Extracellular signal-regulated kinase and the small GTP-binding protein, Rac, contribute to the effects of transforming growth factor-beta1 on gene expression.

The kinases and regulatory proteins that convey signals initiated by transforming growth factor-beta (TGF-beta) to the nucleus are poorly characterized. To study the role of the extracellular signal-regulated kinase (ERK) pathway in this process, we transiently transfected NIH 3T3 fibroblasts with TGF-beta-responsive luciferase reporter genes and expression vectors designed to interrupt this kinase cascade. Mitogen-activated protein (MAP) kinase phosphatase-1 and a dominant negative MAP/ERK kinase 1 mutant reduced stimulation of plasminogen activator inhibitor-1 (PAI-1) promoter activity by TGF-beta1 from 11.5- to 4-fold and 4.9-fold, respectively. Similar results were observed with the type I collagen promoters. TGF-beta1 increased ERK1 activity 4.5-fold at 5 min and 3. 1-fold at 3 h, while Jun kinase and p38 activity were not affected. Cotransfection of a dominant negative mutant of the small G protein, Rac, but not dominant negative Ras, Cdc42, or Rho mutants, reduced the effects of TGF-beta1 on the PAI-1 promoter by approximately half. In support of a role for Rac in signaling by TGF-beta, GTP binding to Rac was increased 3.7-fold following exposure of NIH 3T3 cells to TGF-beta1 for 3 min. These findings indicate that TGF-beta1 modulates gene expression partly through ERK and Rac in NIH 3T3 cells.

sant, thereby promoting tumor growth (6). Consequently, it is of great interest to define the signaling pathways utilized by TGF-␤.
The mechanisms by which TGF-␤ causes changes in gene expression are incompletely understood. Chemical cross-linking with 125 I-labeled TGF-␤ has revealed three distinct types of TGF-␤ receptors, including type I (55 kDa), type II (80 kDa), and type III (280 kDa) receptors that have all been cloned (7,8). The type III receptor is a proteoglycan, betaglycan, that is involved in ligand presentation (9). On the other hand, the type I and II receptors are transmembrane serine/threonine kinases (7,8). Based on a series of experiments with mutant cells lacking one type of receptor and transfections of mutant receptors, it has been ascertained that the type I and II receptors are both required for all responses to TGF-␤ (7,8). Upon ligand binding, the type II receptor phosphorylates the type I receptor in its GS domain, enabling the type I receptor to phosphorylate target substrates (7,8). The ability of a constitutively active type I receptor, with threonine 204 mutated to aspartic acid, to inhibit cellular proliferation and increase gene expression (8) suggests that the type I receptor is primarily responsible for transmitting signals from TGF-␤. However, the relevant substrates are largely unknown. Use of the yeast two-hybrid system has shown that FK506-binding protein-12 (10), TRIP-1 (a WD protein) (11), and farnesyltransferase (12) all interact with TGF-␤ receptors, but their role in signaling has not been demonstrated.
Much more is known about receptor tyrosine kinases and receptors coupled to heterotrimeric G proteins. The signals from these receptors are partly propagated to the nucleus through mitogen-activated protein (MAP) kinases. These serine/threonine kinases characteristically require phosphorylation on both threonine and tyrosine residues to be activated (13). In eukaryotes, MAP kinases form signaling modules made up of a MAP kinase kinase kinase (MAPKKK), a MAP kinase kinase (MAPKK), and a MAP kinase. The first discovered mammalian MAP kinase cascade consists of the MAP kinases, extracellular signal-regulated kinases 1 and 2 (ERK1 and 2), the MAPKKs, MAP/ERK kinase 1 and 2 (MEK1 and 2), and the MAPKKKs, c-Raf and B-Raf (13)(14)(15)(16). Targets of ERK include the kinase p90 rsk , cytoplasmic phospholipase A 2 , and transcription factors such as c-Myc and Elk-1 (13,17).
