Stress-induced stimulation of early growth response gene-1 by p38/stress-activated protein kinase 2 is mediated by a cAMP-responsive promoter element in a MAPKAP kinase 2-independent manner.

The p38/stress-activated protein kinase2 (p38/SAPK2) is activated by cellular stress and proinflammatory cytokines. Several transcription factors have been reported to be regulated by p38/SAPK2, and this kinase is involved in the control of expression of various genes. In human Jurkat T-cells, induction of the early growth response gene-1 (egr-1) by anisomycin is completely inhibited by SB203580, a specific inhibitor of p38/SAPK2a and -b. Northern blot and reporter gene experiments indicate that this block is at the level of mRNA biosynthesis. Using mutants of the egr-1 promoter, we demonstrate that a distal cAMP-responsive element (CRE; nucleotides -134 to -126) is necessary to control egr-1 induction by p38/SAPK2. Pull-down assays indicate that phospho-CRE binding protein (CREB) and phospho-activating transcription factor-1 (ATF1) bind to this element in a p38/SAPK2-dependent manner. In response to anisomycin, two known CREB kinases downstream to p38/SAPK2, MAPKAP kinase 2 (MK2) and mitogen- and stress-activated kinase 1 (MSK1), show increased activity. However, in MK2 -/- fibroblasts derived from mice carrying a disruption of the MK2 gene, the phosphorylation of CREB and ATF1 and the expression of egr-1 reach levels comparable with wild type cells. This finding excludes MK2 as an involved enzyme. We conclude that egr-1 induction by anisomycin is mediated by p38/SAPK2 and probably by MSK1. Phosphorylated CREB and ATF1 then bind to the CRE of the egr-1 promoter and cause a stress-dependent transcriptional activation of this gene.

A group of pyridinyl imidazole compounds have been identified as highly specific inhibitors of p38␣,␤/SAPK2a,b (34 -36). The inhibitory effect of these compounds toward p38/SAPK2 is attributed to binding of the drug to the ATP binding pocket of the kinase (37). The compound SB203580 inhibits p38/SAPK2 with an IC 50 of 0.6 M and exhibits no effect even at 100 M on the activities of 12 other protein kinases tested, including ERK2 and JNK/SAPK1 (35). In the past 2 years, SB203580 has been employed extensively to explore the specific roles of p38/ SAPK2 in cellular responses in a variety of experimental systems (5, 10, 35, 38 -40).
The immediate early gene egr-1 (also known as NGFI-A, zif268, TIS8, and krox24) encodes a transcription factor containing a DNA binding domain formed by three zinc finger motifs. This protein (Egr-1) binds to a specific GC-rich se-quence in the promoter region of many genes to regulate the expression of these target genes, including growth factors and cytokines. egr-1 is rapidly and transiently induced by growth factors and differentiation signals and is functionally involved in cell proliferation and differentiation. egr-1 expression can be induced by tumor necrosis factor-␣ and IL-1 (41), by hydrogen peroxide (42), by insulin (43), and by the protein serine/threonine phosphatase inhibitor okadaic acid (44,45). All of these stimuli are known to activate p38/SAPK2 and JNK/SAPK1. egr-1 has also been reported to be induced by diverse types of DNA-damaging agents including UV light (46), ionizing radiation (47), and AraC (48). Whereas in rat mesengial cells, lysophosphatidic acid-mediated egr-1 gene expression does not depend on p38/SAPK2 but on MEK 1 activation (49), in NIH-3T3 cells the induced egr-1 expression by several stress treatments such as heat shock, sodium arsenite, UV light, radiation, and anisomycin is correlated with p38/SAPK2 and JNK1/SAPK1c activation (50).
Using the inhibitor SB203580 we have demonstrated that stimulation of egr-1 induction in response to anisomycin treatment is dependent on activation of p38/SAPK2. We have identified a CRE in the egr-1 promoter necessary for this stimulation and have shown that binding of CREB and ATF1 to this element is stress-induced and phosphorylation-dependent. Furthermore, we analyzed which CREB kinases downstream to p38/SAPK2 could be involved in egr-1 expression and provided evidence that MSK1 but not MK2 may participate in anisomycin-induced up-regulation of the egr-1 gene.
