Antagonistic Regulation of Type I Collagen Gene Expression by Interferon- g and Transforming Growth Factor- b INTEGRATION AT THE LEVEL OF p300/CBP TRANSCRIPTIONAL COACTIVATORS*

Among the extracellular signals that modulate the synthesis of collagen, transforming growth factor- b (TGF- b ) and interferon- g (IFN- g ) are preeminent. These two cytokines exert antagonistic effects on fibroblasts, and play important roles in the physiologic regulation of extracellular matrix turnover. We have shown previously that in normal skin fibroblasts, TGF- b positively regulates a 2(I) procollagen gene ( COL1A2 ) promoter activity through the cellular Smad signal transduction pathway. In contrast, IFN- g activates Stat1 a , down-reg-ulates COL1A2 transcription, and abrogates its stimulation induced by TGF- b . The level of integration of the two pathways mediating antagonistic collagen regulation is unknown. We now report that IFN- g abrogates TGF- b -stimulated COL1A2 transcription in fibroblasts by inhibiting Smad activities. IFN- g appears to induce competition between activated Stat1 a and Smad3 for interaction with limiting amounts of cellular p300/CBP. Overexpression of p300 restored COL1A2 stimulation by TGF- b in the presence of IFN- g , and potentiated IFN- g dependent positive transcriptional responses. In contrast to fibroblasts, in U4A cells lacking Jak1 and con-sequently unable to activate Stat1 a -mediated responses, IFN-

Transforming growth factor-␤ regulates cellular responses through modulation of transcription of genes encoding cell cycle regulators, extracellular matrix proteins, adhesion molecules, cytokines, and transcription factors (1). One of the most potent effects of TGF-␤ 1 is connective tissue accumulation, achieved in part by stimulation of the transcription of type I collagen genes (2)(3)(4)(5). Thus, TGF-␤ plays crucial roles in embryonic development and organogenesis, and physiologic connective tissue remodeling during wound healing and tissue repair. On the other hand, excessive TGF-␤ activity is implicated in the development of pathological fibrosis, the "dark horse of tissue repair" (6).
TGF-␤ initiates cellular signals through two distinct transmembrane serine-threonine kinase receptors. Upon ligand binding, the activated TGF-␤ receptor complex transiently interacts with receptor-activated Smads which propagate TGF-␤ signals (7). Smad2 and Smad3 are direct substrates of the TGF-␤ receptor kinase, and interact with the common partner Smad4 (8 -10). Smad4-containing heteromeric Smad complexes then translocate from the cytoplasm into the nucleus where they function as transcriptional regulators. Smads2-4 share highly conserved DNA-binding MH1 and transactivating MH2 domains; the latter also mediates Smad interactions with other proteins (11,12). Smad7 contains a characteristic MH2 domain, but lacks the conserved SSXS phosphorylation motif, and its MH1 domain shows marked divergence from that of Smad3 (13). Furthermore, in contrast to receptor-activated Smads, Smad7 stably binds to TGF-␤ receptors and interferes with ligand-induced phosphorylation of Smad2 and Smad3 (14). Because its expression is markedly induced by the ligand, Smad7 appears to serve an autoregulatory negative feedback function in cellular TGF-␤ signaling.
We have shown previously that overexpression of Smad3 in primary skin fibroblasts mimicked the action of TGF-␤, markedly inducing COL1A2 promoter activity in vitro (15). Transactivation was blocked by phosphorylation-defective dominant negative mutants of Smad3, establishing the critical role of endogenous Smads in transducing information from the activated TGF-␤ receptor in these cells. The COL1A2 promoter contains Smad3/Smad4-binding consensus "CAGA boxes" (16), also found in the promoters of PAI-1, junB, and other TGF-␤ inducible genes (17,18), that are necessary and sufficient to mediate transcriptional responses induced by TGF-␤ (19). However, as CAGA boxes are widely distributed in the promoters of mammalian genes, and the affinity and specificity of SMAD binding to these is relatively low, other nuclear factors are likely to contribute to the specific and tight Smad-DNA interactions that are required for transcriptional regulation (20). Cooperation with FAST-1 (21), AP-1 (22,23), TFE3 (24), Sp1 (25), and vitamin D receptor (26) are implicated in Smadmediated transcription of TGF-␤-responsive genes.
