Synergistic activity of STAT3 and c-Jun at a specific array of DNA elements in the alpha 2-macroglobulin promoter.

The transcriptional activity of natural promoters is sensitive to the precise spatial arrangement of DNA elements and their incorporation into higher order DNA-protein complexes. STAT3 and c-Jun form a specific ternary complex in vitro with a synthetic DNA element containing AP1 and SIE sites. These associations are critical for synergistic activation of transcription from a synthetic promoter by STAT3 and c-Jun. Expression of the acute phase protein alpha(2)-macroglobulin is induced in vivo by interleukin-6 (IL-6)-related cytokines; we demonstrate that coordinate interactions among STAT3, c-Jun, and a specific array of DNA elements contribute to activation of the alpha(2)-macroglobulin promoter in response to IL-6 family members. At least five promoter elements are involved in activation: two AP1 sites at -113 to -107 and -152 to -140, an acute phase response element (APRE (SIE)) at -171 to -163, and two AT-rich regions at -143 to -138 and -128 to -123. Synergism between STAT3 or STAT3-C and c-Jun is impaired by mutation of the APRE (SIE) or either AP1 site, as well as by mutations that alter the AT-rich regions or their phasing. Mutations of STAT3 previously shown to disrupt physical and functional interactions with c-Jun do not impair synergy between STAT3-C and c-Jun at the alpha(2)-macroglobulin promoter in HepG2 cells, suggesting that STAT3-C and STAT3 differ with respect to their precise contacts with c-Jun.

Signal transducers and activators of transcription (STATs) 1 play central roles in the induction of gene expression by a diverse variety of signaling pathways involving cytokines, growth factors, and peptide hormones. Cytokine receptor-associated protein-tyrosine kinases of the JAK family phosphorylate STAT proteins upon binding of cytokines to their cognate receptors (1)(2)(3). The phosphorylated STATs then form dimers via their SH2 domains and rapidly translocate from the cytoplasm to the nucleus, where they bind regulatory DNA elements of target genes (4 -7). STAT proteins thereby exert effects on a number of fundamental biological processes, including cell proliferation, differentiation, apoptosis, and development (8,9).
In mammals, the acute phase response (APR) is invoked in response to tissue injury, trauma, or infection (16,22). During the early stages of inflammation serum concentrations of several acute phase response proteins increase as much as 1000fold. These increases are triggered by glucocorticoids or by cytokines such as IL-1, IL-6, tumor necrosis factor ␣, and interferon ␥ (IFN␥) (23,24). A number of APR genes contain both IL-1-and IL-6-responsive elements, suggesting that these signaling pathways act in synergy to induce APR gene transcription (25,26). Glucocorticoids exert a synergistic effect on the IL-6-mediated inflammatory response through a direct interaction between IL-6-activated STAT3 and ligand-bound glucocorticoid receptor (27)(28)(29)(30).
Physiologic enhancers serve as substrates for the assembly of multicomponent complexes containing transcriptional regulators and proteins that modify DNA structure. Such DNAprotein complexes, termed enhanceosomes, exhibit specific patterns of spatial organization. Enhanceosomes in the genes encoding the T-cell receptor ␣-chain and interferon ␤ (IFN␤) have been studied extensively (31)(32)(33)(34)(35). Two pairs of AT-rich sequences positioned in-phase on the IFN␤ enhancer are required for cooperative binding of HMG-I(Y) and enhancement of transcriptional activity in vivo (34). Formation of a stable ternary complex of transcription factors on the T-cell receptor ␣-chain enhancer and enhancer activity in nonlymphoid cells requires the lymphoid-specific, HMG domain-containing protein LEF-1, which facilitates interactions between proteins bound at nonadjacent sites (33). Similarly, STAT3 and Smad1, when bridged by the transcriptional coactivator p300, are reported to exert a synergistic effect on transcription from the glial fibrillary acidic protein promoter, thereby inducing astro-cytic differentiation of primary fetal neural progenitor cells (36).