Cell Transfection and Assay of Reporter Gene Activity-Cells (10 5 ) were transfected by liposome-mediated transfer using Lipo-fectAMINE (Life Technologies, Inc.) and 3 g of the different egr-1 promotor-CAT constructs. To calculate the efficiency of transfection and to normalize CAT activity, 1.5 g of a cytomegalovirus promoter-luciferase construct was co-transfected. After 5 h of transfection, the medium was replaced and the cells were incubated in fresh medium containing or not 10 M SB203580 (Calbiochem) for 30 min, and then 10 g/ml anisomycin (Sigma) was added for an additional 30 min. Alternatively, cells were co-transfected with 2 g of pcDNA3-FLAG-MKK6-EE expressing a constitutively active mutant of MKK6 (19) and, as a control, with 2 g of pcDNA3. Cells were washed three times with serum-free medium and incubated for 24 h. For measurement of reporter gene activity, cells were lysed in 500 l of hypotonic buffer containing 25 mM Tris-HCl, pH 7.5, and 2 mM MgCl 2 by three cycles of freezing and thawing. Lysates were clarified by centrifugation (10 min, 10,000 ϫ g, 4°C), and luciferase activity was determined in 50 l of lysate with a Lumat luminometer (Bertold). For determination of CAT activity, 50 l of cell extracts were incubated with 20 l of 0.01 Ci/l 14 C chloramphenicol, 5 l of 5 mg/ml butyryl coenzyme A (Sigma), 5 l of 2 M Tris-HCl, pH 8.0, and 20 l of H 2 0 at 37°C for 1 h. 200 l of tetramethylpentadecane:xylene (2:1) were then added, and after vortexing, the top organic phase was removed for liquid scintillation counting.
Northern Blot Analysis-Total RNA was prepared from Jurkat Tcells according to Chomczynski et al. (51). 10 g of total RNA were separated on agarose-formaldehyde gels and blotted onto nylon membranes (Hybond TM -Nϩ, Amersham Pharmacia Biotech). Membrane filters were hybridized with [␣-32 P]dATP random-primed cDNA probes prepared from a partial egr-1 cDNA clone (GenBank accession no. AA507023). Hybridized filters were washed under high stringency conditions (0.2ϫ SSC, 0.1% SDS, 65°C) and subjected to bio-imaging using the Cyclone Imaging System and the OptiQuant software package (Packard Instrument Co.). Equal loading of RNA was confirmed by stripping and reprobing the blots with a probe for glyceraldehyde-3phosphate dehydrogenase.
Western Blot-Cell lysates were separated by electrophoresis on 7.5-20% SDS-polyacrylamide gradient gels and transferred to nitrocellulose (Hybond TM -carbon extra, Amersham Pharmacia Biotech) in 25 mM Tris, 40 mM ⑀-aminocaproic acid, and 20% methanol at 1mA/cm 2 for 90 min. Membranes were blocked for 1 h at room temperature and incubated overnight at 4°C with 1:5000 anti-phospho-CREB/ATF1 and anti-CREB rabbit antibodies (New England BioLabs). Immunoreactive proteins were detected according to the enhanced chemiluminescent protocol using 1:5000 horseradish peroxidase-linked anti-rabbit secondary antibody. Blots were exposed to film for 1-10 min.
