including proliferation, differentiation, transformation, and apoptosis [2]. Conversely, deregulation of c-Fos may be linked to diverse pathological conditions including immunological, skeletal, and neurological defects, as well as to oncogenic transformation and tumor progression. c-Fos determines the functions of osteoblasts and osteoclasts, and regulates cytokine release in inflammatory diseases [3]. Overexpression of c-Fos in animals results in chondroblast and osteoblast transformation, which contribute to osteosarcoma formation. In contrast to the widespread view on c-Fos's oncogenic function, c-Fos also serves critical apoptotic and tumor-suppressive functions by upregulating Bim [4,5]. Moreover, mice lacking c-Fos expression have been reported to exhibit abnormal development and function in certain tissues [6].
Accumulating evidence indicates that the phosphorylation of Elk-19s transactivation domain critically regulates the ternary complex formation of Elk-1 with SRF, which occurs through the binding to the serum response element (SRE) within the c-fos promoter, and, thus, phosphorylation serves as a prerequisite for the transactivation function of Elk-1. The transactivation domain of Elk-1 is phosphorylated in most cases by extracellular signalregulated kinase (ERK) [1,8,9], although the phosphorylation may also be mediated by p38 and c-Jun N-terminal kinase (JNK), depending on the stimuli encountered [10]. SRF is also phosphorylated, and its phosphorylation by ERK, CaMKK, or PI3K was reported to increase c-fos expression [11][12][13]. Furthermore, SRF-mediated c-fos expression has been reported to be positively regulated by the transcriptional co-activator MKL1 and the RhoA-dependent pathway [14]. In addition to the SRE, the CREB-response element (CRE) is required for maximal c-fos promoter activation. The phosphorylation of CREB at Ser133 by numerous kinases including RSK, MSK, protein kinase C (PKC), and Akt also plays a crucial role in c-fos transcription [15]. Among these kinases, MSK has the highest affinity for CREB and acts as the primary regulator of CREB activity. Moreover, in addition to SRF-dependent recruitment of Elk-1 to the SRE, Elk-1-mediated recruitment of ERK and MSK to SRE promoter complexes has been reported [16].
Nuclear factor (NF)-kB is a key transcription factor that is tightly regulated by diverse signaling pathways and controls the expression of numerous genes [17,18]. Although NF-kB-dependent regulatory mechanisms acting through Elk-1 and SRF have been well characterized, functional studies linking NF-kB activation to c-fos expression are limited. Elk-1 was previously identified as a downstream target gene of NF-kB, and this function of NF-kB consequently contributes to c-Fos expression induced by serum and other stimuli [19]. Another study using epidermal growth factor (EGF)-stimulated fibroblasts as a model demonstrated that IKKa and p65/RelA contribute substantially to EGF-induced cfos expression independently of the classical IkBa degradation pathway [20]. However, whether the classical NF-kB pathway also similarly upregulates c-fos expression remains unknown. Moreover, whether the NF-kB signal alone is sufficient for inducing c-Fos or whether NF-kB signaling must be functionally coordinated with the ERK pathway to induce c-Fos has not been investigated.
In this study, we addressed the role of the IKK-dependent NF-kB pathway and its crosstalk with ERK and downstream transcription factors (Elk-1, SRF, and CREB) in the control of cfos expression. We compared and analyzed the effects and molecular mechanisms of action of 2 stimuli, phorbol 12-myristate 13-acetate (PMA) and tumor necrosis factor (TNF)-a, which were shown to activate NF-kB and ERK signaling to distinct degrees in diverse cell types. We report that a kB site is present in the mouse, but not human, c-fos promoter, and also that p65 homodimer binding to the c-fos promoter and Elk-1 activation must be coordinated to regulate mouse c-fos transcription. An inability of TNF-a to evoke sufficient ERK activation and the consequent low-level Elk-1 phosphorylation make TNF-a a weak c-Fos inducer compared with PMA.

Immunoblotting Analysis
After the cells were treated with reagents, the medium was aspirated and the cells were rinsed once with ice-cold phosphatebuffered saline (PBS) and then harvested and lysed directly using a 16 sodium dodecylsulfate (SDS) gel-loading buffer (50 mM Tris-HCl at pH 6.8, 2% 2-mercaptoethanol, 0.1% bromophenol blue, and 10% glycerol). Samples were heated at 95uC for 5 min, and equal amounts of whole-cell extracts were subjected to SDSpolyacrylamide gel electrophoresis (PAGE; 8%-15%) and then transferred to Immunobilon-P (Millipore). Membranes were blocked using TBST (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, and 0.1% Tween 20) containing 5% nonfat milk for 1 h at room temperature to minimize nonspecific antibody binding. After hybridization with the primary antibodies, the membranes were washed 3 times with TBST and incubated with HRP-conjugated secondary antibodies for 1 h. After washing 3 times with TBST, protein bands were detected by using an enhanced chemiluminescence (ECL) detection reagent (PerkinElmer or Millipore). b-Actin was used as an internal control.