Direct interactions between the c-Jun and STAT3 proteins have been detected using the yeast two-hybrid system or an in vitro GST precipitation assay (37,38). Our laboratory has reported that c-Jun and a carboxyl-terminal-truncated form of STAT3 (STAT3␤) show synergistic activation of the IL6-responsive ␣ 2 -macroglobulin (␣ 2 -MG) promoter in transfected F9 and EGF-stimulated COS-7 cells (37). The ␣ 2 -MG promoter contains binding sites for STAT3 (acute phase response element (APRE); TTCTGGGAA) and AP1 (TGACTCT) (39). These are separated by 49 base pairs (39 -41). Synergistic activation of the ␣ 2 -MG promoter by the same transcription factors has been also reported upon IL-6 stimulation of HepG2 cells (38).
Here we present evidence that sequence-specific DNA and protein interactions at a specific array of DNA elements in the ␣ 2 -MG promoter are critical for cooperative transcriptional activation of the ␣ 2 -MG transcription by STAT3 and c-Jun. Our observations suggest that members of the IL-6 cytokine family induce formation of an ␣ 2 -MG enhanceosome containing activated transcription factors in a specific spatial arrangement.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Human HepG2 cells were maintained in Eagle's minimal essential medium supplemented with 10% fetal bovine serum (Hyclone). Mouse fibroblast NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were plated at 2 ϫ 10 5 cells per well in 24-well plates. On the following day, cells were transfected with DNA (up to 1.5 g per well) and LipofectAMINE (Life Technologies, Inc.) in triplicate. The plasmid pRL-TK (Promega) was cotransfected to assess transfection efficiency. Twenty hours after transfection, cells were transferred to medium containing 0.5% fetal bovine serum and incubated overnight. Cells were then stimulated by addition of IL-6 (10 ng/ml), LIF (10 ng/ml), or OSM (10 ng/ml) in the presence of 0.5% fetal bovine serum for 16 h or as indicated. IL-6, LIF, and OSM were purchased from R&D Systems. Luciferase was assayed according to the manufacturer's directions (Promega).
Plasmids and Mutagenesis-Plasmids encoding STAT3 and c-Jun in the mammalian expression vector pYN3218 have been described previously (37,39). Expression plasmids for STAT3C, STAT3C-TKR, STAT3C-L148A, and STAT3C-V151A were generated in pYN3218 by site-directed mutagenesis as described (38,42). To produce a plasmid encoding STAT3C-TKR, two rounds of mutagenesis were performed, using oligonucleotides 5Ј-GGTGTCCAGTTTGCCACGGCAGTCAGGT-TGCTG-3Ј and 5Ј-GCCACGGCAGTCGCGTTGCTGGTC-3Ј. To construct plasmid pYN3218/GST-c-Jun, the mouse c-Jun coding sequence was fused in-frame with the glutathione S-transferase (GST) coding sequence in the bacterial expression vector pGEX-4T; the entire GSTc-Jun fragment was then transferred to the mammalian expression vector pYN3218.
Gel-purified oligonucleotides were annealed, labeled with 32 P, and used in EMSA as described previously (17). Briefly, 0.5 ng of 32 Plabeled, duplex oligonucleotides was incubated with purified proteins or nuclear extracts and poly(dI-dC) (1 g) in binding buffer (10 mM HEPES (pH 7.9), 50 mM KCl, 5 mM MgCl, 5 mM DTT, 1 mM EDTA, 24 g of bovine serum albumin, 10% glycerol); reactions (24 l) were incubated at room temperature for 15 min. For supershift experiments, antibody (1 g) to STAT3 (Upstate Biotechnology Inc.), c-Jun (Santa Cruz Biotechnology), or GST (Santa Cruz Biotechnology) was added to the reaction mixture after 15 min of incubation, and incubation was continued 15 min further at room temperature. DNA-protein complexes were resolved on a 4% polyacrylamide gel or as indicated.
Purification of GST-c-Jun and His-STAT Proteins-The pYN3218c-Jun mammalian expression plasmid was transfected into COS-7 cells as described above. At 40 -48 h post-transfection, the cells were harvested in phosphate-buffered saline containing 1 mM EDTA, 1 mM DTT, 2 g/ml aprotinin, 2 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 1% Triton X-100. Total cell lysate was incubated with glutathione-agarose beads at 4°C for 2 h. After washing five times in the same buffer, bound protein was eluted with 10 mM glutathione in Tris-Cl (pH 8.0) and then dialyzed against STAT3 binding buffer. Purified protein was concentrated by ultrafiltration (Amicon), and protein concentration was determined by SDS-polyacrylamide gel electrophoresis and staining with Coomassie Blue.
