Activation of the 9E3/cCAF Chemokine by Phorbol Esters Occurs via Multiple Signal Transduction Pathways That Converge to MEK1/ERK2 and Activate the Elk1 Transcription Factor*

Using primary fibroblasts in culture, we have investigated the signal transduction mechanisms by which phorbol esters, a class of tumor promoters, activate the 9E3 gene and its chemokine product the chicken chemotactic and angiogenic factor. This gene is highly stimulated by phorbol 12,13-dibutyrate (PDBu) via three pathways: (i) a small contribution through protein kinase C (the commonly recognized pathway for these tumor promoters), (ii) a contribution involving tyrosine kinases, and (iii) a larger contribution via pathways that can be interrupted by dexamethasone. All three of these pathways converge into the mitogen-activated protein kinases, MEK1/ERK2. Using a luciferase reporter system, we show that although both the AP-1 and PDRIIkB (a NFκB-like factor in chickens) response elements are capable of activation in these normal cells, regions of the 9E3 promoter containing them are unresponsive to PDBu stimulation. In contrast, we show for the first time that activation by PDBu occurs through a segment of the promoter containing Elk1 response elements; deletion and mutation of these elements abrogates 9E3/chicken chemotactic and angiogenic factor expression. Electrophoretic mobility shift assays and functional studies using PathDetect systems show that stimulation of the cells by phorbol esters leads to activation of the Elk1 transcription factor, which binds to its element in the 9E3 promoter.

It has been known for some time that chemokines play important roles in leukocyte chemoattraction and inflammation (1). More recently, however, it has become increasingly clear that these small cytokines are also involved in wound healing (2), deterrence of retroviral infections (3), and tumorigenesis (4,5). In the latter case, chemokines can potentially act at several steps in the development of tumors, and their action is dependent not only on the stimulant but also on the environment.
Tumors develop as a result of multiple insults and chemokines could play important roles in these events because a number of them are stimulated by phorbol esters, injury, and oncogenes (2), all of which have been shown to be involved in tumor promotion. Agents that promote tumor development are called tumor promoters; they are not themselves carcinogenic but they promote the development of tumors in areas of the body that have been exposed to a carcinogen (6,7). There is extensive literature that demonstrates that phorbol esters are very effective tumor promoters (8,9). Wounding is also a tumor promoter because it can cause cancer to develop at the edges of wounds inflicted in areas that have been exposed previously to a carcinogen (10 -15). The v-src oncogene, which is the transforming protein of the Rous sarcoma virus (a retrovirus that causes tumors in chickens), also has been shown to be a tumor promoter (16). The 9E3 gene and its product, the chicken chemotactic and angiogenic factor (cCAF), 1 are stimulated to high levels by the v-src oncogene and also by phorbol esters and wounding/inflammation (17)(18)(19)(20)(21). Therefore, this chemokine can potentially be a mediator of the tumor-promoting action of these agents. In the case of v-src, during the development of Rous sarcoma virus-induced tumors the expression of the 9E3/ cCAF occurs only in the cells of the tissues surrounding the tumor (19,21). At later stages of tumor growth, when 9E3/ cCAF are expressed abundantly, numerous new blood vessels develop in the area (2,19,21). Because cCAF is angiogenic (22,23) and angiogenesis is very important for tumor growth, it is possible that this chemokine is a mediator of the tumor-promoting actions of the v-src oncogene. In the case of wounding, we have shown that after injury the 9E3 gene is highly expressed shortly after injury and in the granulation (repair) tissue during wound healing (19,21). The persistent expression of 9E3/cCAF during the healing process coupled with its angiogenic properties in tissues that have been exposed to a carcinogen could mediate the tumor promotion stimulated by wounding. The stimulation of 9E3/cCAF to high levels by phorbol esters, again coupled to its angiogenic properties, can potentially mediate the tumor-promoting actions of these molecules.
Until recently, it was believed that gene activation by phorbol esters always involves PKCs (24). PKCs are a family of serine and threonine kinases that contain structural motifs with a high degree of sequence homology. Most PKC isoenzymes have a conserved cysteine-rich (C1) domain at the N terminus of the regulatory domain (25). The C1 domain is involved in binding to diacylglycerol or its potent functional analogues, the phorbol esters, resulting in the translocation of the enzyme to the plasma membrane. The binding also causes a conformational change in PKC that removes the pseudo substrate domain from the active site, allowing substrate binding and catalysis (26). However, several receptors of phorbol esters that are not PKCs now have been identified (25). Examples are n-chimaerin (27), Unc-13 (28), Vav (29), cathepsin-L (30), and protein kinase D (31). Thus it is clear that not all of the cellular effects of phorbol esters are mediated via PKC. As a consequence, it is important to delineate the signal transduction pathways by which chemokines are turned on by these agents so that crucial steps in the activation process can be identified and potentially used as targets for regulation of these genes. We have investigated the pathway of activation of 9E3/cCAF by phorbol esters such as phorbol 12-myristate 13-acetate and phorbol 12,13-dibutyrate (PDBu) in primary fibroblasts, the cells that most highly express 9E3/cCAF in vivo. The work presented here shows that activation of this chemokine by phorbol esters involves multiple signaling pathways that culminate in phosphorylation and activation of the MAP kinase ERK2 followed by activation of the transcription factor Elk1, leading to 9E3/cCAF expression.

MATERIALS AND METHODS
Reagents-The dosages used for particular experiments are indicated in the text or in the figure captions. Bovine thrombin (Sigma) was reconstituted in water and used at 9 units/ml. Calphostin C (100 -200 nM), genistein (15-25 M), H-7 di-HCL (100 -200 nM), tyrphostin AG1478 (5-500 nM), and PD98059 (10 M) were all purchased from Calbiochem and reconstituted in Me 2 SO. Phorbol-12,13-dibutyrate and 4␣-phorbol-12,13-dibutyrate (Biomol) were dissolved in Me 2 SO and used at 5-100 nM. Anti-phosphotyrosine PY20 (Transduction Laboratories) and 4G10 (Upstate Biotechnology Inc.) were used for the phosphotyrosine immunoblots. ECL reagents and secondary antibodies were conjugated to horseradish peroxidase (Amersham Pharmacia Biotech). For each inhibitor or activator, a range of doses was tested to determine the optimal dose for the study in question. The Bradford assay was performed using the DC protein assay kit (Bio-Rad).