In vitro IKK Assay
Cell lysates were precleaned at 4uC for 1 h and then mixed with 0.5 mg of an anti-IKKc antibody overnight. Subsequently, 10 mL of protein A-agarose beads were added and the samples were placed on a rocker for 1 h at 4uC. The immunoprecipitates were washed 5 times with a cold lysis buffer and then twice with a kinase reaction buffer (20 mM HEPES at pH 7.5 and 10 mM MgCl 2 ). The beads were then incubated at 30uC in 40 mL of a kinase reaction buffer supplemented with 10 mCi of [c-32 P] ATP, 2 mM NaF, protease inhibitor cocktails, and 2 mg of GST-IkBa. The reaction products were separated using SDS-PAGE and then examined using autoradiography. GST-IkBa (1-67) was prepared by transforming Escherichia coli strain JM109 with a plasmid encoding the GST-fusion protein. The GST-fusion protein was induced and purified according to a commercial kit protocol (Bulk GST Purification module, GE Healthcare UK Limited, Amersham Place, Little Chalfont, Buckinghamshire, UK).

Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)
Nuclear extracts were prepared from MEFs as previously described [25]. Each gel-shift reaction mixture contained 0.025 pmol of a radiolabeled probe, 1 mg of poly (dI-dC), and 5 mg of nuclear proteins in a buffer of 50 mM NaCl, 30 mM KCl, 10 mM Tris-HCl at pH 7.5, 1 mM EDTA, 1 mM DTT, and 10% glycerol. Competing DNA oligonucleotides and antibodies against p65, p50, p52, RelB, and c-Rel (Santa Cruz Biotechnology) were added to the nuclear extract and incubated for 30 min on ice prior to the DNA binding reaction. After the DNA binding reaction was allowed to proceed at room temperature for 25 min, the samples were electrophoresed through non-denaturing 4% polyacrylamide gels for 1.5 h at 70 V. The gels were then dried and used for autoradiography. Competition experiments were performed by adding 1 pmol of an unlabeled probe. The oligonucleotides used in this study included the NF-kB-binding

Quantitative Polymerase Chain Reaction (qPCR)
The cells were harvested using RNA-Bee isolation reagents (Roche, Mannham, Germany) and total RNA was extracted by following the manufacturer's protocol. We synthesized cDNAs by using MMLV reverse transcriptase purchased from Promega (Madison, WI, USA). To quantify mRNA levels, we performed real-time PCR by using an ABI Prism 7900 (Applied Biosystems, Foster City, CA, USA) and then detected the fluorescence of samples in 96-well plates by using FastStart SYBR Green Master (Roche, Mannham, Germany). Each 25-mL PCR reaction contained the cDNA, the Master Mix, and the following primers:

Immunofluorescence Microscopy
The cells were seeded in 96-well plates and grown until they were 70%-80% confluent. After stimulating with indicated reagents, the medium was aspirated and the cells were fixed using 4% paraformaldehyde in PBS (pH 7.4) for 15 min at room temperature. The samples were washed 3 times with PBS and then incubated for 10 min in PBS containing 0.25% Triton X-100. The cells were then washed in PBS 3 times and incubated for 1 h in PBST containing 1% bovine serum albumin (BSA) to block nonspecific binding of antibodies. The cells were incubated overnight at 4uC with the anti-p65 primary antibody, washed 3 times using PBS (5 min for each wash), and then incubated with FITC-conjugated secondary antibodies diluted in a buffer containing BSA (1%) and DAPI (0.1 mg/mL). After being incubated for 1 h at room temperature in the dark, the cells were washed 3 times with PBST and then subjected to immunofluorescence microscopy to determine the cellular localization of p65.

c-fos Promoter and the NF-kB Activity Assay
To construct the c-fos promoter, we isolated genomic DNA from C57BL/6 mice and amplified the promoter region from +30 to 21070 by performing PCR using KOD hot-start DNA polymerase (Merck KGaA, Darmstadt, Germany) and the primers 59-ATC GGC TAG CCA GTT TCC AGG CTG CAC GTA-39 and 59-ATC GCT CGA GTC TTC TCA GTT GCT CGC TGC-39. A 1.1-kb fragment containing NheI and XhoI sites was then ligated into the PCRII-TOPO vector by using a TA cloning kit (Invitrogen). The insertion fragment was sequenced by using SP6 and T7. The promoter fragment was excised from PCRII-TOPO-c-fos by digesting it with NheI and XhoI, and then cloned to the pGL3-basic vector to produce the pGL3-c-fos promoter. The predicted kB site at 2216/2206 of the mouse c-fos promoter region (GGAGACCCCC) was mutated to GGAGAACACG by using the QuikChange Site-Directed Mutagenesis Kit from Agilent Technologies (Santa Clara, CA, USA). The NF-kB reporter plasmid containing 3 kB-binding sites (pGL2-ELMAluciferase) was provided by Dr. S. L. Hsieh (Yang-Ming University, Taipei, Taiwan).
Following the commercial standard protocol, HEK293 cells and MEFs were cotransfected with a reporter plasmid and a bgalactosidase expression vector (pSVlacZ) by using Lipofectamine 2000 (Invitrogen). After treatment with the indicated reagents, the cells were lysed in a reporter lysis buffer (Promega) and the lysates were reacted with a commercial luciferase substrate provided in the luciferase assay system kit (Promega) and assayed on a microplate luminometer (Packard, Meriden, CT, USA). Luciferase activity was normalized relative to the activity of b-galactosidase and expressed as multiples of the untreated control. In some experiments, p65 2/2 MEFs were transfected with HA-p65 by using Lipofectamine 2000.