CREB-binding protein (CBP) and adenovirus E1A-associated protein p300 are structurally conserved large proteins that function as essential coactivators in several signal transduction pathways. p300/CBP enhance transcription by bridging DNA-bound factors and the basal transcriptional machinery, thereby stabilizing the pre-initiation complex. Furthermore, by nature of their intrinsic histone acetyltransferase enzyme activity, these coactivators acetylate amino-terminal lysine residues of nucleosomal histones (27,28). Local histone hyperacetylation causes nucleosomal relaxation, thus promoting access of transcription factors to target DNA sequences. The transcriptional activity of genes is strongly correlated with their acetylation state (29). The presence of conserved proteinbinding domains enables p300/CBP to interact with distinct classes of regulated transcription factors. Through binding to CREB, the p65 component of NF-B, the c-Jun, and c-Fos components of AP-1, and Stats, p300/CBP integrate converging cellular signaling pathways (30 -35). Receptor-activated Smads interact directly with p300/CBP via the Smad MH2 domain and the p300 COOH-terminal region overlapping the E1A-binding site (36 -41). We previously demonstrated that p300 markedly enhanced TGF-␤-stimulated COL1A2 transcription in fibroblasts (42). By competing with activated Smad3 for limiting amounts of cellular p300/CBP, the adenoviral oncoprotein E1A abrogated TGF-␤ transactivation, indicating the critical role of p300/CBP in collagen gene transcription in fibroblasts.
Interferon-␥ (IFN-␥), a pleiotropic cytokine produced by T cells and NK cells, plays fundamental roles in both innate and acquired immune responses (43). Transcriptional responses induced by IFN-␥ in most cells are mediated through the Jak-Stat pathway (44). Upon stimulation by IFN-␥, tyrosine-phosphorylated cytoplasmic Stat1␣ forms homodimeric complexes that can translocate into the nucleus, and bind directly to palindromic ␥-activated sites (GAS) of IFN-␥-responsive target gene promoters. Stat1␣ thus serves as an essential mediator of IFN-␥-induced transcriptional responses. Stat1␣ physically associates with p300/CBP near its amino-terminal domain; this interaction plays an important functional role in positive regulation of IFN-␥-induced transcriptional responses (45,46). In addition to transcriptional stimulation, IFN-␥ can also negatively regulate the transcription of selected genes, but no common IFN-␥-specific inhibitory elements have been identified. We and others have shown previously that IFN-␥ inhibits the transcription of collagen in fibroblasts independent of Stat1promoter interactions, and abrogates its stimulation induced by TGF-␤ (47)(48)(49)(50)(51). Thus, TGF-␤ and IFN-␥ exert opposite effects on collagen synthesis. Because these two cytokines are secreted by inflammatory cells at sites of tissue injury, their antagonistic interactions regulating collagen synthesis are likely to be of great importance in the maintenance of connective tissue homeostasis.
As well as suppressing the stimulation of collagen transcription, IFN-␥ abrogates other TGF␤ responses, including collagenase-3 expression in epithelial cells (52), perlecan expression in colon carcinoma cells (53), and fibronectin and laminin receptor expression in monocytic cells (54). The basis underlying antagonistic modulation of TGF-␤ signaling by IFN-␥ is incompletely understood. In the present report, we characterized the repression of TGF-␤-stimulated collagen transcription by IFN-␥ in normal skin fibroblasts. The results indicate that the stimulatory effects of TGF-␤ on Type I collagen gene (COL1A2) transcription were abrogated by IFN-␥ through a minimal Smad-binding element of the COL1A2 promoter. Inhibition of TGF-␤ signaling in fibroblasts was not mediated through antagonistic Smad7. The stimulatory effect of TGF-␤ on COL1A2 transcription could be rescued in the presence of IFN-␥ by overexpression of p300. These findings indicate that p300/CBP integrate IFN-␥/TGF-␤-induced signals that positively or neg-atively regulate collagen gene transcription in fibroblasts, and suggest that an increase in activated Stat1␣ in IFN-␥-treated fibroblasts suppressed Smad-mediated transactivation by titrating away the coactivators. The findings provide novel understanding of the physiologically important antagonistic regulation of collagen gene transcription by cytokines.