Cell Culture-Primary cultures of chicken embryo fibroblasts (CEFs) were prepared from 10-day-old chicken embryos as described previously (32). On the fourth day, secondary cultures were prepared by trypsinizing and plating the primary cells in 199 medium containing 0.3% tryptose phosphate broth and 2% donor calf serum at a density of 1.2 ϫ 10 6 /60-mm dish. To study the effects of phorbol esters, 12-O-tetradecanoylphorbol-13-acetate, PMA, or PDBu on 9E3 expression and cCAF production, we used quiescent confluent CEF (qcCEF) cultures and incubated them in serum-free 199 medium containing the specific treatment for varying times (see "Results" for specifics). In general, prior to the addition of the specific activator cells were incubated in serum-free 199 medium containing the appropriate inhibitor for 30 min; incubation at 37°C was continued for 1 h more before removing the medium and replacing it with fresh serum-free medium containing again the specific inhibitor under study. At the end of the incubation period the supernatant was collected and processed as described earlier (33). Pretreatment with phorbol esters to down-regulate PKC activity in the cells was performed for 18 -24 h and then removed and replaced with medium containing the appropriate treatment for 18 h when evaluating cCAF levels. In the cases where the cell extracts were analyzed for activation of the ERKs, the cells were pretreated with the PDBu for 24 h, the inhibitor of MEK1/ERK (PD98059) for 1 h, and then the activator, PDBu, at 5 mM for 5 min. Because qcCEFs are primary cells rather than a cell line, there are small variations in the basal levels of 9E3/cCAF expression from batch to batch of cells. Therefore, for each experiment and/or treatment of a different batch of cells, we used positive (treated cells) and negative (untreated cells) controls.
Western Blotting-Volumes of the cell culture supernatant corresponding to equal amounts of protein in the cell extracts were loaded on 20% polyacrylamide-glycerol gels and electrophoresed at 16 -32 mA for about 3 h. The concentration of the protein was determined using the Bio-Rad DC protein kit. Transfer was performed using a semi-dry transfer apparatus (Millipore). The efficiency and consistency of the transfer was monitored by silver-staining the gel after the transfer. The 9E3 protein was detected using polyclonal antibodies raised in rabbit (34) and enhanced chemiluminescence (ECL) reagents (Amersham Pharmacia Biotech).
Northern Blot Analysis-CEFs were homogenized and total RNA was prepared using triZOL reagent (Life Technologies, Inc.). RNA samples (10 g/ml) were denatured in a formamide-formaldehyde buffer containing ethidium bromide and separated on formaldehyde-agarose gel. After electrophoresis, the RNA was transferred to MagnaGraph nylon membranes (MSI Inc.), which were photographed to visualize the qual-ity of the rRNA and confirm equal loading and even transfer. The RNA was UV-cross-linked to the membrane for 2 min and baked at 80°C for 2 h. Prehybridization was performed for 6 -9 h, and hybridization with a cDNA probe was carried out for 24 h (33).
Phosphotyrosine Immunoblots-Cell extracts were prepared by lysing the cells in boiling 2ϫ Laemmli buffer (containing 100 M sodium orthovanadate) and shearing by passing through a 26-gauge needle 10 times followed by boiling in a water bath for 5 min. The samples were electrophoresed on a 7.5% SDS-polyacrylamide gel and transferred to the nitrocellulose membrane (Schleicher & Schuell) using a wet-transfer system (Bio-Rad) and Towbin's buffer (25 mM Tris, pH 8.3, 192 mM glycine, 0.05% SDS, 2% methanol) for 13 h at 30 V or 5 h at 60 V at 4°C. The blots were blocked for 2 h at 37°C in TBS (25 mM Tris, pH 7.5, 2.7 mM KCl, 137 mM NaCl) containing 0.1% Tween 20, 2% bovine serum albumin, and 0.02% thimerosal (TBST-BSA) and then incubated with the anti-phosphotyrosine antibodies (PY20 at 1:2500, 4G10 at 1:1000 dilutions) in TBST containing 2% bovine serum albumin for 2 h at 37°C. The excess antibodies were washed three times for 5 min each plus a longer 20 min wash with TBST. A second blocking step was done for 1 h with TBST containing 5% nonfat milk. This was followed by incubation with the appropriate secondary antibodies conjugated to horseradish peroxidase (1:5000) for 1 h at room temperature. The washings were performed as described for the primary antibodies. The antibody detection was done using ECL reagents (Amersham Pharmacia Biotech). For reprobing of the blots, the membranes were stripped by incubation in stripping buffer (100 mM 2-mercaptoethanol, 2% SDS, and 62.5 mM Tris-HCl, pH 6.7) at 50°C for 30 min with gentle shaking and then washed (2ϫ for 10 min each) with TBST, following which they were tested by ECL to ensure complete stripping of antibodies. Finally, the membranes were washed again with TBST (three times for 10 min) before proceeding with the reprobing. Microdensitometry analysis was performed by laser densitometric scanning in an LKB microdensitometer.
9E3 Gene Promoter Cloning-Using PCR, we isolated a 1.5-kilobase pair DNA fragment from the immediate 5Ј-upstream region of the 9E3 gene (Ϫ1506 to ϩ32). We used chicken genomic DNA as a template and two oligonucleotides (5Ј-primer, GGATGAATGGCATTTCAGTGCAC; 3Ј-primer, TCGACACTAGAGAGGACAGTCTCCT) for the PCR reactions that were performed under the following conditions: denaturing of the DNA at 94°C for 5 min, 55°C for ϳ20 min to add, and then 94°C for 40 s to ensure that the DNA was denatured before starting the annealing for 2 min at 55°C and extension for 1.5 min at 72°C. The 1.5-kilobase pair PCR product was then cloned using the AT cloning system pCR2.1 (Invitrogen) and sequenced with the UBI Sequenase kit using vector primers and internal primers. The sequence was confirmed to be correct by comparison with the published sequence (35).
