Disruption of the c-JUN-JNK Complex by a Cell-permeable Peptide Containing the c-JUN δ Domain Induces Apoptosis and Affects a Distinct Set of Interleukin-1-induced Inflammatory Genes*

The transcription factor activator protein (AP)-1 plays crucial roles in proliferation, cell death, and the immune response. c-JUN is an important component of AP-1, but only very few c-JUN response genes have been identified to date. Activity of c-JUN is controlled by NH2-terminal phosphorylation (JNP) of its transactivation domain by a family of JUN-NH2-terminal protein kinases (JNK). JNK form a stable complex with c-JUN in vitro and in vivo. We have targeted this interaction by means of a cell-permeable peptide containing the JNK-binding (δ) domain of human c-JUN. This peptide strongly and specifically induced apoptosis in HeLa tumor cells, which was paralleled by inhibition of serum-induced c-JUN phosphorylation and up-regulation of the cell cycle inhibitor p21cip/waf. Application of the c-JUN peptide to interleukin (IL)-1-stimulated human primary fibroblasts resulted in up-regulation of four genes, namely COX-2, MnSOD, IκBα, and MAIL and down-regulation of 10 genes, namely CCL8, mPGES, SAA1, hIAP-1, hIAP-2, pent(r)axin-3, CXCL10, IL-1β, ICAM-1, and CCL2. Only a small group of genes, namely pent(r)axin-3, CXCL10, ICAM-1, and IL-1β, was inhibited by both the c-JUN peptide and the JNK inhibitor SP600125. Thereby, and by additional experiments using small interfering RNA to suppress endogenous c-JUN we identify for the first time three distinct groups of inflammatory genes whose IL-1-induced expression depends on c-JUN, on JNK, or on both. These results shed further light on the complexity of c-JUN-JNK-mediated gene regulation and also highlight the potential use of dissecting signaling downstream from JNK to specifically target proliferative diseases or the inflammatory response.

The transcription factor activator protein 1 (AP-1) 1 was one of the first mammalian transcription factors to be identified, but its physiological functions are still being unraveled. AP-1 is involved in cellular proliferation, transformation, survival, cell death, and the immune response (1,2). AP-1 converts extracellular signals into changes of the expression of target genes, and AP-1 binding sites are found in a large number of genes. AP-1 is not a single protein, but a homo-or heterodimer composed of members of the JUN, FOS, ATF, and other protein families (1,2). With the exception of cell cycle regulatory genes, because of the structural and regulatory complexity of AP-1, the knowledge of AP-1 target genes mediating AP-1 functions is far from complete (1,2). c-JUN is one of the best characterized components of AP-1 (3). Genetic evidence suggests that c-JUN is essential for development and proliferation (1)(2)(3). Activity of c-JUN is controlled at multiple levels, first by changes in gene transcription, mRNA turnover, and protein stability, second by interaction with other transcription factors, and third by phosphorylation of its NH 2 -terminal transactivation domain (1)(2)(3)(4)(5)(6).
The phosphorylation sites required for inducible c-JUN activation have been mapped to serines 63 and 67. A family of 10 highly homologous serine/threonine protein kinases derived from three genes by alternative splicing has been identified that specifically phosphorylate these residues in c-JUN and has therefore been named c-JUN NH 2 -terminal protein kinases (JNK) 1-3 (7,8). So far, no other protein kinases have been identified that phosphorylate the NH 2 terminus of c-JUN (6,7). Importantly, early work demonstrated that JNK not only phosphorylate c-JUN, but also bind to a region called the ␦ domain, which is located immediately NH 2 -terminal of the c-JUN transactivation domain (6 -10). Binding of JNK to c-JUN is a remarkable example of a high affinity signaling complex (6 -10). Further studies of the complex in intact cells showed that it does not require the JNK catalytic activity or the presence of the phospho-acceptor sites in c-JUN (11). Presumably, the c-JUN-JNK interaction serves two purposes, first it provides the specificity of JNK for c-JUN, and second it helps to increase the local concentration of JNK at gene promoters that bind c-JUN, thereby enhancing c-JUN-mediated transcription (4).
Like the c-JUN protein, JNK have been implicated in numerous biological roles in response to growth factors, stress, and inflammatory cytokines, implying that JNK may mediate their gene regulatory effects mainly through c-JUN (12)(13)(14). However, for several reasons this is unlikely. First, JNK phosphorylate other transcription factor substrates, such as ATF-2 (15) and ELK-1 (16); second, individual JNK isoforms display different affinities for c-JUN in vitro (8,9); and third, the phenotypes of c-JUN and JNK knockout mice show no obvious overlapping phenotype (13,17,18). Therefore, one of the most intriguing questions regarding the c-JUN-JNK interaction is which of the many biological processes ascribed to both proteins critically requires a c-JUN-JNK complex.
We have addressed this question by designing a cell-permeable peptide containing the JNK-binding site of human c-JUN. We report here that this peptide specifically disrupts the c-JUN-JNK complex in vitro and in vivo. Thereby we identify a number of hitherto unrecognized genes whose expression depends on the c-JUN-JNK complex.