Chromatin Immunoprecipitation (ChIP) Assay
In vivo binding of the p65 transcription factor to the mouse c-fos promoter was investigated using a ChIP assay according to the EZ-CHIP TM protocol (Merck KGaA, Darmstadt, Germany). MEFs (3610 6 ) were treated for 10 min with 1% formaldehyde diluted in a tissue culture medium at 37uC and then washed twice in ice-cold PBS containing protease inhibitors (1 mM PMSF and 1 mg/mL each of aprotinin and pepstatin A). The cells were pelleted and lysed in an SDS lysis buffer. The samples were sonicated to shear the DNA to lengths of 200-1000 bp and were then diluted 5-fold with a dilution buffer (0.01% SDS, 1.2 mM EDTA, 16.7 mM Tris-HCl at pH 8.1, 1.1% Triton X-100, and 167 mM NaCl). The diluted cell lysates were precleared using salmon-sperm DNA/protein A agarose for 1 h at 4uC and then incubated with 2 mg of a specific antibody or normal immunoglobulin G (IgG) overnight at 4uC to immunoprecipitate protein-DNA complexes from the precleared supernatants; immune complexes were precipitated by using protein A agarose beads. The immunoprecipitated DNA was amplified by performing qPCR using an ABI Prism 7900. In the PCR, primers flanking the kB-binding sites in the mouse c-fos promoter were used, and a 200bp c-fos promoter fragment was detected using sense (59-GGG ACC ATC TCC GAA ATC CTA CAC GC-39) and antisense (59-GGA CTT CCT ACG TCA CTG GGC GGA AC-39) primers corresponding to nucleotide positions 2250 R 250 of the mouse c-fos promoter. Furthermore, we conducted a positive ChIP assay to confirm the stimulating effects of TNF-a on the kB site. The following primers around the kB-binding site in the ICAM-1 promoter were designed: sense 59-GAT GTC CTT TCC GGT TGC AG-39 and antisense 59-CAG GGA AAT TCC CGG AGT ACA-39 [26]. To verify the experimental specificity of the ChIP assay, we chose as a negative control a DNA site lacking the NF-kB-binding element that is present in the Igk promoter [27]; the negative control primers in the Igk gene promoter were the following: sense 59-CTA ATA GTC CCA TGC TCT CC-39 and antisense 59-TCC TTC CTT TGG TGG CTT C-39.
Small Interfering (si)RNA HEK293 cells (3610 6 ) were transfected with 100 nM siRNA targeting the mRNA of human p65 (catalog no. L-003533-00) by using DharmaFECT Transfection Reagents. The control nontargeting siRNA used was a pool of 4 functional nontargeting siRNAs. All siRNAs were obtained from Dharmacon Research (Lafayette, CO, USA). After siRNA transfection for 48 h, the cells were treated with various stimulants and the gene-silencing effects were then evaluated by performing immunoblotting assays.

Statistical Evaluation
Values are expressed as the mean 6 the standard error of the mean (SEM) of at least 3 independent experiments, which were performed in duplicate. An analysis of variance (ANOVA) was used to assess the statistical significance of differences, and P,0.05 was considered statistically significant. Figure 2. The IKK complex is involved in PMA-induced p65 phosphorylation and nuclear localization and c-Fos expression. In most experiments, WT, IKKa 2/2 , IKKb 2/2 , IKKc 2/2 , and p65 2/2 MEFs were treated with TNF-a (20 ng/mL) or PMA (10 nM) for the indicated times. (A, D, F) Total cell lysates were prepared and subjected to western blotting analysis; the antibodies used are indicated. The western blots are representative of 3 independent experiments. (B) Total RNA was extracted, reverse transcribed into cDNAs, and subjected to qPCR analysis. Values were normalized relative to b-actin gene expression and expressed relative to the control sample. * P,0.05 indicates a significant reduction in PMA-induced gene expression in IKK-null MEFs. (C) Distinct response elements and the predicted NF-kB-binding site in the mouse c-fos promoter are shown. Regulatory elements identified in the mouse c-fos promoter are labeled and include the cis-inducible elements SRE, Fos activator protein 1 site (FAP), a glucocorticoid response element (GRE), and CRE. The major CRE is nearest of the sites to the start site within the large box, whereas the minor CREs are further upstream and are denoted using small boxes. Constitutive activity of the c-fos promoter with or without mutation of the kB site was determined in WT and IKK-null MEFs. Data in (B) and (C) are the means 6 SEM from 3 independent experiments. * P,0.05 indicates a significant reduction in basal and PMA-induced promoter activity in IKK-null MEFs. (D) The p65 2/2 MEFs were reconstituted with p65 protein, and HEK293 cells were transiently transfected with control or p65 siRNA (100 nM) for 48 h. Cells were treated with PMA for the indicated times. (E) Immunofluorescent staining of p65 (green) and nuclei (blue) shown by images captured under a fluorescence microscope; similar results were obtained in 2 additional experiments. doi:10.1371/journal.pone.0084062.g002