MATERIALS AND METHODS
Reagents and Cell Culture-All tissue culture reagents were obtained from Life Technologies, Inc. (Grand Island, NY). Recombinant human IFN-␥ was from Genentech Inc. (South San Francisco, CA), and TGF-␤ from Amgen (Thousand Oaks, CA). Primary cell cultures were established from neonatal foreskin by previously described explant techniques (15), and studied between passages 4 and 8. Cells were grown at 37°C in a 5% CO 2 atmosphere in modified Eagle's medium supplemented with 1% or 10% fetal calf serum, 1% vitamins, and 2 mM L-glutamine. The U4A Jak1-deficient cell line, which does not support IFN-␥-induced gene expression (55), and U4A/Jak1 were a kind gift of O. Colamonici. Transforming growth factor-␤ and IFN-␥ were added simultaneously, and cultures were harvested following 24 -48 h incubation. In previous studies under similar conditions, we found that neither TGF-␤ nor IFN-␥ significantly effected cell number or viability in confluent fibroblast cultures (15). Total RNA was isolated with TRIZOL Reagent (Life Technologies, Inc.) following the indicated periods of incubation, and relative levels of mRNA were examined by Northern analysis using radiolabeled Smad3, Smad7, and glyceraldehyde-3-phosphate dehydrogenase cDNA probes. The cDNA-mRNA hybrids were visualized by autoradiography on Kodak X-AR5 films exposed for 24 -48 h with intensifying screens.
Immunoprecipitation and Western Immunoblotting-At the end of the indicated period of incubation, cells were washed with ice-cold phosphate-buffered saline, and whole cell lysates prepared. Lysates were either directly resolved by electrophoresis in polyacrylamide gels, or first immunoprecipitated for 1 h using anti-p300 antibody (Santa Cruz Biotechnology, Santa Cruz CA). Gels were blotted onto polyvinylidene difluoride membranes (Millipore, Bedford, MA), and subjected to immunoblotting with primary antibodies (human anti-p300, anti-Smad7, and anti-actin (C-2) from Santa Cruz; anti-Smad3 from Zymed Laboratories Inc., San Francisco, CA; anti-Type I collagen from Southern Biotech (Birmingham, AL); anti-Stat1 from Transduction Labs (Lexington, KY); and anti-phosphotyrosine Stat1 (Y701) from Upstate Biotechnology (Lake Placid, NY) for 2 h at room temperature. The blots were then washed, followed by incubation with the appropriate horseradish peroxidase-conjugated secondary antibody, and visualized by chemiluminescence.
Cellular Immunofluorescence Imaging-The expression and intracellular localization of endogenous SMADs and Stat1␣ in the presence or absence of IFN-␥ or TGF-␤ was studied by indirect immunofluorescence. For this purpose, fibroblasts were seeded into chamber glass slides and incubated in media with 0.1% fetal calf serum and IFN-␥ added 30 min before TGF-␤. After 2 h incubation, cells were then fixed with methanol, incubated with primary antibodies (anti-Smad3 and anti-Smad4 from Santa Cruz; anti-Stat1␣ from Transduction Labs) for 1 h, followed by horseradish peroxidase-conjugated secondary antibodies (Santa Cruz). After three washes, the slides were stained, and the intracellular distribution of Smads and Stat1 was examined by fluorescence or confocal microscopy. Quantitation was performed in a blinded fashion by scoring 100 fibroblasts in different fields as showing predominantly nuclear or cytoplasmic immunofluorescence.
