Molecular Basis for the Direct Inhibition of AP-1 DNA Binding by 15-Deoxy-Δ12,14-prostaglandin J2*

Cyclopentenone prostaglandins may interfere with cellular functions by multiple mechanisms. The cyclopentenone 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) has been reported to inhibit the activity of the transcription factor AP-1 in several experimental settings. We have explored the possibility of a direct interaction of 15d-PGJ2 with AP-1 proteins. Here we show that 15d-PGJ2 covalently modifies c-Jun and directly inhibits the DNA binding activity of AP-1. The modification of c-Jun occurs both in vitro and in intact cells as detected by labeling with biotinylated 15d-PGJ2 and mass spectrometry analysis. Attachment of the cyclopentenone prostaglandin occurs at cysteine 269, which is located in the c-Jun DNA binding domain. In addition, 15d-PGJ2 can promote the oligomerization of a fraction of c-Jun through the formation of intermolecular disulfide bonds or 15d-PGJ2-bonded dimers. Our results identify a novel site of interaction of 15d-PGJ2 with the AP-1 activation pathway that may contribute to the complex effects of cyclopentenone prostaglandins on the cellular response to pro-inflammatory agents. They also show the first evidence for the induction of protein cross-linking by 15d-PGJ2.

Cyclopentenone prostaglandins may interfere with cellular functions by multiple mechanisms. The cyclopentenone 15-deoxy-⌬ 12,14 -prostaglandin J 2 (15d-PGJ 2 ) has been reported to inhibit the activity of the transcription factor AP-1 in several experimental settings. We have explored the possibility of a direct interaction of 15d-PGJ 2 with AP-1 proteins. Here we show that 15d-PGJ 2 covalently modifies c-Jun and directly inhibits the DNA binding activity of AP-1. The modification of c-Jun occurs both in vitro and in intact cells as detected by labeling with biotinylated 15d-PGJ 2 and mass spectrometry analysis. Attachment of the cyclopentenone prostaglandin occurs at cysteine 269, which is located in the c-Jun DNA binding domain. In addition, 15d-PGJ 2 can promote the oligomerization of a fraction of c-Jun through the formation of intermolecular disulfide bonds or 15d-PGJ 2 -bonded dimers. Our results identify a novel site of interaction of 15d-PGJ 2 with the AP-1 activation pathway that may contribute to the complex effects of cyclopentenone prostaglandins on the cellular response to pro-inflammatory agents. They also show the first evidence for the induction of protein cross-linking by 15d-PGJ 2 .
Cyclopentenone prostaglandins (cyPG) 1 are products of the cyclooxygenase pathway, which are generated by the spontaneous dehydration of certain PG. Dehydration of PGD 2 gives rise to PGJ 2 and 15-deoxy-⌬ 12,14 -prostaglandin J 2 (15d-PGJ 2 ), whereas PGE 2 is converted into PGA 2 (1). cyPG have been detected in situations associated with an increase in cyclooxygenase-2 expression and PG production, such as acute and chronic inflammatory conditions (2)(3)(4). In addition, generation of cyPG has been reported in several experimental models, such as activated macrophages and mesangial cells (3,5). cyPG have received considerable attention during the past decade due to their action as antiproliferative agents and inducers of apoptosis in a variety of normal and cancer cell types (6,7). cyPG also display anti-inflammatory effects both in cellular and in animal models (2,8,9). This has led to the proposal that cyPG, and in particular 15d-PGJ 2 , may contribute to the resolution of inflammation through various mechanisms that include the inhibition of the expression of pro-inflammatory genes, the induction of apoptosis in activated macrophages, and the activation of the transcription factor peroxisome proliferator activated receptor ␥ (PPAR␥) (10,11). Thus, cyPG have been envisaged as potential therapeutic agents against inflammatory and proliferative diseases (1).
cyPG have been found to reduce the activity of transcription factors key for the inflammatory response, like NF-B and AP-1, by acting at multiple levels. 15d-PGJ 2 has been shown to interfere with the signaling pathway leading to NF-B activation through the direct inhibition of the IB kinase (12,13), the PPAR␥-dependent sequestration of coactivators (14), and the impairment of the ability of NF-B to bind DNA (15,16). Inhibition of DNA binding is the consequence of the formation of an adduct between 15d-PGJ 2 and NF-B proteins, which has been demonstrated both in vitro and in intact cells (16). The transcription factor AP-1 is formed by homo-or heterodimers of proteins of the Fra family. 15d-PGJ 2 has been shown to reduce AP-1 activation in several experimental systems (11,17) by mechanisms that may involve activation of PPAR␥ (11,18) and/or inhibition of c-Jun NH 2 -terminal kinase (JNK) (19). However, the possibility of a direct interaction of cyPG with AP-1 proteins has not been explored.
