The Early-Immediate Gene EGR-1 Is Induced by Transforming Growth Factor-β and Mediates Stimulation of Collagen Gene Expression*

Transforming growth factor-β (TGF-β) stimulates collagen synthesis and accumulation, and aberrant TGF-β signaling is implicated in pathological organ fibrosis. Regulation of type I procollagen gene (COL1A2) transcription by TGF-β involves the canonical Smad signaling pathway as well as additional protein and lipid kinases, coactivators, and DNA-binding transcription factors that constitute alternate non-Smad pathways. By using Affymetrix microarrays to detect cellular genes whose expression is regulated by Smad3, we identified early growth response factor-1 (EGR-1) as a novel Smad3-inducible gene. Previous studies implicated Egr-1 in cell growth, differentiation, and survival. We found that TGF-β induced rapid and transient accumulation of Egr-1 protein and mRNA in human skin fibroblasts. In transient transfection assays, TGF-β stimulated the activity of the Egr-1 gene promoter, as well as that of a minimal Egr-1-responsive reporter construct. Furthermore, TGF-β enhanced endogenous Egr-1 interaction with a consensus Egr-1-binding site element and with GC-rich DNA sequences of the human COL1A2 promoter in vitro and in vivo. Forced expression of Egr-1 by itself caused dose-dependent up-regulation of COL1A2 promoter activity and further enhanced the stimulation induced by TGF-β. In contrast, the TGF-β response was abrogated when the Egr-1-binding sites of the COL1A2 promoter were mutated or deleted. Furthermore, Egr-1-deficient embryonic mouse fibroblasts showed attenuated TGF-β responses despite intact Smad activation, and forced expression of ectopic EGR-1 in these cells could restore COL1A2 stimulation in a dose-dependent manner. Taken together, these findings identify Egr-1 as a novel intracellular TGF-β target that is necessary for maximal stimulation of collagen gene expression in fibroblasts. The results therefore implicate Egr-1 in the profibrotic responses elicited by TGF-β and suggest that Egr-1 may play a new and important role in the pathogenesis of fibrosis.

Transforming growth factor-␤ (TGF-␤) stimulates collagen synthesis and accumulation, and aberrant TGF-␤ signaling is implicated in pathological organ fibrosis. Regulation of type I procollagen gene (COL1A2) transcription by TGF-␤ involves the canonical Smad signaling pathway as well as additional protein and lipid kinases, coactivators, and DNA-binding transcription factors that constitute alternate non-Smad pathways. By using Affymetrix microarrays to detect cellular genes whose expression is regulated by Smad3, we identified early growth response factor-1 (EGR-1) as a novel Smad3-inducible gene. Previous studies implicated Egr-1 in cell growth, differentiation, and survival. We found that TGF-␤ induced rapid and transient accumulation of Egr-1 protein and mRNA in human skin fibroblasts. In transient transfection assays, TGF-␤ stimulated the activity of the Egr-1 gene promoter, as well as that of a minimal Egr-1-responsive reporter construct. Furthermore, TGF-␤ enhanced endogenous Egr-1 interaction with a consensus Egr-1-binding site element and with GC-rich DNA sequences of the human COL1A2 promoter in vitro and in vivo. Forced expression of Egr-1 by itself caused dose-dependent up-regulation of COL1A2 promoter activity and further enhanced the stimulation induced by TGF-␤. In contrast, the TGF-␤ response was abrogated when the Egr-1-binding sites of the COL1A2 promoter were mutated or deleted. Furthermore, Egr-1-deficient embryonic mouse fibroblasts showed attenuated TGF-␤ responses despite intact Smad activation, and forced expression of ectopic EGR-1 in these cells could restore COL1A2 stimulation in a dose-dependent manner. Taken together, these findings identify Egr-1 as a novel intracellular TGF-␤ target that is necessary for maximal stimulation of collagen gene expression in fibroblasts. The results therefore implicate Egr-1 in the profibrotic responses elicited by TGF-␤ and suggest that Egr-1 may play a new and important role in the pathogenesis of fibrosis.
Tissue fibrosis, the pathological hallmark of scleroderma/ systemic sclerosis, pulmonary fibrosis, glomerulosclerosis, and other chronic diseases, is a major determinant of morbidity and mortality (1). Diverse fibrotic conditions are characterized by excessive synthesis and accumulation of collagen and other extracellular matrix (ECM) 2 proteins in target organs, resulting in progressive disruption of their architecture. Currently, there are no effective therapies to arrest or reverse the process of fibrosis. Furthermore, despite its enormous clinical impact, the pathogenesis of fibrosis remains poorly understood. A major feature of all forms of fibrosis is fibroblast activation exemplified by the induction of genetic programs for ECM synthesis, resistance to apoptosis, up-regulation of cytokine receptor and adhesion molecule expression, and transformation into contractile myofibroblasts.
The multifunctional cytokine transforming growth factor-␤ (TGF-␤) is a key signal for regulation of connective tissue metabolism. TGF-␤ exerts potent stimulatory effects on collagen synthesis, ECM accumulation, fibroblast proliferation, and myofibroblast differentiation in vivo and in vitro and is implicated in normal wound healing on the one hand and in the pathogenesis of fibrosis on the other hand (2). Although the interaction of TGF-␤ with its cell surface receptors has been extensively investigated, the precise molecular events downstream from the activated receptors that mediate profibrotic responses elicited by TGF-␤ remain to be fully characterized. Signaling by TGF-␤ is initiated when a type II TGF-␤ receptor (T␤RII) at the cell surface binds its ligand and phosphorylates the activin-like kinase 5 (ALK5) type I TGF-␤ receptor (3). Activated ALK5 in turn phosphorylates cytoplasmic Smad2/3, promoting their heterodimerization with the common mediator Smad4 and rapid translocation into the nucleus. Within the nucleus, the Smad2/3/4 complex interacts directly with consensus Smad-binding elements (SBE) of target genes to activate their transcription (4). Nuclear coactivators such as p300/CBP interact with DNA-bound Smads to positively modulate TGF-␤-induced transcriptional responses (5). Recent studies from our laboratory and others have revealed that Smad3 plays a pivotal role in mediating TGF-␤-induced stimulation of collagen gene expression in normal fibroblasts (6 -9).
Contrary to initial predictions that envisioned the Smad pathway as a relatively straightforward linear system for propagating TGF-␤ responses, intracellular TGF-␤ signaling is turning out to be a remarkably complex, dynamic, and contextdependent biological process. For example, recent studies indicate that in addition to the canonical Smad pathway, TGF-␤ also induces alternate signal transduction pathways in cell typespecific manner. Non-Smad signaling mediators shown to be activated by TGF-␤ in various cell types include protein and lipid kinases such as protein kinases A and C, c-Abl, calmodulindependent protein kinase II, the MAPKs, ERK, JNK, and p38, TAK1, and phosphatidylinositol 3-kinase (reviewed in Ref. 10). Each appears to contribute to translating TGF-␤ signals into distinct biological responses in a cell type-and target genespecific manner. These alternate signaling pathways may mediate TGF-␤ responses completely independent of Smads, or they may modulate the amplitude and duration of ligand-dependent Smad activation or engage with the Smad pathway through intracellular cross-talk. In turn, Smads can modulate the expression and activity of other signaling pathways or induce additional transcription factors. In light of these emerging observations indicating a high degree of complexity, a better understanding of the interactions of Smads with other signaling proteins, including those that may be downstream targets of Smads and contribute to TGF-␤ signaling, is of clear significance.