EXPERIMENTAL PROCEDURES
Cells and Materials-HeLa cells stably expressing the tet transactivator protein, kindly provided by H. Bujard, and primary human fibroblasts derived from gingiva were cultured in Dulbecco's modified Eagle's medium complemented with 10% fetal calf serum. [␥-32 P]ATP was purchased from Hartmann Analytics. Rabbit antibodies against phospho(Ser 63 ) c-JUN, c-JUN, phospho(Thr 183 /Tyr 185 ) JNK, and JNK were from Cell Signaling Technology (kits 9250 and 9260). Other rabbit antibodies against c-JUN (H-79) or ERK (C-14) were from Santa Cruz. Horseradish peroxidase-coupled secondary antibodies against rabbit IgG were from Sigma. Glutathione-Sepharose was from Amersham Biosciences. Human recombinant IL-1␣ and GST-JUN were produced as described (19). SP600125 was from Tocris. SMARTpool small interfering (si) RNA oligonucleotides against c-JUN were from Dharmacon.
Plasmids and Transfections-Plasmids pFR-Luc encoding five GAL4-binding sites upstream of a luciferase gene, pFC-MEKK1 encoding the catalytic domain of MEKK1, pFC2-dbd encoding the DNAbinding domain of GAL4 (amino acids 1-147), and pFA2 encoding the transactivation domain of c-JUN (amino acids 1-223) fused to the DNA-binding domain of GAL4 were obtained from Stratagene. Transient transfections by the calcium phosphate method were performed as described (20). For determination of promoter activity, cells (seeded at 1 ϫ 10 5 per well of 24-well plates) were transfected with 1 g of pFR-Luc, 100 ng of pFA2-cJUN, 100 ng of pFC2-dbd, or 50 ng of pFC-MEKK1. Equal amounts of plasmid DNA within each experiment were obtained by adding empty pCS3MT vector to a total amount of 2.2 g of DNA per well. Where indicated 100 M c-JUN peptide was added directly after transfection. 16 -20 h later cells from two wells transfected independently were pooled, lysed, and luciferase reporter gene activity was determined as described (20). For transfection of siRNA oligonucleotides OligofectAMINE TM (Invitrogen) or TransitIT-TKO TM (Mirus) were used, respectively, according to the manufacturer's instructions.
Peptide Synthesis-Peptides (0.1 mmol) were synthesized as COOHterminal amides on TentaGel R RAM resin in an Applied Biosystems Pioneer instrument, using standard Fmoc protocols (21). Briefly, Fmoc derivatives (0.4 mmol) were activated with N-[(dimethylamino)-1H-1,2,3-triazolo [4,5-b]pyridin-1-yl-methylene]-N-methylmethanaminiumhexafluorophosphate N-oxide (0.4 mmol) in the presence of diisopropylethylamine (0.8 mmol). Fmoc group removal was with a mixture of 2% (v/v) piperidine and 2% (v/v) 1,8-diazabicyclo(5,4,0)undec-7-ene in dimethylformamide. To minimize aspartimide formation, Asp in the scrambled sequence was coupled as the Asp-Ser pseudo-proline (22). When the sequence was complete, the resin was washed with methanol and peroxide-free ether, and dried under nitrogen before the addition of dichloromethane (4 ml) and dimethylpropylene urea (0.5 ml). 5(6)-Carboxyfluorescein (0.5 mmol), 7-hydroxy-1-azabenzotriazole (0.5 mmol), and diisopropylcarbodiimide (0.6 mmol) were added and the mixture was shaken gently for 16 h. A ninhydrin test (21) showed that the reaction was complete. Piperidine (0.1 ml) was added and the resin was washed with dimethylformamide, methanol, and diethyl ether before drying for 16 h in vacuo over P 2 O 5 . Peptides were released from the resin by treatment for 3-4 h at room temperature with a mixture of trifluoroacetic acid (9 ml), thioanisole (0.5 ml), dithiothreitol (0.25 g), and triisopropylsilane (0.25 ml) containing 0.15 g of ammonium iodide to prevent oxidation of methionine (23). The spent resin was removed by filtration and washed with a little trifluoroacetic acid. The pooled filtrate was evaporated in vacuo to an oil, which was added dropwise to 45 ml of peroxide-free ether at 0°C. The precipitated peptide was recovered by centrifugation at 720 ϫ g for 3 min and washed three times with 45 ml of ether by resuspension and centrifugation, before drying under a gentle stream of nitrogen gas. Crude peptides were purified by reverse phase high performance liquid chromatography on a column (22 ϫ 250 mm) of Vydac octadecyl-silica (15-20 m particle size) using a linear gradient of acetonitrile in water containing 0.1% trifluoroacetic acid. Fractions containing homogeneous product were identified by analytical high performance liquid chromatography on a column (4.6 ϫ 250 mm) of Vydac octadecyl-silica (5 m diameter). These fractions were pooled and the acetonitrile was removed by rotary evaporation in vacuo. The residue was diluted with 10% (v/v) acetic acid and freeze-dried. The identity of the purified peptides was confirmed by mass spectrometry: TAT-c-JUN peptide, expected mass 4729.6 Da, found 4730.4 Ϯ 0.3 (S.D., 6); TAT-scrambled (scr.) peptide, expected mass 4729.6 Da, found 4729.8 Ϯ 0.2 Da (S.D., 6). The sequence of the TAT peptide control is FlCO-YGRKKRRQRRR-4Abu-NH 2 (M r 2002.3). FlCO is fluoresceinyl made with 5(6)-carboxyfluorescein and 4-Abu is a residue of 4-aminobutyric acid. NH 2 is the COOH-terminal amide.