PMA, but not TNF-a, Induced c-Fos Expression and Elk-1 Phosphorylation in MEFs and HEK293 Cells
PMA, which is structurally similar to diacylglycerol, is a potent PKC activator that is widely used to study the cellular functions of PKC. In MEFs and HEK293 cells, we first compared the effects of PMA and the proinflammatory cytokine TNF-a on c-Fos expression. We determined that PMA (10 nM) robustly upregulated c-Fos protein and gene expression in MEFs and HEK293 cells in a time-dependent manner; by contrast, TNF-a (20 ng/mL) did not similarly upregulate c-Fos expression (Fig. 1A, B). Because c-Fos induction requires the coordinated binding of TCF and SRF to SRE and CREB binding to CRE, we subsequently tested whether stimulating cells with the 2 reagents alters the phosphorylation of Elk-1 (a major TCF member), SRF, and CREB. Both PMA and TNF-a induced Elk-1 phosphorylation in MEFs and HEK293 cells, but the extent of stimulation, which was correlated with their effectiveness in inducing c-Fos, was considerably stronger and sustained with PMA treatment than with TNF-a treatment (Fig. 1A). PMA also elicited stronger and longer-lasting CREB phosphorylation than TNF-a did in MEFs and HEK293 cells. In HEK293 cells, TNF-a did not induce significant CREB phosphorylation (Fig. 1A). By contrast, PMA and TNF-a induced SRF phosphorylation to a similar extent and with similar kinetics in MEFs, and PMA induced slightly higher SRF phosphorylation than TNF-a did in HEK293 cells (Fig. 1A). These results suggest that TNF-a is a poorer inducer of c-Fos than PMA in MEFs and HEK293 cells, and that Elk-1 and CREB might act as major determinants of c-Fos induction.

NF-kB is Involved in PMA-induced c-Fos Expression
We subsequently investigated the link between NF-kB and c-Fos by comparing the effects of PMA on WT, IKKa 2/2 , IKKb 2/2 , and IKKc 2/2 MEFs. We confirmed the deficiency of each IKK isoform (a, b, and c) in the respective IKK-null MEFs and established that p65 was expressed at equal levels in these cells. Our results showed that the marked upregulation of c-Fos protein induced by PMA was abrogated in IKKa 2/2 , IKKb 2/2 , and IKKc 2/2 MEFs ( Fig. 2A). We used real-time PCR to determine whether this inhibition occurred through a transcriptional or a translational mechanism. We observed that c-fos mRNA levels increased substantially within 15-30 min after treatment of WT MEFs with PMA, but this effect was severely diminished in IKKa 2/2 and IKKb 2/2 MEFs (Fig. 2B). Supporting this finding, c-fos promoter activity was reduced in all 3 types of IKK-null MEFs, which suggests that constitutive NF-kB activity is required for c-fos expression. To examine this possibility more closely, we analyzed the mouse c-fos promoter by using the Transcription Element Search System (TESS) and Signal Scan and identified an NF-kB binding site (with a TESS La score of 10.7494) that lies at 2206 to 2216 in the proximal region of the mouse c-fos promoter (Fig. 2C, upper panel). Mutation of the kB -binding sequence reduced the basal promoter activity to 30% of the WT promoter level, and PMA no longer elicited c-fos promoter activation (Fig. 2C, lower panel). Based on these results, we concluded that the kB-binding site is a critical cis-element that controls PMAinduced mouse c-fos promoter activity.
To further confirm the role of NF-kB in c-fos expression, we tested the PMA-induced c-Fos response in p65 2/2 MEFs and p65-silenced HEK293 cells. We determined that PMA-induced c-Fos protein expression was suppressed in p65 2/2 MEFs. Conversely, the loss of c-Fos expression was restored when p65 was ectopically expressed in p65 2/2 MEFs (Fig. 2D, upper panel). However, silencing p65 in HEK293 cells did not affect PMAinduced human c-Fos protein expression (Fig. 2D, lower panel). These results indicate that p65 plays a critical role in the PMAinduced expression of mouse but not human c-Fos.
To examine p65 localization, we used immunofluorescence microscopy. In both WT and IKK-null MEFs, p65 was localized in the cytoplasm before PMA stimulation. After PMA treatment for 30 min, WT MEFs clearly displayed nuclear localization of p65, whereas IKKa 2/2 and IKKc 2/2 MEFs did not exhibit p65 localization in nuclei. In IKKb 2/2 MEFs, although PMA-induced translocation of p65 was markedly reduced, we still detected nuclear p65 in approximately 30% of the population, compared with .90% in the case of WT cells (Fig. 2E). These results suggest that PMA-induced p65 nuclear translocation required each member of the IKK complex, especially IKKa and IKKc.
In addition to the nuclear translocation of p65 following IkBa downregulation, direct IKK-targeted p65 phosphorylation at Ser536 was shown to be critical for p65 nuclear translocation and, in turn, for NF-kB transactivation [28,29]. To test whether IKK isoforms exhibit this activity in response to PMA and TNF-a, we examined their effects on Ser536 phosphorylation in p65. TNF-a induced p65 phosphorylation on Ser536, and this was not altered in IKKa 2/2 or IKKb 2/2 MEFs (Fig. 2F, upper panel). PMA also induced p65 (Ser536) phosphorylation, but unlike in the case of TNF-a, this phosphorylation was markedly reduced in MEFs deficient in each of the IKK isoforms (Fig. 2F, lower panel). Thus, we suggest that, whereas TNF-a-induced p65 (Ser536) phosphorylation occurs through a pathway independent of IKKa/ b, the IKKs play key roles in PMA-induced p65 (Ser536) phosphorylation.

