Analysis of SOCS-3 promoter responses to interferon gamma.

SOCS-3 (suppressor of cytokine signaling 3) is an intracellular protein that is selectively and rapidly induced by appropriate agonists and that modulates responses of immune cells to cytokines by interfering with the Janus kinase/signal transducer and activator of transcription (Jak/STAT) pathway. On the basis of the observations that interferon gamma (IFNgamma) up-regulates SOCS-3 gene and protein expression in primary mouse macrophages, J774 macrophage cell line and embryonal fibroblasts, we investigated which sequences of the 5' SOCS-3 gene are responsive to IFNgamma. By promoter deletion analysis we identified a functional IFNgamma-responsive element, located at nucleotides -72/-64 upstream from the transcription initiation, whose presence and integrity is necessary to ensure responsiveness to IFNgamma. This element contains a STAT consensus binding sequence (SOCS-3/STAT-binding element (SBE)) whose specific mutation totally abolished the responsiveness to IFNgamma. In contrast, discrete deletion of other 5' regions of the SOCS-3 promoter did not substantially modify the inducibility by IFNgamma. Electromobility shift assay analyses revealed that IFNgamma promotes specific DNA binding activities to an oligonucleotide probe containing the SOCS-3/SBE sequence. Even though IFNgamma triggered tyrosine phosphorylation of both STAT1 and STAT3 in macrophages and J774 cells, only STAT1 was appropriately activated and thus found to specifically bind to the SOCS-3/SBE oligonucleotide probe. Accordingly, IFNgamma-induced SOCS-3 protein expression was not impaired in STAT3-deficient embryonal fibroblasts. Taken together, these results demonstrate that the induction of SOCS-3 by IFNgamma depends upon the presence of a STAT-binding element in the SOCS-3 promoter that is specifically activated by STAT1.


SOCS-3 (suppressor of cytokine signaling 3) is an intracellular protein that is selectively and rapidly induced by appropriate agonists and that modulates responses of immune cells to cytokines by interfering with the Janus kinase/signal transducer and activator of transcription (Jak/STAT) pathway. On the basis of the observations that interferon ␥ (IFN␥) up-regulates SOCS-3 gene and protein
expression in primary mouse macrophages, J774 macrophage cell line and embryonal fibroblasts, we investigated which sequences of the 5 SOCS-3 gene are responsive to IFN␥. By promoter deletion analysis we identified a functional IFN␥-responsive element, located at nucleotides ؊72/؊64 upstream from the transcription initiation, whose presence and integrity is necessary to ensure responsiveness to IFN␥. This element contains a STAT consensus binding sequence (SOCS-3/STAT-binding element (SBE)) whose specific mutation totally abolished the responsiveness to IFN␥. In contrast, discrete deletion of other 5 regions of the SOCS-3 promoter did not substantially modify the inducibility by IFN␥. Electromobility shift assay analyses revealed that IFN␥ promotes specific DNA binding activities to an oligonucleotide probe containing the SOCS-3/SBE sequence. Even though IFN␥ triggered tyrosine phosphorylation of both STAT1 and STAT3 in macrophages and J774 cells, only STAT1 was appropriately activated and thus found to specifically bind to the SOCS-3/SBE oligonucleotide probe. Accordingly, IFN␥-induced SOCS-3 protein expression was not impaired in STAT3-deficient embryonal fibroblasts. Taken together, these results demonstrate that the induction of SOCS-3 by IFN␥ depends upon the presence of a STAT-binding element in the SOCS-3 promoter that is specifically activated by STAT1.
Interferon ␥ (IFN␥) 1 is a pluripotent cytokine involved in the regulation of nearly all the different phases of both innate and adaptive immune responses. Produced by activated T and natural killer cells, IFN␥ has a crucial role in several processes, including host defense against viruses and microorganisms, cell proliferation, phagocyte activation, control of apoptosis, promotion of antigen processing and presentation, and T helper type 1 (T H 1) differentiation (1). IFN␥ has a number of activating properties on cells of the immune system but can also exert important immunosuppressive actions by modulating cellular responses to different cytokines and inflammatory stimuli (2). For example, under specific conditions, IFN␥ may inhibit proinflammatory cytokine release by activated human peripheral blood mononuclear cells and polymorphonuclear neutrophils (3,4) and up-regulates the release of cytokine antagonists such as IL-1Ra, type II IL-1 receptors, and IL-18BP from mononuclear phagocytes (5)(6)(7)(8)(9). The biologic activities of IFN␥ are mainly mediated through the regulation of gene expression. This is achieved upon interaction of IFN␥ with its specific cell surface receptor(s) and activation of different intracellular signaling cascades (10). It is well established that the immediate transcriptional responses induced by IFN␥ are achieved primarily through the activation of the Jak/STAT signaling pathway (11). STAT1 plays a major role in mediating the immune and proinflammatory actions of IFN␥ (12,13), but also STAT3 and STAT5 can be activated by IFN␥ in certain cell types (14 -16).