Protein Kinase Assays-Cells were harvested, washed in phosphatebuffered saline, and incubated for 15 min in ice-cold swelling buffer (5 mM Tris, pH 7.4). Then cells were lysed in 20 mM HEPES, pH 7.4, 2.5 mM MgCl 2 , 0.1 mM EDTA, 0.05% Triton X-100, 20 mM ␤-glycerophosphate, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, and 1 mM fresh phenylmethanesulfonyl fluoride (Sigma). Lysates were cleared by centrifugation at 10,000 ϫ g for 15 min at 4°C. Cell extract proteins (0.5 mg) were incubated with 2.5 g of GST-JUN previously immobilized on glutathione-Sepharose beads to adsorb JNK protein kinases. Where indicated, 100 M c-JUN peptide was added to the binding reaction. After incubation for 30 min at 30°C, beads were pelleted, extensively washed in cell lysis buffer, and resuspended in 10 l of the same buffer. Then 10 l of H 2 O and 10 l of kinase buffer (150 mM Tris, pH 7.4, 30 mM MgCl 2 , 60 M ATP, 4 Ci of [␥-32 P]ATP) were added. After 30 min at room temperature SDS-PAGE sample buffer was added, and proteins were eluted from the beads by boiling for 5 min. After centrifugation at 10,000 ϫ g for 5 min, supernatants were separated on 10% SDS-PAGE. Phosphorylated proteins were visualized by autoradiography.
Western Blotting-Cell extract proteins were separated on 10% SDS-PAGE and Western blotting was performed as described (19,20). Blots were stripped prior to reprobing with c-JUN, JNK, or ERK antibodies. Proteins were detected by using the Amersham enhanced chemiluminescence system.
Enzyme-linked Immunosorbent Assay-IL-8 protein concentrations in the cell culture medium were determined using the human IL-8 duo set kit (R&D Systems) exactly following the manufacturer's instructions.
Fluorescence Microscopy-Subcellular distribution of fluoresceinlabeled c-JUN and scrambled peptides was examined by phase-contrast and fluorescence microscopy at ϫ40 magnification using a Zeiss Axiovert 200M microscope. Cells were fixed for 15 min in 4% para-formaldehyde in phosphate-buffered saline. Nuclei of cells were stained with Hoechst 33342. Phase-contrast and fluorescence images of the same cells were collected in separate channels and images were saved as TIF files and processed electronically using Micrografix Picture Publisher Software 8.0.
Determination of Cell Number and DNA Synthesis-Cells were seeded in six-well plates and counted after the indicated treatments in a Neubauer chamber. For determination of DNA synthesis rates, 10 4 cells were seeded in 96-well plates and incubated with 0.5 Ci/well [ 3 H]thymidine (Hartmann Analytics) for the final 4 h of treatment. Radioactivity incorporated into cellular DNA was determined by liquid scintillation counting.
DNA Microarray Experiments-The microarray used in this study contains amino-modified oligonucleotides of 50 bp in length immobilized on panepoxy-coated glass slides (MWG Biotech). The oligonucleotide probes are complementary to several housekeeping genes and to 110 human genes that are strongly regulated during inflammation, which were selected by an extensive literature search using published resources. 2 With a few exceptions, three specific oligonucleotide probes per gene were designed by identifying unique sequences in these genes by a computer-based algorithm developed at MWG Biotech. The specificity of the probes for their respective target genes was then verified in a large number of biological experiments using different cell lines and inflammatory stimuli. 2 Fluorescent cRNA copies of mRNAs of cells treated as indicated in the legend of Table I were prepared by reverse transcription of 5 g of total RNA purified with a Qiagen RNeasy kit. RNA was treated with DNase and used for synthesis of double-stranded cDNA synthesis followed by fluorophore-cRNA synthesis. To be specific, the cDNA synthesis system from Roche and the MEGAscript T7 kit from Ambion were used as directed by the manufacturers, with 100 ng of double-stranded cDNA and 1.25 mM Cy3-UTP or 1.25 mM Cy5-UTP in each cRNA labeling reaction. Labeled cRNAs were hybridized individually to microarrays in pre-prepared hybridization solution (MWG Biotech) at 42°C overnight and then washed sequentially in 2ϫ SSC, 0.1% SDS, 1ϫ SSC, and 0.5ϫ SSC.