Luciferase Reporter Construction-The 1.5-kilobase pair 9E3 promoter region was subcloned into pSL1180 (Amersham Pharmacia Biotech) using the SacI and EcoRV and then subcloned into the pGL3 basic vector (Promega) using XhoI and BglII. Using this vector we obtained constructs with Ϫ66 to ϩ32 bp, Ϫ218 to ϩ32 bp, Ϫ470 to ϩ32bp, and Ϫ683 to ϩ32 bp fragments upstream from the initiation site with restriction enzyme digests and religations. We created the Ϫ493 Elk1 binding element mutant (pmElk1) by PCR-based methods. 5Ј-primer ATacGcgTGCTTTTAATACTGCACCCT was used to substitute the Elk1-conserved binding element (underlined; original sequence, CAG-GAT). The controls for this mutated promoter, Ϫ542 to ϩ32 bp and Ϫ495 to ϩ32 bp, were also obtained using PCR and the appropriate oligonucleotides. The PCR products were cloned into the PCR2.1 vector and sequenced to verify accuracy. KpnI/EcoRV (on pCR2.1) and KpnI/ SmaI (on pGL3 basic) were used for further reporter construction. The reporter vector with the mutated AP-1 binding site (Ϫ114 to Ϫ107) was prepared directly from the p683 construct with the Quickchange sitedirected mutagenesis kit (Stratagene). TGACTCAT was changed into gGcCTtAT, and the mutation was verified by endonuclease digestion (additional HaeIII site introduced by mutagenesis) and confirmed by sequencing. All the pGL3/9E3 promoter subclones were transformed into DH5␣ Escherichia coli (Life Technologies, Inc.) and prepared using the endotoxin-free Maxiprep kit (Qiagen).
Transient Transfection, PDBu Activation, and Culture Lysates Collection-qcCEFs were transiently transfected with 6 g of DNA (4 g of pGL3/9E3P subclone DNA cotransfected with 2 g of PCH110 vector (Amersham Pharmacia Biotech) containing Lac-Z as internal transfection control) for each 35-mm dish using Ca 2 PO 4 -mediated methods without glycerol shock. All the reporter constructs were transfected and assayed in the same batch of cells. The transfected qcCEFs were incubated in modified 199 medium for 36 h at 37°C, 5% CO 2 (24). Then cells were stimulated with PDBu or LPS for 3 h prior to lysis. For the inhibitor experiments, the qcCEFs were incubated with inhibitors for 30 min before PDBu stimulation. Cell extracts were prepared with reporter lysis buffer according to the protocol provided by the manufacturer (Promega) and stored in a Ϫ70°C freezer.
Luciferase and ␤-Galactosidase Assay-Cell extracts were assayed using a Luminometer with an automated injection device (Monolight 2010, Analytical Luminescence Laboratory). The reaction substrate and buffer are parts of the luciferase assay system (Promega). Aliquoted samples were used for the ␤-galactosidase assay according to the procedure for the ␤-galactosidase enzyme assay system (Promega). A minimum of triplicates for each experiment was performed, and the data were expressed as mean light units of luminescence/unit of ␤-galactosidase activity.
PathDetect System-We used a cis-PathDetect system for AP-1 and a trans-PathDetect system for the Elk1. These systems were purchased from Promega, and we followed the protocols described by the company with the exception that we used our modified Ca 2 PO 4 precipitation method for cell transfections that we have optimized in our system for lower stress-induced background, rather than transfections with Lipofectin as recommended in the procedure.
Electrophoretic Mobility Shift Assays (EMSA)-Cells were treated as described earlier and nuclear extracts were prepared as described previously (36). Protein concentration was determined with the DC protein assay kit (Bio-Rad). 10 g of extracted nuclear protein was used for each binding reaction with 2 ng of [␥-32 P]ATP-labeled probe containing the conserved Elk-1 binding element (wild type, ATCAGGATGCTTTTA-ATACTGCACC CT (bold indicates the binding site for the Elk1 transcription factor)) in EMSA buffer (188 mM NaCl, 50 mM HEPES, pH 7.9, 2.5 mM EDTA, pH 8.0, 2.5 mM dithiothreitol, and 20% glycerol). 30 ng of unlabeled probe was used for the competition assay, and 1 l of anti-Phorpho Elk1(Ser-383) antibody (NE Biolab) was added for the supershift assay. Samples were assayed in 6% nondenaturing polyacrylamide gel electrophoresis.

Stimulation of 9E3/cCAF by Phorbol Esters-qcCEFs
do not express the 9E3/cCAF or they express it at very low levels. However, upon stimulation with PDBu, the levels of the 9E3 mRNA increase dramatically. The rise in mRNA is first seen at 7 min, and it peaks at 3-6 h and declines thereafter (Fig. 1A). cCAF production after stimulation by PDBu shows that the protein accumulates in the culture supernatant (Fig. 1B). Accumulation is dose-dependent ( Fig. 1C) and is specific; the Me 2 SO used as solvent for PDBu did not significantly stimulate 9E3/cCAF (Fig. 1

, D and E).
It has been considered for a long time that phorbol esters exert their effects on cells primarily through the activation of PKC isoforms that contain the cysteine-rich (C1) domain to which phorbol esters bind (37). Therefore, we investigated the possibility that stimulation of 9E3/cCAF by phorbol esters occurs via activation of these classical PKCs. We treated cells with PDBu for 24 h to down-regulate total PKC activity by causing proteolytic degradation of these kinases (38 -40) followed by restimulation with PDBu. The result was only moderately reduced expression compared with that stimulated by PDBu without pretreatment ( Fig. 2A). These observations suggest that activation of PKC represents only part of the stimulation of the 9E3/cCAF by phorbol esters. These results were confirmed by treatment of cells with calphostin C, a specific inhibitor of PKC that competes with phorbol esters for the binding of the C1 domain in the regulatory region of PKC, and we found that 9E3 expression and production of cCAF were minimally blocked by this inhibitor (Fig. 2, B and C). Similar results were obtained when the cells were treated with H7 dihydrochloride, a broad spectrum inhibitor of Ser/Thr kinases that also inhibits PKC (Fig. 2, B and C). None of these inhibitors by themselves cause stimulation of 9E3/cCAF (33). In addition, we also found that the 4␣-isomer of PDBu (4␣-PDBu), which does not activate PKC (41), stimulates 9E3 expression and cCAF production almost as efficiently as PDBu (Fig. 3, A  and B). This stimulation, much like that by PDBu, is time- ( 3C) and dose-(not shown) dependent, albeit not as efficient as the stimulation by PDBu. These results taken together strongly suggest that 9E3/cCAF stimulation by phorbol esters occurs primarily via PKC-independent pathway(s).