One of the genes rapidly induced by IFN␥ is SOCS-3 (suppressor of cytokine signaling 3) (17)(18)(19)(20)(21)(22). SOCS-3 is a member of a family of intracellular proteins that negatively regulate responses of immune cells to cytokines by inhibiting the Jak/ STAT pathway (23). Although in vitro overexpression studies have reported that the SOCS proteins are pleiotropic inhibitors of the Jak/STAT pathways activated by several cytokines, targeted gene disruption has shown that these molecules have more specific roles in vivo (20, 21, 24 -28). In this regard, conditional gene targeting studies have recently suggested that SOCS-3 has a nonreduntant function in macrophages and hepatocytes, in which it selectively regulates IL-6 and gp130 signaling (20,21), in accordance with previous observations showing that SOCS-3 binds phosphorylated gp130 and regulates the subsequent activation of STAT3 induced by IL-6 (29 -31). Furthermore, these recent reports revealed that SOCS-3 not only inhibits STAT3-dependent IL-6 signaling but also precludes IL-6 from eliciting STAT1-dependent responses, suggesting that this mediator has a critical role in the balance between these two pathways.
In this study, we investigated the mechanisms regulating SOCS-3 expression by IFN␥ in mouse peritoneal macrophages, J774 mouse macrophage cell line, and embryonal fibroblasts. Analysis of Ϸ7-kb of the genomic 5Ј-flanking region of the mouse SOCS-3 gene revealed that the most proximal STATbinding site present in the SOCS-3 promoter is necessary for IFN␥-induced transcriptional activity. Deletion or point mutation of this STAT-binding element (SBE) completely abrogated IFN␥ action. Accordingly, IFN␥ promoted specific DNA binding activities to an oligonucleotide corresponding to the SOCS-3/ SBE sequence that exclusively contained STAT1. Any contribution of STAT3 was excluded, as demonstrated by the inability of IFN␥-activated STAT3 to bind the SOCS-3/SBE sequence and by the evidence that IFN␥ fully retains the capacity of inducing SOCS-3 expression in STAT3 Ϫ/Ϫ embryonal fibroblasts. These data further reinforce the finding that induction of SOCS-3 expression by IFN␥ is STAT1-dependent. On the basis of these observations, it is conceivable that defining the molecular mechanism and the signaling pathway responsible for the induction of SOCS-3 gene expression in response to IFN␥ might help in clarifying how IFN␥ can modulate cellular responses to different cytokines.
Construction of Reporter Plasmids-The 5Ј genomic region of murine SOCS-3 was cloned from 129 genomic library (33) in pBlueScript (Stratagene, La Jolla, CA). Cloning of the XhoI/NotI fragment into pBS was followed by orientation verification, restriction enzyme digestion, and subcloning of a Ϸ7-kb construct into the KpnI/BamHI sites of the promoterless luciferase reporter vector pGL2basic (Promega, Madison, WI), generating construct 1. The SOCS-3 genomic fragment cloned in construct 1 spans from Ϫ6298 to ϩ884 relative to the transcription initiation, which is defined as ϩ1 (GenBank TM accession number AF314501) and contains the untranslated exon 1 and the fragment of exon 2 upstream the ATG. Truncated forms of SOCS-3 promoter were generated by restriction enzyme digestion of construct 1 as follows: construct 2, deleted the KpnI fragment; construct 3, deleted the KpnI/ NheI fragment; construct 4, deleted the XhoI/SpeI fragment; construct 5, deleted the KpnI/SacII fragment. Construct 6 was generated by KOD Plus 2Љ PCR (Toyobo, Tokyo, Japan) and spans Ϫ50 to ϩ968 nucleotides. 3Ј-deleted construct 7 was generated by removing the proximal SacII/NotI fragment from construct 1. A mutated form of the Ϫ72/Ϫ64 STAT-binding element in construct 5 was produced by substituting the TTCCAGGAA sequence with TTCCAGGTT by site-directed mutagenesis.