cyPG are reactive compounds that possess an ␣,␤-unsaturated carbonyl group in the cyclopentenone ring. This group may react with sulfhydryl groups of cysteine residues of proteins by Michael's addition (20), resulting in an alteration of protein function (16). The DNA binding domain of human c-Jun possesses two functionally important cysteine residues. Cys 269 is located in close contact with DNA, and it is involved in the redox and nitric oxide-dependent regulation of DNA binding by means of specifically targeted S-glutathionylation (21)(22)(23). Cys 320 is located in the leucine zipper region and participates in c-Jun homo-or heterodimerization with proteins of the Fos family through the formation of an intermolecular disulfide bridge (24). Here we show that 15d-PGJ 2 can form a covalent adduct with c-Jun and directly inhibit DNA binding. These observations unveil a novel PPAR-␥-independent mechanism that may contribute to the overall effect of cyPG on the AP-1 signaling pathway.  1 The abbreviations used are: cyPG, cyclopentenone prostaglandins; 15d-PGJ 2 , 15-deoxy-⌬ 12,14 -prostaglandin J 2 ; PPAR, peroxisome proliferator activated receptor; NF-B, nuclear factor-B; EMSA, electrophoretic mobility shift assay; HRP, horseradish peroxidase; Me 2 SO, dimethyl sulfoxide; MALDI-TOF, matrix-assisted laser desorption ionization-time of flight; JNK, c-Jun NH 2 -terminal kinase; DTT, dithiothreitol; r.t., room temperature; PG, prostaglandin; RMC, rat mesangial cells.
Cell Culture-Rat mesangial cells (RMC) (25) and HeLa cells were grown in RPMI 1640 medium (Invitrogen) containing 10% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. NIH-3T3 fibroblasts and RAW264.7 macrophages were grown in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum and antibiotics. Treatment with 15d-PGJ 2 was carried out in serum-and phenol red-free medium. Levels of 15d-PGJ 2 in the cell culture supernatant of RAW264.7 macrophages, incubated in the absence or presence of 10 g/ml bacterial lipopolysaccharide for 24 h, were assessed with an enzyme immunoassay from Assay Designs, Inc. (Ann Arbor, MI) following the instructions of the manufacturer.
Preparation of Total Cell Lysates and Nuclear Extracts-Cells were lysed in 50 mM Tris-HCl, pH 7.5, 0.1 mM EDTA, 0.1 mM EGTA, 0.1 mM ␤-mercaptoethanol, 1% Nonidet P-40, or RIPA buffer containing protease inhibitors for 30 min on ice. DNA was disrupted by several forced passes through a narrow needle, and cellular debris was removed by centrifugation for 5 min at 10,000 ϫ g at 4°C. Nuclear extracts were obtained as described previously in detail (26). Protein content was determined using the BCA protein assay from Pierce for the cell lysates and the Bradford protein assay from Bio-Rad for the nuclear extracts.
Electrophoretic Mobility Shift Assays (EMSAs)-Sense and antisense oligonucleotides corresponding to the AP-1 site from the human preproendothelin-1 promoter (5Ј-TGTTGGTGACTAATAACAC-3Ј), located at Ϫ114 to Ϫ96 bases from the transcriptional start site, were annealed and labeled by incubation with [␣-32 P]dCTP (PerkinElmer Life Sciences) as described previously (26). Two ng of the labeled doublestranded oligonucleotides (ϳ20,000 cpm) were used per binding reaction. For EMSA of recombinant human c-Jun, 0.5 M c-Jun treated with vehicle or 15d-PGJ 2 was incubated for 30 min at r.t. in the presence of labeled oligonucleotide, 0.2 mg/ml bovine serum albumin, and 10 g/ml poly(dI-dC) carrier in binding buffer (20 mM Tris, pH 7.5, 50 mM NaCl, 5 mM MgCl 2 , 1 mM EDTA, 5% glycerol, 0.01% Nonidet P-40). For EMSA of nuclear extracts, incubation reactions containing 4 g of protein were carried out for 20 min on ice. Unless stated otherwise, the DTT final concentration was 0.1 mM. Protein-DNA complexes were separated by electrophoresis on non-denaturing 6% polyacrylamide gels and visualized by autoradiography. For competition experiments, a 200-fold excess of unlabeled AP-1 oligonucleotide or of an unrelated oligonucleotide of sequence 5Ј-GGAGAGGGGATTCCCTGCG-3Ј was added to the binding reaction to ensure the specificity of the AP-1-DNA complexes. The intensity of the complexes formed was quantitated by densitometry of the autoradiographic exposures of gels.