Hybridized arrays were scanned at maximal resolution on an Affymetrix 428 scanner at variable PMT voltage settings to obtain maximal signal intensities without probe saturation. Fluorescence intensity values from TIFF images of Cy3 or Cy5 channels were integrated into one value per probe and normalized by the MAVI software (MWG Biotech) and further analyzed using Imagene 4.2 software (Biodiscovery). Additionally, ratio data from probes with signal intensities of less than 10% of the average signal intensity in one or both channels were excluded from the analyzed data sets.

A Cell Permeable c-JUN Peptide Inhibits the c-JUN-JNK Interaction in Vitro and in Vivo-
The c-JUN transcription factor belongs to the basic region leucine-zipper proteins (3,4,10). It contains COOH-terminal DNA-binding and dimerization domains. The first half of the protein harbors the transactivation domain. This region contains serines 63 and 73 that are inducibly phosphorylated by JNK as well as the JNK-binding (␦) domain (Fig. 1A). We designed a synthetic peptide comprising the ␦ domain (amino acids 33-57) of human c-JUN fused to the protein transduction domain (amino acids 47-57) of the HIV-1 transactivator protein TAT ( Fig. 1A) (24). Protein transduction is a process of unknown mechanism that allows TAT and other proteins to traverse biological membranes in a receptor-and transporter-independent fashion. Thus, fusion to the protein transduction domain can be used to deliver proteins and peptides to intact cells (24). Initial experiments were performed to reveal if the c-JUN peptide had the potential to disrupt a c-JUN-JNK complex using a two-stage assay. First, endogenous JNK are isolated from cells by virtue of their binding to an immobilized recombinant GST-JUN-(1-135) fusion protein (19). Second, after removal of unbound proteins, the presence of active JNK in the complex is detected by phosphorylation of GST-JUN-(1-135) in vitro (Fig. 1B). Addition of c-JUN peptide to this assay efficiently disrupted the binding of activated affinity purified endogenous JNK in vitro as detected by disappearance of in vitro phosphorylation of GST-JUN-(1-135) (Fig. 1B). To evaluate if the peptide also inhibits c-JUN-specific gene expression, we activated a fusion protein of the DNA-binding domain of the yeast transcription factor GAL4 and the c-JUN transactivation domain by co-expression of the protein kinase MEKK1, a strong upstream activator of endogenous JNK (12). This resulted in activation of a luciferase reporter gene driven by a promoter containing five GAL4-binding sites (Fig. 1C). The c-JUN peptide completely inhibited activation of the GAL4-c-JUN protein by MEKK1 (Fig. 1C).
To further verify these results we analyzed the effect of the c-JUN peptide on basal and inducible phosphorylation of endogenous c-JUN. For this purpose we also designed a peptide with a scrambled sequence as shown in Fig. 1A as a control. As shown in Fig. 1D, treatment of synchronized HeLa cells by serum resulted in increased expression of c-JUN and the appearance of two phosphoforms of c-JUN, indicating the presence of at least two phosphorylation states. This is in agreement with the known multisite phosphorylation of c-JUN by JNK that, as outlined above, occurs primarily at serines 63 and 73, but also to a lesser extent at threonines 91, 93, and 95 (4, 6). The c-JUN peptide did not interfere with c-JUN expression. However, it almost completely inhibited the most slowly migrating hyperphosphorylated form of c-JUN (Fig. 1D, upper  panel). In contrast, the serum-induced phosphorylation of three different JNK isoforms was not inhibited by the c-JUN peptide, indicating that the peptide acts immediately downstream from JNK (Fig. 1D, lower panel). The scrambled peptide had no effect on serum-induced c-JUN phosphorylation indicating specificity of the effects observed for the c-JUN peptide. Taken together, the experiments shown in Fig. 1, B to D, indicate that the c-JUN peptide specifically prevented the interaction with and subsequent phosphorylation and activation of c-JUN by JNK in vitro as well as in vivo.
The c-Jun Peptide Causes Apoptosis-The results presented in Fig. 1, B-D, suggested that the c-JUN peptide could be used to identify c-JUN-JNK target genes in cells. Additional experiments with fluorophore-labeled peptides indicated a very efficient transfer in HeLa cells (Fig. 2). More importantly, both peptides evenly distributed within the cells, suggesting that the peptides, like endogenous c-JUN and JNK, located to the cytoplasm as well as to the nucleus (Fig. 2). The transduction of the c-JUN or the scrambled peptide was indistinguishable from that of a peptide containing the TAT sequence alone (Fig. 2). During these experiments cells treated with the c-JUN peptide, but not those exposed to the scrambled peptide, showed increased cell death that occurred at around 200 M and increased at higher doses (Fig. 3A). HeLa cells exposed to the c-JUN peptide started to round up and detach after a few hours. After 24 h of treatment at least 50% of the cells had died (Fig. 3B). This effect was specific for the c-JUN peptide, as it did not occur with the scrambled peptide, the TAT peptide, or in untreated cells. Staining of cells with Hoechst dye indicated fragmentation of nuclei resembling apoptosis (Fig. 3B). The pivotal role of c-JUN and AP-1 in apoptosis is well known, with evidence for both pro-and anti-apoptotic functions of c-JUN (1).