Transient Transfection and Dual Luciferase Assay-J774 (4 ϫ 10 5 cells/well) were seeded in 12-well plates and transfected 24 h later with SuperFect transfection reagent (Qiagen). 3g of each SOCS-3 promoter/ luciferase construct were mixed at a 50:1 ratio with the Renilla-encoding pRL-null vector (Promega) and incubated with 7.5 l of SuperFect transfection reagent according to the manufacturer's instruction. 12 h before stimulation, the cells were split into equal aliquots and replated. The cells were then treated with 100 units/ml IFN␥ (PeproTech, London, UK) or left untreated as a control, for the time indicated. After stimulation, the cells were harvested, washed twice with phosphatebuffered saline, and lysed in 30 l of Passive lysis buffer (Promega) followed by two freeze-thaw cycles. Luciferase assays of both firefly and Renilla reniformis luciferases were performed using a dual luciferase reporter assay system (Promega) according to the manufacturer's instructions, and the enzymatic activities of both luciferases were quantified using a Packar LumiCount Microplate Luminometer (Packard Instrument Co., Meriden, CT). The values of firefly luciferase activity were divided for the R. reniformis luciferase activity, to normalize for differences caused by unequal transfection efficiency.
Northern Blot-After stimulation with IFN␥, J774 were harvested at different times, and total RNA was extracted by the guanidinium isothiocyanate method and processed for Northern blot analysis, as already described (35). Northern blot analysis was performed on 15 g of RNA/lane. Specific SOCS-3 mRNA was detected by autoradiography after Northern blot hybridization with the cDNA probe encoding SOCS-3, labeled by the Ready-to-go kit (Amersham Biosciences).
Electromobility Shift Assay-4 ϫ 10 6 cells/well J774, 20 ϫ 10 6 PEM, and 5 ϫ 10 6 MEF were seeded in 60-mm culture dishes overnight. After IFN␥ or IL-6 stimulation for the times indicated, the cells were harvested, and the nuclear extracts were prepared as described previously (36). Protein-DNA complexes were detected by EMSA analysis as previously described (19). 5 g of nuclear extracts were incubated with a 32 P-labeled double-stranded oligonucleotide probe containing the STAT-binding element (SOCS-3/SBE) located at Ϫ72/Ϫ64 nucleotides of the promoter of the SOCS-3 gene (5Ј-CAGTTCCAGGAA-TCGGGGGGC-3Ј) or with the mutated form of this sequence (5Ј-CAG-TTCCAGGTTTCGGGGGGC-3Ј) for 15 min. Supershift experiments were performed by incubating extracts with 2 g of anti-STAT1, anti-STAT3, or anti-CRE Abs (Santa Cruz Biotechnology, Santa Cruz, CA) for 30 min at room temperature before adding the labeled probe. The reaction mixtures were then subjected to electrophoresis in a 5% nondenaturing polyacrylamide gel, dried, and analyzed in an Instant Imager (Packard Instruments, Meridien, CT).

Induction of SOCS-3 Gene and Protein Expression by
IFN␥-We initially investigated the ability of IFN␥ to induce SOCS-3 protein expression in the mouse macrophage cell line J774 and PEM. SOCS-3 protein expression was not detectable in resting J774 and PEM cultures (Fig. 1, A and B, medium). The addition of 100 units/ml IFN␥ to each cell culture induced high levels of SOCS-3 protein expression, which was maximal within 3 h, remained stable for up to 6 h and declined thereafter (Fig. 1, A and B). Northern blot analysis indicated that barely detectable levels of SOCS-3 mRNA were present under non stimulated conditions in J774 cells (Fig. 1C). Following IFN␥ stimulation SOCS-3 mRNA was rapidly induced, reaching maximum levels within 1 h (Fig. 1C). These results extend at the protein level previous observations showing that IFN␥ is able to transiently induce SOCS-3 mRNA in different cell types (17,18,20,(37)(38)(39).