Polyhistidine-tagged STAT 3 (His-STAT3) protein was expressed and purified as described previously (17). In brief, Sf9 cells were co-infected with recombinant His-STAT3 and Jak1-or Jak2-expressing baculoviruses (17) and harvested at 72 h after infection. The cells were disrupted in lysis buffer (20 mM Hepes (pH 7.9), 100 mM NaCl, 2 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na 3 VO 4 , 1 mM Na 3 MO 4 , 15% glycerol, 0.5% Nonidet P-40, and 10 g/ml leupeptin), supplemented with 15 mM imidazole. Total cell lysate was bound to Ni 2ϩ -Probond (Invitrogen) at 4°C for 2 h. After washing five times with lysis buffer containing 15 mM imidazole, bound protein was eluted stepwise with lysis buffer supplemented with 40 mM, 50 mM, and finally 350 mM imidazole. Fractions containing His-STAT3 protein were pooled, and EDTA was added to a final concentration of 20 mM. Pooled protein was dialyzed overnight against 10 mM HEPES (pH 7.4), 100 mM NaCl, and 0.5 mM DTT. Purified protein was concentrated by ultrafiltration. The final protein concentration was determined by SDS-polyacrylamide gel electrophoresis and staining with Coomassie Blue.

STAT3 and c-Jun Proteins Exhibit Cooperative DNA Bind-
ing Activity in Vitro-STAT3␤, a truncated isoform of STAT3 lacking the carboxyl-terminal 55 amino acid residues, cooperates with c-Jun in transactivation of the ␣ 2 -MG promoter in EGF-stimulated COS-7 cells (37). To understand better the mechanism of cooperativity between STAT3 and c-Jun, we first examined the binding of STAT3 and c-Jun in vitro to an oligonucleotide containing binding sites for both proteins. Polyhistidine-tagged STAT3␤ and GST-fused c-Jun proteins were expressed separately in Sf9 cells and COS-7 cells, respectively. STAT3␤ was activated by co-infection with baculovirus-encoding Jak1 or Jak2. STAT3␤ or GST-fused c-Jun protein each exhibited specific binding to an AP1/SIE-containing DNA probe when assayed by EMSA (Fig. 1A, lanes 1-3). In addition, a second, more slowly migrating species was observed when STAT3␤ and GST-c-Jun were combined with the AP1/SIE probe (Fig. 1A, lanes 6 and 7). This slower species was not seen in control reactions containing STAT3␤ and GST protein ( In comparison, a DNA-protein complex containing c-Jun alone was disrupted by the antibody against c-Jun but not by the anti-STAT3 antibody (data not shown). These results indicated that both STAT3␤ and GST-c-Jun were present in the slow mobility complex. We next examined STAT3␣ and c-Jun proteins for cooperative DNA binding activity in vitro, using the AP1/SIE oligonucleotide as a probe. As was seen for STAT3␤, incubation of the AP1/SIE probe with STAT3␣ and c-Jun was associated with the appearance of a slowly migrating complex (Fig. 1C, compare lanes 2 and 3 with lane 4). Mutation of either DNA binding site abolished formation of the slower complex (Fig. 1C, lanes [5][6][7][8][9][10][11][12], and mutation of both binding sites abolished specific binding altogether (Fig. 1C, lanes [13][14][15][16]. Taken together, these results indicate that the formation in vitro of DNA-protein complexes containing STAT3 and c-Jun requires specific DNA binding by both proteins. AP1 and SIE Elements Synergistically Support Cytokineinduced Transcription from Artificial Promoters-We next at-tempted to correlate the results of in vitro DNA binding experiments with the transcriptional activities of endogenous STAT3 and c-Jun proteins in cytokine-stimulated NIH3T3 or HepG2 cell lines. For this purpose, we assembled a luciferase reporter construct (pGL3-AP1/SIE-luc) in which transcription is driven by an artificial promoter containing an SIE (TTCCCGTAA) and an AP1 site (TGACTCA), separated by 11 bp. Additional reporter constructs contained mutations at the AP1 site (pGL3-AP1(Ϫ)/SIE-luc), the SIE (pGL3-AP1/SIE(Ϫ)-luc), or both (pGL3-AP1(Ϫ)/SIE(Ϫ)-luc). Wild-type or mutant constructs were transfected into NIH3T3 or HepG2 cells, and luciferase activity was assayed in response to stimulation with IL-6, LIF, or OSM (Fig. 2, A and B).