Pull-down Assay-A DNA fragment (nt Ϫ226 to Ϫ109) spanning the distal CRE (nt Ϫ136 to Ϫ124) of the egr-1 promoter was amplified from the wild type or mutated full-length promoter in pCAT vector by polymerase chain reaction using the primers 5Ј-AGGCTTCCCCGAAGCT-GGG-3Ј and 5Ј-CCGGGGAGGACCCGGAGTGACG-3Ј. The polymerase chain reaction product was isolated and biotinylated using biotin-High Prime (Roche Molecular Biochemicals) according to the manufacturer's instructions and incubated (2 g) with 40 g of proteins of nuclear extracts of 293 cells (53) in binding buffer (70 mM KCl, 5 mM MgCl 2 , 4 mM Tris (pH 7.5), 10 mM HEPES (pH 7.9), 1 mM dithiothreitol, 1 mM EDTA, 0.1 mg/ml poly(dI⅐dC), 50 g/ml bovine serum albumin, and 10% glycerol) for 20 min at 4°C. 50 l of pretreated (54) Streptavidinagarose was added, followed by incubation for 30 min at 4°C. The non-bound proteins were discarded by washing three times with 500 l of binding buffer, and the bound proteins were eluted with 30 l of elution buffer (1 M KCl, 5 mM MgCl 2 , 20 mM Tris (pH 7.5), 1 mM dithiothreitol, 1 mM EDTA, 200 g/ml bovine serum albumin, and 10% glycerol), separated by SDS-polyacrylamide gel electrophoresis, and transferred to nitrocellulose membranes. The membranes were probed with antibodies against CREB or phospho-CREB/ATF1 (New England BioLabs). Immunoreactive proteins were detected as described above.

RESULTS
Anisomycin Stimulates a Transient p38/SAPK2-dependent Induction of egr-1-To examine the regulation of egr-1 induction by the p38/SAPK2 pathway, Jurkat T-cells were treated with anisomycin for different times in the presence or absence of the specific p38␣,␤/SAPK2a,b inhibitor SB203580. p38/ SAPK2 activity was determined in cells lysates by kinase assays using His-tagged MK2 as a substrate. p38/SAPK2 activation was already detectable 30 min after stimulation and a high kinase activity was maintained for more than 3 h (Fig. 1A). p38/SAPK2 activation by anisomycin was almost completely blocked in the presence of SB203580 (Fig. 1A, ϩ SB). We also monitored JNK/SAPK1 activity in the cell lysates by using GST-c-jun-(1-79) as a substrate. JNK/SAPK1 shows a kinetics of anisomycin-stimulation similar to p38/SAPK2 (Fig. 1B). However, SB203580 treatment does not inhibit JNK/SAPK1 stimulation in Jurkat T-cells. Total RNA was isolated in parallel and subject to Northern blot analysis (Fig. 1C). egr-1 mRNA was not detectable in cells grown under control conditions. A low level of egr-1 mRNA becomes detectable after 30 min of anisomycin treatment, increases, and reaches its maximum at about 2 h. SB203580 completely blocked anisomycininduced egr-1 mRNA expression. The similar kinetics of p38/ SAPK2 activation and egr-1 expression and, especially, the similar effects of SB203580 indicate that the p38/SAPK2 pathway is directly involved in the regulation of egr-1 expression at the level of mRNA induction.
A Putative cAMP-responsive Element in the egr-1 Promoter Is Essential for p38/SAPK2-dependent Induction-To analyze the mechanism of egr-1 induction by the p38/SAPK2 pathway, human egr-1 promoter-CAT reporter gene constructs were transiently transfected into human 293 cells. The transfected cells were treated with anisomycin in the presence or absence of SB203580, and reporter gene activity was measured (Fig. 2). The full-length promoter (Ϫ600) revealed a high reporter gene activity after anisomycin treatment, which was reduced to approximately 50% in the presence of SB203580 (Fig. 2, Ϫ600). 5Ј-Deletion mutants of the egr-1 promoter, lacking potential binding sites for Egr-1, SP1 (Ϫ480), and serum response factor (Ϫ180), presented a lower reporter gene activity compared with the full-length promoter, but CAT activity showed a similar inhibition by SB203580 treatment (Fig. 2, Ϫ480, Ϫ235, Ϫ180). Further 5Ј-deletion of the egr-1 promoter (Ϫ116) resulted in a loss of the reduction of reporter gene activity by SB203580. A putative cAMP-responsive element (CRE) with the sequence 5Ј-TCACGTCA-3Ј was identified within position Ϫ134 to Ϫ126 in the human egr-1 promoter (55). Site-directed mutagenesis of this potential CRE to a CREB/ATF1 nonbinding sequence 5Ј-TCTCATCA-3Ј (Ϫ600m) (56) within the full-length promoter (Ϫ600) results in a complete loss of SB203580 sensitivity of the CAT reporter gene activity (Fig. 2, Ϫ600m). To make sure that the transcriptional effects observed are not specific for anisomycin or for the JNK/SAPK1 pathway but for the p38/SAPK pathway, we transfected human 293 cells with a constitutive active mutant of the p38/SAPK2 activator kinase MKK6, MKK6 EE (19). As shown in Fig. 2C, MKK6 EE also activates the egr-1 wild type promoter (Ϫ600) but not the mutated one (Ϫ600m). Together with the data from the deletion experiments, these findings demonstrate that p38/SAPK2-dependent egr-1 induction is mediated by the CRE within position Ϫ134 to Ϫ126. These results further indicate that transcription factors binding to this element are targets of the p38/SAPK2 pathway.