Transient Transfection-To measure transcriptional responses to IFN-␥ or TGF-␤, 772COL1A2/CAT consisting of the Ϫ772 to ϩ58 bp segment of human COL1A2 promoter linked to the CAT gene (56), CAGA-COL1A2/luc containing six copies of the Ϫ266/Ϫ258 bp sequence of COL1A2 (17), or SBE4-luc containing four tandem repeats of an 8-bp palindromic consensus Smad-binding consensus sequence (16) were used. Expression vectors for wild type and ⌬HAT-mutant p300 (57), Smad3 (58), and antisense Smad7 cDNA (59), and appropriate control vectors were transfected. As controls, ⌬-152DR-CAT (from J. Ting, University of North Carolina, Chapel Hill, NC) containing an IFN-␥-responsive 152-bp segment of the HLA-DR promoter ligated to the CAT gene (60), and GAS-tk-CAT (from G. Sen, Cleveland Clinic Foundation Research Institute) containing an IFN-␥-responsive 24-bp segment of the IRF-1 promoter ligated to the CAT gene, were used. Transient transfections were performed by calcium phosphate/DNA co-precipitation or using Superfect reagent (Qiagen, Valencia, CA), as described previously (15). The total amount of DNA in each transfection was kept constant by addition of appropriate empty vectors, as re-quired. Following incubation with the indicated cytokines, CAT, and luciferase activities were determined in duplicate and normalized with protein concentration.
Preparation of Nuclear Extracts and Electrophoretic Gel Mobility Shift Assay-Fresh media with 0.2% fetal calf serum and IFN-␥ alone or with TGF-␤ were added to confluent fibroblasts. At the end of the indicated incubation periods, nuclear extracts were prepared according to the method of Andrews and Faller (61), and protein concentrations were determined using the Bio-Rad assay. Double-stranded oligonucleotide probes corresponding to a consensus Smad-binding element (SBE). The probes were end-labeled using T4 polynucleotide kinase (Promega, Madison WI). Electrophoretic gel mobility shift assays were performed as described previously (15).
Statistical Analysis-Statistical differences between experimental groups were determined by analysis of variance, and values of p Ͻ 0.05 by Fisher's test were considered significant.

Antagonistic Regulation of Collagen Transcription by TGF-␤
and IFN-␥, Convergence at the Level of Smad3-To examine the regulation of collagen transcription, normal skin fibroblasts were transiently transfected with COL1A2 promoter-CAT reporter constructs. TGF-␤ caused marked stimulation of CAT activity driven by 772 bp of the COL1A2 promoter. This induction was completely abrogated by IFN-␥, indicating that ciselements necessary for negative regulation of COL1A2 expression were contained within this region of the promoter (Fig. 1A,  left panel). The activity of ⌬152DR-CAT was induced by IFN-␥, demonstrating that the inhibitory effect of IFN-␥ was selective for the COL1A2 promoter (Fig. 1A, middle panel). Changes in the intracellular pool of collagen in response to the cytokines paralleled those in promoter activity (Fig. 1B). Inhibition of collagen synthesis did not require pretreatment with IFN-␥ prior to stimulation. IFN-␥ induced the rapid nuclear translocation of Stat1, an essential mediator of IFN-␥ transcriptional responses (Fig. 2). To further examine Stat1 activation, the consensus Stat1␣-binding GAS oligonucleotide was used as probe in gel shifts. As expected, IFN-␥ treatment of fibroblasts for 15 min resulted in the formation of a GAS-specific DNAprotein complex (Ref. 62; and data not shown). A careful analysis of the proximal COL1A2 promoter failed to identify the presence of consensus Stat1␣-binding GAS elements (TTN 5 AA). Furthermore, in contrast to classical Stat1-mediated transcriptional responses, the inhibitory effect of IFN-␥ on COL1A2 promoter activity was delayed (Ͼ8 h) (51). Therefore, these observations suggested that Stat1␣ activation could not be fully responsible for suppression of COL1A2 transcription by IFN-␥ .
Because we have previously established a fundamental role for Smad3 in TGF-␤ stimulation of COL1A2 transcription in fibroblasts (15), the possibility that the Smad signal transduction pathway was a target for the inhibitory activities of IFN-␥ was considered. To directly examine the involvement of Smad3, the heterologous minimal construct CAGA-COL1A2/luc, which contains six tandem copies of the COL1A2 Smad-binding CAGA element shown to be sufficient to mediate TGF-␤ responses (19) was used. The TGF-␤-induced increase in activity of this promoter was suppressed in IFN-␥-treated fibroblasts (Fig. 1A, right panel). Identical results were obtained with SBE4-luc, containing a consensus SBE (16). IFN-␥ had no effect on endogenous Smad3 or Smad4 mRNA or protein expression (data not shown). Together, these findings indicated that IFN-␥ suppressed TGF-␤ stimulation of COL1A2 transcription mediated by multimerized Smad3-recognition sites, while inducing endogenous Stat1␣ activation. Inhibition did not result from decreased expression or DNA binding of cellular Smads, suggesting instead that IFN-␥ targeted their transcriptional activities.