Analysis of the Interaction between 15d-PGJ 2 and Recombinant c-Jun in Vitro-Full-length recombinant human c-Jun or purified c-Jun fragments were incubated at a final concentration of 5 M in 20 mM Tris-HCl, pH 7.0, 45 mM NaCl, 5 mM MgCl 2 , 0.14 mM ␤-mercaptoethanol, 0.1 mM DTT, 0.01% Nonidet P-40, 1% glycerol in the presence of 15d-PGJ 2 , biotinylated 15d-PGJ 2 , or vehicle (Me 2 SO) for 2 h at r.t. The incubation mixtures were analyzed either by MALDI-TOF mass spectrometry or by SDS-PAGE on 15% polyacrylamide gels followed by staining with Coomassie Blue or Western blot and detection of incorporated biotin with HRP-conjugated streptavidin and ECL (Amersham Biosciences). By using the c-Jun-(223-327) construct as an acceptor, the incorporation of biotinylated 15d-PGJ 2 (from 2 to 20 M) was proportional to the concentration of protein and PG, as estimated from the densitometry of the ECL exposures.
MALDI-TOF Analysis-For analysis of control and 15d-PGJ 2 -treated c-Jun, trifluoroacetic acid was added to the incubation mixtures at a final concentration of 0.1%, and peptides were purified on ZipTip C18 (Millipore, Bedford, MA). The laser desorption/ionization experiments were performed on a BIFLEX III time-of-flight instrument (Bruker-Franzen Analytik, Bremen, Germany) operated in the positive mode. A saturated solution of sinapinic acid in acetonitrile/water (1:2) with 0.1% trifluoroacetic acid was used as the matrix. Equal volumes (0.5 l) of the sample solution and the matrix were spotted on the target and air-dried. External calibration was performed, using the protein calibration standard II (Bruker Daltonics, Bremen, Germany), and samples were analyzed in the linear mode. For analysis of c-Jun monomer and dimer, 15d-PGJ 2 -treated c-Jun was subjected to SDS-PAGE, and gels were stained with Coomassie Blue. Bands corresponding to the monomer and dimer species were excised and digested in-gel with trypsin. The resulting peptides were analyzed by MALDI-TOF mass spectrometry in the reflectron mode. A saturated solution of ␣-cyano-4-hydroxycinnamic acid in acetonitrile/water (1:2) with 0.1% trifluoroacetic acid was used as the matrix. External calibration was performed, using the monoisotopic peaks of angiotensin (m/z 1046.5), adrenocorticotropic hormone (m/z 2,465.2), and the matrix (␣-cyano-4-hydroxycinnamic acid, m/z 379) recorded in a single spectrum. Typically, 50 -100 laser shots were summed into a single mass spectrum for analysis.
Analysis of c-Jun Dimerization-Dimerization assays were performed essentially as described (22). c-Jun-(223-327) was incubated under the conditions described above, except that DTT final concentration was 0.5 mM, in the absence or presence of 10 M 15d-PGJ 2 or 1 mM H 2 O 2 for 30 min at 37°C. Reactions were stopped by the addition of iodoacetamide at a final concentration of 50 mM and a further 30-min incubation at r.t. The reversibility of dimer formation was assessed by addition of 10 mM DTT and incubation at r.t. for 30 min before iodoacetamide addition. Aliquots of the incubation mixtures containing 1.7 g of protein were analyzed by non-reducing SDS-PAGE on 15% polyacrylamide gels. Gels were stained with Coomassie Blue R-250 and analyzed by image scanning.