CREB and ATF1 Are Phosphorylated and Bind to the CRE as a Result of p38/SAPK2 Activation-To determine whether the transcription factors CREB and ATF1 were involved in egr-1 induction, we performed pull-down assays using a biotinylated DNA fragment spanning the distal CRE (nt Ϫ134 to Ϫ126) of the egr-1 promoter and nuclear extracts of Jurkat T-cells stimulated with anisomycin in the presence or absence of SB203580. Western blot analysis performed with phospho-CREB/ATF1 antibodies revealed that CREB and ATF1 are already phosphorylated to a low extent in nonstimulated Jurkat T-cells (Fig. 3A, nuclear extract, C). After anisomycin treatment CREB and ATF1 are highly phosphorylated (Fig. 3A,  nuclear extract, Ani). This phosphorylation is completely blocked by SB203580 treatment (Fig. 3A, nuclear extract, SB). Pull-down assays revealed that phospho-CREB and phospho-ATF1 bind to the wild type CRE of the egr-1 promoter (Ϫ226 to Ϫ109) in anisomycin-treated cells (Fig. 3, A and B, wtCRE,  Ani). In the presence of SB203580, no binding of CREB, phospho-CREB, or phospho-ATF1 could be detected (Fig. 3, A and B, wtCRE, SB). In cells grown under control conditions, phospho-ATF1 binds also to the wild type CRE (Fig. 3A, wtCRE, C). Binding of CREB, phospho-CREB, or phospho-ATF1 was not detected using the mutant CRE (Ϫ226 to Ϫ109, A-132T, G-130A) (Fig. 3, A and B, mCRE). These data indicate a phosphorylation-specific binding of CREB and ATF1 to the CRE. Because it is clear that p38/SAPK2 itself is not able to directly phosphorylate CREB and ATF1, it becomes highly probable that protein serine/threonine kinases downstream to p38/ SAPK2 are involved.
Analysis of the Involvement of Protein Kinases Downstream to p38/SAPK2-The only CREB-kinases downstream to p38/ SAPK2 known so far are MK2 and MSK1. To determine whether MK2 or MSK1 or both kinases are responsible for the anisomycin-induced SB 203580-dependent egr-1 induction, we decided to analyze stress-dependent erg-1 expression in cells lacking MK2 activity. For that reason, MEFs derived from mice with a targeted disruption of MK2 (Ϫ/Ϫ MEF) (68) and from wild type mice (ϩ/ϩ MEF) were treated with anisomycin for 120 min in the presence or absence of SB 203580. p38/SAPK2 and MK2 activation by anisomycin was verified in these cells by kinase assays using cell lysates and His-tagged MK2 or Hsp25 as a substrate. MSK1 activity were determined in parallel by immunocomplex kinase assays using CREBtide as sub- strate. In both ϩ/ϩ and Ϫ/Ϫ MEFs, p38/SAPK2 activity was increased after anisomycin stimulation, whereas in the presence of SB203580 p38/SAPK2 activity was hardly detectable (Fig. 4A). In Ϫ/Ϫ MEFs, no MK2 activity could be detected (Fig.  4B), but MSK1 activity shows the same inducibility as in ϩ/ϩ MEFs (Fig. 4C). By Western blot analysis using a phospho-CREB/ATF1-specific antibody, we demonstrate that CREB as well as ATF1 exhibits an anisomycin-induced SB203580-sensitive phosphorylation in both the Ϫ/Ϫ and ϩ/ϩ MEFs (Fig. 4D). Furthermore, Northern blot analysis reveals a comparable anisomycin-stimulated SB203580-sensitive increase in the egr-1 mRNA level in Ϫ/Ϫ and ϩ/ϩ MEFs (Fig. 4E). Taking together, these data show that MK2 is involved in neither CREB phosphorylation nor induction of egr-1 by p38/SAPK2, making MSK1 a good candidate to mediate CREB and ATF1 phosphorylation after stimulation of the p38/SAPK2 pathway in vivo.