More recently, it was found that cellular stress, such as ultraviolet light and high osmolality, and the inflammatory cytokines, interleukin-1 and TNF␣, activate a separate MAP kinase cascade. This pathway includes the MAP kinase, Jun kinase (JNK), also termed stress-activated protein kinase (SAPK) (18,19), and the MAPKK, SAPK/ERK kinase 1 (SEK1) (20). A parallel pathway that is activated by similar stimuli culminates in the activation of p38 (21). Among the substrates of JNK and p38 are c-Jun, ATF-2, and Elk-1 (22). Growth factors and phorbol esters, which are strong inducers of ERK, activate the JNK and p38 pathways only weakly (18,(21)(22)(23). Thus, JNK and ERK respond to different external stimuli and have different MAPKKs and substrates (14,15). It has been shown that TGF-␤ modestly increases ERK activity in some cell lines (24,25), but it is not known at present whether any of the MAP kinases play a functional role in the initiation of gene expression by TGF-␤.
Small G proteins, such as Ras and the Rho-like proteins, Rac and Cdc42 (26,27), are important regulators of MAP kinase cascades. Growth factor modulation of Ras guanine nucleotide exchange factor activity causes Raf to bind to Ras and eventually activate ERK through MEK1 (28 -30). Rac and Cdc42 control the JNK and p38 pathways (26,27). Consequently, the small G proteins are good candidates to participate in the control of gene expression by TGF-␤. Interestingly, TGF-␤ triggers GTP binding to Ras within minutes in serum-starved epithelial cells (31,32). Specification of ventral cell fate in Xenopus embryos by bone morphogenic protein-4, a member of the TGF-␤ superfamily, is blocked by dominant negative mutants of Ras and Raf (33).
We sought to determine the role of ERK, JNK, and the small G proteins, Ras, Rac, Cdc42, and Rho, in mediating the effects of TGF-␤1 on the promoters of genes that might be involved in TGF-␤-stimulated accumulation of extracellular matrix. Studies using transient transfections and immune complex kinase assays demonstrated a functional role for ERK and Rac in signaling by TGF-␤1.

EXPERIMENTAL PROCEDURES
Plasmids-Expression vectors were prepared by cloning cDNAs into the vector, pcDNA 3 (Invitrogen, San Diego, CA), that uses a cytomegalovirus promoter to drive gene expression. The cDNAs for the amino terminus of human c-Raf-1 (amino acids 1-257), Cdc42, Rho, MEK1, and murine MAP kinase phosphatase-1 (MKP-1) and Rac were obtained by polymerase chain reaction (PCR) using the enzyme, pfu (Stratagene, La Jolla, CA), and commercial cDNA libraries (Stratagene). Mutations were introduced into these cDNAs by overlap extension and PCR, changing serine 17 to asparagine for Rac, Cdc42, and Rho, and lysine 97 to alanine for MEK1. A human Ha-Ras cDNA, with threonine 17 mutated to asparagine, was a gift from Dr. Sean Egan (The Hospital for Sick Children, Toronto, Canada). Sequences were verified by dideoxy sequencing using Sequenase 2.0 (USB, Cleveland, OH).
Luciferase reporter genes containing the human plasminogen activator inhibitor-1 (PAI-1) promoter (Ϫ740 to ϩ44) and the human ␣1(I) collagen gene promoter (Ϫ802 to ϩ51) were prepared by PCR from genomic DNA. They were then subcloned between HindIII and Asp-718 in the luciferase vector, pA 3 LUC, a previous gift of Dr. E. Chester Ridgeway and Dr. William Wood (University of Colorado Health Science Center, Denver, CO). This vector has three upstream SV40 poly(A) sites to prevent read-through transcription (34). The reporter genes for a shorter version of the human PAI-1 gene promoter (Ϫ699 to ϩ64) and the murine ␣2(I) collagen promoter (Ϫ350 to ϩ54) have been previously described (35).