Here, we observed increased cell death by targeting the c-JUN activation by JNK, suggesting that in proliferating HeLa cells the c-JUN-JNK complex plays an anti-apoptotic role. Furthermore, we found that the number of living cells decreased on treatment with the c-JUN peptide (data not shown), suggesting that the c-JUN peptide negatively affected proliferation. Accordingly, cells treated with the c-JUN peptide, but not those treated with the scrambled peptide, showed a significant reduction in DNA synthesis (data not shown) and up-regulation of the cell cycle inhibitor p21 cip/waf (Fig. 3C), indicating that the c-JUN peptide indeed affected proliferation and that apoptosis was very likely the indirect consequence of cell cycle arrest.
Identification of Interleukin-8 as an AP-1 Target Gene That Is Inhibited by the JNK Inhibitor SP600125, but Not by the c-Jun Peptide-Activation of JNK and AP-1 is not restricted to growth factors, but is also of central importance to many genes involved in immune response, inflammation, or tissue remodeling (25). We therefore asked if the c-JUN peptide also interfered with gene expression induced by IL-1, a major pro-inflammatory cytokine.
To test this hypothesis we investigated the IL-1-induced expression of IL-8, a major human chemoattractant protein that is activated by a plethora of external stimuli and whose promoter contains a consensus AP-1-binding site (26). We have previously studied the signal-dependent regulation of IL-8 in great detail and have shown by both expression of JNK antisense RNA or JNK dominant-negative mutants that IL-8 expression requires JNK activation (19,20). In line with these results, treatment of HeLa cells with the novel JNK inhibitor SP60125 resulted in a 50% reduction of IL-1-inducible IL-8 secretion (Fig. 4A), reinforcing our earlier conclusion that the JNK pathway provides an important signal for maximal IL-8 secretion. In sharp contrast, by RT-PCR and enzyme-linked immunosorbent assay, we found that the c-JUN peptide did not inhibit IL-1-inducible IL-8 mRNA expression or protein secretion (Fig. 4, B and C). Fig. 4 were surprising and suggested that signaling from JNK to AP-1 target genes of the inflammatory response diverges at the level of the c-JUN-JNK complex. To further strengthen this conclusion, we screened the expression of 110 genes with known relevance to inflammation, such as cytokines, chemokines, and matrix metalloproteinases (MMP) After that time cells were fixed and phase-contrast and fluorescence microscopic images were recorded in separate channels. (25). This was achieved by a customized DNA microarray developed by our laboratory. On this microarray each oligonucleotide probe has been optimized thoroughly and evaluated for its ability to specifically detect its target gene. 3 In more than 250 experiments, we found that primary cells show a broader spectrum and much stronger expression of genes in response to IL-1 compared with tumor cell lines such as HeLa. 4 We therefore transduced the peptides into human primary fibroblasts derived from gingiva (HuGi) that, like HeLa cells, showed uptake of the c-JUN and scrambled peptides into cytosolic and nuclear compartments with a 100% transduction efficiency (Fig. 5).

Identification of a Novel Set of Distinct Inflammatory Genes That Is Up-or Down-regulated by the c-Jun Peptide-The experiments shown in
As judged by microarray analysis, 61 inflammatory genes were expressed in HuGi cells, 31 of which were induced by IL-1 by at least 2-fold and up to 100-fold as shown in Table I, column 3. To identify the genes that were affected by the c-JUN peptide, cells were pretreated with c-JUN or scrambled peptides and then stimulated with IL-1. We used this comparison for further analysis to specifically identify the genes that were affected by the c-JUN peptide and to exclude potentially unspecific effects on inducible gene expression by treatment of fibroblasts with cell-permeable peptides. Ratios of gene expression obtained from cells stimulated with IL-1 ϩ c-JUN peptide were divided by ratios of gene expression of cells treated with IL-1 ϩ scrambled peptide. This analysis revealed a number of genes that were either down-or up-regulated by the c-JUN peptide, but not by the scrambled peptide, and whose expression is, therefore, specifically affected by disrupting the c-JUN-JNK interaction (Table I, columns 4 and 5). As summarized in Fig. 6, of the IL-1-inducible genes, four were up-regulated by more than 25% by the c-JUN peptide, namely COX-2, MnSOD, IB␣, and MAIL. Ten genes were down-regulated by the c-JUN peptide by more than 25%, including CCL8, mPGES, SAA1, hIAP-2, pent(r)axin-3, hIAP-1, CXCL10, ICAM-1, IL-1␤, and CCL2. CCL2, which is also called MCP-1, was the most strongly affected gene (inhibition of about 90%). This result, as well as the lack of inhibition of IL-8, was confirmed by RT-PCR (Fig. 7). To our knowledge, none of these genes has been shown previously to be dependent on a c-JUN-JNK complex. An immediate question that arose from these results was how far this set of genes would overlap with JNK-dependent genes. For this purpose, we treated HuGi cells with 20 M SP600125, a concentration that in agreement with other studies (27,28) was effective at blocking IL-1-induced phosphorylation of endogenous c-JUN (data not shown). As shown in Table I, column 6, a significant number of genes was affected by the SP600125 inhibitor. Of the 31 IL-1-induced genes, we observed downregulation of 14 and up-regulation of 5 by more than 25% (Fig. 6).