DISCUSSION
The immediate-early gene egr-1 is rapidly and transiently induced in response to multiple mitogenic signals (41,49,56), by the serine/threonin protein phosphatase inhibitor okadaic acid (44,45), by diverse types of DNA damaging agents (46 -48), and by several stress treatments (42,50,57). Recently, it has also been shown that the muscarinic acetylcholine receptors activate the Egr family of transcription factors (58). In contrast to the numerous studies concerning egr-1 induction, the understanding of the signaling pathways contributing to egr-1 activation in response to stress and other stimuli is rather incomplete. In B-lymphocytes, egr-1 induction by antigen receptor stimulation is dependent on the p21 ras pathway (59). The insulin-induced egr-1 expression in 32D cells requires MEK1 activation but not phosphatidylinositol 3-kinase activity (43). In NIH-3T3 cells, stress treatment such as heat shock, sodium arsenite, ultraviolet radiation, and anisomycin induces egr-1 promoter activity involving p38/SAPK2 and/or JNK1/ SAPK1c (50). In contrast, in rat mesangial cells, lysophosphatidic acid-mediated egr-1 gene expression does not depend on p38/SAPK2 but on MEK1 activation (49). In our studies we have used anisomycin, a stress stimulus that does not activate ERKs (29, 32) but does activate JNK/SAPK1 and p38/SAPK2. Similar to the results obtained for NIH-3T3 cells (50), we found that anisomycin induction of egr-1 in human Jurkat T-cells is completely inhibited by SB203580, whereas JNK/SAPK1 activity remains unchanged. We suggest that in these cells p38␣,␤/ SAPK2a,b are the main kinases responsible for egr-1 gene expression after anisomycin stimulation.
The egr-1 promoter contains several putative regulatory elements, including two Sp1-binding sites, five putative SREs, two putative CREs, an Egr-1 binding site, and a tetra-decanoyl phorbol acetate-responsive element (55). In BA/F3 cells, activation of the egr-1 promoter by granulocyte/macrophage-colony stimulating factor is controlled by SREs (nt Ϫ56 to Ϫ116 and Ϫ235 to Ϫ480) through pathways that involve Janus kinase 2 and Ras (60). In TF-1 cells it has been demonstrated that, in addition to binding of serum response factor to the SRE, CREB is also bound to proximal CRE (nt-57 to Ϫ76) of the egr-1 promoter to induce its activity after granulocyte/macrophagecolony stimulating factor or IL-3 stimulation (56). The transcriptional activation of egr-1 by granulocyte/macrophage-colony stimulating factor necessitates the phosphorylation of FIG. 2. The distal CRE (nt ؊134 to ؊126) is necessary for activation of the egr-1 promoter by p38/SAPK2. A, schematic representation of the egr-1 promoter fragments fused to the CAT gene. The putative regulatory elements of the egr-1 promoter are indicated: EBS, Egr-1 binding site; Sp-1, SP1-binding G/C-rich element; SRE, serum response element; CRE, cAMP-responsive element. Successive truncations (Ϫ480, Ϫ235, Ϫ180, and Ϫ116) of the full-length (Ϫ600) promoter were examined. In addition, a loss of function mutation of the distal CRE (nt Ϫ134 to Ϫ126) in the full-length egr-1 promoter (Ϫ600m) was used. B and C, CAT reporter gene assay. B, 293 cells were transfected with the constructs shown in A using the LipofectAMINE reagent. 5 h after transfection, cells were stimulated by anisomycin in the presence (SB ϩ Ani) and absence (Ani) of SB203580 for 30 min. C, 293 cells were transfected with the fulllength promoter construct (Ϫ600) and the mutated construct (Ϫ600m) and co-transfected with pcDNA3-FLAG-MKK6 EE (19) and pcDNA3 as a control. After 24 h, cells were harvested and CAT activity was measured. The data shown (means Ϯ S.D.) were obtained from two different experiments performed in triplicate.