Transfections-NIH 3T3 cells (ATCC) were maintained in Dulbecco's modified Eagle's medium and 10% calf serum (Life Technologies, Inc.). One day prior to transfection, they were plated at 1 ϫ 10 6 cells per 10-cm dish. The cells were transfected with 25 g of DNA (5 g of reporter gene, 15 g of expression vector, and 5 g of the ␤-galactosidase expression vector, pTK␤ (Clontech, Palo Alto, CA)) by calciumphosphate precipitation, as described (35,36). The cells were kept overnight at 35°C in 3% CO 2 and then washed three times with phosphate-buffered saline (PBS) and transferred to serum-free medium that contained Dulbecco's modified Eagle's medium/F12 (Life Technologies, Inc.), 1 g/liter of Albumax (Life Technologies, Inc.), 60 mg/liter of mannose 6-phosphate, 10 mg/liter transferrin and trace metals (Life Technologies, Inc.). Three hours later, 4 ng/ml porcine TGF-␤1 (R and D Systems, Minneapolis, MN) or vehicle were added. The cells were harvested after 21 h and luciferase activity assayed in duplicate as described (35,36). pTK␤ was used to normalize for transfection efficiency. All transfections were repeated three times and the results expressed as the ratio of relative luciferase activity in TGF-␤1-treated cells compared with vehicle-treated cells. The Student's t test was used to eval-uate the data from the transfections and other experiments statistically.
ERK1 and JNK Kinase Activity-ERK1 activity was measured with an immune complex kinase assay using the method of Crespo et al. (37). Semiconfluent NIH 3T3 cells were washed three times with PBS and maintained in serum-free medium for 2 h. The cells were exposed to 4 ng/ml TGF-␤1, vehicle, or UV-C light (80 J/m 2 , in a UV Stratalinker 2400 (Stratagene)) and kept for the indicated times at 37°C. The cells were then washed three times with PBS, lysed with 20 mM Hepes, pH 7.5, 1% Nonidet P-40, 10 mM EGTA, 2.5 mM MgCl 2 , 40 mM ␤-glycerophosphate, 2 mM sodium vanadate, 0.5 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 20 g/ml aprotinin, 1 mM dithiothreitol, and 5 mM benzamidine and centrifuged at 14,000 rpm to eliminate cellular debris. Equal amounts of protein were rotated with 2 g of ERK1 antibody (sc-93, Santa Cruz, Santa Cruz, CA) for 1 h and protein A-agarose (Santa Cruz) for an additional 2 h at 4°C. The immunoprecipitates were washed with 1% Nonidet P-40, 2 mM sodium vanadate in PBS, and 0.5 M LiCl in 100 mM Tris, pH 7.5. The kinase assay was performed with kinase buffer (12.5 mM MOPS, pH 7.5, 7.5 mM MgCl 2 , 20 g/ml aprotinin, 12.5 mM ␤-glycerophosphate, 0.5 mM vanadate, 0.5 mM EGTA, and 0.5 mM sodium fluoride), 20 M ATP, 10 Ci of [␥-32 P]ATP, and 10 g of myelin basic protein (Sigma) for 20 min at 30°C. Following boiling in SDS sample buffer, the products of the kinase reaction were separated on a 14% SDS-polyacrylamide gel and subject to autoradiography with Kodak Biomax film (Kodak). Quantitation was achieved by scanning the film with a Hewlett-Packard Scan-Jet 4c scanner (Hewlett-Packard, Palo Alto, CA) and analysis with NIH Image software.