Responses of SOCS-3 Reporter Gene to IFN␥-To identify the region(s) responsive to IFN␥ in the SOCS-3 gene, we carried out a functional analysis of the SOCS-3 promoter. The murine SOCS-3 gene structure has been previously determined (40) and comprises an untranslated exon 1 (ϩ1 to 299) separated from exon 2 (starting at ϩ856) by an intron (ϩ300 to ϩ855). The main transcription site identified has been referred as ϩ1, and the translation initiation site for murine SOCS-3 has been identified in exon 2 at ϩ946 (Fig. 2) (18). A series of 5Ј-deletions of the mouse SOCS-3 promoter/reporter constructs (schematically represented in Fig. 2) were generated as described under "Experimental Procedures." Construct 1 spans the Ϫ6298 to ϩ884 5Ј genomic region of mouse SOCS-3 linked to the luciferase reporter in pGL2Basic vector. The 5Ј truncations of construct 1 are: construct 2, nucleotides Ϫ3980 to ϩ884; construct 3, nucleotides Ϫ2661 to ϩ884; construct 4, nucleotides Ϫ2056 to ϩ884; construct 5, nucleotides Ϫ456 to ϩ884; and construct 6, nucleotides Ϫ50 to ϩ968. The 3Ј truncation of construct 1 is: construct 7, nucleotides Ϫ6298 to Ϫ457. Basal and IFN␥-induced luciferase activity were assayed after transient transfection of J774 cells with the different constructs together with pRL-null, used to normalize transfection efficiency. Luciferase activity was expressed as fold induction by IFN␥-treated over unstimulated cells (Fig. 3). Stimulation with 100 units/ml IFN␥ for 6 h did not increase the luciferase activity in J774 transfected with pGL2Basic alone (1.3 Ϯ 0.1-fold relative to unstimulated pGL2Basic-transfected cells, n ϭ 3). In contrast, basal luciferase activity was increased in IFN␥-treated cells by 8 Ϯ 2-fold when the full SOCS-3 promoter (Ϫ6298/ϩ884 nucleotides, construct 1) was used (Fig. 3). Progressive 5Ј deletion of the SOCS-3 promoter from Ϫ6298 down to Ϫ454 nucleotides did not significantly alter responsiveness to IFN␥, which remained substantially similar with construct 2 (7 Ϯ 1.5-fold, n ϭ 4), construct 3 (6.96 Ϯ 2-fold, n ϭ 4), construct 4 (6.7 Ϯ 1.7-fold, n ϭ 4), and construct 5 (5.3 Ϯ 2.8-fold, n ϭ 4, p Ͼ 0.05) (Fig. 3). A further 5Ј removal of 404 nucleotides resulted in the complete loss of responsiveness to IFN␥. Indeed, stimulation with IFN␥ caused no further increase of luciferase activity in J774 cells transfected with construct 6 (Ϫ50/ϩ968 nucleotides) as compared with J774 transfected with pGL2Basic alone (1.4 Ϯ 0.08 and 1.3 Ϯ 0.1-fold, respectively) (Fig. 3), indicating that the region from nucleotides Ϫ50 to ϩ968 is likely not involved in promoter activity. Collectively, these data show that the promoter region mediating IFN␥ induction is restricted within nucleotides Ϫ452 to Ϫ50. To further confirm these data, we assayed IFN␥-induced luciferase activity after transfecting J774 with construct 7 (Ϫ6298/Ϫ453 nucleotides) carrying a 3Јdeletion of the full-length promoter that removes the proximal 1381 nucleotides, where the minimal IFN␥-responsive elements are located. Removal of this region reduced luciferase activity of IFN␥-stimulated versus untreated cells to 1.9 Ϯ 1-fold, indicating that the sequences located upstream nucleotide Ϫ454 do not significantly contribute to IFN␥ responsiveness.
Previous analysis of the region encompassing Ϫ454 to Ϫ50 nucleotides revealed the presence of two putative STAT-binding elements (TTX 5 AA, where X is any nucleotide), found at Ϫ95 to Ϫ87 and at Ϫ72 to Ϫ64 (40). The Ϫ72/Ϫ64 proximal STAT consensus element (TTCCAGGAA) had been shown to be essential for leukemia inhibitory factor-induced SOCS-3 promoter transactivation in ACTH-secreting corticotroph AtT-20 cells (40) and to be involved in insulin-induced SOCS-3 expression (41). Therefore, in subsequent experiments, we focused on this SOCS-3/SBE. To analyze whether SOCS-3/SBE was necessary for SOCS-3 induction by IFN␥, we generated a mutated form of this sequence by site-directed mutagenesis, introducing AA 3 TT substitution of Ϫ65/Ϫ64 nucleotides into the Ϫ454/ ϩ884 nucleotides promoter of construct 5 (Fig. 4, Construct 5 mut ), thus destroying the specific TTCCAGGAA. Luciferase activity induced by IFN␥ was then assayed in J774 cells transiently transfected with the luciferase expression plasmid carrying the mutated form of the Ϫ454/ϩ884 SOCS-3 promoter (construct 5 mut ) and was compared with that obtained with the Ϫ454/ϩ884 wild type promoter (construct 5) (Fig. 4). Luciferase activity in response to IFN␥ was completely abrogated in the presence of mutant STAT-binding sequence at any time assayed (Fig. 4).