Smad7-independent Inhibition of COL1A2 Transcription in Fibroblasts-Intracellular cross-talk among cytokines with op- posing effects can be mediated through induction by one cytokine of autocrine mediators that block signaling triggered by the another cytokine. For instance, IFN-␥ abrogates interleukin-4-induced transcriptional responses via endogenous suppressor of cytokine signaling SOCS-1, which prevented STAT6 activation in these cells (63,64). In a similar vein, TGF-␤ stimulation of 3TP-lux transcription is prevented by IFN-␥ or TNF-␣ via induction of antagonistic cellular Smad7, which blocks ligand-induced Smad3 phosphorylation and its attendant events through stable interaction with the TGF-␤ receptors (65,66). These observations suggested a possible mechanism to account for the antagonistic effect of IFN-␥ on TGF-␤-stimulated COL1A2 transcription: induction of an endogenous inhibitor of TGF-␤ signaling by IFN-␥. In contrast to SOCS-1, which has not been shown to suppress TGF-␤ signaling, Smad7 specifically blocked Smad3-mediated responses in fibroblasts (15). Therefore, we sought to examine whether endogenous Smad7 could be implicated in IFN-␥ suppression of TGF-␤-stimulated COL1A2 transcription. The results showed that, whereas it consistently induced rapid and transient increase in Smad7 mRNA expression in HepG2 epithelial cells, IFN-␥ had no detectable effect on the levels of endogenous Smad7 determined by immunocytochemistry and Western immunoblotting, or on Smad7 mRNA expression determined by Northern analysis (data not shown). Our failure to demonstrate induction of Smad7 expression by IFN-␥ in fibroblasts, in contrast to cells of epithelial origin, is consistent with recent report (66).
We next examined the functional consequences of downregulating endogenous Smad7 using an antisense expression plasmid (59). Fibroblasts were transfected with 10 g of 772COL1A2/CAT along with 5 g of antisense Smad7 cDNA, or empty vector. As expected, transactivation of COL1A2 by TGF-␤ was greater in fibroblasts expressing antisense Smad7 than in control fibroblasts transfected with empty vector, indicating that the level of endogenous Smad7 determined the magnitude of the TGF-␤-induced response (Fig. 3A). Smad7 antisense cDNA markedly decreased the amount of cellular Smad7 (Fig. 3B). The inhibitory effect of antisense on cellular Smad7 was unaffected by IFN-␥ (data not shown). Down-regulation of endogenous Smad7 did not abrogate the inhibitory effect of IFN-␥ on TGF-␤-induced COL1A2 promoter in these fibroblasts (Fig. 3A). Similar results were obtained when antisense oligonucleotides were used to down-regulate endogenous Smad7 (data not shown). Taken together, these results indicate that endogenous Smad7 is not responsible for abrogation of TGF-␤-stimulated COL1A2 transcription by IFN-␥ in primary fibroblasts.
IFN-␥ Abrogates COL1A2 Stimulation Downstream from Smad3 Activation-The antagonistic effect of IFN-␥ on Smadmediated COL1A2 transactivation could result from interference with TGF-␤-induced Smad3 activation. Three complementary approaches were undertaken to examine this possibility. First, we sought to determine whether inhibition of Smadmediated transactivation by IFN-␥ altered the DNA binding activity of nuclear proteins. Electrophoretic mobility shift assays with fibroblast nuclear extracts showed that incubation with IFN-␥ slightly enhanced TGF-␤-induced binding of endogenous Smad3 to the consensus SBE (Fig. 4A). To assess the effects of IFN-␥ on the transactivation function of Smad3, fibroblasts overexpressing recombinant Smad3 were treated with IFN-␥. As shown in Fig. 4B, the striking elevation in COL1A2 promoter activity in Smad3-transfected fibroblasts was substantially abrogated, indicating that in these cells IFN-␥ disrupted signaling downstream of the activated TGF-␤ receptor. IFN-␥ did not alter the level of recombinant Smad3 expression (Fig. 4B, inset; and data not shown). Next, the effect of IFN-␥ on the subcellular distribution of endogenous Smad3 was examined. Nuclear import of Smad3, essential for regulating target gene transcription, is highly dependent on its ligandinduced phosphorylation (11). Treatment of the fibroblasts with TGF-␤ caused rapid (Ͻ2 h) accumulation of endogenous Smad3 and Smad4 within the nucleus (Fig. 4C). IFN-␥ by itself had no effect on Smad cellular localization, and pretreatment of the cells failed to prevent TGF-␤-induced nuclear translocation at early or late time points. The inability of IFN-␥ to disrupt ligand-induced nuclear import of activated Smads further indicates that mechanisms distinct from blocking ligand-induced Smad activation underlie the antagonistic effects of IFN-␥ on TGF-␤ stimulation of COL1A2 transcription.