PMA Induces Less IKK Activation than does TNF-a
After determining that the IKK complex and p65 were required for PMA-induced c-Fos expression, we determined the effect of PMA on IKK-dependent NF-kB activation and compared that with the effect of TNF-a. In MEFs, TNF-a treatment increased IKKa/b (Ser176/177) and IkBa phosphorylation within 5 min and sustained IKKa/b (Ser176/177) phosphorylation for 40 min. Concurrently, IkBa downregulation, an index of IKK-mediated IkBa phosphorylation, occurred after 5 min, but recovered by 40 min. Compared with TNF-a, PMA (10 nM) induced the phosphorylation of IKKa/b (Ser176/177) and IkBa more slowly and to a lesser extent (Fig. 3A). Similar disparities in PMA-and TNF-a-dependent IKK activation were also detected in HEK293 cells. TNF-a increased IKKa/b phosphorylation at Ser176/ 177 by 10 min, and this was accompanied by IkBa degradation. By contrast, PMA (10 nM) induced neither IKKa/b (Ser176/177) phosphorylation nor IkBa degradation (Fig. 3B), and increasing PMA concentrations up to 1 mM did not result in an increase in IkBa degradation (Fig. 3B). Because IkBa degradation is mediated by the proteasome degradation system, we pretreated cells with the proteasome inhibitor MG-132 (10 mM) and then examined the effects of the 2 stimuli. Treatment with MG-132 sustained IkBa phosphorylation in response to TNF-a, but did not cause a clear elevation of IkBa phosphorylation after PMA treatment (Fig. 3B). To confirm that the limited effect of PMA on IkBa could be ascribed to weak activation of IKK, we conducted an in vitro kinase assay by using the IKK complex isolated from HEK293 cells and MEFs. TNF-a rapidly induced GST-IkBa (1-67) phosphorylation, and this occurred in parallel with a TNF-adependent increase in IKK phosphorylation and IkBa degradation. Compared with the clear phosphorylation of GST-IkBa(1-67) in response to TNF-a stimulation, the IKK complex showed poor kinase activity toward its substrate after treatment with PMA (Fig. 3C). Collectively, these results suggest that PMA weakly stimulates the IKK activity required for IkBa phosphorylation and downregulation. In agreement with this notion, when we determined NF-kB activity by measuring the kB reporter activity, we observed that TNF-a induced higher NF-kB luciferase activity in HEK293 cells than did PMA, and both actions were blocked by the IKK inhibitor BMS-345541 in a concentration-dependent manner (Fig. 3D). These results strongly indicate that, compared with the potent activator TNF-a, PMA produced a substantially weaker effect on the IKK-NF-kB signaling pathway in both MEFs and HEK293 cells.

