Altered transforming growth factor signaling in epithelial cells when ras activation is blocked.

We have previously demonstrated that growth inhibition of untransformed intestinal epithelial cells by transforming growth factor beta1 (TGFbeta) and TGFbeta2 was associated with a rapid activation of both Ras and extracellular signal-regulated kinase 1 (Erk1) (Mulder, K. M., and Morris, S. L. (1992) J. Biol. Chem. 267, 5029-5031; Hartsough, M. T., and Mulder, K. M. (1995) J. Biol. Chem. 270, 7117-7124). In order to determine whether Ras was required for TGFbeta regulation of both Erk1 and downstream components associated with TGFbeta-mediated growth inhibition, the intestinal epithelial cell (IEC) line IEC 4-1 was transfected with a vector containing a dominant-negative mutant of Ras (RasN17) under the control of an inducible metallothionein promoter. Using two different RasN17-transfected clones treated with ZnCl2, we demonstrate here that induction of Ras expression by at least 4-fold completely abrogated the TGFbeta-mediated activation of Erk1. Moreover, the RasN17-mediated reversal of the TGFbeta effect on Erk1 was dependent upon the level of expression of the dominant-negative protein. ZnCl2 treatment of control cells transfected with the empty vector did not alter Ras expression or the activation of Erk1 by TGFbeta. In order to determine whether the activation of Ras by TGFbeta was required for the growth inhibitory effect of TGFbeta, we examined TGFbeta2 effects on Cdk2-associated histone H1 kinase activity, cyclin A protein expression levels, and DNA synthesis in two intestinal epithelial cell clones transfected with RasN17. In cells expressing RasN17, we observed a 50% reversal of the inhibition of Cdk2 activity, a 78% reversal of the down-regulation of cyclin A protein expression, and a 21% reversal of the inhibition of DNA synthesis by TGFbeta. Collectively, these results indicate that Ras activation is obligatory for TGFbeta-mediated activation of Erk1, whereas it is partially required for the growth inhibitory effect of TGFbeta.

The transforming growth factor-␤ (TGF␤) 1 family currently consists of three mammalian secreted polypeptides (TGF␤ 1 , TGF␤ 2 , and TGF␤ 3 ) that regulate cellular growth, morphogenesis, differentiation, and adhesion (3). TGF␤ exerts these cellular effects through a heteromeric complex of the type I (RI) and type II (RII) TGF␤ receptors, each containing serine/threonine kinase domains that interact in a phosphorylation-dependent manner (4,5). However, little is currently known regarding TGF␤ regulation of cytoplasmic components that are rapidly activated after receptor interaction with ligand. Twohybrid screens have indicated that immunophilin FKBP-12 and farnesyltransferase-␣ specifically bind RI and that a novel protein, termed TGF␤-receptor interacting protein-1, interacts with RII (6 -9). The functional significance of these receptor interacting proteins has yet to be elucidated.
We have reported direct evidence for the rapid activation of cytoplasmic signaling components by TGF␤ in a mammalian cell system. That is, we have shown that both Ras and Erk1 are rapidly activated by TGF␤ 1 and TGF␤ 2 in TGF␤-sensitive epithelial cells but not in TGF␤-resistant cells (1,2). These effects occurred in asynchronous cultures of epithelial cells under conditions where DNA synthesis was inhibited by 95-98% (2). The recent identification of the interaction between RI and farnesyltransferase-␣ mentioned above (7,8) suggests a potential upstream mechanism for the activation of Ras in the TGF␤ signaling pathway. The only other cytoplasmic signaling events that have been shown to be modulated by TGF␤ in untransformed epithelial cells are an activation of protein phosphatase 1 (10), an involvement of protein kinase C in early TGF␤ responses (11,12), and an association of phospholipase C with the elevation of gene expression by TGF␤ (12).