To gain a more complete picture of intracellular signaling networks mediating TGF-␤ responses, we examined the expression profile of genes whose induction was rapidly up-regulated by Smad3 in human skin fibroblasts. We took advantage of a novel experimental strategy for tightly regulated Smad3 expression in transfected primary cells. The results indicated that the zinc finger transcription factor early response gene-1 (EGR-1), also known as KROX24, NGF1-A, or ZIF268, was rapidly induced by Smad3. Furthermore, TGF-␤ also induced Egr-1 expression in normal skin fibroblasts. The levels of Egr-1 protein and mRNA were rapidly and transiently up-regulated by TGF-␤1 in vitro, and Egr-1 promoter activity was enhanced in transiently transfected fibroblasts. Ectopic Egr-1 expression by itself resulted in transactivation of COL1A2 and stimulation of type I collagen synthesis, whereas mutant Egr-1 blocked the response. Stimulation by TGF-␤ involved ligand-dependent interaction of cellular Egr-1 with cis-acting GC-rich DNA elements partially overlapping functional Sp1-binding sites within the proximal COL1A2 promoter and was attenuated in cells lacking Egr-1. The in vivo expression of Egr-1 in skin fibroblasts was markedly elevated by TGF-␤ injection in mice. Together, these results identify Egr-1 as a novel TGF-␤/Smad3 target that up-regulates the expression of collagen genes and plays a critical role in mediating TGF-␤ stimulation. Therefore, Egr-1 appears to be a new and important component of the fibrotic process.

MATERIALS AND METHODS
Cell Cultures and Metabolic Labeling-Cultures of human dermal fibroblasts were established from neonatal foreskins by explant techniques described previously (11). Only early passage (Ͻ8) fibroblasts were studied. Egr-1-null murine embryonic fibroblasts (MEFs) provided by J. Milbrandt (Washington University, St. Louis) were established from embryos of mice with targeted deletion of the Egr-1 gene (12). NIH3T3 mouse fibroblasts and HepG2 cells were obtained from the American Type Culture Collection (Manassas, VA). Cells were routinely maintained in modified Eagle's medium or Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 1% vitamins, antibiotics, and 2 mM L-glutamine and studied at early confluence. To determine the role of Egr-1 in regulation of collagen synthesis, wild-type and Egr-1-deficient MEFs were used in metabolic labeling experiments, as described previously (11). Briefly, confluent wild-type and Egr-1-null MEFs in parallel were incubated in serum-free media supplemented with ascorbic acid (50 g/ml) and ␤-aminoproprionitrile (100 g/ml). Twenty four h later, TGF-␤ (12.5 ng/ml) and [ 14 C]proline (2.5 Ci/ml) were added to the cultures, for the final 24 h. The media were removed and extensively dialyzed, and aliquots normalized for cell numbers were examined by SDS-PAGE. Enhanced gels were visualized by autoradiography, and protein levels were quantitated using Molecular Analyst Image densitometry software (Bio-Rad).
Inducible Smad3 Expression in hTERT-derived Fibroblasts-To construct a tightly regulated inducible system for Smad3 expression, hTERT-BJ1 cells derived from human foreskin fibroblasts stably expressing the human telomerase reverse transcriptase (Clontech) were used. The cells were first transfected with the murine ecotropic receptor, which renders them susceptible to infection with retrovirus, and the modified bacterial lacI repressor, which allows for isopropyl ␤-thiogalactoside (IPTG)-inducible expression of promoters coupled with lac operator sequences (13).
The IPTG-inducible LNXCO3 expression construct contains the retroviral long terminal repeat and three lac operators in a ␤-galactoside regulated promoter. Full-length Smad3 cDNA was generated by PCR (primers 5Ј-GTAAGATCTCAG-CCATGTCGTCCATCCT-3Ј and 5Ј-ATAGCGGCCGCGTC-TCTAAGACACACTG-3Ј), and the coding sequence was digested with NotI and HpaI and cloned into LNXCO3 to yield Smad3-LNXCO3. Upon addition of IPTG, the repressor is removed, and the expression of the insert is induced. Insert-free LNXCO3 vector was used for control infections. Smad3-LNXCO3 and LNXCO3 plasmids were used for retroviral transduction of BOSC23 ecotropic virus packaging cell line. Supernatant were used to infect hTERT-BJ1-LSR1 subline fibroblasts. Smad3-LNXCO3 and LNXCO3-transduced fibroblast populations were then selected with G418 (600 g/ml).
cDNA Array Hybridization and Gene Expression Analysis-At confluence, transduced hTERT-BJ1 fibroblasts were placed in serum-free media with 0.1% bovine serum albumin for 24 h and then exposed to 100 M IPTG (Invitrogen) for 4 h. Total RNA was isolated using the RNeasy kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Agarose gel electrophoresis was performed to monitor the quality of mRNA. Carefully examined aliquots in triplicate were used for preparation of probes and hybridization to HG-U133A GeneChip Human Genome Arrays (Affymetrix, Santa Clara, CA), rep-resenting ϳ11,000 human genes, according to protocols supplied by Affymetrix. GeneChip arrays were scanned by confocal scanner, and data were processed using software supplied by Affymetrix. A 2-fold difference in hybridization intensity was used as a cutoff threshold for genes up-regulated in IPTG-treated fibroblasts.
RT-PCR and Northern Blot Analysis-Total RNA was isolated from confluent fibroblasts using TRIzol reagent (Invitrogen), and integrity of RNA was determined by agarose gel electrophoresis and ethidium bromide staining. For RT-PCR, 1 g of total RNA was reverse-transcribed to first strand cDNA using avian myeloblastosis virus-reverse transcriptase system (Promega, Madison, WI) and subjected to PCR amplification with the following primers: human Egr-1 forward, 5Ј-GACAG-CAACCTTTTCTCCCAGG-3Ј, and reverse, 5Ј-GTTAGGTC-CTCACTTGGGGGAA-3Ј, yielding a 545-bp fragment; and actin forward, 5Ј-ATCTGGCACCACACCTTCTACAATG-AGCTGCG-3Ј, and reverse, 5Ј-CGTCATACTCCTGCTTG-CTGATCCACATCTGC-3Ј, yielding an 833-bp fragment. Egr-1 and actin were amplified together; the PCR products were separated by electrophoresis in 1.5% agarose gels containing ethidium bromide, and bands were visualized using Kodak imaging system. For Northern analysis, RNA was prepared and examined as described (6), using [␣-32 P]dCTP-labeled cDNA probes for human Egr-1, COL1A2, 18 S, glyceraldehyde-3phosphate dehydrogenase, and mouse COL1A1. Following extensive washing of the nitrocellulose membranes, cDNA-mRNA hybrids were visualized by autoradiography on Kodak BioMax film. Signal intensities were quantitated by densitometry, and results were normalized with the RNA levels for glyceraldehyde-3-phosphate dehydrogenase or 18 S in each sample.
Confocal Microscopy-The expression and subcellular localization of native Smad3 and Egr-1 was studied by immunocytochemistry, as described (14). Briefly, fibroblasts (10,000 cells/well) were seeded into 8-well Lab-Tek II chamber glass slides (Nalge Nunc International, Naperville, IL). Following incubation in serum-free media with IPTG (100 M) or TGF-␤1 (1 ng/ml), the cells were fixed with 100% methanol, washed, and incubated with DAPI for nucleus or 10 g/ml primary antibodies to Smad3 or Egr-1 or IgG as control, followed by horseradish peroxidase-conjugated secondary antibodies (all from Santa Cruz Biotechnology). Cells were stained according to the manufacturer's protocol (Tyramide Signal Amplification; PerkinElmer Life Sciences), and fluorescence intensity and subcellular localization were examined by laser scanning confocal microscopy using a Zeiss LSM510 microscope. The data were analyzed using Release 3.2 Zeiss LSM software and plotted as two-dimensional scatterplots to determine Pearson's correlation coefficient, with Ϫ1.0 representing negative correlation, 0 no correlation, and 1.0 complete correlation.