To investigate the possibility that the pathway of activation of 9E3/cCAF by phorbol esters involves tyrosine kinase activation, as it does for stimulation by thrombin (33), we used inhibitors of tyrosine kinases. Herbymicin, an inhibitor of the c-src family of tyrosine kinases, had no inhibitory effect on the PDBu stimulation of 9E3/cCAF (not shown). Similarly, tyrphostin, a selective inhibitor for the epidermal growth factor receptor tyrosine kinase, had essentially no effect on PDBu stimulation of this gene (Fig. 4A), whereas the broad spectrum tyrosine kinase inhibitor genistein produced a distinct decrease in the stimulation of 9E3/cCAF. The same results were observed for cCAF protein levels (not shown). The inhibition of cCAF production when the cells were treated with calphostin C (inhibitor of PKC) and genistein (inhibitor of tyrosine kinases) was additive, but significant cCAF stimulation by PDBu remained (Fig. 4, B and C), suggesting yet other pathways of stimulation for this gene by PDBu.
It has been known for some time that expression of chemokine genes triggered by a variety of stimuli is inhibited by glucocorticoids (42)(43)(44)(45). In most cases tested, these anti-inflammatory agents activate the glucocorticoid receptor that, in turn, interacts with and inactivates transcription factors that are important in chemokine gene expression (46). Dexamethasone is an example of an anti-inflammatory glucocorticoid that inhibits chemokine gene activation (44). Because 9E3/cCAF is stimulated by phorbol esters and plays an important role in the inflammatory response (22), we treated qcCEFs with PDBu in the presence of dexamethasone to determine whether this antiinflammatory agent inhibited 9E3/cCAF stimulated by this phorbol ester. Dexamethasone significantly decreased cCAF production stimulated by PDBu but did not eliminate it (Fig.  5A). Treatment of qcCEFs with PDBu in the presence of dexamethasone, calphostin C, and genistein simultaneously, however, reduced cCAF production to that of the control (Fig. 5B). These inhibitors were all used at doses chosen for optimal activity (see "Materials and Methods") without adversely affecting the cells as judged by their morphology, appearance, adhesion to the culture dish, and percent of cell death, even when applied simultaneously (see fig. 5 legend for doses applied). This is further illustrated by the fact that the combination of inhibitors did not significantly affect the basal level of cCAF produced by the treated cells, which is comparable to the levels of the control (Fig. 5B).

FIG. 2. Activation of 9E3/cCAF by PDBu occurs primarily via PKC-independent pathway(s).
A, immunoblot analysis of cCAF. qcCEFs were pretreated (ϩ) or not (Ϫ) with PDBu (PdiBt) (200 nM) for 24 h to down-regulate PKC. This treatment was followed by a 1-h exposure to thrombin (9 units/ml), an activator of 9E3cCAF known to stimulate this gene independently of PKC (24), or with PDBu (100 nM) followed by removal of the medium and incubation of the cells with fresh serum-free medium for 18 h. Only partial inhibition of the stimulation of cCAF by PDBu was observed when PKC was down-regulated. Signaling Events Leading to 9E3/cCAF Expression Involve ERK2 Activation-As shown above, phorbol esters stimulate the expression of 9E3/cCAF via multiple pathways; a smaller contribution was made by PKC and tyrosine kinases and a more significant contribution from pathway(s) inhibited by dexamethasone. The MAP kinase ERK2 is a convergence point for many different signaling pathways (47)(48)(49). Hence, we investigated if this kinase becomes phosphorylated after stimu-lation by PDBu and if activation of this enzyme is responsible for stimulation of 9E3/cCAF via any of these pathways.
Immunoblot analysis using anti-phosphotyrosine antibodies showed that a protein with molecular mass of 42 kDa was differentially phosphorylated on tyrosines (Fig. 6A) with the maximum phosphorylation occurring at 2 min after stimulation and declining thereafter. We stripped and reprobed the anti-phosphotyrosine blots with anti-ERK2 antibodies and determined that the 42-kDa protein co-migrated with ERK2 (Fig.  6B). To determine if ERK2 phosphorylation/activation is involved in the signaling pathway(s) inhibited by dexamethasone, we treated cells in the presence or absence of this glucocorticoid and found that this inhibitor partially eliminated ERK2 phosphorylation on tyrosines (Fig. 7, A and B), suggesting that the decrease in cCAF production upon treatment with dexamethasone ( Fig. 5B) is linked to ERK2 activation. A specific inhibitor of ERK2 is not available; however, MEK1 (mitogen-activated protein kinase kinase), which can be selectively inhibited by PD98059 (50 -54), is directly upstream from ERK2 in the MAP kinase signaling cascade and is the specific kinase responsible for both phosphorylation and activation of ERK2 (55,56). Therefore, to determine if ERK2 is a convergence point for the various signaling pathways stimulated by PDBu, which lead to expression of this chemokine, we treated cells with PD98059. This inhibitor was highly effective in blocking PDBustimulated phosphorylation of ERK2 (Fig. 7, C and D) and eliminated cCAF production when the cells were treated with PDBu in the presence of PD98059 (Fig. 7E). These results, taken together, strongly suggest that the multiple pathways activated by PDBu, which lead to 9E3/cCAF expression, converge in ERK2. In addition, dexamethasone as well as PD98059 eliminated cCAF production stimulated by 4␣-PDBu (Fig. 7F).
Analysis of the 9E3/cCAF Promoter in Response to PDBu-Further information on the pathways involved in the stimulation of 9E3/cCAF by PDBu can be elicited by determining the active elements of the gene promoter (Fig. 8A). Therefore, we used a promoter region of the 9E3 gene (Ϫ683 to ϩ32 bp), which contains DNA binding elements that can potentially be activated by the pathways identified above, and cloned it into the pGL3 luciferase reporter system (Promega). This region of the promoter (p683) contains the consensus binding sequences for the transcription factors Elk1, AP-1, and PDRIIB (the chicken equivalent of NFB) (57-59) and the CAAT box for C/EBP. We found that when this construct was transfected into our primary normal fibroblasts, expression of luciferase was highly stimulated by PDBu (Fig. 8A, I). To determine the crucial areas of the promoter for activation, we subcloned fragments of the larger promoter region into the reporter system. The intermediate length fragments, Ϫ470 to ϩ32 bp and Ϫ218 to ϩ32 bp, eliminate the two putative Elk1 response elements, and the smallest fragment, Ϫ66 to ϩ32 bp, contains only the TATA box. The Ϫ470 to ϩ32 bp piece showed a small amount of stimulation (twice as much as the control) (Fig. 8A, II), but the other two constructs showed virtually no stimulation upon PDBu treatment (Fig. 8A, III and IV), although the PDRIIB and the AP-1 binding elements are still present in the Ϫ218 to ϩ32 bp construct. Me 2 SO, which was used to dissolve the PDBu and the inhibitors, showed a slight increase of activation of the reporter system over the control (Fig. 8). These results correlate very well with those obtained with northern blot and immunoblot analysis. In addition, we found that the stimulation by PDBu was completely inhibited by PD98059 in a dosedependent manner (Fig. 8B) and that a dose dependence of inhibition was also observed with dexamethasone ( Fig. 8C) but   FIG. 6. Differential phosphorylation of a 42-kDa protein produced by qcCEFs treated with PDBu. Cell extracts of qcCEFs treated with PDBu (100 nM) for the indicated time intervals were resolved on SDS-polyacrylamide gel electrophoresis (7%) followed by immunoblot analysis using antibodies to phosphotyrosines (A) or to ERK2 (B). A, a 42-kDa protein was phosphorylated, reaching a peak at 2 min after stimulation. B, the blot in A was stripped and reprobed with antibodies to ERK2 showing that ERK2 co-migrates with the differentially phosphorylated protein. This immunoblot also illustrates the even loading of protein in the various samples of the blot.