Taken together, our SOCS-3 promoter analysis indicated that (i) the region from nucleotides Ϫ50 to ϩ968 (construct 6) is likely not involved in promoter activity; (ii) the 5Ј truncated constructs 1-5 display similar inducibility by IFN␥, indicating that the region upstream nucleotide Ϫ454 is not responsible for IFN␥-induced SOCS-3 promoter activity; (iii) the region responsive to IFN␥ is localized at nucleotides Ϫ454 to Ϫ50, as indicated by the complete loss of responsiveness to IFN␥ upon its removal (in construct 6 and construct 7); and (iv) within the region from Ϫ454 to Ϫ50, a STAT-binding element from nucleotides Ϫ72 to Ϫ64 is fully responsible for the induction of the SOCS-3 reporter gene in response to IFN␥.

STAT1-dependent Induction of SOCS-3 by IFN␥-SOCS-3
induction has been shown to be strictly dependent on both STAT1 and STAT3 activation in leukemia inhibitory factorstimulated AtT-20 cells (40) and on STAT3 activation in granulocyte colony-stimulating factor-activated granulocytes (42). Because our data indicated that the induction of the SOCS-3reporter gene by IFN␥ was dependent on a STAT-binding sequence, we determined which was/were the STAT(s) involved in the activation of the SOCS-3 promoter by IFN␥. As shown in Fig. 5A, analysis of whole cell lysates prepared from both J774 cells and PEM stimulated with IFN␥ revealed maximal tyrosine phosphorylation of STAT1 and STAT3 within 20 min. Accordingly, EMSA performed with nuclear extracts from resting and IFN␥-stimulated J774 cells and PEM revealed strong DNA binding activities to the oligonucleotide probe containing the SOCS-3/SBE sequence only in activated macrophages (Fig.  5B, lanes 2 and 6). Importantly, disruption of the SOCS-3/SBE consensus sequence resulted in the complete loss of any DNA binding activities (Fig. 5B, lanes 4 and 8), suggesting that SOCS-3 induction by IFN␥ relies on a direct and specific binding of STAT(s) to the SOCS-3/SBE sequence. Indeed, preincubation of nuclear extracts from IFN␥-treated J774 and PEM cells with Abs against STAT1, but not against STAT3, fully displaced the DNA-protein complex (Fig. 5C, lanes 3, 5, 9, and 11), demonstrating that the nuclear factor(s) binding to the SOCS-3/SBE motif is represented by STAT1 only. Nevertheless, the SOCS-3/SBE sequence remains perfectly able to bind tyrosine-phosphorylated STAT3, as observed, for instance, in J774 cells or PEM stimulated with IL-6 ( Fig. 5D). Three distinct SOCS-3/SBE-binding complexes (A, B and C) were in fact detected in nuclear extracts from IL-6-treated macrophages, the most abundant ones being complexes A and C in nuclear lysates of IL-6-stimulated J774 cells (Fig. 5D, lane 3) and complex A in PEM-derived nuclear lysates (Fig. 5D, lane 10). The presence of these three specific DNA-binding complexes is consistent with the formation of STAT1/STAT3 homo-and heterodimers, as revealed by the incubation with Abs against STAT1 and/or STAT3, which abolished the formation of complexes A (STAT3 homodimers) and C (STAT1 homodimers) in J774 cells and PEM (Fig. 5D, lanes 4 -6 and 11-13) and complex B (STAT1/STAT3 heterodimers) in J774 cells (Fig. 5D,  lanes 4 -6). The very faint complex B in IL-6-activated PEM lysates was displaced neither by anti-STAT1 nor by anti-STAT3 Abs (Fig. 5D, lanes 11-13). Taken together, these data not only show that, at least in response to IL-6, STAT3 can bind to the SOCS-3/SBE element, but also exclude the possibility that the lack of binding of the IFN␥-activated STAT3 to the SOCS-3/SBE probe is due to the selectivity of this DNA sequence for STAT1.