In contrast to the effects of TGF␤ on cytoplasmic signaling components, a direct association between TGF␤ modulation of nuclear cell cycle components and the growth inhibitory effects of TGF␤ has been demonstrated. The G 1 cell cycle events that have been shown to be mediated by TGF␤ in untransformed epithelial cells include an inhibition of p34 cdc2 synthesis, phosphorylation levels, and kinase activity (13)(14)(15), a reduction of cyclin-dependent kinase (Cdk) 4 synthesis (16), and an inhibition of mRNA expression of Cdk2, Cdk4, CKShs1, and cyclins E and D (17)(18)(19)(20). Additionally, TGF␤ has been reported to induce the expression of the cyclin-dependent kinase inhibitors p27 kip1 , p21 cip1 , and p15 ink4 and to enhance the association of these cyclin-dependent kinase inhibitors with the relevant Cdk complexes (21)(22)(23)(24)(25). Finally, a direct association between the growth inhibitory effect of TGF␤ and the ability of TGF␤ to suppress both Cdk2 activity and cyclin A mRNA expression has been described in several epithelial cell types, including keratinocytes, lung epithelial cell lines, and mammary epithelial cells (21,22,(25)(26)(27). In addition, down-regulation of either Cdk2 activity or cyclin A expression by TGF␤ has been used as a general indicator of the TGF␤-mediated growth inhibitory effect (25,27).
As discussed above, our previous results have indicated that activation of the cytoplasmic signaling components Ras and Erk1 occurred in parallel with the growth inhibitory effects of TGF␤. However, these cytoplasmic events have not yet been linked to TGF␤ regulation of the nuclear cell cycle components that mediate the growth inhibitory effect of TGF␤. The dominant-negative Ras mutant RasN17 has been used to determine whether blockade of the normal function of endogenous Ras proteins alters growth factor regulation of cellular events; these events include activation of Erks and regulation of gene expression, cell growth, and differentiation (28 -32). In the current report, we have utilized this dominant-negative Ras protein to demonstrate that the activation of Ras by TGF␤ is involved in the following events in untransformed epithelial cells: activation of Erk1, inhibition of Cdk2 activity, downregulation of cyclin A protein expression, and inhibition of DNA synthesis. Our results indicate that activation of Ras occurs upstream and is required for the activation of Erk1 by TGF␤. Further, Ras activation is partially required, but is not sufficient, for the TGF␤-mediated effects on Cdk2 activity, cyclin A protein expression, and DNA synthesis.
Transfection/Selection Procedure-The pM 2 NRasN17 plasmid containing the dominant-negative Ras (RasN17) under the control of a metallothionein promoter, as well as the empty vector (pM 2 N), have been described previously (31). The untransformed IEC 4-1 cell line was derived from IEC-18 cells as described previously (1,33). IEC 4-1 cells (1 ϫ 10 6 cells) were electroporated in basal medium (SM) with 5 g of either the pM 2 N or pM 2 NRasN17 plasmids at 400 or 450 V and 960 microfarads (34,35). Two days following transfection, G418 (131 g/ml; true concentration) was added to the medium for selection of positively transfected cells. G418-resistant colonies were pooled by trypsinization after 11-12 days of selection. The pool containing RasN17 was then cloned by limiting dilution and the N17C6, N17C5, and N17E3 clones were used for further studies. The IEC 4-1 cells and resulting transfected cells were maintained as described previously (1,33).
Induction of RasN17 Protein and Western Blot Analysis of RasN17 Expression-N17C6, N17E3, and M2N cells were plated at a density of 6660 cells/cm 2 in 25-cm 2 flasks. One day later, cells were rinsed once with Tris-buffered saline, pH 7.4, and once with SM. Cells were then incubated in the absence (control) or presence of ZnCl 2 (100 M was found to be optimal for induction of Ras expression) for 36 h. Western blot analysis of Ras protein expression was then performed using the Ras-10 antibody, as described previously (36).
Ras Activation-N17C6, N17C5, N17E3, and M2N cells were plated at a density of 6660 cells/cm 2 in 75-cm 2 flasks. Cells were then incubated in the presence or absence of ZnCl 2 for 36 h, as described above. After this time period, cells were treated with EIT (E ϭ EGF (10 ng/ml) ϩ I ϭ insulin (20 g/ml) ϩ T ϭ transferrin (4.0 g/ml)) in serum-free medium for 3 min. Analysis of Ras bound to GTP and GDP nucleotides was performed using the Ras thin layer chromatography assay as described previously (36).
In Vitro Erk1 Activity Assay-N17C6, N17C5, and M2N cells were plated and incubated with ZnCl 2 as for the Ras activation assay. Cells were then treated with TGF␤ (10 ng/ml) or EIT (described above) for 10 and 5 min, respectively. Erk1 in vitro kinase activity was determined as described previously (2).