Transient Transfection Assays-Several reporter constructs were used in transient transfection assays. Constructs harboring the truncated COL1A2 promoter with 5Ј ends at Ϫ772, Ϫ353, and Ϫ108 bp fused to chloramphenicol acetyltransferase (CAT) were described previously (15). Site-directed mutagenesis was performed using the QuickChange mutagenesis kit (Stratagene, La Jolla, CA) and involved substitution mutations to modify the Egr-1-binding site (EBS) sequences of the Ϫ772COL1A2 promoter. Sequencing confirmed that each construct was correct. The minimal Egr-1-responsive promoter plasmid pEBS1 4 -luc containing four binding sites for Egr-1 was provided by G. Thiel (University of Saarland, Germany) (16). The pEgr-1-luc construct containing 1.2 kb of the Egr-1 promoter was provided by E. Adamson (Burnham Institute, La Jolla, CA). Cells at early confluence were transfected with the reporter constructs along with the Egr-1 expression vector pCMV-Egr-1 or pCMV-Egr-1⌬mt (lacking the first and part of the second zinc fingers) provided by M. Goldring (Beth Israel Deaconess Medical Center, Harvard Medical School, Boston) using calcium phosphate or SuperFect Transfection kit (Qiagen, Valencia, CA) as described (6). Transfected cells were incubated in serum-free media containing 0.1% bovine serum albumin for 24 h, followed by various concentrations of TGF-␤1 for another 24 h. At the end of the incubations, cells were harvested, and the cell lysates, normalized for protein concentrations, were assayed for CAT activities, as described (6). Luciferase activities were assayed using the dual-luciferase reporter assay system (Promega, Madison, WI). In some experiments, a construct containing the Renilla luciferase pRL-TK (Promega) was cotransfected along with the reporter constructs and used as control for transfection efficiency. Transient transfection experiments were performed in triplicate and repeated at least twice.
Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays (EMSA) were performed as described previously (6). Briefly, confluent cultures of serum-starved fibroblasts were incubated with TGF-␤1 (12.5 ng/ml) for the indicated periods, and nuclear extracts were prepared. Double-stranded synthetic DNA oligonucleotides (IDX Technologies, Coralville, IA) were end-labeled using T4 polynucleotide kinase and [␥-32 P]ATP and incubated with nuclear extracts (5 g) for 30 min. Radiolabeled oligonucleotides corresponding to COL1A2 promoter sequences spanning Ϫ147 to Ϫ118 bp (EBS1) and Ϫ310 to Ϫ285 bp (EBS2) were used as probes, along with oligonucleotides containing consensus and mutant Egr-1 and Sp1 sequences (Santa Cruz Biotechnology). To establish the specificity of transcription factor binding to the DNA probes, competition experiments using 100-fold molar excess of unlabeled oligonucleotides were performed. For supershift experiments, antibodies specific for Egr-1 or Sp1, Sp3, and preimmune IgG (Santa Cruz Biotechnology) were added to nuclear extracts in binding reaction mixture at 4°C for 60 min before addition of radiolabeled oligonucleotide probes and electrophoresis. All experiments were repeated at least three times.
Chromatin Immunoprecipitation (ChIP) Assays-Chromatin immunoprecipitation assays were performed using the ChIP assay kit (Upstate Biotechnology, Inc.) following the manufacturer's instructions. Briefly, skin fibroblasts were incubated in the presence or absence of TGF-␤1 (12.5 ng/ml) for 60 min and harvested. Formaldehyde was added to the cultures (final concentration 1%) to cross-link chromatin. Cell lysates were then prepared and sonicated on ice to break chromatin DNA to an average length of ϳ500 bp. Aliquots of lysates containing 100 -200 g of protein were used for immunoprecipitation of Egr-1-chromatin complexes with antibodies (5 g) to Egr-1 (588), Sp1 (C-59), and Sp3 (D-20) (all from Santa Cruz Biotechnology) or normal rabbit IgG. DNA was recovered, and PCR amplification of the captured sequences was performed using primers complementary to the COL1A2 promoter region harboring both putative Egr-1-binding sites. The primer sequences were sense primer, 5Ј-CTACAGGG-CACAGGTGAGG-3Ј, and antisense primer, 5Ј-AAAGCC-CGGATCTGCCCTA-3Ј, to generate a 422-bp amplification product. DNA samples were analyzed by electrophoresis in 2% agarose gels stained with ethidium bromide.
Egr-1 Expression in Vivo-Six-week-old female C3H mice (Taconic Farms, Germantown, NY) weighing 20 -25 g received a subcutaneous injection of TGF-␤1 (250 ng) or phosphatebuffered saline into the shaved back. Twenty four h later, mice (three per group) were euthanized, and the injected skin was removed and processed for histological examination (18) or immunohistochemical analysis (19). Sections were incubated with antibodies to Egr-1 (C-19) (Santa Cruz Biotechnology) at a final dilution of 1:50. Substitution of the primary antibody with irrelevant IgG and preincubation of primary antibody with blocking peptide served as negative controls. Staining was repeated for each sample at least three times. Fibroblasts were identified based upon their characteristic spindle-shaped morphology. A minimum of 30 cells in several microscopic fields in each slide at ϫ400 magnification was scored as positive or negative for Egr-1 by three independent examiners blinded to the treatment. The ratio of positive cells/total number of cells was calculated. Mice were cared for in compliance with the princi-ples of the National Institutes of Health/Association for Assessment and Accreditation of Laboratory Animal Care, and with the approval of the animal care review committee of the University of Illinois.
Statistical Analysis-Experiments were repeated at least three times with consistent results. Statistical significance was determined using the unpaired Student's t test. Values of p Յ 0.05 were considered to be significant.

Identification and Validation of EGR-1 as a Novel SMAD3
Target Gene in Fibroblasts-To identify cellular genes whose expression is directly induced by Smad3 in normal fibroblasts, we established a novel system for tightly regulated Smad3 expression. For this purpose, an IPTG-inducible retroviral vector containing SMAD3 or insertless vector was used in parallel to stably transfect human fibroblasts immortalized with telomerase reverse transcriptase (hTERT). Under routine culture conditions, hTERT fibroblasts maintained normal skin fibroblast characteristics, including spindle-shaped morphology, TGF-␤ induction of ␣-smooth muscle actin expression, and synthesis of type I collagen, for up to 80 serial passages in vitro (data not shown), and have an infinite life span (20). The hTERT-BJ1-LNXCO-Smad3 subline is a derivative of hTERT fibroblasts where Smad3 expression can be turned on using the physiologically neutral agent IPTG. Nontransfected hTERT fibroblasts, as well as hTERT fibroblasts transfected with Smad3 or empty vector, displayed detectable levels of native Smad3 (Fig. 1A). However, exposure of hTERT-BJ1-LNXCO3-Smad3 fibroblasts to IPTG resulted in substantial and specific induction of Smad3, detectable as early as 4 h (data not shown) and showing time-dependent accumulation with maximal 5-fold increase in Smad3 levels at 24 h (Fig. 1A). Furthermore, increased Smad3 accumulation in IPTG-induced fibroblasts was associated with its phosphorylation and nuclear accumulation, whereas no Smad3 activation could be demonstrated in hTERT fibroblasts transfected with empty vector (Fig. 1, A and B). In contrast to Smad3, the levels of endogenous Smad4 in Smad3-transfected fibroblasts were not modulated by IPTG, thus serving as a useful additional control.
Next, total RNA was isolated from Smad3-or insertless vector-transfected hTERT fibroblasts incubated with IPTG for 4 h, a time point chosen because preliminary experiments established it as the earliest point when Smad3 protein accumulation became evident, and subjected to analysis using Affymetrix U133A GeneChip Arrays. Global gene expression patterns are shown in Fig. 1C. The results identified several genes whose expression was induced Ն2-fold in Smad3-expressing hTERT fibroblasts at this early time point. Genes showing early up-regulation by Smad3 included EGR-1, calreticulin, RAB14, MIF, and SGK, as well as CYR61, SMAD7, and CTGF, which had been shown previously to be Smad/TGF-␤-inducible genes in fibroblasts (21,22). Of these, EGR-1 was initially chosen for further study, because this inducible zinc finger transcription factor was originally identified as an early-immediate gene induced in fibroblasts by a variety of environmental signals, and had been

Egr-1 Regulation of Type I Collagen Gene
shown to play important roles in inflammatory and hypoxic responses, wound healing, and vascular injury.