Our results confirm the reported suppression of MMP-1, Red arrows indicate apoptotic morphology, respectively. In C, total RNA from HeLa cells treated as in A was isolated and p21 cip/waf and tubulin mRNA transcripts were amplified by RT-PCR, separated by agarose gel electrophoresis, and visualized by ethidium bromide staining. (27,28) and also identify novel genes inhibited by SP600125, such as MnSOD, PAI-2, and others (Fig. 6). SP600125 also caused inhibition of IL-8 (CXCL8) expression in HuGi cells, whereas the c-JUN peptide had no effect (Fig. 6), confirming the observations made in HeLa cells (Fig. 4). In addition, we also identify genes that are up-regulated, such as c-JUN, indicating that in HuGi cells, JNK negatively affects c-JUN expression (Fig. 6).

MMP-3, and COX-2 by SP600125
A very interesting result emerging from these experiments is that only a small group of genes is inhibited by both the c-JUN peptide and SP600125 (Fig. 6). These genes are pent(r)axin-3, CXCL10, ICAM-1, and IL-1␤. Taken together, we have identified several novel target genes of c-JUN or JNK by means of the JNK inhibitor SP600125 or the c-JUN peptide containing the JNK-binding ␦ domain.
The biological role of c-JUN as suggested by the results of these experiments was further confirmed by suppression of endogenous c-JUN protein by double-stranded siRNA molecules (Fig. 8). Significant reduction of c-JUN protein in HeLa cells (Fig. 8B) causes apoptosis (Fig. 8A). These effects were specific, as they did not occur with transfection reagent alone (Fig. 8, A and B) or with irrelevant siRNA directed against luciferase (data not shown). Furthermore, application of c-JUN siRNA to human gingival fibroblasts inhibited IL-1-induced CCL2 expression by 50% but did not affect IL-8 expression (Fig. 8C). Compared with the c-JUN peptide (Table I and Fig.  6) siRNA against c-JUN were somewhat less effective in CCL2 suppression (Fig. 8C). This most likely results from the difference between transduction efficiency of the cellpermeable peptide, which is 100% (Figs. 2 and 5), versus transfection efficiency of siRNA, which in HuGi was about 50 -80% (data not shown). Accordingly, there was still some c-JUN protein detectable (Fig. 8D) that is likely to account for the residual IL-1-induced CCL2 expression. In conclusion, based on the compelling evidence provided here, we identify for the first time three distinct groups of inflammatory genes whose IL-1induced expression depends on c-JUN, on JNK, or on both proteins (Fig. 6). DISCUSSION The transcription factor AP-1 and its component c-JUN are of central importance for enabling cells to respond to environmental changes. Recently, the usage of genetically altered mice and cells derived from them has unraveled crucial functions for AP-1 as a regulator of cell life and death (1,2). However, very little information is available from these model systems on the role of AP-1 in inflammation and infection, despite the fact that AP-1-binding sites are found in many genes activated during innate and adaptive immune reactions (1,2). Furthermore, because of its complex regulation, the importance of AP-1 for human disease is still unclear (29). It is, therefore, desirable to develop pharmacological inhibitors to dissect signaling pathways leading to activation of AP-1 to conclusively identify AP-1 target genes (1, 29).
In addressing this question, we have focused on the long known specific interaction of c-JUN with its activating kinase

TABLE I
Comparison of the effects of the c-JUN peptide and the JNK inhibitor SP600125 on IL-1-inducible inflammatory gene expression Human primary fibroblasts were treated with IL-1 (10 ng/ml) for 4 h, or left untreated (0). In parallel cells were treated with 400 M c-JUN peptide or 400 M scrambled peptide for 1 h, or, with 20 M SP600125 for 30 min prior to stimulation with IL-1 for 4 h. Thereafter total RNA was isolated from all samples and used to prepare double-stranded cDNA followed by cRNA synthesis. cRNA was labeled with Cy3 and hybridized independently to DNA microarrays containing amino-modified oligonucleotide probes representing 110 genes relevant to inflammation as well as a number of housekeeping genes. Each gene was detected by three independent oligonucleotide probes. Fluorescence intensities of bound cRNAs were recorded, normalized, and used to identify 61 inflammatory genes that were significantly expressed. Alterations imposed by the different treatments on the IL-1-inducible or on the basal expression of these genes were determined as ratio of relative gene expression compared to unstimulated cells. For each comparison shown in columns 3-6, the mean ratio Ϯ S.E. of normalized dye fluorescence intensities from at least two independent experiments was calculated. Genes are ordered according to their relative induction by IL-1 as shown in column 3. In columns 1 and 2 GenBank™ accession numbers and gene names, respectively, are provided for identification. JNK (6). Here, we report that a cell-permeable peptide containing the minimal JNK-binding domain of human c-JUN efficiently and rapidly enters cells and affects c-JUN-specific gene expression. When applied to spontaneously growing HeLa cells, the peptide, like suppression of endogenous c-JUN protein by siRNA, caused apoptosis (Figs. 3, A and B, and 8). The pivotal role of c-JUN and AP-1 in apoptosis is well known, but the underlying mechanism is controversial with evidence for both, pro-and anti-apoptotic functions of c-JUN, depending on the cellular context (1, 2). Here, we observed increased cell death by targeting the c-JUN