In Vitro Cdk2 Histone H1 Kinase Assay-N17C6, N17E3, and M2N cells were plated and treated with ZnCl 2 as for the Ras activation assay. Cells were then treated with TGF␤ 2 for 6 h prior to determination of histone H1 kinase activity (16). Briefly, lysates were precleared with protein A-Sepharose for 30 min at 4°C. Total protein levels were normalized for each sample, and lysates were adjusted to a final concentration of 1 mM dithiothreitol before immunoprecipitation with 4 g of anti-Cdk2 antibody at 4°C for 2 h. Immune complexes were collected on protein A-Sepharose beads, washed five times with NET-N buffer (50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 200 mM EDTA, and 0.5% Nonidet P-40), and then twice with histone H1 kinase buffer (50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol). Beads were then resuspended in 50 l of H1 kinase buffer containing 10 M ATP, 10 g of histone H1, and 10 Ci of [␥-32 P]ATP and were incubated for 30 min at 30°C. Reactions were stopped by addition of 50 l of 2 ϫ SDS loading buffer, and samples were boiled for 5 min before separation of proteins by SDS-polyacrylamide gel electrophoresis (12%). Quantitation of radiolabeled histone H1 was determined by scanning of dried gels on a Betagen betascope 603 blot analyzer (Betagen Corp., Waltham, MA).
Cyclin A Western Blot Analysis-N17C6, N17E3, and M2N cells were plated and incubated with ZnCl 2 as for the Ras activation assay. Cells were then treated with TGF␤ 2 (10 ng/ml) for 12 h, followed by lysis in the assay buffer described previously (22). Total cell lysates were separated by SDS-polyacrylamide gel electrophoresis (10%); proteins were transferred to polyvinylidene difluoride. Cyclin A protein was detected by immunoblotting the polyvinylidene difluoride with a rabbit polyclonal cyclin A antibody (SC-751) for 1 h at 25°C, followed by a second incubation with rabbit IgG-horseradish peroxidase conjugate for 1 h at 25°C. Cyclin A protein was visualized by enhanced chemiluminescence.
Thymidine Incorporation-M2N, N17E3, and N17C6 cells were plated at a density of 1 ϫ 10 4 cells/cm 2 in 12-well dishes. The next day, cells were washed three times with SM and were incubated in the absence or presence of ZnCl 2 (100 M) for 60 h. After this time period, TGF␤ 2 (10 ng/ml) was added to the appropriate wells, and cells were incubated for an additional 24 h. Incorporation of [ 3 H]thymidine into DNA was then determined as described previously (2). The results are presented as x Ϯ S.E. (n ϭ 3).

Characterization of IEC 4 -1 Cells Transfected with the Dominant-negative RasN17-
We have previously shown that the IEC 4-1 cell line is exquisitely sensitive to the growth inhibitory effects of TGF␤ (1,33) and that TGF␤ treatment of these cells resulted in a rapid activation of the Ras and Erk1 signaling components (1, 2). Here we determined whether Ras activation was required for TGF␤ regulation of specific downstream events, including the activation of Erk1 and the inhibition of Cdk2 activity, cyclin A protein expression, and DNA synthesis. Accordingly, a dominant-negative RasN17 mutant, under the control of an inducible metallothionein promoter (31), was transfected into IEC 4-1 cells by electroporation. RasN17 transfectants were selected by treatment of transfected cultures with G418, and the resulting colonies were pooled and then cloned by limiting dilution.
In addition, IEC 4-1 cells were transfected with the empty vector (pM 2 N) by electroporation, and positive transfectants were selected in the presence of G418. The resulting cell line was designated M2N and was used as a control. In order to determine if ZnCl 2 treatment altered endogenous Ras protein expression, Western blot analysis of Ras proteins using a pan-Ras antibody was performed on cellular lysates obtained from M2N cells grown in the absence or presence of ZnCl 2 (100 M, 36 h). As shown in Fig. 1, A and B, ZnCl 2 treatment (100 M, 36 h) of M2N cells did not alter endogenous Ras protein expression. Thus, Western blot analysis with a pan-Ras antibody can be used to measure ZnCl 2 -induced alterations of RasN17 protein expression in the RasN17-transfected cells.
