A Novel c-Jun-dependent Signal Transduction Pathway Necessary for the Transcriptional Activation of Interferon γ Response Genes*

The biological effects of interferon γ (IFNγ) are mediated by interferon-stimulated genes (ISGs), many of which are activated downstream of Janus kinase (JAK)/signal transducer and activator of transcription 1 (STAT1) signaling. Herein we have shown that IFNγ rapidly activated AP-1 DNA binding that required c-Jun but was independent of JAK1 and STAT1. IFNγ-induced c-Jun phosphorylation and AP-1 DNA binding required the MEK1/2 and ERK1/2 signaling pathways, whereas the JNK1/2 and p38 mitogen-activated protein kinase pathways were dispensable. The induction of several ISGs, including ifi-205 and iNOS, was impaired in IFNγ-treated c-Jun–/– cells, but others, such as IP-10 and SOCS3, were unaffected, and chromatin immunoprecipitation demonstrated that c-Jun binds to the iNOS promoter following treatment with IFNγ. Thus, IFNγ induced JAK1- and STAT1-independent activation of the ERK mitogen-activated protein kinase pathway, phosphorylation of c-Jun, and activation of AP-1 DNA binding, which are important for the induction of a subset of ISGs. This represents a novel signal transduction pathway induced by IFNγ that proceeds in parallel with conventional JAK/STAT signaling to activate ISGs.

and cell cycle progression through the induction of IFN-stimulated genes (ISGs) (1). Typically, IFN␥ signaling proceeds through the ligation of surface receptors, sequential phosphorylation of JAKs and STAT1 on tyrosine 701 (Tyr-701), and translocation of STAT1 homodimers to the nucleus to bind IFN␥-activated sequences (GASs) found in the promoter regions of many ISGs (1,2). However, other signaling pathways are important for IFN␥ signaling. For example STAT1 requires phosphorylation on Ser-727 to achieve full transcriptional activity (3,4). Additionally, microarray analysis revealed that approximately one-third of ISGs were still regulated by IFN␥ in the absence of functional STAT1 (5)(6)(7), and STAT1 Ϫ/Ϫ mice were more resistant to virus infection than mice lacking expression of the IFN␥ and IFN␣/␤ receptors (5,8). Thus, IFN␥ can induce expression of ISGs by STAT1-independent mechanisms, and the activity of multiple intracellular pathways acting in parallel play an important role in the biological response to IFN␥.
Aside from the JAKs, other kinases are activated in response to IFN␥, including MAP kinases (9 -11), phosphatidylinositol 3-kinase and AKT (12), Pyk2 (11), calcium/calmodulindependent protein kinase II (13), and protein kinase C isoforms (14,15). However, these enzymes have predominantly been assessed for their ability to trigger phosphorylation of STAT1 on Ser-727, and little is known about their potential impact upon STAT-independent gene transcription. IFN␥ can activate transcription factors other than STAT1, including, class II trans-activator (CIITA) (16 -18), CCAAT enhancer-binding protein (CEBP)-␤ (19 -21), and interferon-responsive factors (IRFs) (22,23), but the induction of these factors occurs downstream of JAK-STAT1 signaling. There are conflicting reports regarding the involvement of the IKK/IB/NF-B pathway in STAT1-independent IFN␥ signaling (24 -26). Recently, it was shown that activation of IKK␣ and IKK␤ in the absence of NF-B activity was important for the induction of a subset of ISGs following treatment with IFN␥ (26). However the signaling proteins and transcription factors downstream of IKK activation by IFN␥ were not identified, and although STAT1 was phosphorylated on Tyr-701 and Ser-727 in IFN␥-treated IKK␣/ ␤-deficient cells, it was not determined whether this pathway activated the expression of ISGs in the absence of STAT1.