Rac GTP Binding-The method of de Vries-Smits et al. (38), previously employed to measure GTP binding to Ras, was adapted for Rac. In preliminary studies it was determined that 500 units/ml of streptolysin O (Sigma) was required to obtain maximum permeabilization of NIH 3T3 cells as measured by release of radioactivity into the media following a 5-h incubation with [ 32 P]dCTP. Cells were plated at 2 ϫ 10 6 cells per 10-cm dish. The next day the cells were washed three times with PBS and placed in serum-free media. Eight hours later the cells were washed three more times with PBS, once with permeabilization buffer (39) (10 mM Pipes, 120 mM KCl, pH 7.2, 2.5 mM MgCl 2 , 30 mM NaCl, 0.5 mM CaCl 2 , 2 mM EGTA, 1 mM ATP, 1 mM K 2 HPO 4 ) and placed in permeabilization buffer at 37°C. TGF-␤1 (10 ng/ml) or vehicle was added for 1 min and then streptolysin O (500 units/ml) and [ 32 P]GTP (10 Ci/dish) were added for a further 2 min. The cells were then lysed in 25 mM Tris-HCl, pH 7.4, 1% Triton X-100, 137 mM NaCl, 5 mM MgCl 2 , 5 mM KCl, 1 mM EGTA, 1 mM Na 2 PO 4 , 100 M GDP, 100 M GTP, 10 mM benzamidine, 40 g/ml leupeptin, 40 g/ml aprotinin, and 10 g/ml soybean trypsin inhibitor. The lysate was centrifuged at 14,000 rpm and detergents (0.5% sodium deoxycholate and 0.005% SDS, final concentrations) and NaCl (0.5 M, final concentration) were added to the supernatant. An overnight incubation at 4°C was carried out with Rac1 antibody (sc-14, Santa Cruz) and protein A-Sepharose. The immune complexes were then washed seven times with 50 mM Hepes, pH 7.4, 500 mM NaCl, 5 mM MgCl 2 , 0.1% Triton X-100, and 0.005% SDS and counted in a scintillation counter.

Regulation of the PAI-1 and Type I Collagen Gene
Promoters by TGF-␤1 Involves ERK-We chose fibroblasts to investigate the signaling mechanisms used by TGF-␤, because of the participation of this cell type in fibrotic diseases. Since at least part of the effects of TGF-␤ on gene expression are exerted at the level of transcription (see references in Ref. 35), transient transfections of reporter genes linked to TGF-␤-dependent pro-moters were performed. The cell line, NIH 3T3, used in these experiments has been previously shown to be suitable for studying responses of reporter genes to TGF-␤ (35,40,41).
Previously, we and others (35,42,43) have shown that TGF-␤1 increases the luciferase activity of reporter genes driven by the PAI-1 promoter. To investigate a possible role for ERK in this effect of TGF-␤1, two cDNAs that are well established inhibitors of ERK activation, a dominant negative MEK1 mutant and the phosphatase MKP-1, were cotransfected with a PAI-1 promoter-reporter gene. Dominant negative MEK1, which has lysine 97 mutated to alanine rendering it catalytically inactive, specifically disrupts ERK but not JNK activation (14 -16, 30). MKP-1 (also called CL100, HVH1, 3CH134, and Erp) is a dual specificity protein phosphatase that was previously found to attenuate induction of gene expression by ERK (44). As shown in Fig. 1A, TGF-␤1 stimulated Ϫ699 to ϩ64, PAI-1 promoter activity 11.5-fold as measured by a luciferase reporter gene. The use of a larger PAI-1 promoter (Ϫ740 to ϩ44) that includes an additional TGF-␤-responsive element, from Ϫ726 to Ϫ703 (45), increased the effect of TGF-␤1 by only 25% (not shown). MKP-1 and the dominant negative MEK1 mutant reduced induction of PAI-1 promoter activity by TGF-␤1 to 4-and 4.9-fold, respectively (Fig. 1A). In contrast, MKP-1 and dominant negative MEK1 essentially abolished the effects of platelet-derived growth factor (PDGF) on the c-Fos promoter, which is known to be dependent on ERK (22) (Fig.  1B). These results indicate that ERK activation accounts for part of the effects of TGF-␤1 on the PAI-1 promoter.
Because TGF-␤ has been shown, in some but not all reports (40,43,46), to augment PAI-1 and type I collagen promoter activity through different transcription factors, we analyzed the actions of TGF-␤1 on the type I collagen promoters as well. As shown in Fig. 1A, TGF-␤1 increased the activity of the ␣2(I) collagen gene promoter linked to a luciferase reporter gene 4-fold, as previously reported (35) (Fig. 1A). These effects were diminished by MKP-1 and the dominant negative MEK1 mutant (Fig. 1A). TGF-␤1 provoked only a 1.5 Ϯ 0.13-fold increase in the activity of the ␣1(I) collagen promoter, and this was eliminated by cotransfection of MKP-1 (0.86 Ϯ 0.04-fold stimulation, n ϭ 9, p Ͻ 0.01). Thus, ERK plays a role in the response of the type I collagen gene promoters to TGF-␤1 as well.