Binding of Biotinylated 15d-PGJ 2 to c-Jun in Intact Cells-HeLa or NIH-3T3 cells were incubated in the presence of 10 M biotinylated 15d-PGJ 2 for 2 h. Cells were lysed in RIPA, and biotinylated proteins were purified by adsorption onto Soft-Link avidin (Promega, Madison, WI) following the manufacturer's instructions. The presence of c-Jun, or other proteins of interest, in the eluate was assessed by Western blot. In addition, cell lysates were immunoprecipitated with anti-c-Jun antibodies, and the incorporation of biotinylated 15d-PGJ 2 into the immunoprecipitated protein was monitored by Western blot and detection with HRP-conjugated streptavidin.

15d-PGJ 2 Inhibits AP-1 DNA Binding Activity Both in Intact
Cells and in Vitro-15d-PGJ 2 has been reported to inhibit the transcriptional activity of AP-1 and to reduce its DNA binding activity in several cell types (11,19,27), through PPAR␥-dependent or -independent mechanisms. We were interested in studying whether the direct modification of AP-1 constituents by 15d-PGJ 2 may be involved in the inhibitory effect. Treatment of HeLa cells with 10 M 15d-PGJ 2 reduced the DNA binding activity of AP-1, as evaluated by EMSA, by ϳ30% (Fig.  1A). A similar inhibitory effect was evidenced in RMC (Fig. 1B). c-Jun and c-Fos proteins are the major components of the AP-1-DNA complex detected in nuclear extracts from cytokinestimulated RMC as deduced from the blocking effect of specific antibodies (Fig. 1C). The decrease in AP-1-DNA complex formation elicited by 15d-PGJ 2 was not the result of a reduction in the nuclear levels of c-Jun and c-Fos proteins, either in RMC (Fig. 1D) or in HeLa (not shown). Interestingly, the inhibitory effect could also be observed in a cell-free system by incubating nuclear extracts isolated from cytokine-stimulated cells with 15d-PGJ 2 in vitro ( Fig. 2A). The extent of inhibition was similar when extracts from HeLa cells, in which AP-1 activity derives primarily from the c-Jun protein (28, 29) ( Fig. 2A), or from RMC (not shown) were used. In order to explore whether 15d-PGJ 2 directly interacts with AP-1 proteins, we used recombinant c-Jun as a model. Incubation of c-Jun with 15d-PGJ 2 led to a marked inhibition of the ability of the purified protein to bind DNA (Fig. 2B). These results show that 15d-PGJ 2 directly interferes with c-Jun DNA binding.
15d-PGJ 2 Forms a Covalent Adduct with c-Jun-To characterize the interaction between 15d-PGJ 2 and c-Jun, we used a biotinylated 15d-PGJ 2 derivative. Incubation of recombinant c-Jun with biotinylated 15d-PGJ 2 led to the formation of a c-Jun-15d-PGJ 2 adduct that was resistant to electrophoresis under denaturing conditions, as evidenced by Western blot and detection with HRP-conjugated streptavidin (Fig. 3A). Labeling of c-Jun was completely prevented by excess 15d-PGJ 2 , suggesting that this interaction is specific. The incorporation of 15d-PGJ 2 was dose-dependent and could be detected with nanomolar concentrations of this cyPG (Fig. 3B). We have shown above that the interaction of 15d-PGJ 2 with c-Jun precludes DNA binding. We then explored whether DNA binding by c-Jun would prevent incorporation of 15d-PGJ 2 . As shown in Fig. 3C, preincubation of c-Jun with an oligonucleotide containing the consensus site for binding of AP-1 markedly reduced incorporation of biotinylated 15d-PGJ 2 . In contrast, a mutated AP-1 oligonucleotide or an oligonucleotide with a consensus sequence for binding of NF-B did not display this protective effect. These results suggest that the interaction of 15d-PGJ 2 with c-Jun involves a region closely implicated in DNA binding. The region of c-Jun that is in close contact with DNA contains a cysteine residue (cysteine 269), which has been shown to be specifically targeted by S-glutathionylation under oxidative and nitrosative stress conditions, an interaction that results in inhibition of DNA binding (22,23). Preincubation of c-Jun with the thiol reagent iodoacetamide blocked incorporation of biotinylated 15d-PGJ 2 (Fig. 3D). This confirms the requirement of free cysteine residues for attachment of this cyPG. Moreover, preincubation of c-Jun with oxidized glutathione or nitrosoglutathione under conditions previously found to specifically induce the glutathionylation of cysteine 269 also precludes binding of biotinylated 15d-PGJ 2 (Fig. 3D). These results point to the involvement of cysteine 269 of c-Jun in the interaction with 15d-PGJ 2 .