IFN-␥ treatment of fibroblasts caused rapid Stat1 nuclear accumulation (Fig. 2); and increase in its DNA binding activity (62). To directly determine the role of Stat1␣ in mediating repression of TGF-␤-induced COL1A2 transcription by IFN-␥, the response of Jak1-deficient U4A cells was examined. These cells are a genetically defined system commonly used to delineate IFN-␥ signaling (55). Treatment of U4A cells with IFN-␥ had little effect on COL1A2 promoter activity in the presence of TGF-␤ (Fig. 5). In contrast, U4A cells rescued with stable expression of exogenous Jak1 (U4A/Jak1) demonstrated normal Stat1␣-mediated responses, and IFN-␥ caused Ͼ70% decrease in TGF-␤-stimulated CAT activity. As expected, the activity of the GAS-driven reporter was markedly up-regulated by IFN-␥ in the parental, but not in the signaling defective, cells (data not shown). We therefore conclude that Stat1␣ activity is required for IFN-␥ abrogation of TGF-␤-induced, Smad3-mediated COL1A2 transactivation. Because Stat1 activation is rapid and precedes repression of COL1A2, it is likely to represent an early step in a series of events culminating in transcriptional repression.
TGF-␤ and IFN-␥ Signals Are Integrated at the Level of p300/CBP Coactivators-Certain responses mediated by IFN-␥/Stat1␣ and TGF-␤/Smad3 pathways are critically dependent on p300/CBP coactivators (9, 36 -42, 45). We reasoned, therefore, that the antagonistic effect of IFN-␥ on Smad-mediated TGF-␤ transactivation of COL1A2 could result from competition between Smad3 and IFN-␥-activated signal transducers such as Stat1␣ for interaction with a limiting cellular pool of p300/CBP. Consistent with this possibility, we found that overexpression of p300 in fibroblasts rescued TGF-␤ stimulation of COL1A2 promoter activity (Fig. 6A) and endogenous collagen accumulation (Fig. 6B) in the presence of antagonistic IFN-␥. In contrast, a mutant form of p300 defective in histone acetyltransferase activity failed to relieve IFN-␥ repression (data not shown); indicating that modulation of COL1A2 transcription involved the histone acetylase function of p300. p300 enhanced transactivation of a minimal GAS promoter by low concentrations (10 units/ml) of IFN-␥, suggesting a significant in vivo functional interaction between p300 and activated endogenous Stat1␣ in fibroblasts (Fig. 6C). To examine if the antagonistic effect of IFN-␥ on TGF-␤-stimulated COL1A2 transactivation could be due to down-regulation of cellular p300 levels, immunoblotting was performed. The results showed that the pool of endogenous p300 was slightly increased in IFN-␥-treated fibroblasts, indicating that IFN-␥ interfered with the function, and not the amount, of p300 in these cells (Fig. 6D). Consistent with this notion, we also found that IFN-␥ had no detectable effect on the levels or subcellular distribution of endogenous p300 (data not shown). Taken together, these results suggested that p300 was present in limiting amounts in fibroblasts, and the ability of IFN-␥ to inhibit Smad3 transactivation resulted from competition for p300 by Stat1␣ or other transcription factors induced by IFN-␥.