We previously found that IFN␥ could stimulate AP-1 DNA binding activity (27) and herein investigated the molecular events leading to the activation of AP-1 following treatment with IFN␥. IFN␥ induced rapid phosphorylation of c-Jun and concomitant activation of AP-1 DNA binding activity that was independent of JAK1 and STAT1 but dependent on c-Jun expression. Activation of AP-1 did not require p38 MAP kinases or c-Jun N-terminal kinase (JNK)1/2 but did require the MEK1/2-ERK1/2 pathway. We identified ISGs (ifi205, iNOS) that were not activated by IFN␥ in c-Jun Ϫ/Ϫ cells, suggesting that AP-1 DNA binding activity was an essential regulator of these genes. IFN␥-induced expression of ifi205 (but not iNOS) was suppressed in STAT1 Ϫ/Ϫ cells; additionally, IFN␥-induced iNOS transcription was observed in JAK1-deficient MEFs. In contrast, the IFN␥-mediated induction of IP-10 and SOCS3 was suppressed in STAT1 Ϫ/Ϫ cells but was unaffected by perturbation of signaling through MEK1/2, ERK1/2, and c-Jun. Chromatin immunoprecipitation (ChIP) assays revealed that IFN␥ induced rapid binding of c-Jun to the iNOS promoter. Our data show for the first time that, in addition to activation of JAK/ STAT1 signaling, IFN␥ can mediate phosphorylation of c-Jun and activate AP-1 DNA binding through the ERK MAP kinase pathway independently of STAT1.
Western Blot and Electrophoretic Mobility Shift Assay (EMSA)-Western blot using whole cell lysates and EMSAs with nuclear lysates were performed as previously described (27). The sequence of the GAS oligonucleotide from the Fc␥RI promoter region was 5Ј-TAGGGATTTACGGGAAATTGAT-GAAGCTGATC-3Ј and c-Cis-inducible element from IRF1 5Ј-GCCTGATTTCCCCGAAATGATGA-3Ј, whereas the wt and mutant AP-1 oligonucleotides were described previously (27). For supershift experiments, 2 g of antibody was added to the binding reaction for 1.5 h prior to incubation with probe.
Gels were dried and visualized by autoradiography on x-ray film (Kodak).
Chromatin Immunoprecipitation-ChIP assays were performed essentially as described previously (31) using 5 g of anti-c-Jun antibody or rabbit IgG control antibody (Santa Cruz Biotechnology). The abundance of specific sequences in ChIP EMSAs were performed using radiolabeled oligonucleotides containing AP-1 consensus sequence (A), a GAS consensus sequence (B), and nuclear extracts from wt MEFs treated with 100 IU/ml IFN␥ for the indicated times. C, competition EMSAs were performed using excess unlabeled wild type (wt) oligonucleotides and mutant oligonucleotides containing mutations critical for AP-1 binding and nuclear lysates from wt MEFs treated with 100 IU/ml IFN␥ for 30 min. AP-1 DNA binding activity was assessed by EMSA using nuclear extracts from MEFs treated with 1-1000 IU/ml IFN␥ for 30 min (D) and early passage primary MEFs treated with 100 IU/ml IFN␥ for 15-30 min (E). All data presented are representative of at least three independent experiments. samples was quantitated using the SYBR Green dye detection method (Applied Biosystems, Warrington, UK). Primers used for PCR reactions were: iNOS (forward, 5Ј CCC AGC CCA ATT ACT TGA TTT 3Ј; reverse, 5Ј CGT GTT TTG CCC TTG TCT GAG 3Ј) flanking the region (Ϫ1150 to Ϫ1046) in the iNOS promoter, SOCS3 (forward, 5Ј GCT GAA TGG TCC TAC GTC CCT T 3Ј; reverse, 5Ј TAC AGT TCC AAG CAT CCC GTG 3Ј) flanking the region (Ϫ565 to Ϫ539) in the SOCS 3 promoter were designed using Primer Express 2 software. Threshold cycle numbers (Ct) were measured in the exponential phase for all samples. Ct values were converted to relative values using the equation (10 ϫ 10 10 /(2 Ct ) ϭ C⌬t) for all ChIP samples. Final relative values for enrichment were calculated as percentage of input.