In order to determine the level of induction of RasN17 in the clones, we examined the ability of ZnCl 2 treatment to induce Ras protein expression. As indicated in Fig. 1, Ras protein levels in ZnCl 2 -treated (100 M, 36 h) N17C6 cells were as-sessed by Western blot analysis of cellular lysates using a pan-Ras antibody. Ras protein expression was increased to a level 4-fold above that in untreated N17C6 cells (Fig. 1, C and  D). Moreover, as a positive control to demonstrate that induction of the dominant-negative protein resulted in an alteration of endogenous Ras function, we investigated the ability of RasN17 to block EGF ϩ insulin ϩ transferrin (EIT)-induced activation of Ras in N17C6 cells. Previous studies have demonstrated that expression of RasN17 resulted in a blockade of EGF-and insulin-induced activation Ras (37). As shown in Fig.  2, EIT-induced activation of Ras in the control M2N cells was not altered by ZnCl 2 treatment (Fig. 2, left side). In contrast, EIT treatment of N17C6 cells, in the absence of ZnCl 2 , resulted in a 3.2-fold elevation in the % GTP bound to Ras. After ZnCl 2 induction of Ras by 3-4-fold in RasN17-transfected cells, a 40 -50% reduction in the proportion of Ras bound to GTP after EIT addition was detected (Fig. 2, right side). Similar results were obtained in two other clones, N17C5 and N17E3. Thus, ZnCl 2 treatment of RasN17-transfected cells, but not of cells transfected with vector only, resulted in both induction of RasN17 protein expression and a reduction in the ability of EIT to activate Ras.
We have previously shown that the growth-stimulatory combination of factors EIT stimulated Erk1 activity within 5 min of treatment of cycling IEC 4-1 cells (2). Moreover, RasN17 expression has been reported to inhibit EGF-and insulin-induced activation of Erk1 (31,38). Therefore, in order to further characterize the function of RasN17, in terms of its ability to alter EIT-induced activation of Erk1 in the N17C6 cells, we investigated whether expression of RasN17 in this clone would alter Erk1 phosphorylation of MBP following EIT treatment. As shown in Fig. 3, A and B, EIT treatment of cycling N17C6 cells, in the absence of ZnCl 2 , resulted in a 14-fold activation of Erk1. In contrast, when RasN17 expression was induced by at least 4-fold, Erk1 was activated by EIT to levels that were only 5-fold above base-line values. Thus, RasN17 inhibited growth-stimulatory factor activation of Erk1 in the N17C6 clone by 60%, compared with that in cells not expressing the mutant protein. This level of inhibition of EIT-induced Erk1 activation is within the range of that previously reported for RasN17 blockade of Erk1 activation by a variety of growth factors (31, 38 -40). Thus, the induction protocol employed here is suitable for examining the effect of RasN17 expression on TGF␤ regulation of downstream components.
Inhibition of TGF␤ 2 Activation of Erk1 by RasN17 Expression in Cycling Intestinal Epithelial Cells-We have previously demonstrated that treatment of untransformed epithelial cells with the growth inhibitor TGF␤ resulted in a rapid activation of Erk1 (2). In addition, a more dramatic activation of other mitogen-activated protein kinase (MAPK) family members has been observed in other epithelial-derived cell types 2 (41). Here we assessed the requirement for Ras activity in the activation of Erk1 by TGF␤. For these experiments, exponentially proliferating cultures of N17C6 cells were incubated in serum-free medium in the absence or presence of ZnCl 2 (100 M) for 36 h prior to TGF␤ (10 ng/ml) addition. Cell lysates were immunoprecipitated with either nonimmune rabbit IgG (nonspecific control, previously described in Ref. 2) or with an antibody specific for Erk1 (SC-93). Erk1 activity was then determined by in vitro phosphorylation of MBP, as described under "Experimental Procedures." Phosphorylation of MBP was visualized by autoradiography and quantitated by Betagen scanning. Additionally, equal loading of Erk1 was revealed by immunoblotting transferred proteins with the Erk1-specific SC-93 antibody (previously described in Ref. 2). As Fig. 3, A and C, depicts, TGF␤ 2 treatment of proliferating cultures of N17C6 cells, devoid of ZnCl 2 , resulted in a 2.1-fold increase in Erk1 activity within 10 min of growth factor addition. In contrast, Fig. 3, A and C, indicates that ZnCl 2 induction of RasN17 expression by at least 4-fold resulted in complete abrogation (100% blockade) of the Erk1 activation by TGF␤ 2 . Similarly, in a second clone (N17C5), induction of RasN17 expression by at least 5-fold (EIT-induced Ras activation was reduced by 40 -50%) resulted in complete inhibition of TGF␤-mediated Erk1 activation (data not shown). When the results from all RasN17-transfected clones treated with either TGF␤ 1 or TGF␤ 2 were evaluated, the RasN17-mediated abrogation of Erk1 activation by TGF␤ was statistically significant (p Ͻ 0.01, Student's t test). Furthermore, Erk1 activation by TGF␤ 1 and TGF␤ 2 was not altered by ZnCl 2 treatment in the M2N control cells. Hence, these results indicate that in exponentially proliferating epithelial cells, Ras is an essential upstream signaling component for Erk1 activation by TGF␤.