To clarify why the IFN␥-activated STAT3 does not bind the SOCS-3/SBE oligonucleotide, we initially compared the relative ability of IFN␥ and IL-6 to induce STAT3 activation/nuclear translocation in both J774 cells and PEM and then determined whether and how the latter phenomena correlated with cytokine-induced SOCS-3/SBE binding activities. Western blot experiments, depicted in Fig. 6A, demonstrate that the amount of tyrosine-phosphorylated STAT3 present in nuclear lysates of IL-6-treated cells is much higher than that induced by IFN␥. In contrast, tyrosine phosphorylation of STAT1 triggered by IFN␥ was much stronger than that elicited by IL-6 (Fig. 6A). Furthermore, because STAT3 can be phosphorylated in a regulated and inducible manner on serine, other than tyrosine, residues (an event that can affect STAT3 function (43,44)), we also analyzed whether IFN␥ was defective in promoting STAT3 serine (Ser 727 ) phosphorylation. However, as shown in Fig. 6A, this appeared not to be the case, even though the ability of IFN␥ to induce STAT3 serine phosphorylation was, once again, very weak as compared with that determined by IL-6.
To establish whether the low amount of nuclear STAT3 detected in IFN␥-treated samples (Fig. 6A) reflects an impaired nuclear translocation or a reduced STAT3 activation, we compared the levels of tyrosine-phosphorylated STAT3 in cytoplasmic and nuclear fractions of cells treated with either IFN␥ or IL-6 in the same experiment. Quantitative analysis of these data revealed that following IFN␥ stimulation, total tyrosinephosphorylated STAT3 is approximately four times, whereas nuclear tyrosine-phosphorylated STAT3 is eight times, lower than those induced by IL-6 treatment (Fig. 6B). Altogether, these data made clear that, compared with IL-6, IFN␥ proves to be a very poor inducer of both tyrosine phosphorylation and nuclear translocation of STAT3, raising the possibility that the nuclear levels of IFN␥-activated STAT3 are not high enough to used as control, and subsequently mixed with 32 P-labeled SOCS-3/SBE probe as in B. DNA-protein complexes were analyzed in nondenaturing acrylamide gels and revealed by autoradiography. One experiment representative of four in the case of J774 cells and of two in the case of PEM is shown for each panel.

STAT1-dependent Activation of SOCS-3 Gene Promoter by IFN␥
determine its binding to the SOCS-3/SBE sequence. To verify this hypothesis, we compared the degree of STAT3 activation with the amount of SOCS-3/SBE-STAT3 complex formation in the nuclear extract of J774 cells stimulated with IL-6 ( Fig. 7). As shown in Fig. 7, the formation of the SOCS-3/SBE-STAT3 complexes fully depends on the concentration of IL-6 used to stimulate the cells (Fig. 7A) and perfectly correlates with the degree of IL-6-dependent STAT3 activation (Fig. 7B). The same experiment shows that tyrosine-phosphorylated STAT3 induced by 5 units/ml IL-6 forms a SOCS-3/SBE-STAT3 complex that is barely detectable in EMSA analysis (Fig. 7A, lane 3). Even more so, binding of STAT3 to the SOCS-3/SBE sequence is not detectable in nuclear extracts of J774 cells stimulated with IFN␥ (Fig. 7A, lane 2), which in fact triggers levels of STAT3 tyrosine phosphorylation much weaker than those induced by any dose of IL-6 used (Fig. 7B). Taken together, the data provide convincing evidence that the amounts of activated STAT3 that translocate to the nuclei upon IFN␥ stimulation are not sufficient to promote a detectable STAT3-binding complex to the SOCS-3/SBE oligonucleotide.