Treatment of HepG2 cells with OSM or IL-6 resulted in a 2to 3-fold induction of AP1/SIE-luc reporter expression, whereas LIF had little effect on reporter activity (Fig. 2B, AP1/SIE-luc). The IL-6 and OSM induction ratios were lower than those seen in NIH3T3 cells; this difference may be due in part to the higher basal activity of the AP1/SIE-luc reporter in HepG2 cells. Mutation of the c-Jun or STAT3 binding sites reduced induction to between 10 to 25% that seen with the AP1/SIE-luc plasmid (Fig. 2B, AP1(Ϫ)/SIE-luc and AP1/SIE(Ϫ)-luc). Cytokine-induced transcription from the artificial AP1/SIE promoter was completely abolished by mutations of both STAT3 and AP1 binding sites in HepG2 cells (Fig. 2B, AP1(Ϫ)/SIE(Ϫ)luc). These results suggested that cooperative binding of STAT3 and c-Jun might underlie the synergism between SIE and AP1 elements in supporting induction of AP1/SIE-luc transcription by cytokines.
Mutation of the APRE site of the ␣ 2 -MG promoter (Fig. 3A, SIEm) abolished cytokine induction of luciferase expression (Fig. 3B, ␣2MG-SIEm-luc), consistent with the interpretation that STAT3 binding to the APRE site is essential for activation of the ␣ 2 -MG promoter by IL-6 or OSM. To determine whether an intact APRE could support IL-6-mediated ␣ 2 -MG transcription in the absence of an AP1 site, the c-Jun binding site in the ␣ 2 -MG promoter was mutated (Fig. 3A, AP1m). Despite the presence of an intact STAT3 binding site, transcription of the AP1 mutant reporter was decreased to 30% of the levels supported by the intact ␣ 2 -MG sequence in the presence of IL-6 or OSM (Fig. 3B, ␣2MG-AP1m-luc). Transcription of the reporter construct containing intact AP1 and SIE sites was 3-fold greater than the additive effects of the AP1 and SIE sites alone, suggesting that STAT3 and AP1 binding sites cooperate in activation of the ␣ 2 -MG promoter by IL-6 or OSM Upon inspection, the Ϫ190 to Ϫ100 region of the ␣ 2 -MG promoter was found to contain an additional potential AP1 site (AP1*, at Ϫ152 to Ϫ140) and two AT-rich regions (AT1 and AT2, at Ϫ143 to Ϫ138 and Ϫ128 to Ϫ123, respectively) (Fig.  3A). A duplex oligonucleotide corresponding to the AP1* site and flanking sequences from the ␣ 2 -MG promoter was found by EMSA to bind purified c-Jun protein (Fig. 3C, lanes 2 and 4  versus lane 1, no protein added). Binding is specific, because it was abolished in the presence of excess unlabeled, wild-type AP1* oligonucleotide, but not by a mutant (AP1*m) oligonucleotide (Fig. 3C, lanes 3 and 5). A number of mammalian proteins, including those of the high-mobility group (HMG), are known to bind AT-rich regions (44). A duplex oligonucleotide spanning both AT-rich sites of the ␣ 2 -MG promoter was specifically bound by one or more proteins from nuclear extracts of HepG2 cells maintained in serum with or without OSM (Fig.  3D, lanes 2 and 3 versus lane 1, in which no protein was added). This binding was abolished by excess unlabeled, wild-type oligonucleotide but not by an oligonucleotide carrying mutant AT-rich sites (Fig. 3D, lanes 4 and 5).