FIG. 7. ERK2 activation is involved in the stimulation of cCAF by PDBu.
A, dexamethasone (Dex) (100 nM) inhibits the phosphorylation of ERK2 when qcCEFs are stimulated by PDBu (PdiBt). B, the same blot when probed with antibodies to ERK2 shows equal loading of protein. C, effects of PD98059, a specific inhibitor of MEK1 kinase that phosphorylates and activates ERK2. Cell-extracts were analyzed by immunoblotting using anti-phosphotyrosine antibodies. Inhibition of the MEK1 kinase prevents phosphorylation of ERK2. D, the same blot was stripped and reprobed with anti-ERK2 antibodies to show the amounts of the ERK2 protein in each lane. E, blocking phosphorylation of ERK2 inhibited the stimulation of cCAF in cells treated with PDBu. F, stimulation by 4␣-PDBu is completely inhibited by both dexamethasone and PD98059. total inhibition was not obtained; the highest dose still left twice as much activation as the control.
The results presented above lead us to conclude that the region of the promoter between Ϫ683 and Ϫ470 bp contains the dominant elements important for activation of 9E3/cCAF by PDBu. This region contains two elements that are highly conserved for binding of the Elk1 transcription factor with one starting at Ϫ534 bp and the other at Ϫ493 bp. Elk1 is an important substrate for ERK2 that we have shown here is critical for expression of 9E3/cCAF after stimulation of the cells by PDBu. Therefore, we deleted the Elk1 starting at Ϫ534 bp and mutated the one starting at Ϫ493 bp to obtain a construct containing Ϫ493 to ϩ32 bp (pmElk1) and performed transfec-tion studies similar to those described above. The activity of the two promoters containing the two Elk1 elements (Ϫ683 to ϩ32 bp and Ϫ542 to ϩ32 bp) was not significantly different. The promoter containing only the first Elk1 element (Ϫ495 to ϩ32 bp) showed a significant decrease in activity, whereas the mutated promoter (pmElk1) showed a very low level of activity that was similar to that obtained with the Ϫ470 to ϩ32 bp construct (Fig. 9A). These results suggest that the two Elk-1 elements are the major regulatory elements for cCAF expression when cells are stimulated by phorbol esters.
To complement the mutagenesis studies, we tested whether the Elk1 transcription factor is activated in response to PDBu. For these experiments, we used a PathDetect reporter system FIG. 8. Constructs of the 9E3 gene promoter, their responsiveness to PDBu and to the inhibitors PD98059 and dexamethasone. A, schematic representation of four reporter gene constructs containing the 9E3 promoter region up to Ϫ683 bp and showing the presence of consensus DNA binding elements for several transcription factors. Controls consisted of the basic pGL3 vector containing the appropriate construct transfected into qcCEFs, in the presence or absence of Me 2 SO but without stimulation by PDBu (PdiBt). The promoter activity for each construct after stimulation by PDBu is shown as luciferase activity normalized to ␤-galactosidase activity and to the control values. Me 2 SO, which was used to dissolve the PDBu and the inhibitors, did not cause significant activation, corroborating the results obtained with the northern blot and immunoblot analysis. The largest construct p683 (I) conferred a large increase in luciferase gene activity upon stimulation by PDBu. The construct with the Elk1 DNA binding elements deleted, p470 (II), showed a much reduced response to PDBu (only twice the activity of its control). The two smallest constructs p218 and p66 (III and IV, respectively) showed virtually no response to PDBu. B, PD98059 lowered the stimulation of pGL3/9E3p683 by PDBu to below the level of the control in a dose-dependent manner. C, dexamethasone also inhibited activation of transcription of pGL3/9E3p683 by PDBu in a dose-dependent manner, but even at the highest doses expression remained above the control. The experiments depicted in this figure were performed several times, but here we show a representative experiment because we use primary cells and significant variations in activation occur from batch to batch of cells. The bars represent S.E. of three samples/condition in the same batch of cells. We identified the potentially functional elements known to be important in regulation of chemokine transcription by the Transcription Element Search Software (81) assay. developed by Stratagene and applied it to our system (Fig. 9B). Cells were cotransfected with the pFR-Luc construct, which contains the Gal4 DNA binding sequence upstream of the luciferase gene, and with the pFA-Gal4dbd-Elk1AD construct (pFA-dbd-Elk1), which is an expression vector for a fusion protein containing the Gal4 DNA binding domain (Gal4dbd) and the Elk1 activation domain (Elk1AD). The Gal4dbd part of this fusion protein binds to the Gal4 element on pFR-Luc, and the Elk1AD portion of the fusion protein is capable of activat-ing the luciferase gene. As a negative control we used an expression vector for the Gal4 DNA binding domain alone (pFC-dbd), and as a positive control we used an expression vector for MEK1 (pFC-MEK1). Our results show that when pFC-Gal4dbd was cotransfected with the reporter construct (pFR-Luc) and subsequently treated with PDBu, there was a very small activation of the luciferase gene, which was not significantly different from the control (Fig. 9B, b); the same was observed if cotransfection was performed with pFA- FIG. 9. Elk1 is the major response element/transcription factor involved in PDBu stimulation of 9E3/cCAF. A, a construct (pmElk1) was prepared by deletion of the Elk1 element starting at Ϫ534 bp and a mutation of the one starting at Ϫ493 bp. The basal transcription level of this construct was the same as that of the p470 and was much lower than that of the normal p683 construct or the controls p542 and p495. The loss of the Elk-1 elements results in loss of responsiveness to PDBu and to PD98059 inhibition. B, Elk-1 transcription factor responds to PDBu (PdiBt) stimulation. Cells were cotransfected with a reporter construct containing six Gal4 elements in a series immediately upstream of the luciferase gene (pFR-Luc) and with the expression vector for a fusion protein (pFA-Gal4dbd-Elk1AD) containing Gal4dbd and the Elk-1 protein activation domain (Elk-1AD). In addition to the transfection of pFR-Luc alone (a), two other controls were used: (i) cells were cotransfected with pFR-Luc and the expression vector for the Gal4dbd and then stimulated by PDBu (b), and (ii) cotransfection was done with pFR-Luc and the expression vector for the fusion protein in the absence of stimulation by PDBu (c). We observed no activation in either case. Cotransfection of the cells with pFR-Luc and pFA-Gal4dbd-Elk1AD and stimulation with PDBu resulted in a 6-fold increase over the control (d). This latter stimulation can be inhibited to the background level by 25 M PD98059 (e), the same dose used in other p683 studies. As a positive control, MEK-1, the kinase that activates ERK2 was cotransfected with pFA-Gal4dbd-Elk1AD and pFR-Luc. Overexpression of MEK-1 in the absence of stimulation by PDBu resulted in a 3-fold increase in activity