The Effect of Partial Induction of RasN17 Expression on Erk1 Activation by Different TGF␤ Isoforms-It was of interest to determine whether complete abrogation of Erk1 activation by TGF␤ 2 was dependent upon the level of RasN17 induction. Accordingly, Erk1 activity assays were performed in N17C6 cells under conditions that resulted in a reduced level of induction of RasN17 expression, compared with that utilized for Fig.  3. Moreover, since we have previously demonstrated that TGF␤ 1 also activated Erk1 (2), we examined whether Ras was similarly involved in the regulation of Erk1 by this TGF␤ isoform. In these experiments, proliferating cultures of N17C6 cells, grown in the absence or presence of ZnCl 2 , were treated for 10 min with TGF␤ 1 (10 ng/ml) or TGF␤ 2 (10 ng/ml). In the absence of ZnCl 2 , Erk1 enzymatic activity in the N17C6 cells was increased by 2.4-and 2.0-fold after addition of TGF␤ 1 and 2 R. S. Frey and K. M. Mulder, manuscript submitted for publication.

FIG. 1. Effect of ZnCl 2 treatment on Ras protein expression in control M2N cells (A and B) and in the N17C6 cell clone (C and D).
Proliferating cultures of intestinal epithelial cells, stably transfected with either empty vector (M2N) or RasN17 (N17C6), were incubated in the absence or presence of ZnCl 2 (100 M) for 36 h, as described under "Experimental Procedures." A and C, cell lysates were separated by SDS-polyacrylamide gel electrophoresis, and Western blotting was performed using the Ras-10 antibody, as described under "Experimental Procedures." Proteins were detected by enhanced chemiluminescence and exposure of the blots to autoradiographic film. B and D, plots of results of densitometric analysis of the blots depicted in A and C, respectively. TGF␤ 2 , respectively (Fig. 4). However, under conditions in which RasN17 expression was induced to a level only 2-fold above that in untreated cells, TGF␤ 1 and TGF␤ 2 -mediated Erk1 activation was inhibited by only 34 and 20%, respectively (Fig. 4). These results indicate that inhibition of Erk1 activation by TGF␤ 1 and TGF␤ 2 is dependent upon the level of RasN17 expression. That is, induction of RasN17 expression must be greater than 2-fold above that in uninduced controls in order to efficiently block Erk1 activation by TGF␤. Similarly, previous reports have demonstrated that RasN17 blockade of the effects of growth-stimulatory factors and of the differentiation factor nerve growth factor was dependent upon the level of expression of the mutant protein (28,29,39,42).

Partial Blockade of TGF␤ 2 Inhibition of Cdk2 Activity and Cyclin A Expression by RasN17 Expression in Cycling
Intestinal Epithelial Cells-In order to determine whether the Ras/ MAPK pathway was required for TGF␤ regulation of cell cycle components that mediate growth inhibition by TGF␤, we chose to examine Cdk2 activity and cyclin A protein expression. Both of these cell cycle components have been linked to the growth inhibitory effects of TGF␤ (16,21,(25)(26)(27). Moreover, TGF␤mediated decreases in Cdk2 activity and cyclin A mRNA expression have been utilized previously as general markers for TGF␤-mediated growth inhibition (25,27).