The p65 Homodimer Binds to a kB Site in the Mouse c-fos Promoter
To elucidate the transcriptional function of NF-kB in regulating mouse c-fos expression, we designed oligonucleotides containing this NF-kB-binding site (2226 to 2196) and conducted EMSAs.  Two complexes were formed before and after PMA stimulation in WT MEFs (Fig. 4A). Complex 2 exhibited nonspecific binding because the binding was inhibited when we used cold probes of consensus NF-kB, NF-Y, and AP-1 sequences (Fig. 4A). By contrast, the Complex 1 signal that was enhanced after PMA stimulation was specific to NF-kB (Fig. 4A), and when the p65 antibody was added, a clear supershifted band beyond Complex 1 was detected (Fig. 4B), suggesting that p65 was bound to this complex. We also added antibodies against other NF-kB family members to identify the binding partner of p65. However, antibodies against p50, p52, RelB, and c-Rel did not produce a supershift after PMA stimulation (Fig. 4B), suggesting that the p65 homodimer is involved in c-fos regulation. Furthermore, nuclear extracts of PMA-treated IKKa 2/2 and IKKb 2/2 MEFs failed to interact with oligonucleotides containing the NF-kB binding site (Fig. 4C), and the p65 homodimer that bound to the NF-kBbinding site of the mouse c-fos promoter was not affected by U0126 treatment (Fig. 4D). These results suggest that the binding of the p65 homodimer to the mouse c-fos promoter is required for PMAdependent c-fos induction.
To verify that p65 binding is required for PMA-induced c-fos promoter activation, we then performed a ChIP assay by using a specific antibody against p65 to immunoprecipitate formaldehydefixed chromatin. Basal and PMA-induced increases in the binding of p65 to the kB site (2206 to 2216) of the c-fos promoter were reduced in the absence of IKKa or IKKb (Fig. 4E, upper panel), and TNF-a treatment was unable to induce p65 binding to the kB site of the c-fos promoter. To confirm that these distinct effects are gene specific, we used the kB site of the ICAM-1 gene in the ChIP assay. TNF-a has been shown to be capable of upregulating ICAM-1 expression in MEFs [30,31]. As expected, TNF-a increased the binding of p65 to the ICAM-1 promoter, and this effect was diminished in the absence of IKKs (Fig. 4E, lower  panel). To demonstrate the specificity of the results of the ChIP assay, we used an Igk gene sequence lacking the kB site as a negative control [27]. Neither PMA nor TNF-a affected the weak (approximately 20% of the specific binding) and non-specific effect of the p65 antibody on the Igkpromoter. Thus, we concluded that p65 can directly interact with the kB site of the mouse c-fos promoter in MEFs treated with PMA but not TNF-a.

The ERK Signaling Pathway is a Prerequisite for c-Fos Expression
In addition to our observation that NF-kB contributes to c-fos expression, MAPKs have been reported to play key roles in controlling the induction of immediately early genes. Therefore, we examined the phosphorylation statuses of MAPKs in MEFs and HEK293 cells stimulated with TNF-a or PMA to determine whether transcription factors activated by MAPKs also contribute to the differential effects of TNF-a and PMA on c-fos induction. TNF-a and PMA induced the phosphorylation of ERK, p38, and JNK in MEFs (Fig. 5A) and HEK293 cells (Fig. 5B) to different extents. The major difference in the effects was that PMA was more potent than TNF-a in activating ERK, which was observed as a more sustained and dramatic increase in ERK phosphorylation in response to PMA than in response to TNF-a. By contrast, JNK and p38 were activated similarly by PMA and TNF-a.
To investigate the involvement of MAPKs and downstream signals in the induction of c-Fos expression, we used inhibitors of PKC and MAPKs. Treatment with PKC inhibitors (Ro318220, Ro320432, and LY333531) and a MEK inhibitor (U0126) abolished PMA induction of c-Fos protein expression in MEFs and HEK293 cells. However, p38 and JNK inhibitors (SB203580 and SP600125, respectively) did not alter the expression levels of c-Fos (Fig. 5C). Confirming these findings, c-fos promoter activity stimulated by PMA in HEK293 cells was blocked by the MEK inhibitor, U0126, whereas TNF-a did not increase c-fos promoter activity (Fig. 5D).
To test whether the signals downstream from ERKs were involved in PMA-induced c-Fos expression, MEFs were pretreated with inhibitors of MSK (H-89) and RSK (BI-D1870). Like U0126, only the MSK inhibitor clearly blocked PMA-induced c-fos promoter activation in HEK293 (Fig. 5D) and protein expression in MEFs (Fig. 5E). We also observed that Elk-1 phosphorylation was inhibited after treatment with U0126 but not H-89 or BI-D1870. By contrast, all inhibitors reduced SRF phosphorylation, and U0126 and H89 abolished this phosphorylation. Moreover, PMA-induced CREB phosphorylation was inhibited after the cells were treated with U0126 and H-89 (Fig. 5E). These results suggest that Elk-1 and CREB, but not SRF, play a major role in PMAinduced c-Fos expression.