To directly examine the in vivo interaction of endogenous Smad3 with p300, fibroblasts were transfected with a p300 expression plasmid and treated with TGF-␤ alone, or together with IFN-␥ for 30 min, a time point sufficient for induction of p300 interaction with both Stat1 and Smad3 (36,45). At the end of the incubations, whole cell lysates from fibroblasts transfected with p300 expression plasmid and treated with IFN-␥ and/or TGF-␤ for 30 min were immunoprecipitated with p300 antibody and following electrophoresis, examined by immunoblot with antibodies to Smad3. As shown in Fig. 7A, the TGF-␤-induced interaction of p300 with endogenous Smad3 appeared to be modestly reduced upon treatment of the fibroblasts with IFN-␥. Next, the interaction of Stat1␣ with p300 was examined. Immunoprecipitated proteins were examined by immunoblot with antibodies to Stat1␣ or tyrosine-phosphorylated Stat1␣. As shown in Fig. 7B, IFN-␥ strongly enhanced the interaction of p300 with tyrosine-phosphorylated Stat1␣, and this interaction was not altered by TGF-␤. These results demonstrate that in primary fibroblasts, p300 interacted with both activated Smad3 and activated Stat1␣ in vivo; in a liganddependent manner. Whereas IFN-␥ reduced the interaction of p300 with endogenous Smad3, TGF-␤ was unable to modulate the association of p300 with activated Stat1␣. DISCUSSION The production of ECM proteins by fibroblasts must be coordinated and strictly regulated. During dynamic processes of tissue remodeling such as wound healing, fibroblasts are subject to simultaneous signaling by distinct combinations of cytokines, growth factors, and other regulatory molecules. These extracellular signals provide fibroblasts with information that synergistically or antagonistically influence the expression of target genes. Type I collagen is the major structural component of the connective tissue of the skin and other organs. Because its excessive accumulation results in fibrosis, the dynamic equilibrium between signals that stimulate and those that inhibit collagen synthesis must be carefully maintained. The transcription of COL1A2, one of the best characterized responses in fibroblasts, is stimulated by TGF-␤ and inhibited by IFN-␥ (47)(48)(49)(50)(51). Therefore, the intracellular cross-talk between the signaling pathways activated by these two functionally antagonistic cytokines is of substantial interest.
The results presented here provide evidence for the nuclear integration of TGF-␤ and IFN-␥ signaling at the level of the shared cofactors p300/CBP. Stat1␣, a critical transducer of IFN-␥ responses, is one of a large group of transcription factors that employ p300/CBP to bring about their effects on transcription. Interactions between Stat1␣ and p300/CBP can be liganddependent or constitutive, indicating that both unphosphorylated monomeric, and phosphorylated dimeric Stat1␣ can interact with p300/CBP. We have previously demonstrated that in primary fibroblasts, endogenous p300/CBP is required by activated Smad3 to stimulate COL1A2 transcription (42). TGF-␤ induces direct binding of Smad3 to p300/CBP (36 -42). In fibroblasts exposed to TGF-␤ and IFN-␥ simultaneously, activated Stat1␣ and Smad3 are therefore likely to compete with each other for interaction with p300/CBP. Because p300/ CBP is present in limiting amounts in these cells, the net effect of simultaneous signaling by the two antagonistic cytokines on target gene transcription appears to be determined by the levels of the coactivators, and the relative affinity of their interaction with Smad3 and IFN-␥ signal transducers. Although we cannot exclude the possibility that integration of antagonistic signals at the level of p300 involves direct inhibition of p300 function by IFN-␥/Stat1␣, rather than its simple sequestration, the functional synergism between p300/CBP and IFN-␥ shown in Fig. 6C suggests that this is not the case. Further studies to examine the effect of IFN-␥ on p300/CBP histone acetyltransferase activity are currently in progress. The model for integration of antagonistic transcriptional signals through their competition for nuclear coactivators is illustrated in Fig. 8. This model proposed here is reminiscent of the regulation of the scavenger receptor gene by IFN-␥ and macrophage colony-stimulating factor via p300/CBP (45). The well documented antagonistic effect of nuclear receptors on AP-1mediated signaling likewise is ascribed to competition for limiting amounts of cellular cofactors (67). Furthermore, while this paper was under review, several reports have demonstrated that antagonistic regulation of Smad-mediated transcriptional responses by TNF-␣ is also mediated through competition for cellular p300 (68,69).