First, we confirmed that IPTG-dependent Smad3 induction in hTERT fibroblasts was associated with increased Egr-1 mRNA expression by RT-PCR analysis. The results using RNA preparations from independent experiments validated the array results, indicating a substantial increase in Egr-1 mRNA levels in hTERT-BJ1-Smad3 fibroblasts treated with IPTG for 4 h, with some elevation persisting at 24 h (Fig. 1D). In contrast, no induction of Egr-1 mRNA by IPTG could be demonstrated in fibroblasts transduced with empty vector in parallel (data not shown). Western blot analysis of whole cell lysates indicated time-dependent accumulation of Egr-1 protein induced by IPTG in hTERT-BJ1-Smad3 fibroblasts, with a Ͼ2-fold increase at 4 h and a 5-fold increase at 24 h (Fig. 1E). Egr-1 was predominantly localized within the nuclear fraction (data not shown).
For this purpose foreskin fibroblasts were made quiescent by serum starvation and then stimulated with TGF-␤1 for up to 4 h. Total RNA was extracted and examined by Northern blot analysis. The results showed that TGF-␤1 induced a rapid increase in Egr-1 mRNA levels in these cells ( Fig. 2A). A maximal 3-fold induction was noted at 30 min, followed by a progressive decline and a return to basal levels by 120 min. Western blot analysis of cytosolic and nuclear fractions using polyclonal antibody to Egr-1 revealed a progressive time-dependent accumulation of Egr-1 protein (82 kDa) that peaked at 30 min and persisted for up to 120 min (Fig. 2B). Furthermore, TGF-␤ induced the phosphorylation of native Egr-1, with maximal effect noted at 60 min (Fig. 2C). Nuclear accumulation of Egr-1 upon TGF-␤ treatment of the fibroblasts was confirmed by quantitative immunofluorescence microscopy (Fig. 2D).
Next, the in vivo effect of TGF-␤ on Egr-1 protein expression in the skin was examined. Young C3H mice received a single subcutaneous injection of TGF-␤1 or phosphatebuffered saline in parallel, and 24 h later the lesional skin was excised and examined by immunohistochemistry. The results showed prominent Egr-1 accumulation in the skin from TGF-␤-injected mice (Fig. 2E). Strong intracellular Egr-1 staining was evident in fibroblastic cells and appeared to be largely localized within the nucleus. In contrast, little Egr-1 could be detected in the dermis of mice that received phosphate-buffered saline injections. Quantitative examination of multiple microscopic fields indicated that the number of Egr-1-positive fibroblasts was 2.5-fold greater in the dermis from TGF-␤-injected compared with control mice (Fig. 2E, lower panel ).
To examine if TGF-␤-induced stimulation of Egr-1 synthesis was associated with changes in Egr-1 DNA binding activity, EMSAs were performed. For this purpose, nuclear extracts were prepared from fibroblasts treated with TGF-␤ for the indicated periods and analyzed by EMSAs using radiolabeled Egr-1 consensus oligonucleotide probes. Compared with untreated fibroblasts, fibroblasts exposed to TGF-␤ showed markedly enhanced DNA-protein complex formation that was maximal at 30 min and persisted at 120 min (Fig. 3, left panel ). The specificity of the TGF-␤-induced complex was confirmed by its disappearance in the presence of excess unlabeled consensus Egr-1 but not mutated Egr-1 FIGURE 1. Egr-1 expression is induced by Smad3. Confluent hTERT fibroblasts stably transfected with inducible Smad3 or empty vector were incubated in serum-free media with or without IPTG (100 M). A, after 24 h, whole cell lysates were examined by Western analysis using the indicated antibodies. B, after 4 h, fibroblasts were fixed, stained by DAPI or fluorescein isothiocyanate-labeled antibodies to Smad3, and examined by immunofluorescence confocal microscopy. Merged images are shown. C, total RNA isolated following 4 h of incubation was applied to Affymetrix HU-133A GeneChip arrays. Two-dimensional scatterplot of gene expression values for all genes on the chips is shown. Closed arrowhead, Egr-1; open arrowhead, Smad3. D, RNA from the same samples was subjected to RT-PCR using primers specific for EGR-1 (545 bp) or actin (833 bp), followed by gel electrophoresis. Signal intensities determined by densitometry are shown as means Ϯ S.E. from several independent determinations (lower panel ). E, whole cell lysates were examined by Western analysis with the indicated antibodies. Signal intensities from several experiments were determined by densitometry (lower panel ). JULY 28, 2006 • VOLUME 281 • NUMBER 30 oligonucleotide competitors (Fig. 3, middle panel ). Furthermore, preincubation of nuclear extracts with increasing amounts of anti-Egr-1 antibodies blocked the formation of the complex, whereas preimmune IgG had no effect (Fig. 3,   right panel). These findings indicated that exposure of normal fibroblasts to TGF-␤ resulted in the activation of cellular Egr-1, evidenced by its increased binding to a consensus DNA element.

Egr-1 Regulation of Type I Collagen Gene
To examine the mechanistic basis for the regulation of EGR-1 gene expression by TGF-␤, transient transfection assays were performed. The promoter of the EGR-1 gene harbors multiple regulatory elements, including AP-1-binding sites, serumresponse elements, potential Smad-binding elements, EBS, and other GC-rich sequences. Serum-starved mouse NIH3T3 fibroblasts were transfected with full-length Egr-1 promoterluciferase reporter constructs and empty vector, followed by incubation with TGF-␤ for 24 h and assay of cell lysates for luciferase activities. The results showed that TGF-␤ induced a 5-fold increase in Egr-1 promoter activity in transfected fibroblasts (Fig. 4). Taken together, the results from these studies indicated that TGF-␤ induced a rapid and transient increase in Egr-1 mRNA and protein expression in dermal fibroblasts in vitro and in vivo, and the response was mediated at least in part through the stimulation of EGR-1 gene transcription.
Egr-1 Stimulates COL1A2 Promoter Activity and Collagen Synthesis-To examine the effect of Egr-1 on collagen regulation, foreskin fibroblasts were transfected with expression plasmids for wild-type Egr-1 or a mutated Egr-1 lacking one and a half-zinc fingers (Egr-1⌬mt), along with the 772COL1A2-CAT reporter construct. Following 24 h of incubation of the cultures in serum-free media in the presence or absence of TGF-␤, the transfected cells were harvested and cell lysates were assayed for their CAT activities. As shown in Fig. 5A, overexpression of Egr-1 in these fibroblasts was able to enhance COL1A2 promoter activity in a dose-dependent manner in the absence of exogenously added TGF-␤, indicating that ligand was not required to activate the ectopically expressed Egr-1 protein. In contrast, mutated Egr-1 or empty vector failed to stimulate COL1A2 promoter activity. Transient expression of ectopic Egr-1 resulted in stimulation of type I procollagen synthesis by normal fibroblasts (Fig. 5B, upper panel). Next, transfected fibroblasts were incubated with TGF-␤ for 24 h. The results shown in Fig. 5B (lower panel) indicate that Egr-1-induced stimulation of COL1A2 promoter activity was further enhanced in the presence of TGF-␤1. In contrast, a mutant Egr-1 expression plasmid partially abrogated the TGF-␤ response, suggesting that endogenous Egr-1 was required.