Ribonuclease Protection Assay-MEFs were stimulated as indicated, and RNA was isolated with TRIzol (Invitrogen) according to the manufacturer's protocols. 5 g of RNA was hybridized to the Jun/Fos ribonuclease protection assay probe set (BD Biosciences, San Diego, CA) and ribonuclease protection assays performed using a ribonuclease protection assay kit according to the manufacturer's protocols (BD Biosciences). Results were visualized by autoradiography on x-ray film and quantitated on a phosphorimaging device using FX software (Bio-Rad). Statistical significance was tested using one-way analysis of variance testing with OriginLab version 7.5 software (Northampton, MA).

AP-1 Is Rapidly and Transiently Activated by IFN␥-Using
human tumor cell lines, we previously demonstrated that IFN␥ was capable of activating protein complexes that bound AP-1 FIGURE 2. Activation of AP-1 by IFN␥ in MEFs is independent of JAK1 and STAT1. EMSAs were performed using radiolabeled oligonucleotides containing an AP-1 consensus sequence (A) or a GAS consensus sequence (B) and nuclear extracts from wild type and STAT1 Ϫ/Ϫ MEFs treated with 100 IU/ml IFN␥ for the indicated times. C, EMSAs were performed using radiolabeled oligonucleotides containing an AP-1 consensus sequence and nuclear extracts from wild type and JAK1 Ϫ/Ϫ MEFs.

FIGURE 3. IFN␥-induced AP-1 binding complexes depend on expression of c-Jun.
A, nuclear lysates from wt MEFs treated with 100 IU/ml IFN␥ for 30 min were incubated with specific antibodies raised against c-Jun, JunB, JunD, the Fos family, or ATF-2 prior to performing EMSAs using radiolabeled oligonucleotides containing an AP-1 consensus sequences. B, EMSAs were performed using radiolabeled oligonucleotides containing an AP-1 consensus sequence and nuclear extracts from wt MEFs (1st-3rd lanes) and c-Jun Ϫ/Ϫ MEFs (4th-6th lanes) treated with 100 IU/ml IFN␥ for the indicated time. C, wt MEFs were treated with 1000 IU/ml IFN␥ for 15 and 30 min, and whole cell extracts were used for Western blotting with an antibody against c-Jun phosphorylated on Ser-63 (top panel). The blot was stripped and reprobed with antibodies against total c-Jun (bottom panel). EMSAs were performed using radiolabeled oligonucleotides containing an AP-1 consensus sequence and nuclear extracts from wt MEFs (A) and JNK1/2 Ϫ/Ϫ MEFs treated with 100 IU/ml IFN␥ for the indicated times. B, wt MEFs pretreated for 30 min with a cell-permeable control peptide or a cell-permeable JNK inhibitory peptide followed by the addition of 100 IU/ml IFN␥ for the indicated times.

c-Jun-dependent Activation of IFN␥ Response Genes
sites (27). We confirmed this in immortalized MEFs by EMSA using an oligonucleotide probe containing a consensus AP-1 site. After treatment of cells with IFN␥, we observed a rapid and transient increase in AP-1 DNA binding activity that returned to basal levels by ϳ120 min (Fig. 1A). The rapid induction of AP-1 DNA binding activity after IFN␥ treatment was concomitant with the induction of GAS DNA binding activity (Fig. 1B). The IFN␥-induced AP-1 binding activity was specific, as it was competed by excess unlabeled oligonucleotides containing wild type (wt) (but not mutant) AP-1 binding sites (Fig. 1C). AP-1 activation by IFN␥ was dose-dependent and could be induced by doses as low as 1 IU/ml (Fig. 1D). This effect is not confined to immortalized or transformed cells, as activation of AP-1 DNA binding activity by IFN␥ was also seen in primary (early passage) MEFs (Fig. 1E).