activation by JNK, suggesting that in proliferating HeLa cells the c-JUN-JNK complex plays an anti-apoptotic role. Two models are currently used to explain the role of c-JUN and the AP-1 complex in apoptosis. Either AP-1 is required for expression of pro-or anti-apoptotic regulators of apoptosis, or, AP-1 functions as a homeostatic regulator that keeps cells in a certain proliferative state in response to growth factors. Inhibition of AP-1 then results in cell cycle arrest and subsequent removal by apoptosis of cells unable to re-enter the cell cycle (1,2). The data presented in Fig. 3 strongly suggest that the latter scenario is evoked by the c-JUN peptide, which by disrupting the c-JUN-JNK interaction might prevent serum-induced c-JUN phosphorylation during the cell cycle. Interestingly, our data also suggest that c-JUN NH 2 -terminal phosphorylation (JNP), which is inhibited by the peptide, rather than the c-JUN expression level, which is not affected by the peptide, is more important for this effect to occur (Fig. 1D). This assumption is strongly supported by observations showing that JNP increases during G 2 -M transition in HeLa cells, whereas c-JUN levels remain stable (30). Also, fibroblasts from mice expressing a non-phosphorylatable c-JUN S63A/S73A mutant have a defect in proliferation (31). Unlike c-JUN Ϫ/Ϫ animals, c-JUN S63A/S73A transgenic mice are viable, further indicating that JNP mediates only a part of the spectrum of c-JUN-dependent functions (31). Additional evidence for a selective role of JNP in cell proliferation and apoptosis is provided by experiments using antisense oligonucleotides to inhibit JNK, which causes the same phenotype as observed by using the c-JUN peptide, namely inhibition of tumor cell growth and up-regulation of the cell cycle inhibitor p21, followed by apoptosis (32)(33)(34).
Thus, the cell-permeable c-JUN peptide described here is a novel molecular tool that can be used to acutely and specifically inhibit JNP. It should provide an important tool to address the role of c-JUN in cell cycle control in those situations where cells cannot be genetically manipulated, or, may even be applicable as an efficient means of treating human diseases such as tumors that require c-JUN-JNK for proliferation and survival.
Based on the compelling evidence of c-JUN for a positive regulatory role in cell proliferation, the strong effects of the c-JUN peptide described in the first part of our study might be expected. We therefore did not attempt to identify other c-JUN target genes in addition to p21 involved in cell cycle control or apoptosis.
Rather we extended our investigations on the biological effects of this peptide to other potential AP-1 target genes, namely those of the inflammatory response. We have previously demonstrated in a number of studies that JNK are crucial for the expression of IL-8 and IL-6 in epithelial cells, such as HeLa, KB, or HEK293 (19,20). In addition, studies using JNK1-or JNK2-deficient animals, or, a novel JNK inhibitor, SP600125 (27,28), have clearly established an important role FIG. 6. Divergent effects of the c-JUN peptide or the JNK inhibitor SP600125 on IL-1-inducible gene expression of human primary fibroblasts. A, genes that were up-regulated at least 2-fold by IL-1, as shown in Table I, were selected. Specific effects of the c-JUN peptide on IL-1 inducible expression of these genes were identified by comparing the ratios of IL-1 ϩ c-JUN peptide (Table I, (Table I, column 5). For SP600125, ratios of IL-1 (Table I, column 3) were compared with IL-1 ϩ SP600125 (Table I, column 6). Values above 125% (black bars) or below 75% (hatched bars) indicate the relative induction or suppression by more than 25% caused by the c-JUN peptide or SP600125, respectively. Genes, whose IL-1-inducible expression is not changed are shown in gray. Genes are ordered according to the effects of the c-JUN peptide. B, summary of the three groups of genes that are either up-(ϩ) or downregulated (Ϫ) by the c-JUN peptide, by SP600125, or by both. FIG. 7. The c-JUN peptide inhibits IL-1-induced mRNA expression of CCL2 (MCP-1), but not of IL-8, in human primary fibroblasts. Human gingival fibroblasts were treated as described in the legend of Table I and CCL2, IL-8, and tubulin mRNA expression was analyzed by RT-PCR.
for the JNK pathway in regulating a wide spectrum of other inflammatory genes such as MMP-1 (collagenase-1), MMP-3, MMP-13, COX-2, IL-2, interferon-␥, or tumor necrosis factor-␣ (27,28). Like collagenase, which was one of the first AP-1 response genes identified (35), these genes contain AP-1-binding sites. From this it may be deduced that JNK might generally regulate transcription of inflammatory genes by phosphorylating c-JUN, keeping in with the paradigm of the JNK-c-JUN-AP-1 signaling pathway (1)(2)(3)(4). However, the results of our study suggest that this is clearly an oversimplification. Like collagenase, the IL-8 promoter contains a typical AP-1binding site that is required for JNK-mediated IL-8 transcription (20,26). However, in contrast to the JNK inhibitor SP600125, or to expression of antisense RNA to JNK (19), the c-JUN peptide did neither inhibit IL-8 mRNA expression nor protein secretion (Fig. 4) at concentrations at which it inhibited serum-induced c-JUN phosphorylation (Fig. 1D) and caused apoptosis (Fig. 3). IL-8 expression was also not affected by suppression of endogenous c-JUN protein by siRNA (Fig. 8, C  and D). These results prompted us to investigate many more genes with relevance to inflammation in terms of their sensitivity to SP600125 or the c-JUN peptide.