A computer-assisted analysis of the proximal COL1A2 promoter region revealed the presence of two potential binding sites for Egr-1. The more proximal of these, termed EBS1, is located at position Ϫ141 to Ϫ133 bp adjacent to a previously described Sp1-binding site; and the more distal EBS2 at position Ϫ303 to Ϫ295 bp, is flanked by two Sp1-binding sites. A schematic diagram of the proximal region of the human COL1A2 promoter is shown in Fig. 5C, indicating the localization of the two EBS and a consensus Smad-binding element within this region; the nucleotide sequences of the EBS sites are underlined, with the substitution mutations used in EMSA probes and transfection constructs indicated in lowercase. To examine the functional role of these putative Egr-1 recognition sites in mediating transactivation induced by Egr-1, COL1A2  . TGF-␤ activates nuclear protein binding to consensus Egr-1 DNA sequences. Nuclear extracts were prepared from fibroblasts exposed to TGF-␤ for the indicated periods, and equal aliquots of proteins were used in electrophoretic mobility shift assays with radiolabeled double-stranded oligonucleotides harboring the consensus Egr-1 sequences. Left panel, time-dependent DNA-protein complex formation. Middle panel, nuclear extracts (TGF-␤1 for 60 min) were examined using indicated unlabeled excess oligonucleotides as competitors. Right panel, antibody (Ab) interference. Nuclear extracts were preincubated with antibodies to Egr-1. PI, preimmune IgG.  JULY 28, 2006 • VOLUME 281 • NUMBER 30

Egr-1 Regulation of Type I Collagen Gene
promoter constructs harboring 5Ј truncations linked to the CAT reporter were cotransfected along with Egr-1 expression plasmid. The results from multiple transfection experiments indicated that deletion of both EBS1 and EBS2 led to complete loss of Egr-1-induced transactivation (Fig. 5D). Next, substitution mutations were introduced in the EBS sequences of the 772COL1A2-CAT construct by site-directed mutagenesis, resulting in disruption of Egr-1 binding in EMSAs (Fig. 6B, right  panel). The mutant constructs were then used in transient transfection assays. The results showed that disruption of either EBS1 or EBS2 resulted in partial loss of Egr-1 transactivation, whereas disruption of both completely abrogated this response (Fig. 5D).

TGF-␤ Induces Egr-1 Binding to COL1A2 Egr-1-binding Elements-
The ability of TGF-␤ to induce expression of Egr-1 in fibroblasts, coupled with the presence of two functional EBS sequences within the COL1A2 promoter, and the ability of mutant Egr-1 to block TGF-␤-induced transactivation together suggested that Egr-1 could contribute to the transcriptional activation of the collagen gene elicited by TGF-␤ in fibroblasts. To explore this notion, we first used EMSAs to examine how the binding of endogenous Egr-1 to its putative DNA recognition sites within the COL1A2 promoter was regulated by TGF-␤. Nuclear extracts from TGF-␤1-treated fibroblasts were incubated with radiolabeled EBS1 or EBS2 oligonucleotides, followed by electrophoresis. The results of EMSAs revealed that TGF-␤ rapidly induced the formation of two distinct complexes with the labeled EBS2 probe (Fig. 6A, left  panel). The complexes disappeared in the presence of excess unlabeled wild-type, but not mutated, consensus Egr-1 sequence oligonucleotides (Fig. 6A, middle panel), and by interference assays with antibodies to Egr-1 (Fig. 6A, right panel). Supershifting with antibody to Sp1 (Fig. 6A, right panel, lane 7) reduced the intensity of two of the bands, and antibody to Sp3 (lane 8) produced a strong supershift.
Next, labeled oligonucleotides harboring the Egr-1 consensus sequence were used as probes. The results showed that nuclear extracts from TGF-␤-stimulated fibroblasts formed a DNA-protein complex (Fig. 6B, left panel) that could be effectively competed away by unlabeled excess consensus Egr-1 (Fig.  3, middle panel) or EBS2 oligonucleotides (Fig. 6B, left panel,  lanes 3-5) but not by mutated Egr-1 oligonucleotides (lanes 6 -8). The complex was also competed away by the m8 mutant EBS2 with disrupted Sp1 site (Fig. 6B, right panel, lane 5) but not by EBS2 mutants with disrupted Egr-1 sites (lanes 2 and 4).
As shown in Fig. 5C, the EBS2 site of the COL1A2 promoter is flanked by two partially overlapping GC-rich Sp1-binding sites, and together with a more 3Ј Smad-binding element, this region spanning Ϫ313/Ϫ255 has been implicated previously in mediating TGF-␤ stimulation (23). The potential contribution of Sp1 to complex formation with EBS2 oligonucleotide probes was examined by gel shift assays using labeled Sp1 consensus oligonucleotide probes. The results showed that nuclear extracts from TGF-␤-treated fibroblasts formed multiple complexes of differing mobilities with labeled Sp1 consensus oligonucleotide probes (Fig. 6C). This is consistent with the complex binding pattern of Sp1 family members (24,25). Excess unlabeled EBS2 (Fig. 6C, lanes 2-4) and consensus Sp1 (lanes 5-7)

Egr-1 Regulation of Type I Collagen Gene
but not mutated Sp1 (lane 14) oligonucleotides were effective in ablating binding activity, whereas the m8 oligonucleotide (lanes 8 -10 and 13) that harbors substitutions disrupting the Sp1 site failed to compete effectively. Together, these findings indicate that the EBS2 site of COL1A2 included functional recognition sites for Egr-1 as well as Sp1 family members, and DNA complexes formed by these transcription factors migrated close to one another.
To further confirm the interactions of Egr-1 and Sp1/Sp3 with the EBS2 region of the COL1A2 promoter and to investigate their modulation by TGF-␤, DAPA experiments were performed. For this purpose, biotinylated oligonucleotides corresponding to EBS2 were incubated with nuclear extracts, and DNA-protein complexes were recovered and examined by Western analysis. The results clearly demonstrated the presence of Egr-1, Sp1, and Sp3 in the precipitated complex from untreated fibroblasts (Fig. 6D). Nuclear extracts from TGF-␤stimulated fibroblasts showed an increase in the quantity of Sp1 and Egr-1 in the complexes precipitated by the EBS2 sequence, whereas Sp3 remained unchanged. These results further suggest that TGF-␤ stimulation of fibroblasts was accompanied by accumulation of Egr-1 and Sp1 in EBS2-associated complexes.
To determine whether direct binding of Egr-1 to COL1A2 enhancer sequences occurred in intact cells in vivo, chromatin was cross-linked with formaldehyde, followed by immunoprecipitation with antibodies to Egr-1. Endogenous Egr-1 levels were induced in foreskin fibroblasts by stimulation with TGF-␤ for 60 min under serum-free conditions. Following cross-linking, chromatin was sonicated, and PCR was used to detect COL1A2 promoter-specific DNA that had been pulled down by antibodies to Egr-1, Sp1, or Sp3. The results of ChIP assays indicated that the DNA fragment bracketed by the PCR primers contained functional recognition sites that bound Egr-1 in TGF-␤-stimulated live cells, because the anti-Egr-1 antibody immunoprecipitated DNA fragments specifically detected by PCR (Fig. 7A). Furthermore, immunoprecipitation with antibodies to Sp1/Sp3 pulled down the same DNA fragment. Consistent with the results from DAPA experiments, clear constitutive binding of Sp3 to the DNA probe was observed, without further enhancement in TGF-␤-stimulated fibroblasts (Fig.  7A). The inability of nonimmune IgG to precipitate EBS-containing DNA fragments provided a negative control. These results indicated that Egr-1, along with Sp1 family members, can directly bind to EBS consensus sequence-containing DNA fragment from the COL1A2 promoter, and it does so in live cells induced by TGF-␤ to express endogenous Egr-1.