IFN␥ Activation of AP-1 Is Independent of JAK1 and STAT1-IFN␥-mediated activation of AP-1 DNA binding was independent of either JAK1 or STAT1 expression, because it could still be detected in JAK1 Ϫ/Ϫ or STAT1 Ϫ/Ϫ MEFs (Fig. 2). Interestingly, we observed that both basal and IFN␥-induced AP-1 DNA binding activities were enhanced in the STAT1-and JAK1-deficient cells. To determine whether other STAT complexes could compensate for the loss of STAT1 in our system, we studied GAS DNA binding activity in wt and STAT1 Ϫ/Ϫ cells using the GAS element from the Fc␥RI gene (Fig. 2B). GAS binding species were detected in nuclear extracts from wt MEFs treated with IFN␥ for 15-30 min but not in extracts from STAT1 Ϫ/Ϫ cells, indicating they were dependent on the presence of STAT1. Similar data were obtained using the GAS element from the IRF-1 promoter (supplemental Fig. S1A) and the SIE from the c-fos promoter (data not shown). Furthermore, we detected no increase in expression or phosphorylation of STAT3 in IFN␥-treated STAT1 Ϫ/Ϫ cells (supplemental Fig.  S1B). We therefore propose that we have identified an IFN␥induced, JAK1-and STAT1-independent signaling pathway leading to activation of AP-1 DNA binding.
c-Jun Is a Critical Component of the IFN␥-induced AP-1 Complex-AP-1 transcription factors are composed of Jun family (c-Jun, JunB, and JunD) homodimers, or Jun/Fos (c-Fos, FosB, Fra1, and Fra2) or Jun/ATF2 heterodimers (32). Ribonuclease protection assays showed that the Jun family genes and Fra1 and Fra2 were equivalently expressed in resting MEFs; however, almost no FosB or c-Fos mRNA expression was detected (supplemental Fig. S2, A and B). Moreover, none of the AP-1 subunit genes were induced by IFN␥ (supplemental Fig. S2, C-E). Supershift assays with anti-  JANUARY 12, 2007 • VOLUME 282 • NUMBER 2

JOURNAL OF BIOLOGICAL CHEMISTRY 941
bodies specific for c-Jun, JunB, or JunD impaired the formation or migration of the IFN␥-induced AP-1/oligonucleotide complex, whereas antibodies against Fos family proteins or ATF2 had no effect (Fig. 3A). Expression of c-Jun was critical for the formation of the IFN␥-activated AP-1 DNA binding complex, as no AP-1 DNA binding activity was detected in c-Jun Ϫ/Ϫ MEFs treated with IFN␥ (Fig. 3B). GAS binding activity was still evident in IFN␥-stimulated c-Jun Ϫ/Ϫ cells (data not shown), indicating that the JAK/STAT pathway remained intact. Optimal DNA binding by c-Jun requires phosphorylation of serine 63 (Ser-63) and serine 73 (Ser-73) (33). A minimal basal level of Ser-63-phosphorylated c-Jun was detected in untreated wt MEFs, which was robustly enhanced following exposure of cells to IFN␥ (Fig. 3C). In contrast, we observed no consistent change in total c-Jun expression in response to IFN␥. Taken together, these data demonstrate that treatment of cells with IFN␥ results in rapid phosphorylation of c-Jun and formation of an AP-1 DNA binding complex that contains c-Jun as an essential component.