Thereby we identify for the first time a distinct group of genes whose IL-1-induced expression is either specifically enhanced or inhibited by the c-JUN peptide (Table I, Fig. 6). Because we demonstrated in initial experiments that the c-JUN peptide inhibits c-JUN phosphorylation in vitro as well as in vivo (Fig. 1) we conclude that this group of genes requires JNP for activation and is sensitive to disruption of a c-JUN-JNK complex by the c-JUN peptide. Very interestingly, the JNK inhibitor SP600125 affects a significantly larger set of genes, suggesting that JNK may regulate theses genes by substrates other than c-JUN (Fig. 6). Only four genes are suppressed by both the c-JUN peptide and the JNK inhibitor, suggesting that they are activated by the "classical" JNK-c-JUN signaling cascade (Fig. 6).
Interestingly, we also identified genes whose expression was suppressed by the c-JUN peptide, but not by the JNK inhibitor (Table I, Fig. 6). The most drastic example of this group was CCL2, also called MCP-1, a chemokine whose IL-1-induced expression was impaired by 90% by the c-JUN peptide (Table I, Figs. 6 and 7), or, by siRNA directed against endogenous c-JUN protein (Fig. 8, C and D), but was unaffected by SP600125. One explanation for this discrepancy might be that c-JUN bound to the CCL2 promoter is phosphorylated by a particular JNK isoform that is not efficiently inhibited by SP600125. It is not known if SP600125 inhibits all 10 JNK isoforms in intact cells, but it is a less potent inhibitor of JNK3 in vitro (27,28).
The selectivity of IL-1 response genes to the c-JUN peptide might also be explained by a number of other considerations. To induce apoptosis and to inhibit phosphorylation of c-JUN in intact cells we had to use concentrations of the cell-permeable peptides around 400 M. These doses were similar to concentrations that were required in another study to inhibit the Nemo-IKK interaction (36). All JNK isoforms phosphorylate c-JUN, but they vary in their affinity to c-JUN (8,9) and it is possible that we affected only the less stable c-JUN-JNK complexes with the concentrations of c-JUN peptide used in this study.
It is also possible that c-JUN-JNK complexes with different sensitivity to the c-JUN peptide might result from interactions of the c-JUN-JNK complex with proteins that may weaken or stabilize the c-JUN-JNK interaction, such as the recently discovered novel interaction partners of c-JUN, namely RNA helicase RHII/GU (37), or of JNK, namely Sab(SH3BP5) (38).
The JNK signaling cascade is organized into modules formed by a complex of JNK, MKK4, or MKK7 and one of the many mitogen-activated protein kinase kinase kinases that activate JNK, such as MEKK1. Accessibility of the c-JUN peptide to JNK within these multiprotein complexes may vary according to the nature of the docking domains and scaffold proteins that tether the JNK signaling module (39,40).
Finally, hitherto undetected c-JUN protein kinases that do not bind to the ␦ domain may contribute to c-JUN activation. In this case the c-JUN peptide would be unable to interfere with c-JUN phosphorylation. The existence of additional JUN kinases has been suggested by several groups (30,31,41). Alone or in combination these mechanisms may render genes more or less susceptible to the c-JUN peptide.
Many of the genes affected or not affected by the c-JUN peptide contain known AP-1 sites. An important conclusion from our results, therefore, is that the sole presence of AP-1 sites is insufficient to predict if a gene is a target of a c-JUN-JNK signaling complex. Further experiments are required to identify the underlying mechanisms that result in sensitivity of the different genes identified here to either the c-JUN peptide, or to the JNK inhibitor. These results may than provide a basis to understand the complexity of transcriptional regulation of the inflammatory genes via AP-1.
In summary, we have identified several novel target genes of c-JUN or JNK by means of the c-JUN peptide containing the JNK-binding domain, by the JNK inhibitor SP600125, or by siRNA directed against c-JUN. With regard to inflammation, these genes fall into three distinct groups whose IL-1induced expression depends on c-JUN, JNK, or on both proteins (Fig. 6B).
Collectively, our results show that at the level of the c-JUN-JNK complex signals from growth factors or inflammatory cytokines can be specifically disrupted by a cell-permeable peptide to block proliferation and survival, or, inflammatory gene expression, respectively (Fig. 9). The results shed further light on the complexity of c-JUN-JNK-mediated gene regulation and also highlight the potential use of dissecting signaling downstream from JNK to specifically target proliferative diseases or the inflammatory response.