We then asked whether the AT-rich regions could cooperate with AP1 or SIE elements in formation of DNA-protein complexes on the ␣ 2 -MG promoter. Four prominent protein complexes (I-IV) were formed with a 105-base pair DNA fragment containing the minimal ␣ 2 -MG promoter (Ϫ190 to Ϫ100) in nuclear extracts of OSM-treated, HepG2 cells (Fig. 3E, lane 1). Formation of complex I was dependent on the AT-rich elements, as this species was not observed with probes containing mutated AT-rich regions (Fig. 3E, lane 4). Formation of complexes II and IV required the AT-rich and AP1 sites (Fig. 3E,  lanes 3 and 4), consistent with formation of protein-DNA complexes involving the AP1 and AT-rich elements of the ␣ 2 -MG promoter. The yield of complex IV was somewhat diminished by mutation of the SIE site (Fig. 3E, lane 2). Formation of complex III did not require the SIE, AP1, or AT-rich regions of the ␣ 2 -MG promoter (Fig. 3E, lanes 1-4). (The appearance of novel species in lanes 3 and 4 could represent binding to nonphysiologic sites generated by the AP1 or AT mutations.) Taken together, these results suggested the involvement of one or more specific macromolecular complexes, involving interactions between AT-rich and AP1 elements, in activation of the ␣ 2 -MG promoter by IL-6 or OSM.
STAT3-C and c-Jun Synergistically Activate Transcription from the ␣ 2 -MG Promoter through Specific DNA Binding-To examine the molecular basis of synergism between AP1 and APRE sites in the ␣ 2 -MG promoter, we transiently cotrans-fected HepG2 cells with expression vectors encoding STAT3␣, c-Jun, and the pGL3-␣2MG-luc reporter construct. Overexpression of c-Jun alone had little effect on OSM-stimulated reporter gene transcription relative to the vector control, whereas overexpression of STAT3␣ alone increased transcriptional activity by about 30% (Fig. 4A). Cotransfection of STAT3␣ and c-Jun had a synergistic effect, increasing OSM-stimulated luciferase expression 2-fold, relative to the vector control (Fig. 4A). This stimulatory effect, however, was relatively small compared with the deleterious effects of mutations at the AP1 and APRE sites (Fig. 3B), possibly because the effects of transfected STAT3 and c-Jun were blunted by endogenous STAT3 or c-Jun. To overcome this problem, further experiments employed STAT3-C, a highly active form of STAT3 that constitutively dimerizes, binds to DNA, and activates transcription (42). As previously reported (42), overexpressed STAT3-C exhibited constitutive transcriptional activity when assayed using an artificial SIE-containing promoter (data not shown). We proceeded to ask whether overexpressed STAT3-C acts synergistically with c-Jun in activating transcription from the ␣ 2 -MG promoter in HepG2 cells. In the absence of cytokine stimulation, STAT3-C was unable to stimulate transcription from the ␣ 2 -MG promoter (Ϫ190 to Ϫ100) (Fig. 4B, Unstim.). This stands in contrast to a report that STAT3-C constitutively activates transcription from the ␣ 2 -MG promoter (Ϫ1151 to ϩ54) in 293T cells (42); the difference may reflect the presence of additional regulatory elements outside of the Ϫ190 to Ϫ100 interval. Increasing amounts of c-Jun had no effect on transcription of ␣ 2 -MG-luc in the presence or absence of STAT3-C in unstimulated HepG2 cells (Fig. 4B, Unstim.). Transcription of pGL3-␣2MG-luc was induced 6-fold by OSM stimulation, and this was unaffected by increasing amounts of c-Jun expression (Fig. 4B, ϩOSM, vector). Expression of STAT3-C alone was associated with a 20% increase in the level of OSM-induced transcription; this was stimulated further upon coexpression of increasing amounts of c-Jun, indicating transcriptional synergism between c-Jun and STAT3-C (Fig. 4B, ϩOSM, Stat3-C). Similar synergism was observed in HepG2 cells stimulated with IL-6 (data not shown).