The Activation of AP-1 by IFN␥ Is Independent of JNK and p38 MAP Kinase but Requires ERK-JNK is the primary kinase responsible for the stimulated phosphorylation of c-Jun at Ser-63 and -73 (34). IFN␥-induced AP-1 DNA binding activity was detected in JNK1/2 double knock-out MEFs with similar kinetics and magnitude of induction to that seen in wt MEFs (Fig. 4A). Treatment with a specific peptide inhibitor of JNK (35) did not affect IFN␥-induced activation of AP-1 (Fig. 4B); however, this inhibitor effectively suppressed JNK-dependent activation of AP-1 following stimulation with tumor necrosis factor ␣ (supplemental Fig. S3, A  and B). In contrast to tumor necrosis factor ␣, IFN␥ did not stimulate the phosphorylation of JNK1 and JNK2 (supplemental Fig. S3C), providing further supporting evidence that these kinases do not play a role in the IFN␥-induced activation of AP-1. p38 MAPK plays an important role in type I IFN signaling (9,36); however, its role in IFN␥ signaling is unclear (37,38). We saw no phosphorylation of p38 MAPK following treatment with IFN␥, and SB203580 (a pharmacological inhibitor of p38 MAPK) had no effect on IFN␥-induced activation of AP-1 DNA binding (supplemental Fig. S4).
A previous study identified ERK1/2 as the kinases responsible for c-Jun phosphorylation in response to phorbol-ester or epidermal growth factor stimulation (39). Therefore, we tested whether the MEK/ERK pathway could mediate IFN␥-induced phosphorylation of c-Jun and activation of AP-1 DNA binding activity using the MEK1/2 inhibitor PD98059. PD98059 almost completely ablated the activation of AP-1 DNA binding activity following IFN␥ treatment (Fig. 5A), suggesting that the MEK1/2-ERK1/2 pathway lies upstream of AP-1 DNA binding. This was supported by Western blot showing that ERK1/2 phosphorylation was detected within 5 min of IFN␥ treatment (Fig. 5B) that was completely blocked by PD98059 (Fig. 5C). Moreover, PD98059 completely suppressed IFN␥-induced phosphorylation of c-Jun (Fig. 5D). These and supporting data obtained using the chemically distinct MEK1/2 inhibitor UO126 (supplemental Fig. S5) indicate that IFN␥ can activate the MEK1/2-ERK1/2 signaling pathway resulting in c-Jun phosphorylation and induction of AP-1 DNA binding.
AP-1 Is Critical for IFN␥-mediated Induction of ifi205 and iNOS-To delineate the functional significance of the novel STAT-independent, IFN␥-stimulated MEK1/2-ERK1/2-AP-1 pathway that we had identified, we sought to identify ISGs that were regulated through this pathway. We selected candidate genes on the basis of their responses to IFN␥ treatment in the presence or absence of STAT1 or AP-1 (3, 6, 7, 40) and performed quantitative reverse transcription-PCR to determine changes in expression in response to IFN␥. We identified Ifi205 and iNOS as genes significantly activated in IFN␥-treated wt cells but not in IFN␥-treated c-Jun Ϫ/Ϫ cells, suggesting that c-Jun is essential for their induction by IFN␥ (Fig. 6, A and B). In contrast, induction of SOCS3 and IP-10 by IFN␥ was equivalent in wt and c-Jun Ϫ/Ϫ cells (Fig. 6, C and  D). These experiments identify ifi205 and iNOS (but not

c-Jun-dependent Activation of IFN␥ Response Genes
SOCS3 or IP-10) as genes requiring AP-1 activity for induction following treatment with IFN␥.
JAK1 and STAT1 Are Dispensable for IFN␥-mediated Induction of iNOS-We next determined whether the IFN␥-stimulated, c-Jun-dependent induction of ifi205 and iNOS occurred in the absence of the classical JAK1/STAT1 pathway. iNOS was maximally induced in wt MEFs following 6 h of exposure to IFN␥ (Fig. 7A), and a corresponding increase in iNOS protein was observed by Western blot (data not shown). In STAT1 Ϫ/Ϫ (Fig. 7A) or JAK1 Ϫ/Ϫ (Fig. 7E) MEFs, IFN␥ induced iNOS mRNA to the same levels seen in wt cells; however, maximal induction was observed following 1 h of exposure to IFN␥ and decreased slightly by 6 h, indicating that both JAK1 and STAT1 are dispensable for the induction of iNOS by IFN␥. In contrast, ifi205 was significantly induced in wt cells exposed to IFN␥ for 1 or 6 h but not in STAT1 Ϫ/Ϫ cells (Fig. 7B). Thus, unlike iNOS, IFN␥-mediated induction of ifi205 requires both AP-1 and STAT1. The induction of SOCS3 and IP-10 seen in IFN␥treated wt cells was significantly attenuated in STAT1 Ϫ/Ϫ cells (Fig.  7, C and D).