The interaction between 15d-PGJ 2 and c-Jun was next studied by mass spectrometry. Untreated c-Jun-(223-327) showed a peak of m/z ϭ 13,479, which corresponds to the calculated molecular mass of the hexahistidine-c-Jun-(223-327) construct (22) and a peak of m/z ϭ 6,768 (doubly charged) (Fig. 4A). Treatment of c-Jun with 15d-PGJ 2 resulted in the appearance of a peak of m/z ϭ 13,828 and its corresponding doubly charged peak. This mass is compatible with the formation of a covalent adduct between 15d-PGJ 2 (mass 316.5) and c-Jun (expected m/z of the adduct 13,795.5), plus an excess of 32 Da that may be accounted for by oxidative modifications of two additional residues (expected m/z 13,827.5, see below). Interestingly, a peak of m/z ϭ 13,634 was also detected in the 15d-PGJ 2 -treated c-Jun sample. This could be indicative of the formation of a c-Jun dimer containing one 15d-PGJ 2 molecule, the doubly charged species of which would appear at that position of the spectrum (expected m/z ϭ 13,637). Observation of the higher molecular mass region of the spectrum (Fig. 4B)  c-Jun preparation of 315 and 620 Da, compatible with the incorporation of one and two 15d-PGJ 2 molecules, respectively, into a putative c-Jun dimer.
15d-PGJ 2 Induces c-Jun Dimerization-Dimerization of c-Jun molecules through the formation of intermolecular disulfide bonds has been shown to occur under oxidative conditions, and it has been proposed as one of the potential mechanisms involved in redox regulation of AP-1 DNA binding (30). To assess the effect of 15d-PGJ 2 on c-Jun dimer formation, we separated c-Jun monomers from covalently linked dimers by non-reducing SDS-PAGE, by a method established previously (22,23). Incubation of c-Jun constructs with 15d-PGJ 2 induced the appearance of higher molecular weight species compatible with dimer and, in a smaller proportion, multimer formation (Fig. 5A). The electrophoretic mobility of the 15d-PGJ 2 -induced c-Jun dimer was different from that induced by H 2 O 2 treatment. The 15d-PGJ 2 -induced dimer appeared as a diffuse doublet, whereas H 2 O 2 treatment elicited predominantly the appearance of the lower component of this doublet. The proportion of dimer formation was 13 and 17% for 15d-PGJ 2 and H 2 O 2 treatment, respectively (estimates from three independent experiments). H 2 O 2 -induced c-Jun dimer formation could be reversed by treatment with excess DTT. In contrast, 15d-PGJ 2 -elicited dimerization was only partially abrogated by DTT. These observations suggest that 15d-PGJ 2 may induce both reversible and non-reversible dimerization of c-Jun.
cyPG may form Michael adducts with cysteine residues of glutathione or proteins. However, the possibility that 15d-PGJ 2 may induce protein cross-linking has not been raised. To address this question we treated c-Jun with 15d-PGJ 2 and assessed the appearance of higher molecular weight species by SDS-PAGE under reducing conditions (Fig. 5B). 15d-PGJ 2 treatment induced the appearance of c-Jun dimer in a dose-dependent fashion (Fig. 5B, right panel). Concentrations as low as 2 M 15d-PGJ 2 induced significant dimer formation, and a maximal effect was attained with concentrations between 10 and 100 M. c-Jun dimerization could be detected as early as 5 min after addition of 15d-PGJ 2 , and it was maximal after 2 h of incubation (not shown). The ability of 15d-PGJ 2 to induce c-Jun dimerization may be due to the presence of several electrophilic carbons in the molecule, two of which can simultaneously form Michael adducts with thiols. In accordance with this, PGA 1 , which is also a cyclopentenone but possesses only one electrophilic carbon, did not induce c-Jun dimerization (Fig. 5C).