We next asked whether the synergism between STAT3-C and c-Jun was dependent on the presence of their specific binding sites in the ␣ 2 -MG promoter. When either c-Jun binding site was mutated singly, transcriptional activity was coordinately reduced to about 15-20% that of the intact promoter (Fig. 4C, compare ␣2MG-AP1m-luc or ␣2MG-AP1*m-luc with ␣2MG-luc). Neither mutation, however, abolished synergism between STAT3-C and c-Jun. This indicates that a single intact AP1 site is sufficient to support a functional, cooperative interaction between c-Jun and STAT3-C. Nonetheless, the presence of both AP1 sites is essential for full promoter activity (Figs. 3B and 4C). As expected, when both AP1 sites were mutated, synergism between STAT3-C and c-Jun was abolished, although weak STAT3-C-dependent, c-Jun-independent promoter activity remained (Fig. 4C, ␣2MG-AP1*mAP1m-luc). As observed above (Fig. 3B), mutation of the APRE site nearly abolished transcriptional activity, although cotransfection of STAT3-C and increasing amounts of c-Jun was associated with a slight increase in expression (Fig. 4C, ␣2MG-SIEm-luc). The latter result could reflect possible recruitment of STAT3-C into the ␣ 2 -MG regulatory complex through direct protein-protein interactions. Upon mutation of the SIE and either AP1 site, no cooperative transcriptional activation was observed (Fig. 4C, ␣2MG-AP1mSIEm-luc and ␣2MG-AP1*mSIEm-luc). The dependence on specific cis-acting elements observed for STAT3-C was confirmed with the wild-type protein. Synergism between wild-type STAT3 and c-Jun in OSM-treated HepG2 cells exhibited a similar pattern of dependence on the AP1 sites and the APRE (Fig. 5D, compare ␣2MG-luc with ␣2MG-SIEm-luc, ␣2MG-AP1*mAP1m-luc, and ␣2MG-AP1mSIEm-luc).
Synergism between STAT3-C and c-Jun at the ␣ 2 -MG Promoter Requires Specific Phasing of AT-rich Regions-Functional interactions between transcriptional regulatory proteins can be exquisitely sensitive to alterations in their spatial orientation. Because AT-rich regions of DNA exhibit intrinsic curvature (46 -48) and are often binding sites for DNA bending proteins (49,50), we asked whether the AT-rich regions of the ␣ 2 -MG promoter contribute to the synergy between STAT3-C and c-Jun (Fig. 5A). Mutation of the first AT-rich region (AT1m) reduced overall transcription levels to about 10 -15% those of the intact construct, although synergism between STAT3-C and c-Jun was maintained (Fig. 5A, compare ␣2MG-AT1m-luc and ␣2MG-luc). Mutation of the second AT-rich region (AT2m) had less of an inhibitory effect than mutation of AT1 (reduction to about 30% of pGL3-␣2MG-luc activity); cooperativity between STAT3-C and c-Jun was preserved (Fig.  5A, ␣2MG-AT2m-luc). Mutation of both AT-rich regions, however, abolished synergy between STAT3-C and c-Jun as well as basal transcriptional activity (Fig. 5A, ␣2MG-AT1/2m-luc).
The center-to-center distance between the two AT-rich regions in the ␣ 2 -MG promoter is 15 bp, placing them on opposite sides of the DNA helix. To determine whether the helical phasing of these AT-rich regions affects the functional interaction between STAT3-C and c-Jun, we constructed a variant ␣ 2 -MG promoter in which the center-to-center distance between the two AT-rich regions is 10 bp, placing them on the same side of the DNA helix. To maintain the overall integrity of the ␣ 2 -MG promoter, the first AT-rich region was shifted 5 bp to the right (Fig. 5B, ␣2MG-Shift-luc). This alteration resulted in a 5-fold inhibition of luciferase activity, relative to the intact pGL3-␣2MG promoter in HepG2 cells; moreover, synergy between STAT3-C and c-Jun was impaired (Fig. 5C, compare ␣2MG-Shift-luc and ␣2MG-luc). Dependence on the presence and phasing of the AT-rich sites was also observed in OSM-treated HepG2 cells expressing wild-type STAT3 and c-Jun (Fig. 5D, compare ␣2MG-luc with ␣2MG-AT1/2m-luc and ␣2MG-Shiftluc). These results suggest that the phasing of the AT-rich regions of the ␣ 2 -MG promoter as well as AT-rich DNA sequences are critical for the maintenance of transcriptional synergism between STAT3 or STAT3-C and c-Jun.