of the reporter gene when compared with the control (f). C, gel shift analysis using nuclear extracts from primary cells overexpressing Elk1 from pCMV-Elk1 (60). Nuclear extracts from cells treated with PDBu and incubated with the radiolabeled oligonucleotide containing the sequence of the Elk1 binding element present in the 9E3 gene resulted in three band shifts (lane 1); competition with excess unlabeled oligonucleotide (lane 2) and incubation with radiolabeled mutated oligonucleotide (lane 3) show specificity of binding. Shifted bands are indicated with arrowheads. In lane 4 the extract was incubated with an antibody directed to a peptide containing phosphoserine 383, which is phosphorylated only when Elk1 is activated. Binding of this antibody produced the supershift indicated by a double arrowhead. When the nuclear extracts from cells treated with PDBu in the presence of the inhibitor PD98059 were incubated with the radiolabeled oligonucleotide and the same Elk1 antibody, a supershift did not occur (lane 5). Each experiment depicted here was performed at least three times in three different batches of primary cells. The bars represent S.E. of the three samples analyzed/condition. Gal4dbd-Elk1AD in the absence of stimulation by PDBu (Fig.  9B, c). However, when this phorbol ester was used to stimulate the cells cotransfected with the pFA-Gal4dbd-Elk1AD, we observed a 4-fold increase in activation (Fig. 9B, d), which was eliminated by PD98059 (Fig. 9B, e), the inhibitor for MEK1. Cotransfection of a MEK1 expression vector with the reporter construct and the expression vector for the fusion protein Gal4dbd-Elk1AD but in the absence of PDBu stimulation caused significant activation of the reporter gene (Fig. 9B, f). Therefore, overexpression of MEK1 activated its substrate ERK2 followed by activation of the ELK1 transcription factor. The results show that PDBu can stimulate activation of the Elk1 transcription factor leading to gene expression.
To determine if Elk1 activated by PDBu binds specifically to its element in these primary fibroblasts, we performed EMSA and supershifts using antibodies specific for activated Elk1. Our normal primary cells express very low levels of the Elk1 transcription factor making it difficult to obtain strong shift and supershift bands. Therefore, we overexpressed Elk1 using pCMV-Elk1 (60), treated the cells with PDBu, and incubated the nuclear extracts with a radiolabeled 27-mer oligonucleotide containing a sequence identical to that of the Elk1 elements present in our promoter. As observed by others (60,61), EMSA showed multiple shifted bands (Fig. 9C, arrowheads). These shifted bands were competed out with cold oligonucleotide (Fig.  9C, lane 2) and disappeared when incubated with the mutated radiolabeled oligonucleotide (Fig. 9C, lane 3). Incubation with a specific antibody to activated Elk1, which is phosphorylated on serine 383 (60), resulted in a supershift of two of the Elk1 bands to a common band (Fig. 9C, lane 4, double arrowhead). Because Elk1 can bind to its element even when it is not activated (61) to verify that the supershifted band represents activated Elk1, we treated the same batch of primary cells with PDBu in the presence of the inhibitor PD98059. As described above, PD98059 is a specific inhibitor of MEK1 that directly activates ERK2, which phosphorylates/activates Elk1. Therefore, the Elk1 transcription factor bound to the oligonucleotide should not be phosphorylated on Ser-383 and should not be recognized by the antibody specific to this Ser. When the nuclear extracts from these cells were incubated with the antibody against activated Elk1, there was no supershift (Fig. 9C,  lane 5). Thus, EMSA confirms that Elk-1 binds specifically to its element in the 9E3 gene promoter and that this transcription factor is activated by PDBu. The results of transcription activation taken together with those on signal transduction show that PDBu stimulates MEK1/ERK2, which results in the activation of the Elk1 transcription factor and stimulation of 9E3/cCAF. Because it has been previously shown that phorbol esters activate the NFB and AP-1 transcription factors (62-64) although it is not the case in our transfection experiment (Figs. 8A and 9A), we tested the possibility that the PDRIIB and the AP-1 binding elements were somehow inactivated by alterations of our constructs during the PCR reaction. Sequence analysis showed that they contain no mutations. Further testing of the PDRIIB was obtained by treating the cells transfected with the Ϫ218 to ϩ32 bp construct with LPS, which is known to activate the CINC chemokine gene through NFB (45). Treatment of this region of the 9E3 gene promoter with LPS activated the reporter system (Fig. 10A), showing that PDRIIB is functional in the 9E3/cCAF promoter but does not respond to PDBu. Testing of the functionality of the AP-1 binding element was performed by cotransfecting the fibroblasts with the p683 construct and the expression vectors containing the c-Fos and c-Jun cDNAs. The latter vectors, when expressed, should form an active AP-1 transcription factor com-plex (60 -65). To determine if the expressed AP-1 transcription factor was functional in our normal fibroblasts, the cis-Path-Detect system (Stratagene) for the AP-1 binding element was used. This system contains seven AP-1 elements in series in a Cis-reporter backbone containing the luciferase reporter gene (pAP-1-Luc). The Cis-reporter by itself caused virtually no activation of the luciferase gene (Fig. 10B, a), whereas in the presence of PDBu (PdiBt) stimulation alone (Fig. 10B, b), c-Fos ϩ c-Jun (Fig. 10B, c), or c-Fos ϩ c-Jun ϩ PdiBt (Fig. 10B, d) there was a 3-4-fold increase in activation. MEK kinase was used as a positive control to verify that the Cis-reporter system was working correctly (Fig. 10B, e). Thus, the AP-1 transcription factor complex made from the expression vectors is functional in our cells. However, when cells were cotransfected with p683 and the expression vectors for the AP-1 complex, we found that c-Fos and c-Jun individually or in combination do not stimulate expression of the reporter gene above that of the control (Fig. 10C, c, e, and g). When these cells were simultaneously stimulated with PDBu, there was an increase in the expression of the reporter to slightly lower levels (Fig. 10C, d, f, and h) than those of the positive control (cells transfected with the p683 alone and treated with PDBu) (Fig. 10C, b). These results show that overexpression of a functional c-Fos⅐c-Jun complex (that can bind to AP-1 response elements and activate transcription) does not activate the reporter gene in the context of our promoter. To further establish the lack of involvement of AP-1 in 9E3/cCAF expression, we used sitedirected mutagenesis to mutate the AP-1 binding element in the context of the p683 promoter (pmAP-1). In comparison to the native p683, the p683 with the mutated AP-1 element showed a small decrease in luciferase activity upon stimulation with PDBu, demonstrating that AP-1 in not significantly involved in the activation of 9E3/cCAF by this tumor promoter.