15d-PGJ 2 Binds to Cysteine 269 of c-Jun-To identify the site of modification of c-Jun by 15d-PGJ 2 , c-Jun was treated with 15d-PGJ 2 as in Fig. 5B. After SDS-PAGE, the protein bands corresponding to the monomer and dimer were excised, subjected to in-gel digestion with trypsin, and analyzed by MALDI-TOF mass spectrometry ( Fig. 6A and Table I). Analysis of the monomer band showed the presence of a peak of m/z ϭ 595, which is compatible with the incorporation of one 15d-PGJ 2 molecule into the Cys 269 -Arg 270 dipeptide (expected m/z ϭ 594.62). These results indicate that 15d-PGJ 2 binds to the cysteine residue located in the basic DNA binding domain of c-Jun (cysteine 269). Analysis of the dimer band showed the presence of a peak of m/z ϭ 873, which is compatible with a dimer composed of two dipeptides containing Cys 269 and Arg 270 , cross-linked through the cysteine residues by one 15d-PGJ 2 molecule (expected m/z ϭ 872.74), as depicted in the inset of Fig. 6A. This peak was absent from the monomer sample. The presence of 15d-PGJ 2 in the c-Jun dimer was also confirmed by incubation of c-Jun with biotinylated 15d-PGJ 2 and analysis of the resulting species by SDS-PAGE under reducing conditions. Incorporation of biotinylated 15d-PGJ 2 could be observed both in the monomer and in the c-Jun dimer, as evidenced by Western blot and detection with HRP-conjugated streptavidin (Fig. 6B). We then assessed the role of 15d-PGJ 2 addition to cysteine 269 in the inhibition of c-Jun DNA binding. As shown in Fig. 6C, the inhibition by 15d-PGJ 2 was strongly reduced when a C269S mutant of the c-Jun construct was used. This mutant showed improved DNA binding activity over the wild type construct, as reported previously (21), and was also more resistant to the inhibition observed under strong oxidizing conditions achieved by treatment with H 2 O 2 . This indicates that modification of cysteine 269 is relevant for 15d-PGJ 2elicited inhibition, whereas oxidative modifications of other parts of the molecule do not account for the inhibitory effect.
Biotinylated 15d-PGJ 2 Binds to c-Jun in Intact Cells-We next studied the interaction between 15d-PGJ 2 and c-Jun in intact cells. Incubation of cells in the presence of biotinylated 15d-PGJ 2 led to the incorporation of the biotinylated PG into several polypeptides, the most intensely labeled of which displayed apparent molecular masses of 90, 45-55, and 34 kDa. The labeled polypeptides were clearly distinguishable from endogenous biotinylated proteins, which showed molecular masses of ϳ120 and 80 kDa (Fig. 7A). Immunoprecipitation of c-Jun from biotinylated 15d-PGJ 2 -treated cells showed the incorporation of biotin into the c-Jun band (Fig. 7B). This signal was absent in the immunoprecipitates from cells treated with non-biotinylated 15d-PGJ 2 . In addition, c-Jun from biotinylated 15d-PGJ 2 -treated cells, but not from cells incubated with 15d-PGJ 2 , was retained on avidin beads, as shown by Western blot after avidin pull-down. Similar results were obtained using NIH-3T3 fibroblasts (Fig. 7B)

and HeLa cells (not shown).
These results indicate that 15d-PGJ 2 modifies c-Jun in intact cells. To assess the selectivity of this modification, we explored the presence of other transcription factors with cysteines important for function in the avidin-binding fraction. As shown in Fig. 7C, c-Fos, which is highly homologous to c-Jun and possesses a cysteine residue in its DNA binding domain, was detected in the avidin pull-down, whereas PPAR␥ was not. In order to explore the potential significance of this finding, we explored whether c-Jun modification occurred under conditions close to those encountered in a cellular model of inflammation. For this purpose we first estimated the levels of 15d-PGJ 2 in the culture medium of RAW264.7 macrophages. Under our experimental conditions 15d-PGJ 2 levels increased from 1 Ϯ 0.01 nM in basal conditions to 524 Ϯ 84 nM (average of four determinations Ϯ S.E.) after stimulation with bacterial lipopolysaccharide. We then incubated cells in the presence of 500 nM biotinylated 15d-PGJ 2 . This concentration was sufficient to elicit the modification of c-Jun in intact cells (Fig. 7D). DISCUSSION cyPG have been proposed to act as regulators of inflammatory and proliferative responses. The results herein show that the cyPG 15d-PGJ 2 can directly interact with the component of the transcription factor AP-1, c-Jun, leading to an inhibition of its ability to bind DNA. In addition, 15d-PGJ 2 participates in the formation of intermolecular cross-links that result in the dimerization of c-Jun under in vitro conditions. 15d-PGJ 2 has been shown to inhibit AP-1 DNA binding in various experimental systems, but the mechanisms involved have not been fully elucidated. We have observed that 15d-PGJ 2 reduces AP-1 DNA binding activity in several cell types, an effect that is not associated with a reduction in the nuclear levels of the AP-1 constituents c-Jun and c-Fos. Our results also show that 15d-PGJ 2 reduces the formation of AP-1-DNA complexes in identical aliquots of isolated nuclear extracts, therefore acting in a manner independent from the amount or nature of the AP-1 proteins present in the assay. Moreover, 15d-PGJ 2 also inhibits recombinant c-Jun DNA binding, thus suggesting the involvement of a direct interaction between the PG and the protein in the inhibitory effect. These observations may help to explain the results of previous reports (31) in which 15d-PGJ 2 induced a decrease in AP-1-DNA binding in the presence of unaffected or even increased nuclear levels of c-Jun protein.