Although the retinoblastoma protein and the proto-oncogene c-myc have been cited as important TGF␤ growth-regulatory elements, direct links between TGF␤-mediated growth inhibition and modulation of these nuclear components by TGF␤ have not been observed in all cell types. While a reduction in the level of hyperphosphorylated retinoblastoma protein by TGF␤ has been linked to the anti-proliferative effects of this polypeptide in keratinocytes and lung epithelial cells (43,44), retinoblastoma protein levels were very low in IEC 4-1 cells, and the phosphorylation state was not modulated by TGF␤. 3 Moreover, retention of TGF␤ responsiveness despite a loss of functional retinoblastoma protein has been demonstrated in other systems (45)(46)(47). Additionally, in cycling epithelial cells, down-regulation of c-myc by TGF␤ can lead to cell cycle arrest through a loss of gene transcription required for G 1 progression (44, 48 -52). However, this TGF␤-mediated suppression of cmyc by TGF␤ does not appear to be necessary for TGF␤-regulated growth inhibition in all epithelial cell types, including IECs (53)(54)(55).
We first examined the effects of RasN17 expression on the ability of TGF␤ 2 to inhibit Cdk2-associated histone H1 kinase activity in the N17C6 clone and in one additional clone (N17E3). For these experiments, Cdk2 was immunoprecipitated from total cell lysates, and Cdk2 activity was determined by in vitro phosphorylation of histone H1, as described under "Experimental Procedures." Treatment of N17C6 and N17E3 (Fig. 5) cells with TGF␤ 2 (5 ng/ml) for 6 h (in the absence of ZnCl 2 ) resulted in a 55 and 60% inhibition of Cdk2 activity, respectively, compared with that in untreated controls. This level of inhibition is within the range previously reported for TGF␤ suppression of Cdk2 activity in mouse mammary epithelial and mink lung epithelial cells (15,16). In contrast, after 3 P. A. Zipfel and K. M. Mulder, unpublished data.

FIG. 2. Blockade of EIT-induced activation of Ras by RasN17 in the N17C6 cells.
Cycling N17C6 cells were treated with or without ZnCl 2 (100 M) for 36 h, as described in Fig. 1, followed by incubation with or without EIT for 3 min. The positions of cold GTP and GDP standard are indicated on the left, and the % GTP-bound Ras was calculated as described under "Experimental Procedures." Guanine nucleotides bound to Ras were visualized by autoradiography, and the % reversal of EIT activation of Ras was determined as indicated in the text.

FIG. 3. Effect of inducible expression of RasN17 on TGF␤-and EIT-induced activation of Erk1 in proliferating cultures of N17C6 cells.
A, exponentially proliferating cultures of N17C6 cells were treated with or without ZnCl 2 (100 M) for 36 h, as described for Fig. 1. Under these conditions, RasN17 protein expression levels were induced by at least 4-fold. Cells were incubated with or without TGF␤ 2 (10 ng/ml, 10 min) or EIT (5 min). Erk1 was immunoprecipitated from cell lysates with an antibody specific for Erk1 (SC-93). In vitro phosphorylation of MBP by Erk1 was analyzed as described under "Experimental Procedures," and phosphorylation of MBP was visualized by autoradiography. B and C, plots of Betagen scan of results in A.
induction of RasN17 expression, TGF␤ 2 suppressed Cdk2 activity in the N17C6 and N17E3 (Fig. 5) clones by only 27 and 32%, respectively, relative to the levels in control cells receiving no TGF␤ 2 . Thus, induction of RasN17 resulted in a 51 or 47% reversal of TGF␤ 2 inhibition of Cdk2 activity in the N17C6 and N17E3 clones, respectively (Table I). Our cumulative data indicated that this effect was statistically significant (p Ͻ 0.01). In contrast, ZnCl 2 treatment of M2N control cells did not affect the ability of TGF␤ 2 to inhibit Cdk2 activity (Table I). These results indicate that Ras is at least partially required for the TGF␤-mediated inhibition of Cdk2 activity.