DISCUSSION
Although it is clear that chemokines play important roles in tumorigenesis, very little is known about the way tumor-promoting agents activate chemokine genes. We have used the 9E3 gene and its product cCAF as a model to study the activation of chemokines in primary fibroblasts by one class of tumor promoters, the phorbol esters. We show the following: (i) 9E3/ cCAF stimulation by PDBu is only moderately inhibited by down-regulation of C1 domain-containing PKCs or by inhibition of this enzyme with the specific inhibitor calphostin C and that stimulation can also be achieved by 4␣-PDBu, the isomer of PDBu that does not activate PKC; (ii) general tyrosine kinase inhibitors, although more effective than PKC inhibitors, also did not completely abolish the activation of 9E3/cCAF by PDBu; (iii) dexamethasone is much more effective than PKC or the tyrosine kinase inhibitor in inhibiting the activation of this chemokine by PDBu; (iv) complete inhibition of 9E3/cCAF was not accomplished until the three types of inhibitors were applied simultaneously; (v) inhibition of MEK1/ERK2 by PD98059 also completely eliminated 9E3/cCAF expression; (vi) Elk1 binding elements and the Elk1 transcription factor are crucial for activation of this chemokine gene by PDBu, whereas the AP-1 and the PDRIIB binding elements, although potentially fully functional, are not responsive to the activation of the 9E3/cCAF by PDBu.
Our studies show that all three of these signaling pathways, C1 domain PKCs, tyrosine kinases, and those mediated by dexamethasone, converge to MEK1/ERK2. The activation of ERK2 via multiple signaling pathways is consistent with the recent demonstration that full activation of ERKs, in particular in fibroblasts, may require the cooperation of various signaling pathways (66,67). Furthermore, ERK1 and ERK2 are central transducers of extracellular signals from hormones, growth factors, and cytokines (55,63) and are known to be points of convergence of signals emanating from tyrosine kinase-coupled receptors (68) and G-protein-coupled receptors (69). The results presented here further these studies and show that ERK2 can also be a point of convergence of several signaling pathways generated independently of cell surface receptors.
The inhibition of PDBu stimulation of 9E3/cCAF by dexamethasone is intriguing, because this glucocorticoid is primarily known for its inhibition of the NFB transcription factor, and yet our results show that PDRIIB is not involved in the stimulation of our chemokine gene by PDBu. Similar studies have shown that 12-O-tetradecanoylphorbol-13-acetate cannot activate CINC (a closely related chemokine gene in rats) through NFB, although this gene can be activated by LPS through this transcription factor (45). We also show here that the PDRIIB element in the 9E3 promoter responded to LPS. Taken together, these results suggest that PDBu does not stimulate 9E3/cCAF in normal fibroblasts via the PDRIIB and that dexamethasone is not acting through the inhibition of this transcription factor.
Dexamethasone can also inhibit cytokine gene expression by binding to and activating the glucocorticoid receptor (70), which, in turn, can bind to the AP-1 (c-Fos⅐c-Jun) complex and prevent it from binding to DNA and activating gene expression (71). Although it has been shown previously that the AP-1 site for the 9E3 gene promoter can function in Rous sarcoma virustransformed cells (58,59), we show here that the AP-1 element of the 9E3 promoter is not responsive to the stimulation of this gene by PDBu. Overexpression in normal fibroblasts of a functional c-Fos⅐c-Jun complex that can bind to AP-1 response FIG. 10. Activity of the PDRIIB (NFB-like) and the AP-1 binding elements in response to PDBu stimulation. A, stimulation of the pGL3/9E3p218 and pGL3/9E3p66 by LPS. LPS stimulated transcription from the p218 promoter because this promoter contains the PDRIIB DNA binding element. p66 contains only the TATA box therefore, stimulation was insignificant. B, testing the functionality of the c-Fos and c-Jun expression systems in normal fibroblasts. For these studies a promoter containing 7 AP-1 binding elements in a series in front of the luciferase reporter gene (cis-PathDetect system from Stratagene) was used. Overexpression of c-Fos and c-Jun in the absence (c) or presence (d) of PDBu (PdiBt), caused a 4 -5-fold increase in transcription of the reporter system. The functionality of the cis-PathDetect system was shown by cotransfection of the cells with an expression vector for MEK kinase, which functions as the positive control for the system (e). C, the activity of the AP-1 binding element in the 9E3 promoter was studied by transfection of the cells with p683 and the expression vectors for c-Fos, c-Jun, or both. c-Fos and c-Jun individually or in combination do not stimulate expression of the reporter gene above that of the control (c). When these cells were simultaneously stimulated by PDBu there was an increase in expression of the reporter to slightly lower levels (d, f, and h) than those of the positive control (b). D, to further determine the lack of involvement of AP-1 in 9E3/cCAF expression, the AP-1 binding element was mutated within the context of the p683 promoter (pmAP-1) by site-directed mutagenesis using Quickchange (Stratagene), and the mutation was verified by restriction enzyme digestion and sequencing. Our results show only an insignificant reduction in activation of the reporter system when the AP-1 binding element was mutated, demonstrating that in primary normal fibroblasts activation of 9E3/cCAF by PDBu does not involve AP-1. Each experiment was performed at least twice with two different batches of cells. The bars represent S.E. of three samples/condition. elements and activate transcription does not activate the reporter gene in the context of the 9E3 promoter AP-1 binding element. Furthermore, mutation of the AP-1 binding element within the context of the 9E3/cCAF promoter did not result in significant reduction of activity of the reporter system. These results taken together demonstrate that AP-1 is not importantly involved in the stimulation of 9E3/cCAF by PDBu. Therefore, it is possible that the AP-1 binding element is inaccessible for activation of this chemokine gene in normal fibroblasts. Interestingly, we have identified a consensus binding sequence for the Oct-1 repressor near the AP-1 binding element. Oct-1 is a member of the POU domain family and has been shown to repress the expression of several genes (72)(73)(74)(75). Therefore, it is possible that this repressor in normal cells inhibits the ability of the activated c-Fos⅐c-Jun complex to bind to its element and/or to activate gene expression, whereas in transformed cells this repression is lifted. Work is in progress 2 to analyze the role of this repressor in the tightly regulated expression of 9E3/cCAF in normal cells to compare with the constitutive expression observed in transformed fibroblasts.