In contrast, a distinct model has been proposed to account for the suppression of other TGF-␤-induced transcriptional responses by IFN-␥ (65) or TNF-␣ (66). In examining the antagonistic regulation of 3TP-lux activity by TGF-␤ and IFN-␥, Ulloa et al. (65) concluded that IFN-␥ suppression of Smadmediated responses was mediated by endogenous Smad7 functioning as autocrine inhibitor of TGF-␤ signaling. Similarly, the opposing activities of TNF-␣ on TGF-␤-induced transactivation of 3TP-lux or SBE4-luc were shown to be mediated by Smad7 (66). In the present studies using primary skin fibroblasts, we were unable using a variety of approaches to detect induction of Smad7 expression by IFN-␥; furthermore, downregulation of endogenous Smad7 levels by antisense failed to abrogate the inhibitory effects of IFN-␥ on TGF-␤-stimulated transcription in these cells. The findings suggest that expression and autoregulatory function of ligand-induced Smad7 may show cell-lineage specific differences.
Two additional lines of evidence are provided here to exclude the role of antagonistic Smad7 in mediating suppression of TGF-␤ responses by IFN-␥. First, we demonstrated that IFN-␥ was capable of abrogating COL1A2 transcriptional stimulation FIG. 7. Ligand-induced interaction of p300 with activated cellular Smad3 or Stat1␣ in vivo. Confluent fibroblasts were transfected with p300 expression plasmids. Following incubation in the presence or absence of IFN-␥ or TGF-␤ for 30 min, whole cell lysates were prepared and immunoprecipitated with antibodies to p300. Equal amounts of immunoprecipitated proteins were analyzed by immunoblot with antibodies to: A, Smad3 and p300; or B, tyrosine-phosphorylated Stat1␣ or Stat1␣. Last lane, control IgG.
FIG. 8. Antagonistic regulation of COL1A2 transcription in primary fibroblasts by TGF-␤ (؉) and IFN-␥ (؊). The two cytokines exert opposing effects on gene expression through competition between Smads and STAT1 for limiting amounts of the shared cellular coactivators p300/CBP. induced by Smad3, indicating that the antagonistic cross-talk between TGF-␤ and IFN-␥ occurred downstream of the ligandbound TGF-␤ receptors. Furthermore, treatment of the fibroblasts with IFN-␥ did not prevent TGF-␤-induced nuclear accumulation of pathway-restricted Smads, indicating that, in distinct contrast to the endogenous Smad7 autoinhibition model proposed by Ulloa et al. (65), negative regulation of Smad signaling by IFN-␥ in fibroblasts occurred distal to Smad3 activation or Smad3-Smad4 complex nuclear translocation. Our observations are consistent with a model whereby competition of regulated transcription factors for p300/CBP interaction, which occurs within the nucleus, but not with Smad7mediated blockade of the activation of pathway-restricted Smads by TGF-␤ receptors. Finally, endogenous Smad7 induction by TGF-␤ abrogated the stimulation of COL1A2 promoter activity (15), indicating that antagonistic Smad7 serves to limit positive regulation of transcription, rather than mediating ligand-induced negative regulation, in primary fibroblasts.
In summary, the present results provide compelling evidence for the role of p300/CBP in antagonistic regulation of COL1A2 transcription by TGF-␤ and IFN-␥. The results indicate that in normal skin fibroblasts, the intracellular signaling pathways triggered by TGF-␤ and IFN-␥ are integrated at the nuclear level, and appear to involve a competition between activated Stat1␣ and Smad3 for interactions with limiting amounts of cellular p300/CBP. Further characterization of the molecular basis by which IFN-␥ inhibits SMAD-mediated transcriptional activity in fibroblasts will facilitate the design of specific inhibitors of Smad signaling, which may be therapeutically useful in scleroderma and other fibrotic disease characterized by excessive TGF-␤ responses. Taken together with recent reports, our observations suggest that antagonistic signals regulating cell functions are integrated by distinct mechanisms that are context-and lineage-specific.