TGF-␤ Induces Egr-1 Transcriptional Activity-To examine if induction of Egr-1 synthesis by TGF-␤ resulted in transcriptionally active Egr-1, we examined the regulation of an Egr-1responsive minimal promoter using the reporter plasmid pEBS1 4 -luc, which contains four binding sites for Egr-1 derived from the Egr-1 gene promoter (16). HepG2 cells were transiently transfected with pEBS1 4 -luc construct and incubated with TGF-␤ for 24 h. The results of transfection assays showed that TGF-␤ induced a dose-dependent increase in EBS-driven luciferase activity (Fig. 7B), indicating that transcriptionally active Egr-1 was produced in the transfected HepG2 cells as a result of TGF-␤ treatment. . TGF-␤ enhances Egr-1 DNA binding to COL1A2 promoter. Equal aliquots of nuclear extracts from foreskin fibroblast incubated with TGF-␤1 for indicated periods were used in electrophoretic mobility shift assays (A-C ) or DNA affinity precipitation assays (D). A, radiolabeled EBS2 oligonucleotides were used as probes. Left panel, time-dependent DNA-protein complex formation. Middle panel, unlabeled oligonucleotides harboring consensus or mutated Egr-1-binding site sequences were used as competitors. Right panel, antibody interference. Nuclear extracts (TGF-␤1 for 60 min) were preincubated with antibodies (Ab) to Egr-1, Sp1, and Sp3, or preimmune IgG (PI) (middle panel ). SSp1, supershifted Sp1; SSp3, supershifted Sp3. B, radiolabeled oligonucleotides harboring Egr-1 consensus binding sites were used as probes with excess unlabeled EBS2 oligonucleotides or EBS2 mutant oligonucleotides with disrupted Egr-1 (m3 and m4) or Sp1 (m8) sites as competitors. C, labeled Sp1 consensus oligonucleotides were used as probes with excess unlabeled oligonucleotides as competitors. NS, nonspecific. D, DNA affinity precipitation assays. Nuclear extracts from fibroblasts exposed to TGF-␤ for the indicated periods were incubated with biotin end-labeled oligonucleotides harboring the EBS2 sequence or poly(dI/dC), and DNA-bound proteins were isolated and examined by Western analysis with indicated antibodies, as described under "Materials and Methods." JULY 28, 2006 • VOLUME 281 • NUMBER 30 TGF-␤ Stimulation of COL1A2 Promoter Activity through EBS-The two EBS sites of the COL1A2 promoter are located between Ϫ310 and Ϫ118 bp. To directly examine the functional role of Egr-1 in COL1A2 promoter transactivation induced by TGF-␤, fibroblasts were transfected with 772COL1A2-CAT or constructs harboring serial 5Ј truncations, along with Egr-1 expression plasmids, followed by incubation of the cultures with TGF-␤ for 24 h. The results from multiple transfection assays showed that promoter constructs with 5Ј ends extending to Ϫ353 bp displayed a Ͼ3-fold increased activity in fibroblasts exposed to TGF-␤, but deletion of the promoter to Ϫ108, resulting in elimination of both Egr-1-binding sites, was associated with complete loss of stimulation (Fig. 8, upper panel). Next, mutant 772COL1A2-CAT constructs harboring disrupted EBS sequences were used as reporters. The results of transient transfection assays indicated that disrupting Egr-1 DNA binding by site-directed mutagenesis of either EBS1 or EBS2 resulted in partial loss of stimulation of the transfectants by TGF-␤, and disruption of both EBS1 and EBS2 almost completely abolished the TGF-␤ response (Fig. 8, lower  panel), indicating that Egr-1 binding to both recognition sites was involved in full stimulation of COL1A2 promoter activity.

Egr-1 Regulation of Type I Collagen Gene
Reduced Collagen Stimulation in Egr-1-null Fibroblasts-The failure of TGF-␤ to enhance the activity of COL1A2 promoter constructs functionally lacking both EBS sequences suggested a critical role for inducible Egr-1 DNA binding in mediating the stimulatory response to TGF-␤. The functional role of Egr-1 in TGF-␤ stimulation of collagen synthesis was directly examined using Egr-1-deficient embryonic fibroblasts. For this purpose, confluent MEFs derived from Egr-1-null and wild-type mouse embryos were incubated in parallel with TGF-␤ for 24 h. The proliferation rates of both MEF strains were similar, as reported previously (12). At the end of the incubations, fibroblasts were harvested, and mRNA expression was examined by Northern blot analysis. The results showed that although TGF-␤ induced a 60 -80% increase in COL1A1 and COL1A2 mRNA levels in wild-type MEFs, no change in mRNA levels was noted in Egr-1-null MEFs incubated with TGF-␤ (Fig. 9A). The defective TGF-␤ response was not because of impaired activation of the Smad pathway, as Smad2 phosphorylation was induced by TGF-␤ with comparable kinetics in both wild-type and Egr-1deficient fibroblast strains (Fig. 9B). In contrast to collagen, mRNA levels for PAI-1, another important TGF-␤-inducible ECM protein, were increased to a comparable degree in both wild-type and Egr-1-null MEFs incubated by TGF-␤ (data not shown).
The role of Egr-1 in regulation of COL1A2 promoter activity by TGF-␤ was examined in transient transfection assays. The results showed that although treatment of transfected wildtype MEFs with TGF-␤ resulted in a 3.5-fold increase in CAT activity, no induction was observed in Egr-1-null fibroblasts, indicating that despite intact Smad signaling pathways, cellular Egr-1 was required for optimal stimulation of collagen gene expression by TGF-␤ (Fig. 9C). To examine if the TGF-␤ responses could be rescued by ectopic Egr-1, Egr-1-null MEFs were cotransfected with Egr-1 expression vector or empty vector, along with 772COL1A2-CAT, and CAT activities were determined following incubation of the cultures with TGF-␤ for 24 h. The results from multiple transfection assays showed FIGURE 7. TGF-␤ enhances Egr-1 transcriptional activity. A, ChIP assays. Foreskin fibroblasts incubated with TGF-␤1 for 60 min were formaldehyde-cross-linked, and chromatin was immunoprecipitated with antibodies to Egr-1, Sp1, or Sp3 or preimmune IgG, followed by PCR amplification with COL1A2-specific primers. PCR products were analyzed by agarose gel electrophoresis. Input genomic DNA was used as positive control. Results from a representative experiment are shown. B, HepG2 cells were transfected with Egr-1-responsive minimal promoter construct pEBS1 4 -luc. Following incubation with indicated concentrations of TGF-␤1 for 24 h, cells were harvested, and lysates were assayed for luciferase activities. The results of triplicate determinations from two separate experiments, normalized with Renilla luciferase, are shown as the means Ϯ S.E. *, p Ͻ 0.05. that forced expression of Egr-1 was able to restore TGF-␤ stimulation of COL1A2 promoter activity in Egr-1-null MEFs in a dose-dependent manner (data not shown). Next, the regulation of type I collagen synthesis by TGF-␤ was examined in Egr-1null MEFs by metabolic labeling, which allows determination of newly synthesized collagenous proteins. Confluent fibroblasts were incubated with [ 14 C]proline for up to 24 h, and radioactive isotope incorporation into proteins secreted into the culture media was examined by SDS-PAGE. The results showed that although the synthesis of type I procollagen was increased by TGF-␤ Ͼ4-fold compared with untreated controls in wild-type MEFs, only an ϳ2-fold stimulation was noted in Egr-1-null MEFs (Fig. 9D). Similar results were obtained when levels of cellular collagen were examined by Western blot analysis (data not shown).

DISCUSSION
The pleiotropic cytokine TGF-␤ has fundamental roles in embryonic development, cell growth and differentiation, regulation of immune and inflammatory responses, angiogenesis, apoptosis, and tissue repair and regeneration (3). In mesenchymal cells, TGF-␤ is a potent inducer of type I collagen gene transcription, and the expression of many other ECM molecules is also up-regulated, allowing TGF-␤ to function as the master regulator of connective tissue remodeling during physiological tissue repair such as wound healing, but also implicating it in the pathogenesis of fibrotic conditions associated with excessive ECM synthesis and accumulation leading to destruction of affected organs (26).
Although it is now clear that TGF-␤ responses involve a plethora of complex and cell type-specific mechanisms, the Smads have emerged as crucial mediators for propagating TGF-␤ signals inside most cells. Recent studies have elucidated the cellular mechanisms that couple TGF-␤-induced fibroblast stimulation to the up-regulation of collagen gene expression. These results indicated that ligand-dependent phosphorylation of Smad2 and Smad3 results in their nuclear import, interaction with conserved enhancer DNA sequences within the COL1A2 promoter region, recruitment of coactivators such as p300, and induction of transcription (reviewed in Ref. 27). Stimulation of collagen gene expression by TGF-␤ is Smad-dependent, as demonstrated by its abrogation in the presence of Smad7, the antagonist Smad, or by pretreatment of fibroblasts with a small molecule inhibitors of ALK5 kinase (6,28,29). Furthermore, TGF-␤-induced stimulation is reduced in fibroblasts from mice with targeted deletion of the Smad3 gene, and bleomycin-induced skin fibrosis is attenuated in Smad3-null mice compared with their wild-type littermates (19). Studies with models of pulmonary, renal, and skin fibrosis have yielded similar results (30 -32). Together, these observations establish that Smad3 is an essential intracellular mediator of fibrotic responses (33).