In addition to its role in the inhibition of NFB and AP-1, dexamethasone also inhibits the phosphorylation of the Raf1 kinase (76,77). Raf1 is known to act upstream from MEK1 and ERK2 in the MAP kinase cascade (77,78) and, therefore, can be importantly involved in ERK2 activation. Our ongoing work in deciphering the signal transduction pathways of stimulation of 9E3/cCAF has shown that Raf1 activation is required for 9E3/ cCAF expression stimulated by phorbol esters, 3 suggesting that dexamethasone acts on Raf1 to inhibit MEK1/ERK2 activity and 9E3/cCAF expression.
ERK2 is known to activate gene expression via a variety of transcription factors including Elk1 (79 -81). Our data show that 80 -90% of the activation of 9E3/cCAF by PDBu occurs in the Ϫ543 to Ϫ470 bp region of the 9E3 promoter (Figs. 8A and 9A), which contains two Elk1 responsive elements; deletion and mutation of the two Elk1 elements present in this region of the promoter eliminated this effect. Using a Gal4-Elk1 fusion protein, we also show that PDBu stimulation leads to the activation of the Elk1 transcription factor and subsequent activation of the reporter system (Fig. 9B). EMSA with nuclear extracts of cells overexpressing Elk1 shows that in vitro the oligonucleotide sequence representing the Elk1 binding elements in the promoter of 9E3/cCAF was specifically bound by nuclear protein(s). The shift band can be retarded by an antibody specific to Ser-383 phosphorylated/activated Elk1, demonstrating that after treatment with PDBu the factor that binds to the elements is activated Elk1 (Fig. 9C). The interesting supershift of two bands to the same heavier band by the antibody to activated Elk1 is the subject of ongoing work in our laboratory. 3 Using transcription activation reporter systems, site-directed mutagenesis, heterologous expression systems, and EMSA, we have shown that the major transcriptional activation pathway for 9E3/cCAF expression upon stimulation by PDBu occurs via ERK2/Elk1.
The remaining 10 -20% of activation of 9E3/cCAF by PDBu occurs almost exclusively in the Ϫ470 to Ϫ218 bp section of the promoter (Figs. 8A and 9A). This region does not contain a consensus sequence for the Elk1 element, but it does contain a closely related Ets sequence, which could be responsible for this smaller activation. It has been shown that several Ets domain binding factors are targets for the ERK proteins (80), making it possible that ERK2 activates one of the Ets-binding proteins, which in turn could be responsible for the small level of acti-vation we observed with this shorter promoter.
When taken together, our results lead to a potential model for activation of 9E3/cCAF by phorbol esters in primary normal fibroblasts. Complete inhibition of PDBu stimulation of 9E3/ cCAF by PD98059 (downstream inhibitor of the MAP kinase cascade) indicates that all three signaling pathways go through MEK1/ERK2, whereas dexamethasone (upstream inhibitor of the MAP kinase cascade) inhibits only about 80% of the stimulation, suggesting that one of the less active pathways (PKC or tyrosine kinase activated) is not inhibited by dexamethasone. Thus, the major pathway by which PDBu stimulates 9E3/cCAF and one of the two minor pathways must merge together before the point of dexamethasone inhibition, and the third pathway involved must merge in the cascade after the point of dexamethasone inhibition. Dexamethasone is known to interfere with Raf1 (76,77), which is the kinase that acts upon MEK1 in the MAP kinase signaling cascade, hence the minor pathway independent of dexamethasone inhibition could merge either at the point of MEK1 or immediately before it. Our observation that both dexamethasone and PD98059 completely inhibit 9E3/cCAF stimulation by 4␣-PDBu (which does not activate PKC), whereas dexamethasone only partially inhibits stimulation by PDBu, suggests that the dexamethasoneindependent pathway is the pathway involving PKC. Alternatively, PKC can stimulate Raf1 activation in a way that cannot be inhibited by dexamethasone.
In summary, we show for the first time that phorbol esters can activate a chemokine gene via multiple pathways that involve simultaneously a dexamethasone-sensitive pathway and to a much lesser extent tyrosine kinases and C1 domaincontaining PKCs. These pathways all culminate in ERK2 activation and in dominant stimulation of a major regulatory region of the gene that contains two Elk1 consensus binding sites. Furthermore, our functional studies show that stimulation of this gene by phorbol esters leads to activation of the Elk1 and that this transcription factor binds to its elements in the 9E3/ cCAF promoter. We conclude that the predominant signaling pathway for the activation of 9E3/cCAF by PDBu goes through the MAP kinase cascade enzymes MEK1/ERK2 and involves activation of the transcription factor Elk1. The existence of multiple pathways to activate chemokine genes is important because these are immediate early response genes that are expressed in response to stresses such as tumors and injuries. Agonists that use several pathways are more versatile in turning on genes potentially allowing for a more reliable response.