Mutations in the c-Jun Interaction Regions of STAT3-C Do Not Impair Synergism at the ␣ 2 -MG Promoter-Putative c-Jun interaction domains have been mapped to two regions of STAT3 (38). Point mutations within these interacting regions (L148A and V151A in region 1 and T346A,K348A,R350A (TKR) in region 2) inhibited physical interactions between STAT3 and c-Jun in vitro and impaired cooperativity between STAT3 and c-Jun in the induction of transcription from the ␣ 2 -MG promoter by IL-6 in vivo (38). To test whether these putative interaction regions are required for synergism between STAT3-C and c-Jun in OSM-inducible transcription from the ␣ 2 -MG promoter, we introduced identical mutations into STAT3-C (STAT3-C (TKR), STAT3-C (L148A), and STAT3-C (V151A); Fig. 6A). Surprisingly, the STAT3-C mutants retained the ability to synergize with increasing amounts of c-Jun in the induction of ␣ 2 -MG transcription by OSM. In contrast, transcriptional synergism between STAT3 and c-Jun was impaired by the TKR mutation (Fig. 6B), as previously described (38). These data indicate that the interaction regions disrupted by the L148A, V151, and TKR mutations, although essential for cooperative interaction of STAT3 with c-Jun, are dispensable for synergistic activation of the ␣ 2 -MG promoter by STAT3-C and c-Jun in OSM-stimulated HepG2 cells. DISCUSSION In this report we have demonstrated that coordinate interactions among STAT3, c-Jun, and a specific array of DNA elements contribute to cytokine-mediated activation of the minimal ␣ 2 -MG promoter. Three classes of DNA sequence ele- ments are essential for synergism between STAT3 and c-Jun in the Ϫ190 to Ϫ100 region of the ␣ 2 -MG promoter: a single APRE site at Ϫ171 to Ϫ163, a pair of AP1 sites at Ϫ152 to Ϫ140 and Ϫ113 to Ϫ107, and a pair of AT-rich sites at Ϫ143 to Ϫ138 and Ϫ128 to Ϫ123. Significantly, mutations of STAT3 previously shown to disrupt a physical interaction with c-Jun do not impair transcriptional synergism between STAT3-C and c-Jun at the ␣ 2 -MG promoter.
Synergism between STAT3 and c-Jun at Synthetic and Natural Promoters-Using a synthetic promoter containing adjacent AP1 and SIE sites, we observed formation of a specific ternary complex containing STAT3, c-Jun, and DNA. DNA sequence-specific binding of STAT3 and c-Jun to this synthetic promoter was associated with synergistic activation of transcription in NIH3T3 and HepG2 cells in response to the related cytokines IL-6, OSM, and LIF. Synergism between STAT3 and c-Jun was also observed at the natural, ␣ 2 -MG promoter, although the relative contribution of STAT3 to transcriptional activation, as revealed by mutational analysis of cis-acting elements, was somewhat greater than seen with the synthetic promoter. The differential contributions of STAT3 and c-Jun in the activation of synthetic and natural promoters within the same cell line may in part reflect differences in the intrinsic affinities of these promoters for STAT3 and c-Jun. Moreover, the ␣ 2 -MG promoter responded differently to cytokines in fibroblast-derived NIH3T3 cells versus hepatocyte-derived HepG2 cells. In NIH3T3 cells, transcription from the ␣ 2 -MG promoter was poorly inducible in comparison to the synthetic promoter (data not shown), suggesting that the ␣ 2 -MG promoter may be more sensitive than the synthetic promoter to cell-to-cell variation in the amount of active STAT3 and c-Jun or in the amounts of undefined costimulatory factors.
Cis-regulatory Elements of the ␣ 2 -MG Promoter Mediate Transcriptional Activation by IL-6 or OSM-Activation of the ␣ 2 -MG promoter by IL-6 is principally dependent on DNA sequence within 209 bp upstream of the transcription start site (40,45). This region contains a sequence motif (5Ј-TTCTGG-GAA-3Ј; Ϫ171 to Ϫ163) that is conserved among various APR genes (40,51). At least five regulatory elements in the minimal ␣ 2 -MG promoter are essential for induction of transcription by IL-6 or OSM, including the following: 1) Two AP1 sites and an APRE site. Our observations suggest that an array of proteins bound at AP1 and APRE sites is essential for full activation of the ␣ 2 -MG promoter. When this array was compromised by mutation, synergy between c-Jun and STAT3 was eliminated and transcriptional activity was substantially impaired. 2) Two AT-rich sites. Intrinsic DNA bends at AT tracts of four residues or greater occur at diverse regulatory elements (reviewed in Ref. 47). Two or three appropriately spaced AT tracts can serve as high affinity binding sites for proteins that bend DNA, such as HMG-I(Y) (34,44). In the ␤-interferon promoter, correct helical phasing of AT-rich, HMG-I binding sites is critical for stabilization of a higher order complex of transcriptional activators by protein-DNA and protein-protein interactions (32,34). In the ␣ 2 -MG promoter, mutation of both AT-rich elements or disruption of correct phasing between these elements impaired induction of transcription by IL-6 and OSM and abolished synergy between STAT3 or STAT3-C and c-Jun. Moreover, mutation of the AT-rich elements abolished in vitro formation of AP1-dependent DNA-protein complexes at the ␣ 2 -MG promoter.