In addition to signal transduction through the Smad pathway, TGF-␤ also elicits cellular responses that involve non-Smad proteins. These alternate signaling pathways may regulate Smad-mediated responses through intracellular cross-talk, or they may operate via Smad-independent parallel signaling (reviewed in Ref. 34). Several nonreceptor protein kinases can induce Smad activation in response to TGF-␤, as well as to other ligands. For example, activation of Erk1/2, JNK, or phosphatidylinositol 3-kinase by TGF-␤ induces Smad phosphorylation (35)(36)(37)(38), and epidermal growth factor and hepatocyte growth factor cause Smad2 phosphorylation via receptor tyrosine kinases (39). It is now clear that the MAPK pathways contribute to certain biological TGF-␤ responses. In mesangial FIGURE 9. Impaired TGF-␤ collagen responses in Egr-1-null fibroblasts. Fibroblasts from Egr-1-null and wild-type MEFs were incubated in parallel with TGF-␤1 for 15 min or 24 h. A, total RNA was prepared, and mRNA levels were determined by Northern analysis. Representative autoradiograms are shown. Lower panel, signal intensities for COL1A1 and COL1A2 mRNA, determined by densitometric analysis of Northern blots from two independent experiments and expressed as the means Ϯ S.E. B, MEFs were incubated with TGF-␤ for the indicated periods, and whole cell lysates were examined by Western analysis using antibodies to Smad3, phospho-Smad2, or vimentin. C, confluent MEFs transiently transfected with 772COL1A2-CAT were incubated with indicated concentrations of TGF-␤1 for 24 h, and cell lysates were assayed for CAT activities. The results of triplicate determinations from two independent assays are shown as the means Ϯ S.E. Open bars, wild-type MEFs; closed bars, Egr-1-null MEFs. D, following radiolabeling of cultures for 24 h, collagen synthesis was analyzed by SDS-PAGE of proteins secreted into the culture media. A representative autoradiogram is shown. The results of densitometric analysis of the relative intensities of the procollagen bands of duplicate samples from two separate experiments are shown as means Ϯ S.E. Open bars, untreated cultures; closed bars, TGF-␤-treated cultures.

Egr-1 Regulation of Type I Collagen Gene
fibroblasts, collagen stimulation by TGF-␤ was dependent on ERK activation (37). We showed that p38 MAPK mediated the suppressive effects of TGF-␤ on cell cycle progression in hematopoietic progenitor cells (40). The MAPK pathways can also propagate TGF-␤ responses in a Smad-independent manner. Although the mechanisms of protein kinase activation by the TGF-␤ receptors, and their biological consequences and relative significance remain poorly characterized, it is clear that important biological activities of TGF-␤ such as apoptosis and induction of epithelial-to-mesenchymal trans-differentiation depend on non-Smad signaling (41).
To gain a better understanding of TGF-␤ regulation of collagen under physiological and pathological conditions, we explored the potential roles of novel TGF-␤ signaling intermediates. In the present study, we used a robust system for tightly regulated Smad3 expression in immortalized hTERT-BJ fibroblasts to identify EGR-1 as a novel Smad3 target. By RT-PCR and Western blot analyses, we confirmed that EGR-1 gene expression was stimulated by Smad3. EGR-1 is a short-lived immediate early response gene located on chromosome 5q31 that encodes an 80-kDa modular zinc finger transcription factor (42). Although negligible in normal cells, EGR-1 expression is elicited in a rapid and transient manner upon stimulation of fibroblasts, endothelial cells, and other cell types by extracellular stimuli associated with stress and injury, such as growth factors, cytokines, hypoxia, ultraviolet light, shear stress, or mechanical injury. The modular Egr-1 protein contains a zinc finger DNA binding domain and both transactivation and repression domains. Egr-1 preferentially binds to GC-rich elements with the consensus GCG(G/T)GGGCG sequence present close to the transcription start site in many promoters. Binding of Egr-1 to its target DNA sequences can induce or repress at least 80 genes (43). Although Egr-1 binds to DNA as a monomer, and does not require the presence of other proteins to modulate transcription, other DNA-binding transcription factors as well as transcriptional coactivators can directly interact with and possibly modulate the transcriptional activity of Egr-1. Interacting factors include Sp1, p53, and p300/CBP. Expression of Egr-1 is regulated at the transcriptional level by multiple environmental stimuli. The Egr-1 gene promoter contains five serum-response elements, as well as binding elements for Sp1, NF-B, and Smad (44). Activation of the ERK-mediated MAPK pathway stimulates Egr-1 promoter activity, and Egr-1 synthesis induced by serum, platelet-derived growth factor, or phorbol 12-myristate 13-acetate appears to be almost completely dependent on ERK (45). Activation of p38-mediated signaling cascades may also control EGR-1 gene transcription. The Egr-1 protein itself can bind to the Egr-1 gene promoter via EBS elements, resulting in down-regulation of its own transcription. This potential negative feedback loop may explain the transient nature of Egr-1 induction observed in normal cells (46). Remarkably, Egr-1 can stimulate transcription of p300, which in turn induces acetylation and stabilization of Egr-1 (47).
The broad range of biological responses modulated by Egr-1 includes control of synaptic plasticity, female reproduction, cell growth and senescence, differentiation and survival, angiogenesis, and connective tissue metabolism (44). To a significant degree, the specific response is determined by the cellular and environmental context. Unexpectedly, Egr-1-null mice appear normal with the exception of infertility in homozygous null females, possibly explained by the retention of sufficient Egr-1like activity because of redundancy with other members of the Egr-1 family (12). Nevertheless, Egr-1 has a key role in physiological and pathological responses. For example, Egr-1 functions as a master switch for triggering inflammatory, coagulation, and vascular injury processes in response to ischemia, and deletion of the Egr-1 gene resulted in aberrant response to vascular injury and enhanced survival in Egr-1-null mice (48,49). Moreover, tumor progression in a mouse prostate cancer model was significantly impaired in the absence of Egr-1 (50).
In this study, we examined the regulation, expression, and function of Egr-1 within the context of TGF-␤-mediated profibrotic responses. The results demonstrate that in serumstarved fibroblasts, TGF-␤ induced Egr-1 expression with rapid and transient kinetics. The stimulation was mediated at least in past at the transcriptional level, as indicated by transient transfection assays. Maximal Egr-1 mRNA expression was noted as early as 30 min after TGF-␤ stimulation, suggesting that Smadindependent mechanisms may have been involved. Indeed, consistent with other reports (51,52), we found that the stimulatory response was completely blocked by the ERK inhibitor U0126, whereas SB431542, a selective ALK5 kinase inhibitor (28,29), had only modest inhibitory effect (data not shown).
Induction of Egr-1 in normal fibroblasts was associated with increased in vitro binding of endogenous Egr-1 to consensus DNA sequences and transactivation of a minimal Egr-1-reponsive construct. Furthermore, transient expression of ectopic Egr-1 in the fibroblasts was by itself sufficient to stimulate type I collagen synthesis and the activity of a transfected COL1A2 promoter, indicating that the induced transcription factor did not require ligand activation. These findings are in accord with previous reports identifying Col1a2 as one of the genes showing elevated expression in mouse synovial fibroblasts stably transfected with Egr-1 (53). The proximal region of the human COL1A2 promoter harbors two GC-rich sequences similar to consensus Egr-1 binding EBS elements found in the promoters of other Egr-1-inducible genes (54,55).