To examine additional reasons for the lack of DNA binding activity by IFN␥-activated STAT3 to the SOCS-3/SBE oligonucleotide, we further considered the possibility that IFN␥ could somehow block the ability of STAT3 to bind to the SOCS-3/SBE sequence. To verify this hypothesis, J774 cells were stimulated with IL-6, alone or in combination with IFN␥, and STAT3 activation was analyzed by EMSA and Western blot. As shown in Fig. 8A, IL-6-induced complex A (consisting of STAT3 homodimers; see Fig. 5A) turned out to be completely withdrawn from binding to the SOCS-3/SBE probe when the cells are stimulated with IL-6 together with IFN␥ (Fig. 8A, lane 4), despite similar levels of STAT1 activation (Fig. 8, A and B). However, under the same experimental conditions, nuclear translocation of tyrosine/serine-phosphorylated STAT3 triggered by IL-6 is not modified by IFN␥ (Fig. 8B), indicating that the lack of STAT3 binding to the SOCS-3/SBE is not due to a defective activation of the upstream signaling pathway leading to STAT3 activation and nuclear translocation. Furthermore, the addition of equal amounts of nuclear extracts from IFN␥activated cells to nuclear extracts from IL-6-activated cells did not modify IL-6-dependent STAT3 DNA binding activity (Fig.  8C, lane 4). This result demonstrates that the disruption of the IL-6-induced STAT3-DNA complex determined by IFN␥ is not caused by the elevated amounts of STAT1 competing for the same sequence.
How IFN␥ prevents IL-6-activated STAT3 from binding to SOCS-3/SBE probe remains to be addressed. One of the mechanisms able to affect the binding of a transcription factor to its DNA recognition sequence is represented by its physical interaction with other different transcription factors, an event that has been described to occur in several situations and that has become a commonly recognized model of action through which gene regulation can be both inhibited and activated. In this context, STAT3 has been reported to be able to interact with FIG. 6. Differential ability of IFN␥ and IL-6 to activate STAT1 and STAT3 phosphorylation and nuclear translocation. A, nuclear cell lysates of J774 cells and PEM stimulated for the times indicated with 100 units/ml IFN␥ or 200 units/ml IL-6 were subjected to Western blot analysis using specific anti-phosphotyrosine STAT1 and STAT3 and antiphosphoserine STAT3 antibodies. The blots were subsequently stripped and reprobed with antibody anti-nuclear protein lamin A/C. One experiment representative of four in the case of J774 cells and of three in the case of PEM is shown for each panel. B, graphic representation of the cytoplasmic and nuclear STAT3-YP levels induced by IFN␥ or IL-6. Nuclear and cytoplasmic cell lysates prepared from the same cells, stimulated with IFN␥ or IL-6, were separated on the same SDS/PAGE and immunoblotted with anti-phosphotyrosine STAT3 antibodies. After stripping, the blots were reprobed with anti-nuclear lamin A/C or anti-cytoplasmic actin antibodies. STAT3-YP levels were quantified using the Odyssey Infrared System and normalized for the relative lamin A/C or actin amounts. One experiment representative of two with similar results is shown. different partners, for instance NF-B p65 homodimers, c-Jun, and GRIM-19, that can either promote or prevent STAT3-dependent transcriptional activity (45)(46)(47)(48)(49)(50)(51). Based on these observations, it is tempting to speculate that STAT3, once activated by IFN␥, is prevented from binding to the SOCS-3/SBE oligonucleotide because it might interact with other transcriptional factors activated at the same time by IFN␥.
To convincingly exclude any STAT3 involvement in the IFN␥dependent induction of SOCS-3, we investigated the ability of IFN␥ to induce SOCS-3 protein expression in mouse embryonal fibroblasts deleted of STAT3 (32). Besides hematopoietic cells, fibroblasts are also targets of IFN␥. Furthermore, there is increasing evidence that fibroblasts exert an important role in the orchestration of immune response (52). As shown in Fig.  9A, in wild type MEFs (STAT3 fl/fl) IFN␥ activates STAT1 and STAT3 tyrosine phosphorylation, whereas in STAT3-deficient MEFs (STAT3 ⌬/⌬) only STAT1 tyrosine phosphorylation is activated by IFN␥. Gel shift analysis showed that IFN␥ induced DNA binding activities to the SOCS-3/SBE probe both in wild type and STAT3 ⌬/⌬ MEFs (Fig. 9B). Under these experimental conditions, SOCS3 protein expression is induced by IFN␥ in wild type and ⌬/⌬ MEFs with similar kinetics and at comparable levels (Fig. 9C), indicating that STAT3 activation does not take part in the mechanisms of SOCS-3 induction triggered by IFN␥.