We have observed that 15d-PGJ 2 directly binds to c-Jun at a cysteine residue located in the DNA binding domain, cysteine 269. This interaction occurs by means of Michael's addition, a covalent linkage between the cysteine sulfhydryl group and the electrophilic carbon in the cyclopentenone ring of the PG. Modifications of this cysteine residue have been shown to result in loss of the DNA binding ability (21)(22)(23). Other members of the Jun and Fos families possess equivalent cysteine residues in their DNA binding domains. Moreover, the DNA binding domain of other transcription factors, including NF-B, p53, CREB, or c-Myb, possess critical redox-sensitive cysteine residues that may be potential targets for modification by cyPG. Our results suggest that c-Fos is also modified by 15d-PGJ 2 . However, although both p65 and p50 NF-B subunits have been shown to be targets for cyPG (15,16), neither p53 (32) nor PPAR␥ (this work) react to an appreciable extent with biotinylated cyPG in intact cells. Taken together, the available studies indicate that protein modification by cyPG is not a random process but displays some specificity determined by factors potentially related to protein or cell context.
One important issue regarding many of the reported effects of cyPG is whether they can be considered either of biological or pharmacological significance. The levels of 15d-PGJ 2 measured in cell-free exudates or extracellular media in several experimental models of inflammation are in the nanomolar range (2,3). However, most of the biological effects of cyPG have been observed using micromolar concentrations (13,(32)(33)(34). Several considerations have been made to reconcile these discrepancies. First, it is known that cyPG accumulate inside cells, particularly in cell nuclei (35); therefore, concentrations measured in the culture medium would be expected to be lower than those present intracellularly. Second, cyPG readily form adducts with glutathione or proteins. This has been shown to result in the underestimation of the generation of cyPG when only the free forms are measured (36). Third, as discussed previously (32), although individual eicosanoids with ␣,␤-unsaturated ketone substituents may not occur in micromolar concentrations at the site of inflammation, inflammatory exudates contain a variety ␣,␤-unsaturated aldehydes and ketones with reactivity similar to that of 15d-PGJ 2 , which can reach concentrations of 10 M in tissues under conditions of oxidative stress. Our results indicate that covalent protein modification can occur in the presence of concentrations of 15d-PGJ 2 in the range of those measured in the culture medium of activated macrophages, thus suggesting the relevance of this mechanism in the effects of cyPG. Further studies will be needed to ascertain whether concentrations of cyPG sufficient to elicit the anti-inflammatory or antiproliferative effects described in most studies may occur locally under pathophysiological conditions.