FIG. 3. CREB and ATF1 bind to the distal CRE in a phosphorylationdependent manner. Nuclear extracts from control (C), anisomycin (Ani)-treated, or anisomycin-and SB203580 (SB)-treated 293 cells were incubated with a biotinylated DNA fragment spanning the wild type CRE (wtCRE, nt Ϫ226 to Ϫ109) or the mutated CRE (mtCRE, Ϫ226 to Ϫ109, 132 A3 T, 130 G3 A). Bound proteins were identified by Western blotting using antibodies against phospho-CREB/ATF1 (A) and CREB (B). As a further control, nuclear extracts used for the binding assay were analyzed. The positions of molecular weight marker proteins are indicated on the left. CREB on serine 133 (61) but does not require activation of protein kinase A (62). In addition, it has been demonstrated that p38/SAPK2 and JNK/SAPK1 induce egr-1 promoter activity through Elk-1 (50). For human 293 cells, we demonstrate a p38/SAPK2-dependent activation of the egr-1 promoter only when the distal CRE (nt Ϫ134 to Ϫ126) is able to bind CREB and ATF1. Furthermore, we show that a point mutant of the full-length promoter (Ϫ600m), which is no longer able to bind phospho-CREB and -ATF1 to the distal CRE, completely lacks SB203580-sensitive and MKK6 EE -stimulatable promoter activity. These data strongly suggest that CREB/ATF1 binding to the distal CRE is crucial for induction of egr-1 by the p38/ SAPK2 pathway. However, it can not be excluded that other transcription factors and signaling pathways contribute to anisomycin-stimulated egr-1 expression in 293 cells in a SB203580-insensitive or MKK6-independent manner. The observation that, for the full-length promoter, about 50% of anisomycin-stimulated activity can not be inhibited by SB203580 as well as the reduction of the overall egr-1 promoter activity obtained as a result of deletion of the SREs (nt Ϫ480 to Ϫ235) may indicate a further contribution by JNK/SAPK1 through phosphorylation of Elk-1 as described (50). ATF1 and CREB are members of the CREB/ATF family implicated in cAMP-and calcium-induced transcriptional activation. Although CREB can bind to DNA as a homodimer (63), two other bZIP proteins, ATF1 and CREM, which are highly related to CREB, are known to form hetero-dimers with CREB (64 -66). We have shown that the phosphorylation of CREB and ATF1 after p38/SAPK2 activation allows their binding to the egr-1 promoter and causes a transient induction of egr-1.
Two downstream kinases of p38/SAPK2, MK2 and MSK1, have been implicated in the phosphorylation of CREB after activation of the p38/SAPK2 pathway. Whereas CREB has been shown to be a poor substrate for MK2 in vitro (30), MK2 has been demonstrated to be involved in stress-and fibroblast growth factor-dependent activation of CREB and ATF1 (67). Using cells that lack MK2 activity, we demonstrate that this enzyme is not necessary for the phosphorylation of CREB and ATF1 in vivo. Because the CREB-kinase MSK1 (28) is highly activated in a SB203580-dependent manner in MK2 Ϫ/Ϫ cells after anisomycin treatment, this enzyme is a good candidate for direct phosphorylation of these transcription factors. However, very recently a third CREB-kinase controlled by p38/SAPK2, RSK-B (52), was identified, which at this time can not be excluded as a contributor to anisomycin-induced egr-1 expression.