IFN␥ Induces Binding of c-Jun to the iNOS Promoter in Vivo-Our experiments indicated that the MEK/ERK pathway and c-Jun were critical for the expression for ISGs such as iNOS, and the kinetics of the response were consistent with this being a direct transcriptional response downstream of c-Jun activation. We therefore investigated whether IFN␥ activated the binding of c-Jun to the iNOS promoter using chromatin immunoprecipitation. The 1.7-kb regulatory region of the murine iNOS promoter contains two consensus AP-1 binding sites (40,41), and we therefore chose PCR primers flanking these sequences. We also studied the IFN␥-responsive region of the SOCS3 promoter as an example of a c-Jun-independent gene promoter. Chromatin immunoprecipitation assays were performed using wild type and c-Jundeficient MEFs after a 30-min exposure to IFN␥. We observed a marked increase in c-Jun binding to the iNOS promoter following treatment of cells with IFN␥ (Fig. 8A,  lanes 3 and 4). No binding was detected using a control antibody in the immunoprecipitation reaction (Fig. 8A, lane 2) or in c-Jun-deficient lysates immunoprecipitated with the anti-c-Jun antibody (Fig.  8A, lanes 5 and 6), demonstrating specificity of the reaction. When quantified by quantitative real-time PCR, there was an 18-fold enrichment of c-Jun bound to the iNOS promoter following IFN␥ treatment (Fig. 8C). Consistent with the data shown in Figs. 6 and 7, no increase in binding of c-Jun to the SOCS3 promoter was observed following treatment with IFN␥ (Fig. 8B). These experiments demonstrate that c-Jun binds to the promoter of the iNOS gene following IFN␥ treatment.

DISCUSSION
Induction of ISGs through JAK/STAT1 plays an important role in mediating the biological effects of IFN␥ (1). However, other signaling pathways functioning in parallel with JAK/ or STAT1 Ϫ/Ϫ (black bars) MEFs were treated with 100 IU/ml IFN␥ for either 1 or 6 h. RNA was extracted, cDNA synthesized, and quantitative real-time PCR performed with primers specific for iNOS (A), ifi205 (B), IP-10 (C), or SOCS3 (D). E, wild type (white bars) or JAK1 Ϫ/Ϫ (black bars) MEFs were treated with 100 IU/ml IFN␥ for either 1 or 6 h. RNA was extracted, cDNA synthesized, and quantitative real-time PCR performed with primers specific for iNOS. Fold induction represents the mean fold change in transcription (untreated (Unt) cells were arbitrarily called 1) of three independent experiments, and error bars are the S.D. from the mean in these experiments. Statistical significance of fold gene induction relative to untreated samples was determined using one-way analysis of variance testing. *, p Ͻ 0.05; **, p Ͻ 0.01.