TGF-␤ has been reported to increase ERK activity in rat glomerular mesangial cells (25) and intestinal epithelial cells (24) but not in Swiss 3T3 fibroblasts (47). To further support a functional role for ERK in the effects of TGF-␤1 on NIH 3T3 fibroblasts, ERK1 activity was measured by immune complex assay using myelin basic protein as a substrate. Administration of TGF-␤1 to NIH 3T3 cells increased ERK1 activity 4.5 Ϯ 0.9-fold at 5 min (p Ͻ 0.01, n ϭ 6), 1.8 Ϯ 0.2-fold at 10 min (n ϭ 10, p Ͻ 0.01), and 3.1 Ϯ 0.2-fold at 3 h (n ϭ 10, p Ͻ 0.01). One of these experiments is depicted in Fig. 2A. Under the same experimental conditions PDGF caused a 10-fold increase in ERK1 activity at 10 min (Fig. 2B), indicating that TGF-␤1 is a relatively weak activator of ERK1.
Protein kinase C (PKC) regulates ERK (29,37,48,49), and PKC is postulated to be involved in signaling by TGF-␤ (50,51). Therefore , and a ␤-galactosidase expression vector, pTK␤. The cells were then stimulated with TGF-␤1 (4 ng/ml) or vehicle and luciferase assayed after 21 h in serum-free media. The results are expressed as fold stimulation by TGF-␤1 for each construct. Each value (average Ϯ S.E.) represents the results from three independent experiments. The asterisk indicates that the fold stimulation by TGF-␤1 when a particular construct was transfected was significantly less than that in cells transfected with pcDNA 3 (Student's t test, p Ͻ 0.01). B, similar experiments were performed with the c-Fos promoter fused to a luciferase reporter gene and the addition of 10 ng/ml PDGF.

FIG. 2. TGF-␤1 increases ERK1 activity.
A, duplicate dishes of NIH 3T3 cells were treated with TGF-␤1 (4 ng/ml) or vehicle for the indicated amount of time. ERK1 was immunoprecipitated and its activity assayed using [␥-32 P]ATP and myelin basic protein as a substrate. The products were subjected to SDS-PAGE and autoradiography. B, similar experiments were performed with 10 ng/ml PDGF. veh, vehicle. serine 73 by JNK (18,19). Therefore, JNK is a good candidate to transduce signals generated by TGF-␤ to the PAI-1 promoter. Accordingly, we examined JNK activity by immune complex assay. TGF-␤1 did not induce JNK activity at 10 min (Fig. 3A), 3 h (Fig. 3A), or at a number of other time points up to 3 h (not shown). In contrast UV-C light caused a 7-fold increase in JNK activity at 10 min (Fig. 3B). Similarly, the activity of a related kinase, p38, was stimulated 4-fold by UV-C light, but was not increased by TGF-␤1 at 10 min (Fig. 4) or at later time points (not shown).