Independent Signaling Pathways of PMA-induced ERK and NF-kB Activation
After determining the involvement of ERK-Elk and IKK in PMA-induced c-Fos expression, we assessed the potential crosstalk between IKK and ERK signaling in mediating c-Fos expression. The phosphorylation status of ERK was examined in IKK-null MEFs stimulated with PMA. ERK activation in response to PMA in IKKa 2/2 and IKKb 2/2 MEFs was comparable to that in WT MEFs (Fig. 6A), and no defect in PMA-induced Elk-1 phosphorylation was observed in IKKa 2/2 , IKKb 2/2 , or IKKc 2/2 MEFs (Fig. 6A). These results suggest that PMA activated ERK and Elk-1 independently of IKKs. Conversely, we also noted that U0126 did not affect TNF-a-or PMA-induced IKKa/b (Ser176/177) phosphorylation when ERK activity stimulated by PMA was inhibited (Fig. 6B). Neither PMA-induced IkB degradation nor p65 phosphorylation was altered in the presence of U0126 (Fig. 6B), and pretreatment with U0126 also did not alter PMAinduced p65 nuclear localization in MEFs (Fig. 6C). Furthermore, U0126 treatment did not affect PMA-induced NF-kB luciferase activity in HEK293 cells (Fig. 6D). Discussion PMA-induced IKK Activation and IkBa Phosphorylation were Weaker than those Induced by TNF-a IKKa/b phosphorylation at Ser176/177 has been demonstrated previously to contribute to enzyme activation and induction of NF-kB signaling. In this study, we compared the abilities of TNFa and PMA to activate IKK and NF-kB. TNF-a was determined to be a stronger activator of IKKa/b-NF-kB than PMA based on the following data. First, in both MEFs and HEK293 cells, TNF-a induced more efficient IkBa phosphorylation at Ser32 than PMA did. Ser32 is a well-known IKKb target that constitutes the major element of canonical NF-kB activation. Second, proteasomemediated degradation of IkBa that accompanied the protein's phosphorylation was more clearly detected in cells treated with TNF-a than in cells exposed to PMA. Third, consistent with the effect on IkBa, measurement of in vitro IkBa phosphorylation demonstrated that TNF-ainduced robust IKK activity in HEK293 cells and MEFs. By contrast, the effects of PMA were minimal. Fourth, stronger IKK-dependent NF-kB activity was detected in a reporter assay in response to TNF-a than in response to PMA.

IKKa and IKKb Mediate PMA-induced p65 Phosphorylation, Nuclear Translocation, and NF-kB Activation
In addition to the nuclear translocation of p65 that is required for NF-kB activation, the phosphorylation of p65 at Ser536 participates in NF-kB activation. The phosphorylation of p65 (Ser536), which was proposed to be mediated by numerous kinases, including IKKa/b, p38, RSK1, TBK1, and Akt [32][33][34][35][36][37], occurs within the transactivation domain of p65 and may induce a conformational change that lower the interaction of p65 with IkBa, the protein that mediates NF-kB nuclear exportation [33,34]. Moreover, the phosphorylation of p65 at Ser536 also results in the displacement of the corepressor SMRT-HDAC3 complex and increases p300/CBP recruitment [38]. These findings collectively indicate the existence of an alternative NF-kB activation pathway that is independent of IKKb-mediated IkB degradation.
In this study, we observed that both PMA and TNF-a induced p65 (Ser536) phosphorylation within 30 min. TNF-a-induced p65 (Ser536) phosphorylation showed no defects in IKKa 2/2 and IKKb 2/2 MEFs, but was inhibited by treatment with SB203580 (data not shown), suggesting that p38 contributes to this phosphorylation. By contrast, a deficiency in each of the IKK isoforms potently reduced PMA-induced p65 phosphorylation and, accordingly, PMA-induced p65 nuclear localization was impaired in IKKa 2/2 and IKKc 2/2 MEFs. Because PMAinduced IkBa phosphorylation was not clearly detected in MEFs, we suggest that the nuclear localization of p65 induced by PMA, unlike that caused by treatment with TNF-a, does not result from IKK-mediated IkB degradation, but can be ascribed to IKKa/bmediated p65 phosphorylation.