c-Jun-dependent Activation of IFN␥ Response Genes
STAT1 are required for a complete IFN␥ response (5-7), as IFN␥ can still induce ISGs, mediate anti-viral responses, and regulate cell proliferation in the absence of STAT1 (5,6,8). We have demonstrated that, in addition to activating STAT1, IFN␥ stimulates the phosphorylation and activation of ERK1/2 through MEK1/2, resulting in the downstream phosphorylation of c-Jun, activation of AP-1 DNA binding, and the induction of genes such as iNOS and ifi205 (Fig. 9). Importantly, we used chromatin immunoprecipitation assays to demonstrate the binding of c-Jun to the iNOS promoter following treatment with IFN␥. This signaling pathway was engaged by IFN␥ in a range of human and mouse tumor cell lines, immortalized and primary MEFs, and in JAK1 Ϫ/Ϫ or STAT1 Ϫ/Ϫ MEFs. This indicated that, although IFN␥-mediated activation of the pathway was independent of JAK1 and STAT1, it was also fully active in JAK1/STAT1-expressing cells and thus did not merely serve as a compensatory mechanism in the absence of either JAK1 or STAT1. Moreover, the stimulation of AP-1 DNA binding activity following IFN␥ treatment occurred rapidly and independently of the transcriptional induction of any AP-1 subunit genes, which is consistent with this pathway being a primary response pathway. Although the IFN␥-stimulated activation of AP-1 is independent of the classical JAK1-STAT1 pathway (and possibly all STAT proteins), we consistently observed that the loss of these proteins augmented basal levels of AP-1 DNA binding. It is possible that this is due to the loss of a negative regulation of IFN␥ signaling controlled by genes in a JAK1-STAT1-dependent manner; however, this needs to be determined experimentally.
We have identified three classes of ISGs that differentially utilize the JAK/STAT1 and MEK/ERK/c-Jun signaling pathways in response to IFN␥. Induction of genes such as IP-10 and SOCS3 by IFN␥ was significantly attenuated in STAT1 Ϫ/Ϫ cells but was unaffected in c-Jun Ϫ/Ϫ cells, demonstrating the importance of STAT1 for the activation of these ISGs. We recently found that ifi204 had similar transcriptional requirements (data not shown). In contrast, iNOS was induced by IFN␥ in wt, JAK1 Ϫ/Ϫ , and STAT1 Ϫ/Ϫ cells but not in c-Jun Ϫ/Ϫ cells. Previously, it has been demonstrated that STAT1, NF-B, IRF-1, and IRF-8 can play important roles in the induction of iNOS by the combined treatment of IFN␥ and lipopolysaccharide in macrophages (42). However, the utilization of different transcription factors in this response can be cell type-dependent, as IRF-1 Ϫ/Ϫ pancreatic islet cells, chondrocytes, and hepatocytes (but not macrophages) showed normal iNOS induction in response to FIGURE 8. Treatment of cells with IFN␥ stimulates binding of c-Jun onto the iNOS but not the SOCS3 promoter. Wild type or c-Jun Ϫ/Ϫ MEFs were treated with 100 IU/ml IFN␥ for 30 min, and ChIP assays were performed with either control rabbit immunoglobulin (␣rIg) or a rabbit polyclonal antibody against c-Jun (␣c-Jun). Immunoprecipitated chromatin was used as a template to detect c-Jun bound to the iNOS (A) and the SOCS3 promoter (B). C, quantitative real-time PCR was used to quantitate the enrichment of c-Jun on the iNOS promoter in untreated cells (white bars) or cells treated with IFN␥ for 30 min (black bars). Statistical significance of enrichment of c-Jun on the iNOS promoter relative to untreated samples over three independent experiments was determined using one-way analysis of variance testing. *, p Ͻ 0.05. Binding of IFN␥ to its cognate receptor stimulates phosphorylation and activation of Jak1 and Jak2 resulting in phosphorylation of STAT1 at Tyr-701. Ligation of IFN receptors also results in the phosphorylation and activation of ERK1/2 through MEK1/2. Activated ERK1/2 can phosphorylate STAT1 on Ser-727, although other serine kinases can also perform this function. Fully active STAT1 can translocate to the nucleus and induce expression of ISGs such as IP-10 and SOCS3. Activated ERK1/2 is necessary for the phosphorylation and activation of c-Jun, resulting in the formation of an active AP-1 DNA binding complex that can bind to the promoters of ISGs, such as iNOS, and induce gene expression. IFN␥-activated AP-1 and STAT1 can also functionally cooperate to induce other ISGs, such as ifi205.