The results presented in this work demonstrate that SOCS-3 induction by IFN␥ is achieved via activation of STAT1, which in turn binds to the SOCS-3/SBE element. Our data are apparently in contrast with recent works reporting, by microarrays analysis, that IFN␥ up-regulates SOCS-3 gene expression in STAT1-deficient mouse macrophages and hepatocytes (22,37). However, the levels of SOCS-3 mRNA induced by IFN␥ in STAT1-deficient macrophages were significantly reduced as compared with macrophages derived from wild type mice (37). Furthermore, in STAT1 Ϫ/Ϫ mouse hepatocytes, the induction of SOCS-3 mRNA by IFN␥ was enhanced and sustained, and it correlated with a prolonged IFN␥-dependent STAT3 activation as compared with wild type cells (22). These observations suggest that, in the absence of STAT1, activation of the alternative STAT3-dependent pathway, leading to induction of SOCS-3 expression, is favored. Indeed, an increased and prolonged activation of STAT3 in the absence of STAT1 has been described in the brain of STAT1-deficient mice in response to IFN␣ (53). The observation that the SOCS-3/SBE sequence is not selective for recruiting activated STAT1 but is also able to bind tyrosine-phosphorylated STAT3 confirms previous observation (40) and indicates that, in the absence of STAT1, an FIG. 7. Comparative analysis of the degree of STAT3 activation and the amount of SOCS-3/SBE:STAT3 complex formation. J774 cells were stimulated with 100 units/ml IFN␥ or with the indicated doses of IL-6 for 20 min. The nuclear lysates were analyzed for STAT3 tyrosine and serine phosphorylation in Western blot (A) and for binding to the 32 P-labeled SOCS-3/SBE probe in EMSA analysis (B). One experiment representative of three is shown.
FIG. 8. IFN␥ inhibits the binding of IL-6-activated STAT3 to the SOCS3/SBE element. A, 5 g of nuclear extracts of J774 cells stimulated with 200 units/ml IL-6 and/or 100 units/ml IFN␥ were incubated with 32 P-labeled SOCS-3/SBE probe and subjected to EMSA analysis. One experiment representative of three is shown. B, the same nuclear lysates analyzed in A were separated on SDS/PAGE and immunoblotted with anti-phosphotyrosine STAT3 antibodies. The blots were stripped and reprobed with anti-phosphoserine STAT3 and subsequently with anti-phosphotyrosine STAT1 antibodies. Equal protein loading was confirmed by probing the membrane with anti-nuclear protein lamin A/C. C, equal amounts of nuclear extracts prepared from untreated, IFN␥or IL-6-treated J774 cells were mixed together as indicated, incubated with the 32 P-labeled SOCS-3/SBE probe and subjected to EMSA analysis. The experiment shown is representative of three performed with similar results. enhanced and prolonged STAT3 activation will ensure the expression of SOCS-3, an intracellular mediator whose absence has been shown to have profound influence on macrophage responses (20,21,28). It would be interesting to analyze whether in the absence of STAT1 the expression of STAT3-dependent genes is quantitatively enhanced in response to IFN␥, similarly to the switch to STAT1-dependent gene expression observed in response to IL-6 in cells lacking STAT3 (32).
Our result that the induction of SOCS-3 expression by IFN␥ is STAT1-dependent is relevant in the context of the emerging concept that IFN␥/STAT1 and IL-6/STAT3 are mutually antagonists and negatively regulate each other through the induction of SOCS proteins (22). Indeed, although SOCS-1 and SOCS-3 are both induced by IFN␥ and IL-6 and can both inhibit functional responses to each of these cytokines when overexpressed in vitro (54), studies conducted with mice in which SOCS-1 or SOCS-3 genes have been selectively inactivated indicated that their actions are more restricted and specific in vivo (20,21,24,25,28). It appears that SOCS-1 and SOCS-3 have complementary roles in regulating cytokines signaling: SOCS-3 Ϫ/Ϫ cells show hyper-responsiveness to IL-6 but not to IFN␥ (20,21), whereas the opposite is true with SOCS-1 Ϫ/Ϫ cells (24,25). The biological consequences of SOCS-3 expression are not limited to the inhibition of STAT3-dependent IL-6 responses but are also responsible of preventing STAT1dependent IL-6 alternative signaling pathway. In fact, the ab-sence of STAT3 or of SOCS-3 has been found to be determinant for a prolonged activation of STAT1 in response to IL-6 and the subsequent induction of IFN␥-like responses (20,21,32). Our finding that SOCS-3 expression induced by IFN␥ is achieved via STAT1 activation is relevant in that it points out that a STAT1-dependent pathway is utilized by IFN␥ to prevent amplification of IFN␥-like responses triggered by IL-6. This consideration should be taken into account in studies aimed at selectively blocking one specific signaling pathway.