The Small G Protein, Rac, Participates in the Induction of PAI-1 Promoter Activity by TGF-␤1-Given the involvement of ERK in activation of the PAI-1 promoter by TGF-␤1, we next examined the role of Ras. Rho-like small G proteins were considered as well, since they are also involved in signal transduction (26,27). Dominant negative mutants were constructed that have threonine or serine 17 mutated to asparagine. Previous studies have shown that this type of mutation in Ras causes it to bind GDP preferentially to GTP, allowing the mutant to titrate out guanine nucleotide exchange factors (52). Cotransfection of these mutants with the PAI-1 promoter-reporter gene revealed that only dominant negative Rac, but not dominant negative Ras, Cdc42, or Rho mutants, reduced induction of PAI-1 promoter activity by TGF-␤1 (Fig. 5A). Fold stimulation of luciferase activity by TGF-␤1 was decreased by dominant negative Rac to 44% of that in cells transfected with the empty expression vector. Base-line activity of the PAI-1 promoter was suppressed by dominant negative Ras, as reported by others (53) (not shown), but fold stimulation by TGF-␤1 was not diminished. As a second test of the Ras pathway, a plasmid expressing dominant negative Raf-1 was transfected. This mutant consisted of the amino-terminal regulatory domain that was previously shown to disrupt the downstream effects of Raf-1 (37,48,54,55). Cotransfection of this dominant negative Raf-1 mutant also did not reduce the effects of TGF-␤1 (Fig.  5A). To ensure that the expression vectors for dominant negative Ras and Raf-1 were functional, we tested their ability to block stimulation of the c-Fos promoter by epidermal growth factor (EGF). Both constructs completely eliminated the effect of EGF on this promoter in NIH 3T3 cells (Fig. 5B).
To determine whether TGF-␤1 affects GTP binding to Rac, we adapted an assay that was previously used successfully for Ras. In these earlier studies, qualitatively similar results were generally obtained by labeling intact cells with [ 32 P]orthophosphate or by measuring binding of [ 32 P]GTP to Ras in permeabilized cells (31,38,56). Therefore, appropriate conditions for the use of streptolysin O, a widely used cell permeabilization agent (39), were established by measuring the release of radioactivity into the media following a 5-h incubation with [ 32 P]dCTP. NIH 3T3 cells were treated with TGF-␤1 for 1 min and then 500 units/ml streptolysin O and [ 32 P]GTP were added for a further 2 min. Rac was immunoprecipitated, and following extensive washing the amount of GTP bound to Rac was estimated by scintillation counting. Exposure of NIH 3T3 cells to TGF-␤1 resulted in a 3.7 Ϯ 0.83-fold increase (n ϭ 8, p Ͻ 0.01) in GTP binding to Rac compared with vehicle-treated cells. No effects of TGF-␤1 were seen in the absence of streptolysin O, with an irrelevant antibody, or in the presence of excess cold GTP (not shown). FIG. 3. TGF-␤1 does not affect JNK activity. A, duplicate dishes of NIH 3T3 cells were treated with TGF-␤1 (4 ng/ml) or vehicle (veh) for the indicated amount of time. JNK1 was immunoprecipitated and its activity assayed using [␥-32 P]ATP and GST-Jun (amino acids 1-79) as a substrate. The products were subjected to SDS-PAGE and autoradiography. B, similar experiments were performed by exposing the cells to UV-C light (80 J/m 2 ) and then keeping them at 37°C for the indicated times.

FIG. 4. p38 activity is increased by UV light but not by TGF-␤1.
Duplicate dishes of NIH 3T3 cells were treated with TGF-␤1 (4 ng/ml), vehicle (veh), or UV-C light (80 J/m 2 ) and kept at 37°C for the indicated amount of time. p38 was immunoprecipitated and its activity assayed using [␥-32 P]ATP and GST-ATF-2 (amino acids 1-96) as a substrate. The products were subjected to SDS-PAGE and autoradiography. cont., control.
FIG. 5. The small GTP-binding protein, Rac, is involved in the induction of PAI-1 promoter activity by TGF-␤1. A, NIH 3T3 cells were cotransfected with plasmids containing the PAI-1 promoter (Ϫ699 to ϩ64), either the empty expression vector (VEC), pcDNA 3 , or expression vectors for dominant negative Ras, Raf-1, Cdc42 (CDC), Rac, or Rho mutants, and a ␤-galactosidase expression vector, pTK␤. The cells were then stimulated with TGF-␤1 (4 ng/ml) or vehicle and luciferase assayed after 21 h in serum-free media. The results are expressed as fold stimulation by TGF-␤1 for each construct. Each value (average Ϯ S.E.) represents the results from three independent experiments. The asterisk indicates that the fold stimulation by TGF-␤1 when a particular construct was transfected was significantly less than that in cells transfected with pcDNA 3 (Student's t test, p Ͻ 0.01). B, similar experiments were performed with the c-Fos promoter fused to a luciferase reporter gene, dominant negative Ras and Raf-1 mutants (DNRAS and DNRAF, respectively), and the addition of 100 nM EGF. DISCUSSION TGF-␤ was previously reported to modestly increase ERK activity in some cell types (24,25), but the functional relevance of this response was not determined. Evidence is presented for the first time that ERK contributes functionally to changes in gene expression mediated by TGF-␤1. Expression of two different cDNAs, MKP-1 and a dominant negative MEK1 mutant that were shown by others to disrupt the ERK pathway, reduced the effects of TGF-␤1 by approximately half. These findings were corroborated by the observation that TGF-␤1 significantly increased ERK1 activity in NIH 3T3 cells.