NF-kB Signaling Coordinates with ERK-Elk-1 and ERK-MSK-CREB Signaling to Mediate c-fos Expression
The TCF-SRF complex has been demonstrated to play a critical role in c-fos expression [39]. Similarly, MSK-dependent CREB activation has been shown to be involved in c-fos expression [2,15]. Elk-1 is a component of TCF and the DNA-binding capacity of Elk-1 increases after being phosphorylated on Ser383 in its C domain, thus undergoing conformational changes [40]. Phosphorylation of Elk-1 regulates its DNA binding and coordi-nation with SRF for the transcriptional activation of c-fos [1,8,10]. In this study, we determined that Elk-1 and CREB activated by ERK and the ERK downstream target MSK, respectively, were essential for c-fos expression. The considerably weaker effect of TNF-a than of PMA on both Elk-1 and CREB activation accounts for the minimal ability of TNF-a to induce c-Fos. Furthermore, unlike the serum-induced c-Fos expression that requires SRF and MKL1, a potent transcriptional coactivator of SRF [14,20], the action of PMA action might be independent of SRF (see below).
How NF-kB contributes to c-fos expression remains poorly understood. EGF was shown to induce c-Fos expression in MEFs through an IKK-dependent mechanism [20]. Another study showed that Elk-1 is an NF-kB-regulated gene [19]. In this study, we demonstrated a crucial role of NF-kB in mouse c-Fos upregulation. First, we noted that PMA-induced c-Fos expression at both the gene and protein levels was markedly diminished in IKK-and p65-null MEFs, which suggests that NF-kB regulates c-Fos expression. Second, our data revealed that the ERK-Elk-1 and IKK-p65-NF-kB axes independently regulated c-Fos expression. Neither ERK nor Elk-1 activation in response to PMA was affected in IKK-null MEFs. Moreover, applying U0126 did not affect the PMA induction of events that depend on NF-kB activation: IKK activation, IkBa degradation, and p65 phosphorylation, nuclear translocation, and DNA binding and activation. Third, we used TESS to identify a previously uncharacterized NF-kB-binding site in the 2206 to 2216 region of the mouse c-fos promoter. Analyzing DNA binding by using EMSA and ChIP assays demonstrated that the p65 homodimer can bind to this promoter region. A specific DNA-recognition mode of the NF-kB p65 homodimer that is distinct from the classical binding mode of the p65-p50 heterodimer has been identified structurally [41,42]. Functionally, the p65 homodimer was shown to be present in PMA-stimulated Jurkat T cells [43] and was identified to bind the promoter of IL-8 [44]. In addition to this effect of PMA, the critical roles of the p65 homodimer in biological function have been demonstrated, such as in inducing neutrophilic inflammation in airway cells [45] and in mediating ICAM-1 expression and neutrophil adhesion in endothelial cells [46]. Thus, the binding of the p65 homodimer to the mouse c-fos promoter is another example that supports the function of the p65 homodimer in transcriptional control of genes. A key finding in this study was that the NF-kB-binding site identified in the mouse c-fos promoter is not present in the human gene. However, Fujita et al. demonstrated that the NF-kB p50-binding site in the first intron of c-fos contributes to its expression in response to TNF-a [47]. Currently, we cannot exclude this possibility and thus must further investigate whether or not the intragenic NF-kB binding site is involved in PMA-induced c-fos expression in future studies. Moreover, in contrast to a report showing that ERK-dependent recruitment of NFI to the human c-fos promoter is required for gene expression [48], we observed no NFI-binding site in the mouse c-fos promoter. Thus, ERK-NFI and NF-kB regulate c-fos expression differently in mice and humans.
This study also showed that NF-kB activation alone is insufficient for inducing c-Fos and that this induction requires adequate ERK-mediated Elk-1 and CREB phosphorylation. This conclusion is primarily based on the action of TNF-a. Although TNF-a was a considerably stronger activator of NF-kB than PMA in MEFs and HEK293 cells, TNF-a poorly induced c-Fos, which can be explained by the inability of TNF-a to induce strong and sustained activation of ERK and the essential downstream transcription factors Elk-1 and CREB. Moreover, because TNFa could not induce NF-kB binding to the c-fos promoter, we suggest that an additional regulator is required for PMA to stimulate the transcriptional function of NF-kB, and that NF-kB coordinates the activation of Elk-1 and CREB in the upregulation of mouse c-fos transcription.
We demonstrated that ERK was required for Elk-1 phosphorylation and activation, but that the downstream kinases MSK and RSK were not required. In addition to Elk-1 phosphorylation, which plays an indispensable role in c-Fos protein expression (as mentioned above), the contribution of SRF phosphorylation to c-Fos expression remains disputed. In PMA-stimulated MEFs, we determined that SRF phosphorylation was suppressed by inhibitors of ERK (U0126), MSK (H-89), and RSK (BI-D1870), although only the ERK and MSK inhibitors attenuated c-Fos expression. Thus, we suggest that the phosphorylation of SRF by multiple kinases is not required for PMA-induced c-Fos expression. Moreover, because TNF-a induced SRF phosphorylation for a longer period than it activated ERK, we suggest that an ERKindependent kinase is involved in the TNF-a-dependent phosphorylation of SRF. Because no specific inhibitor of MSK is currently available and H-89 inhibits kinases other than MSK [49,50], the H-89 data must be interpreted with caution.
In summary, this study provides strong evidence that links NF-kB activation to c-Fos induction. We have demonstrated for the first time that, in addition to the widely recognized requirement of the ERK-Elk-1 and ERK-MSK-CREB signaling pathways in the control of c-Fos expression, the p65 homodimer, which bound to the c-fos promoter when stimulated by PMA, is indispensable for mouse c-fos transcription. This mechanism relies mainly on the members of the IKKa/b/c complex and on their enzymaticactivity-mediated p65 phosphorylation and nuclear localization (Fig. 7). However, the NF-kB activation that occurs without sufficient Elk-1 and/or CREB phosphorylation under TNF-a stimulation does not upregulate c-Fos. These results provide key evidence linking 2 vital transcription systems (NF-kB and c-Fos) in the control of various aspects of cellular functions, and these findings warrant further investigation.