Like other Egr-1-inducible promoters, the human COL1A2 promoter is characterized by the close proximity of binding sites for the zinc finger proteins Egr-1 and Sp1. Regulation of transcription by these two transcription factors is complex, with their synergistic interaction in some genes but mutual competition in other genes. Previous studies established that the COL1A2 promoter TGF-␤-response element includes binding sites for Sp1 family transcription factors, and demonstrated an important functional role of these sites in constitutive COL1A2 expression and TGF-␤-induced stimulation of transcription (8,23,24). In many genes, Egr-1 activates transcription by displacing Sp1 binding. Such competition between Egr-1 and Sp1 for GC-rich binding sites has been demonstrated clearly for the gene encoding PDGF A chain (57). We established that the potential Egr-1-binding sites of the COL1A2 promoter served as bona fide EBS using EMSA, as well as combination of DAPA and ChIP assays. We demonstrated that TGF-␤ induced enhanced DNA binding of native Egr-1 to the EBS2 site in vitro and in vivo. Furthermore, Sp1 and Sp3 could also be detected in the transcriptional complex. Enhanced Egr-1 DNA binding activity did not appear to be associated with a corresponding diminution of Sp1 binding in DAPA and ChIP assay. These observations suggest that Sp1 and Egr-1 bound to adjacent sites of a composite element in the COL1A2 promoter in a noncompetitive manner, raising the possibility that Egr-1 bound to DNA could stabilize the transcriptional complex, including Sp1 and Smad3.
Egr-1 has been shown to induce TGF-␤ synthesis in vitro and in vivo (58 -60). Our observation that COL1A2 promoter constructs harboring mutations disrupting EBS could not be fully activated by TGF-␤, despite the presence of an intact Smad-binding element, indicates that Egr-1 stimulation of COL1A2 transcription was not due to induction of endogenous TGF-␤. The direct role of Egr-1 in mediating TGF-␤-induced stimulation of collagen gene transcription was supported by several lines of evidence. First, we established the requirement for inducible Egr-1 binding to its cognate cis-acting elements by functional assays using substitution and truncation mutants of the COL1A2 promoter. Importantly, these results clearly indicated that disruption of EBS was associated with complete loss of the TGF-␤ response despite an intact Smad-binding element, suggesting that Smad2/3 and Egr-1 are both required for full stimulation of collagen gene transcription by TGF-␤. Second, we showed that a mutant Egr-1 substantially abrogated the ability of TGF-␤ to stimulate COL1A2 promoter activity in transiently transfected fibroblasts. Third, TGF-␤ stimulation of collagen gene expression was significantly compromised in Egr-1-deficient MEFs, despite the fact that ligand-induced activation of the Smad pathway appeared to be preserved in these cells. Taken together, these results suggest that Egr-1 is part of a novel and apparently Smad-independent intracellular TGF-␤ signaling pathway that is required, along with Smad, for mediating full stimulation of the collagen gene.
Accumulating evidence points to an important role for Egr-1 in fibrosis. Elevated Egr-1 expression was detected in synovial fibroblasts in the collagen-rich sub-synovial space in patients with rheumatoid arthritis (61), and in the fibrous caps of atherosclerotic vascular lesions (62). The levels of Egr-1 protein and/or mRNA were elevated in the fibrotic kidneys in rats with ureteral obstruction (63), and in mice with genetic deficiency of type IV collagen, an animal model for Alport syndrome associated with kidney fibrosis (64). The expression of Egr-1 correlated with tissue fibrosis in a model of peritoneal adhesions (65) and in pulmonary artery fibroblasts in hypoxic animals (66). A rapid rise in Egr-1 was observed in excisional wounds, and administration of Egr-1 into the wounds accelerated their closure (67). A recent study demonstrated that Egr-1 mRNA and protein levels were elevated in lung tissues from patients with emphysema (68). Expression of Egr-1 colocalized with collagen, connective tissue growth factor, and TGF-␤ within the same stromal sites. Global gene expression analysis revealed elevated Egr-1 mRNA levels in skin fibroblasts from patients with diffuse scleroderma (69). Similarly, we have observed elevated Egr-1 expression in lesional skin from scleroderma patients. 3 In other studies, a marked increase in lesional fibroblast Egr-1 levels in vivo was found in mice injected with bleomycin. 4 Targeting the Egr-1 gene with specific DNA enzymes in a rat model of ureteral obstruction resulted in down-regulation of myofibroblast marker expression in vivo and reduced fibrosis (63). Furthermore, cardiac hypertrophy and fibrosis induced by isoproterenol infusion were attenuated in Egr-1-null mice compared with littermates (70). Fibrosis of the lungs induced by TGF-␤ or interleukin-13 was markedly attenuated in Egr-1-null mice (60,71).
The mechanisms responsible for elevated Egr-1 expression in fibrotic tissue are currently unknown. Egr-1 induction is normally short-lived due in part to the repressive actions of Egr-1mediated negative feedback loops. Sustained Egr-1 expression may indicate defective feedback regulation in fibrosis. Sustained Egr-1 stimulation may be induced by tissue hypoxia or TGF-␤, both of which have been shown to induce Egr-1 gene expression (48,59), and both conditions are known to prevail in fibrotic tissue. Egr-1 mediates hypoxia-induced fibroblast proliferation, and blockade of Egr-1 by antisense oligonucleotides reduced the response (66). The synthesis of key mediators of the fibrotic response is induced through Egr-1. These include TGF-␤ (58, 72) and its cell surface receptors (73), as well as thrombospondin, which activates latent TGF-␤ (74), basic fibroblast growth factor (75), and platelet-derived growth factor (57). In vivo, Egr-1 cDNA delivery in the skin induced localized TGF-␤ production (67). In addition, Egr-1 directly stimulates smooth muscle ␣-actin expression and myofibroblast trans-differentiation (63), as well as transcription of several ECM genes contributing to the development of fibrosis, including, in addition to collagen, fibronectin (52,76), TIMP-1 (77), as well as ICAM-1 and other adhesion molecules (78). In contrast to our present findings, previous studies reported that suppression of collagen synthesis in osteoblast-like cells by FGF2 (79) or by interleukin-1␤ in immortalized chondrocytes was also mediated through Egr-1 (25), suggesting that target gene regulation through Egr-1 was cell type-specific.
The present results indicate that TGF-␤ induced the expression of Egr-1 in fibroblasts in vitro and in vivo. In turn, multiple studies have demonstrated that Egr-1 stimulates TGF-␤ synthesis in a variety of mesenchymal cell types and enhances their sensitivity to TGF-␤ (59). Thus, Egr-1 may serve dual intracellular functional roles as both a target regulated by TGF-␤, as well as mediator for enhanced TGF-␤ gene expression and target cell responsiveness, indicating the potential existence of an autocrine stimulatory loop. The present findings demonstrate that Egr-1 plays an essential role in propagating TGF-␤ signals leading to collagen gene stimulation. Tissue expression of Egr-1 is elevated during pathological matrix remodeling and fibrosis, possibly due to its sustained local induction by TGF-␤ and hypoxia, as well as additional stress-and injury-associated stimuli. Transient expression of Egr-1 represents an acute response to injury and is likely to play a significant role in wound healing, in line with the prevailing view of Egr-1 as part of the physiologic stress response program (44). In contrast, sustained Egr-1 expression may contribute to fibrosis and other pathological responses (80). Taken together with the consistent association between Egr-1 expression and fibrosis, and the ability of Egr-1 to transactivate multiple genes whose products play important roles in fibrosis, the results suggest that Egr-1 is a key mediator of the process. In light of its crucial role in triggering, amplifying, and sustaining pathological fibrotic responses, targeting Egr-1 represents an attractive novel approach to therapy. Indeed, the amelioration of TGF-␤or interleukin-13-induced lung fibrosis in Egr-1-null mice provides powerful support for the soundness of such an approach (60,71). Multiple pharmacological agents in current use, including the glitazone antidiabetic peroxisome proliferator-activated receptor-␥ agonists (81)(82)(83), statin drugs (84), and curcumin (56), all appear to suppress Egr-1 expression or function. These and similar agents may therefore ultimately find novel clinical utility as anti-fibrotic therapeutics.