Hartsough et al. (24) detected an early 2-fold stimulation of ERK activity by TGF-␤ at 10 min in proliferating but not quiescent epithelial cells. The cells used in the present experiment continue to proliferate while they are being transfected, presumably establishing the conditions for TGF-␤ to activate ERK. The requirement for cellular proliferation for TGF-␤ to influence ERK may reflect the existence of more than one pathway for ERK activation as demonstrated by the synergistic effects of phorbol esters and insulin on this process (57). However, a recent study in Swiss 3T3 cells found that TGF-␤ stimulated cellular growth without eliciting any increase in ERK activity (47). These results in Swiss 3T3 cells may differ because of the prolonged incubation of the Swiss 3T3 cells in serum-free media (40 -48 h) that was used to achieve cellular quiescence in that study. In addition, the results in Swiss 3T3 cells are not comparable with the present study, since stimulation of gene expression by TGF-␤ was not evaluated.
Recently, a novel serine/threonine kinase, TGF-␤-activated kinase (TAK1), was shown to complement an Ste11 mutation in yeast, like Raf and MEK kinase 1 (58). The activity of TAK1 was increased by TGF-␤ in MC3T3-E1 cells, and a dominant negative mutant of TAK1 abrogated stimulation of PAI-1 promoter activity by TGF-␤ in transient transfections (58). Further studies will be required to establish whether TAK1 is activated by TGF-␤ in NIH 3T3 cells. TAK1 was reported to phosphorylate the JNK activator, SEK1, in vitro, and modest JNK activation by TGF-␤ was noted in preliminary form in MC3T3-E1 cells (58). However, JNK did not appear to contribute to the effects of TGF-␤1 in NIH 3T3 cells in the present experiments, since TGF-␤1 did not increase JNK or p38 activity. Consistent with this conclusion is the lack of observed effect of the dominant negative Cdc42 mutant, which can interrupt the JNK pathway when overexpressed (26,27). Overexpression of MKP-1 in transient transfections has been shown to be able to block the activation of JNK, p38, and ERK (44). We found that MKP-1 inhibited the effects of TGF-␤1 to only a slightly greater extent than dominant negative MEK1, which is specific for ERK, further supporting the idea that signaling by TGF-␤ may be cell-specific, involving JNK in only some cell types.
In cells that are growth-arrested by TGF-␤, the activity of a 78-kDa serine-threonine kinase, that could possibly be TAK1, was induced by TGF-␤ (59). This latter kinase is unlikely to explain any of the present results, since it was reported not to be activated by TGF-␤ in NIH 3T3 cells.
Intriguingly, both PKC and TGF-␤ regulate gene expression and can increase or decrease cell growth depending on the context. This similarity in actions has lead some investigators to postulate that the effects of TGF-␤ are mediated through PKC (50,51). Calphostin partially blocked the ability of TGF-␤ to increase the activity of the PAI-1 promoter-reporter gene in this study. This suggests that signal transduction by TGF-␤ may have occurred partly through PKC, which could then have activated ERK and Rac (22,60).