In order to further examine the nature of the Ras requirement for TGF␤-mediated effects on the cell cycle, we chose to determine whether RasN17 expression would affect the ability of TGF␤ 2 to inhibit cyclin A protein expression, as was previously reported (22). For these studies, the TGF␤ effects on cyclin A protein expression levels in total cell lysates were examined by Western blot analysis, as described under "Experimental Procedures." As Fig. 6 depicts, in the absence of ZnCl 2 , treatment of M2N and N17C6 with TGF␤ 2 for 12 h resulted in an inhibition of cyclin A protein expression by 80 and 92%, respectively, compared with that in untreated control cells. Although incubation of M2N control cells with ZnCl 2 (100 M) did not reverse the effect of TGF␤ on cyclin A protein expression (Fig. 6A), ZnCl 2 induction of RasN17 in N17C6 cells treated with TGF␤ 2 for 12 h resulted in only a 37% inhibition of cyclin A protein expression (Fig. 6B). Thus, inhibition of cyclin A expression by TGF␤ was reversed by 60% (Table II). In a second clone (N17E3), TGF␤ 2 treatment resulted in an inhibition of cyclin A protein expression by 88% in the absence of ZnCl 2 , while only a 6% suppression was detected after induction of RasN17. Thus, a more dramatic reversal of the TGF␤ 2 -mediated inhibition of cyclin A protein expression was detected in this RasN17-transfected clone (94% , Table II). Taking data regarding all clones into account, the reversal of the TGF␤mediated suppression of cyclin A expression by RasN17 induction was statistically significant (p Ͻ 0.05). Thus, as for Cdk2 activity, Ras is partially required, but is not sufficient, for the TGF␤-mediated inhibition of this cell cycle protein.
It is noteworthy that the basal levels of Cdk2 activity and of cyclin A protein expression in the N17E3 clone were reduced upon incubation with ZnCl 2 (Fig. 5 and data not shown). This result is indicative of the clonal heterogeneity that was observed in the RasN17-transfected clones with regard to the ZnCl 2 effect on basal levels of these components, because other clones (i.e. N17C6) did not display this effect (see Fig. 6B). However, although this decrease in basal levels was substantial (sometimes as much as 50% of control levels), under other conditions, Cdk2 activity and cyclin A protein expression could be inhibited by TGF␤ to a greater extent than that observed by ZnCl 2 alone. For example, a 24-h TGF␤ 2 treatment of the N17E3 cells can suppress cyclin A expression to levels below the limit of detection. Taken together, these data suggest that the ZnCl 2 -dependent decreases in the basal levels of Cdk2  activity and of cyclin A expression observed in some of the clones did not preclude detection of any potential TGF␤-mediated decreases, during the RasN17 induction, beyond those mediated by ZnCl 2 alone. In addition to examining the ability of RasN17 induction to alter TGF␤ repression of both Cdk2 activity and cyclin A expression (events directly associated with suppression of cellular proliferation by TGF␤), we attempted to assess the effect of RasN17 expression on TGF␤-mediated inhibition of DNA syn-thesis. These studies were difficult to perform due to the heavy metal toxicity observed during the additional time required to detect the dominant-negative Ras effect on TGF␤-mediated growth inhibition. M2N cells treated in this manner displayed only a 5 Ϯ 0.9% reversal of the inhibition of DNA synthesis by TGF␤ in the presence of ZnCl 2 . In contrast, RasN17 induction resulted in a 21 Ϯ 7% reversal of the TGF␤-mediated inhibition of DNA synthesis in the N17E3 clone. This reversal was significantly different from that observed in the M2N control cells, as determined by the Student's t test (p Ͻ 0.05). A similar percent reversal was observed in the N17C6 clone. Hence, the cumulative results of our studies regarding Cdk2, cyclin A, and DNA synthesis suggest that Ras activation plays a role in TGF␤-mediated growth inhibition. However, the data also indicate that Ras is not sufficient for the regulation of these events by TGF␤. In this regard, Ras-independent pathways would be expected to be involved in the TGF␤-mediated growth inhibitory response in epithelial cells. DISCUSSION The results in this report demonstrate that Ras activation is obligatory for TGF␤ regulation of Erk1. That is, expression of the dominant-negative Ras mutant RasN17 in IEC 4-1 cells completely abrogated the ability of TGF␤ to activate Erk1. Furthermore, the extent of the RasN17 inhibition of Erk1 activation by TGF␤ was dependent upon the level of RasN17 protein that was expressed. Additionally, blockade of Ras activation resulted in a 50% reversal of the suppression of Cdk2 activity, a 78% reversal of the inhibition of cyclin A protein expression, and a 21% reversal of the inhibition of DNA synthesis by TGF␤. Collectively, these results demonstrate that RasN17 blockade of Ras activation alters the ability of TGF␤ to regulate a number of events leading to TGF␤-mediated growth inhibition.