The results described here unveil a novel aspect of the post-translational modification of proteins by cyPG as is the ability of 15d-PGJ 2 to induce c-Jun cross-links. This effect is apparently due to the formation of covalent bonds between two re-FIG. 6. 15d-PGJ 2 binds to cysteine 269 of c-Jun. A, c-Jun was incubated in the presence of 15d-PGJ 2 as described above and subjected to SDS-PAGE. The protein bands corresponding to the monomer (upper panel) and dimer (lower panel) species were isolated and digested in-gel with trypsin, and the resulting peptides were analyzed by MALDI-TOF mass spectrometry. The monoisotopic mass of some of the detected peptides is given. The amino acid sequence is shown in Table I. Asterisks mark the position of the peaks corresponding to putatively 15d-PGJ 2 -modified c-Jun fragments, the structure of which is shown in insets. B, c-Jun was incubated in the absence or presence of biotinylated 15d-PGJ 2 , and incorporation of 15d-PGJ 2 into the monomer and dimer forms was assessed as described in the legend of Fig. 3. Results shown are representative of four assays. C, c-Jun wild type and C269S constructs were incubated in the presence of 15d-PGJ 2 or H 2 O 2 as described in Fig. 5, after which DNA binding activity was analyzed by EMSA. Results are representative of three assays.    active carbons in the PG molecule and cysteine residues of two c-Jun monomers. Interestingly, other cyPG, which also possess two electrophilic carbons, like 9-deoxy-⌬ 9 ,⌬ 12 (E)-PGD 2 , have been described to form a conjugate with two molecules of glutathione (37). c-Jun cross-linking occurs in vitro in the presence of low micromolar concentrations of 15d-PGJ 2 . This process is saturable, reaching maximal dimer formation in the presence of 10 -50 M 15d-PGJ 2 under our assay conditions. Higher 15d-PGJ 2 concentrations result in less dimer formation. This may be due to a negative effect of a more extensive c-Jun modification on the possibility that one 15d-PGJ 2 molecule binds two c-Jun monomers simultaneously. The observations that 15d-PGJ 2 -induced c-Jun cross-linking occurs in the presence of 0.5 mM DTT and is resistant to reducing agents suggest that this effect could also occur in the presence of the cellular antioxidant defenses. It is known that oxidative stress-induced lipid peroxidation can give rise to reactive lipid aldehydes that may induce protein cross-linking and aggregation, a process that impairs protein degradation and has important implications in the pathophysiology of inflammatory and neurodegenerative diseases (38). Thus, it would be interesting to explore whether protein cross-linking by 15d-PGJ 2 or related cyPG could contribute to protein aggregation in pathophysiological settings.
The transcription factor AP-1 plays a central role in the inflammatory response. It is activated by cytokines and in turn contributes to the induction of matrix metalloproteases, cyclooxygenase-2, inducible nitric-oxide synthase, and interleukin-8, among other genes. The regulation of AP-1 occurs both at transcriptional and post-transcriptional levels (for reviews see Refs. 39 and 40) (Scheme 1). A variety of stimuli, including growth factors, cytokines, and cell stress, induce the expression of AP-1 proteins through signaling pathways that involve the activation of JAK and/or mitogen-activated protein kinases and of transcription factors that act on the c-Jun or c-Fos promoters. In addition, increased transactivating activity and association with coactivators is achieved after phosphorylation of c-Jun proteins at specific serine residues by JNK. AP-1 activity is also regulated by redox-dependent mechanisms. The reduced state of critical cysteine residues present in the DNA binding domain of AP-1 proteins is essential for DNA binding (21). Levels of intracellular GSH and reducing enzymes, such as thioredoxin, are key to maintain AP-1 proteins in the active form (41).
15d-PGJ 2 can interact with the AP-1 activation pathway at several levels (Scheme 1). Both positive and negative effects of 15d-PGJ 2 on extracellular signal-regulated kinase and JNK activities have been reported in various experimental systems (19,(42)(43)(44). The activity of p38 has also been shown to be modulated by this cyPG (45). These effects could result in the regulation of the levels of AP-1 proteins by transcriptional mechanisms. The modulation of JNK activity could also lead to changes in c-Jun phosphorylation and AP-1 transactivating activity. In addition, 15d-PGJ 2 can affect AP-1 through binding to PPAR␥, which can interfere with AP-1 activation or compete for limiting amounts of coactivators such as CBP/p300 (46,47). Finally, 15d-PGJ 2 can modulate AP-1 function by redox-dependent mechanisms. Incubation of cells with 15d-PGJ 2 induces oxidative stress and intracellular glutathione depletion (34). Moreover, cyPG can directly modify and inhibit both thioredoxin reductase (32) and thioredoxin (48), leading to a decrease in the activity of thioredoxin-dependent transcription factors, such as AP-1. In this work we have shown that 15d-PGJ 2 can also modify a critical cysteine residue in c-Jun resulting in a direct inhibition of DNA binding. The net effect of 15d-PGJ 2 on AP-1 signaling will be the result of its interaction with different potential targets along this pathway, and this, in turn, will be subject to cell-specific factors including the levels of antioxidant defenses and the redox and activation state of the cells. In addition, c-Jun provides the first example of a protein undergoing 15d-PGJ 2 -induced cross-linking. The extent and consequences of this modification will be addressed in future studies.