lipopolysaccharide/IFN␥ (43,44). Our data confirm observations by others (45,46) that iNOS expression is enhanced following treatment with IFN␥ alone, although we found induction of iNOS in MEFs occurred within 6 h of IFN␥ treatment, whereas studies in macrophages did not detect iNOS expression until 24 h post-IFN␥ treatment (45,46). A third type of response was represented by the induction of ifi205 by IFN␥ that required both STAT1 and c-Jun (as did MIP-1␤, data not shown). It is possible that STAT1 and AP-1 function cooperatively on the ifi205 and MIP-1␤ promoters to transactivate these genes; indeed functional interactions between STATs and other transcription factors are known to occur. For example STAT3 and c-Jun cooperate to induce expression of ␣ (2) macroglobulin (47). Moreover, although IFN␥ induced expression of C/EBP␤ in STAT 1 Ϫ/Ϫ MEFs (6), STAT1 was required for transactivation of C/EBP␤ target genes but not DNA binding (48). Although our data indicate that at least three classes of ISGs exist, genome-wide expression profiling will be required to confirm this and may identify other groups of ISGs, such as those repressed by STAT1, AP-1, or both factors following IFN␥ treatment.
Our data show that IFN␥-induced activation of AP-1 DNA binding was independent of both JAK1 and STAT1. Others have reported that hyperactivation of STAT3 by IFN␥ can compensate for the loss of STAT1 and mediate transcription of ISGs (49). Although our data does not entirely exclude this possibility, it does present strong evidence that this is not the case in our experimental system. First, AP-1 activity and corresponding iNOS expression was not only detected in IFN␥-stimulated STAT1 Ϫ/Ϫ cells but also JAK1 Ϫ/Ϫ cells in which IFN␥ is incapable of activating STAT3 or GAS binding (49,50). Second, STAT1 Ϫ/Ϫ MEFs consistently lacked all of the IFN␥-induced GAS binding complexes observed in wild type cells across a range of oligonucleotide probes derived from promoters of different ISGs. Third, the induction of IP-10, SOCS3, and ifi204 was attenuated in STAT1 Ϫ/Ϫ cells, suggesting that, at least for these genes, the absence of STAT1 could not be compensated. Finally, in contrast to a previous study (49), we saw no increase in protein levels or the magnitude and duration of phosphorylation of STAT3 on Tyr-705 in STAT1 Ϫ/Ϫ cells.
We are yet to identify the molecular events that occur upstream of MEK1/2 following stimulation with IFN␥. Our data suggest that JAK 1 is dispensable; however, it does not rule out the possibility of compensation from other JAK kinases (especially JAK 2), which is a focus of continuing studies. IFN␥ can also stimulate a large number of signal transduction proteins including Src family kinases c-Src (49) and Fyn (51), CrkL and its downstream small G protein Rap1 (52), phosphatidylinositol 3-kinase and its effector kinase AKT (12), protein kinase C ␦ (15), and Pyk2 (11). The role of these molecules in IFN␥ signaling has mainly been studied in the context of Ser-727 phosphorylation on STAT1; however, all of these proteins are upstream of ERK1/2 in other systems using various cytokines to stimulate signal transduction and thus are also potential activators of the IFN␥-induced MEK/ERK/AP-1 pathway described herein.
It is becoming increasingly clear that a comprehensive response to IFN␥ involves multiple signaling pathways that cooperate to stimulate transcription of ISGs (53). The IFN␥induced MEK/ERK/c-Jun signaling pathway that we have deciphered provides important additional information regarding the molecular events that may underpin IFN␥ biology. Although the JAK/STAT1 pathway plays a fundamental role in the biological response to IFN␥, it is evident that proteins, such as MAP kinases that can phosphorylate STAT1 on Ser-727, are also required for the full activation of STAT1 (7,53). ERK1/2 and other IFN␥-induced serine/threonine kinases can phosphorylate substrates other than STAT1 (7,53), such as c-Jun (this study), and these may prove to play an important and, so far, undescribed role in IFN␥-induced gene expression and biology.