Similar to the results reported herein, a requirement for Ras activity in specifying biological responses other than growth stimulation or mitogenesis has been demonstrated previously. That is, activation of Ras by nerve growth factor results in differentiation of PC12 cells (29). Moreover, Ras activation by the TGF␤ superfamily member activin was reported to induce the formation of mesoderm in Xenopus embryos (32). An involvement of Ras in controlling growth inhibition by other mechanisms has also been demonstrated previously. For example, the introduction of oncogenic Ras into rat embryo fibroblasts and Schwann cells caused growth arrest in both cell types (56,57). Thus, our results are an extension of those previously published with regard to other biological systems.
One may speculate that the activation of the Ras/MAPK pathway by TGF␤ results from activation of tyrosine kinase receptors, secondary to the secretion of growth stimulatory factors in response to TGF␤. In this regard, TGF␤ has been reported to induce the secretion of both transforming growth factor-␣ and platelet-derived growth factor in epithelial-derived cell types (61,62). Growth factors such as these are known to promote activation of both tyrosine kinase receptors and the Ras/MAPK signaling pathway. However, the conten-  tion that secretion of these growth factors is responsible for the Erk1 activation detected in TGF␤-treated epithelial cells is unlikely due to the rapid kinetics observed for the induction of Ras and Erk activation by TGF␤ (within 3-5 min) (1,2). Thus, the effect of TGF␤ on this pathway would appear to be a direct one.
As stated above, TGF␤-mediated Erk1 activation was completely abolished by RasN17 expression. This result suggests that Ras-independent pathways are not involved in the regulation of Erk1 activity by TGF␤. This finding contrasts with receptor tyrosine kinase signaling through the Erks, for which both Ras-dependent and -independent cascades lead to activation of these signaling components (58,59). Thus, it appears that while similar components are activated by mitogenic and growth inhibitory factors, differences exist in the mode of regulation of these components by the mitogenic receptor tyrosine kinases and the growth inhibitory receptor serine/threonine kinases. It is these differences that may mediate the divergent effects elicited by these receptor types.
Whereas TGF␤ activation of Erk1 is completely dependent upon Ras, we provide evidence that Ras is only partially required for inhibition of Cdk2 activity and of cyclin A expression by TGF␤. These results indicate that Ras-independent pathways contribute to the regulation of these nuclear components by TGF␤. Such Ras-independent signaling mechanisms may include those utilizing protein kinase C, phospholipase C, or even other members of the MAPK superfamily (i.e. the c-Jun N-terminal kinases) 2 (58 -60). In addition, unique TGF␤ pathways initiated by novel TGF␤ receptor binding proteins (such as TGF␤-receptor interacting protein-1) are also likely to contribute to the growth inhibitory effects of TGF␤ (9).
Taking into account the results reported here, as well as those previously published (1,2,63,64), it appears that both TGF␤ and growth stimulators activate the Ras/MAPK pathway, despite the different outcomes that occur with regard to cellular proliferation. A number of possibilities exist to explain these findings. First, the primary function of the activation of the Ras/MAPK pathway by TGF␤ may be to transcriptionally activate select classes of genes (i.e. TGF␤ 1 itself and/or other genes containing AP1 sites) as proposed previously (2), implying that the effect of this pathway on growth inhibition may be indirect. Second, in analogy to growth inhibitory pathways in yeast (65,66), the molecular determinants that specify growth responses may lie between the MAPK and Cdk/cyclin families of intracellular mediators in the signaling cascade. Along these lines, there is evidence that Raf/MAPK activation leads to up-regulation of p21 cip1 expression in mammalian cells, an effect mediating growth inhibition by the anti-cancer agent taxol (67). Third, synergistic signaling pathways may be activated simultaneously with the Ras/MAPK pathway to alter the final outcome. 4 In the case of TGF␤, such pathways may include those involving the mad genes or those utilizing TGF␤ signaling intermediates that are currently unknown 5 (68,69). Future studies, including those directed at the identification of novel TGF␤ signaling components, may provide clues as to which of these possibilities are correct.