A number of investigations have been directed at examining which transcription factors are responsive to TGF-␤. There is evidence that AP-1 is an important determinant of TGF-␤ effects on the PAI-1, TGF-␤1, fibronectin, TIMP-1, and artificial promoters (see references in Ref. 35). Intermediate involvement of ERK in the effects of TGF-␤ on AP-1 would be consistent with the known ability of ERK to phosphorylate Elk-1 (22), which binds to the serum response element in the c-Fos promoter. Another component of AP-1 is JunB; and activation of the PAI-1 and ␣2(I) collagen promoters by TGF-␤ partly depends on JunB (35,61). It is possible that ERK induction by TGF-␤ could also serve to increase JunB transcription (62). Some studies of the ␣2(I) and ␣1(I) collagen promoters implicate the transcription factor, SP1, in mediating the effects of TGF-␤ (40,46). Further studies will be required to determine whether SP1 is modulated by ERK. Much data implicate Ras and Raf-1 in controlling ERK activity in a variety of species including genetic studies of photoreceptor specificity in Drosophila and vulval differentiation in Caenorhabditis elegans (13,16,29,30). TGF-␤ increases GTP binding to Ras in cells that are growth-inhibited by this cytokine (31,32). Despite clear evidence that the effects of TGF-␤1 were partly due to ERK, we found that dominant negative Ras or Raf-1 mutants did not interfere with TGF-␤1 action. Conversely, the effects of EGF on the c-Fos promoter were blocked by these mutants, indicating they are functional. Therefore, TGF-␤1 appears to activate ERK through a pathway that bypasses Ras in NIH 3T3 cells. A number of other reports support the existence of nontraditional pathways for ERK activation. For example, an analysis of NIH 3T3 cells found that B-Raf (which also binds to Ras) and an unknown kinase, rather than Raf-1, accounted for the majority of cytoplasmic activity capable of activating MEK1 (63). Several investigators have reported stimulation of ERK activity by certain ligands to be independent of Ras or Raf-1 (28,48,64,65). Also, it has been proposed that Raf-1 may be activated directly by protein kinase C-mediated phosphorylation (66). The dominant negative Raf-1 mutant used in the present studies may not have been able to prevent this mode of direct activation, as opposed to Raf-1 activation by Ras-Raf-1 interaction.
The role of Rho-like small G proteins in signaling provoked by TGF-␤1 was explored in this study. Our data implicates Rac, for the first time, as being involved in the TGF-␤1 pathway. A dominant negative Rac mutant, but not dominant negative Ras, Cdc42, or Rho mutants, decreased stimulation of PAI-1 promoter activity upon exposure of the cells to TGF-␤1. The significance of the slight increases in fold stimulation of PAI-1 promoter activity by TGF-␤1 when expression vectors for dominant negative Cdc42 or Ras were transfected is unclear. Importantly, addition of TGF-␤1 to NIH 3T3 cells increased GTP binding to Rac within minutes. Reported studies in which epitope-tagged kinases were cotransfected with constitutively active or dominant negative Rac or Cdc42 mutants suggest that these small G proteins regulate the JNK but not the ERK pathway (26,27). In contrast, the results described herein demonstrate that signal transduction by TGF-␤1 involves Rac and ERK but not JNK. An analogous situation is that phorbol esters activate Rac and ERK but increase JNK activity only weakly (21-23, 26, 60). It was not possible to cotransfect the dominant negative MEK1 mutant with the Rac mutant, since maximal effects of the mutant Rac construct required a large input of plasmid. This may be due to inefficient expression of dominant negative Rac, which has been noted previously (27). Possible roles for Rac in signaling by TGF-␤1 include regulation of phosphoinositide metabolism (67) or reorganization of the cytoskeleton and modulation of associated signaling molecules such as p125 fak (68).
In summary, we have started to define the mechanisms of TGF-␤1-triggered gene expression in fibroblasts by demonstrating functional involvement of ERK and Rac. An improved understanding of the intermediate kinases and regulatory proteins that participate in signal transduction by TGF-␤ should eventually lead to identification of the relevant immediate substrates of the serine/threonine kinase domain of the type I TGF-␤ receptor. This will have potentially important implications with respect to elucidating the fibrogenic actions of TGF-␤.