Intrinsic bending of DNA by these AT-rich sequences may impart a particular spatial relationship between critical DNA binding sites (e.g. the AP1 sites and the APRE) that is essential for transcriptional synergy. It is possible that repositioning of an AT-rich site interferes directly with a productive interaction between STAT3 and c-Jun at the ␣ 2 -MG promoter, perhaps through steric hindrance by a protein bound to the AT-rich site.
Alternatively or in addition, proteins bound to the AT-rich sites in the ␣ 2 -MG promoter may make contacts that are sensitive to their phasing on the DNA helix. An antibody specific for HMG-I(Y) failed to react with the protein complex bound to AT-rich regions of a 2 -MG promoter (data not shown). It remains possible that a member of the HMG protein family other than HMG-I(Y) is bound to these AT-rich sites .
Formation of a Multicomponent Transcription Complex at the ␣ 2 -MG Promoter-Transcriptional activation at natural enhancers is exquisitely sensitive to the precise arrangement of DNA elements and proteins within them. For example, upon viral infection, three transcription factors (an ATF2/c-Jun heterodimer, an IRF family member, and the p50/p65 NF-B heterodimer) bind to the four distinct regions of the ␤-interferon promoter. This array of factors, together with HMG-I(Y) bound at three AT-rich sequences, synergistically activates transcription of the ␤-interferon gene. Such multicomponent complexes have been termed enhanceosomes (reviewed in Ref. 52). In response to IL-6-related cytokines, the ␣ 2 -MG promoter is also activated through synergistic interactions of discrete transcrip- tion factors with distinct DNA binding sites; these features and the requirement for properly phased AT-rich regions are reminiscent of previously characterized enhanceosomes.
A physical interaction between STAT3 and c-Jun has been demonstrated in vitro, and evidence that this interaction functions in transcriptional activation has been presented (38). Several mutations in STAT3 were found to disrupt binding to c-Jun in vitro, and these mutations impaired synergy between STAT3 and c-Jun at the ␣ 2 -MG promoter in HepG2 cells (38). In work presented here we introduced identical mutations into STAT3-C, which forms dimers spontaneously and exhibits stronger transcriptional activity than STAT3. To our surprise, similar transcriptional synergy was observed between c-Jun and mutant or wild-type forms of STAT3-C, despite the ability of these mutations to abolish synergy between c-Jun and STAT3. STAT3-C was created by introduction of two cysteine substitutions within the carboxyl-terminal loop of the STAT3 SH2 domain, permitting dimer formation in the absence of tyrosine phosphorylation (42). The effects of these cysteine substitutions on STAT3 structure are unknown. It is possible that regions of STAT3-C other than those previously identified in STAT3 contact c-Jun and establish synergy at the ␣ 2 -MG promoter or that the interactions responsible for synergy between STAT3-C and c-Jun in our experiments may be indirect. The synergistic interactions between c-Jun and STAT3 or STAT3-C depend, however, on a similar set of cis-acting elements, suggesting that these interactions share a common structural basis. The resistance of synergy between STAT3-C and c-Jun to the L148A, V151A, and T346A,K348A,R350A mutations may reflect a structural alteration in STAT3-C that also contributes to its oncogenicity.
In summary, our experiments suggest that multiple DNA elements and their interactions with specific proteins in the ␣ 2 -MG promoter provide a framework for enhanceosome formation and synergistic transcriptional activation in response to cytokine stimulation. It will be important to identify additional proteins that contribute to cytokine-induced activation of this promoter and the elements with which they interact. More challenging will be to define the spatial organization of this higher order complex and its assembly in response to cytokine stimulation.