Two Different Signal Transduction Pathways Can Be Activated by

THE JOURNAL OF B~omrclu. CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269, No. 18, Issue of May 6, PP. 13231-13237, 1994 Printed in USA.

Two Different Signal Transduction Pathways Can Be Activated by Transforming Growth Factor p l in Epithelial Cells*

(Received for publication, October 8, 1993, and in revised form, January 27, 1994)

Zhongfa Yan, Sidney Winawer, and Eileen Friedman$ From the Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, New York 10021

Signal transduction initiated by transforming growth factor p l (TGFp1) was studied in two sublines of the same colon carcinoma cell line, which respond in opposite ways to TGFp1, by proliferation or by growth inhibition. TGFPl activates ras proteins within 5 min of addition when it acts to inhibit growth but not when it acts as a mitogen. In both cases TGFpl also rapidly modulates the activities of three protein kinases, detected by their in gel kinase activity on the mitogen-activated protein kinase ( M A P kinase) substrate, myelin basic protein (MBP). When TGFpl acts as a mitogen for U9 cells, it increases the activity of MBP kinases of 57,105, and 130 kDa within 10 min of the addition without detectably activating ras proteins. When TGFPl inhibits the growth of HD3 cells, it activates ras proteins and the 57-kDa MBP kinase within 5 min but inhibits the activity of the 105- and 130-kDa MBP kinases. In HD3 cells ras activation occurred in two signal transduction pathways, one from TGFPl leading to growth inhibition and one from epidermal growth factor (EGF) leading to proliferation. In addition to ras proteins, EGF activates a different set of MBP kinases in HD3 cells than does TGFp1, MBP kinases of 85,57, and 44 kDa. The latter is likely to be the 44-kDa MAP kinase extracellular signal- regulated kinase (erk) 1, because EGF treatment of HD3 cells activates erkl by increasing its phosphotyrosine level. Therefore, in two closely related epithelial cell lines TGFpl activates two different signal transduction pathways, one ras-dependent and one ras-independent, and modulates the activities of a set of MBP kinases.

A series of recent studies has demonstrated a central role for rus proteins in signal transduction initiated by activation of growth factor receptors, such as the EGF’ receptor and the insulin receptor, leading to cell proliferation (1-9). This common signaling pathway, including GRB2, guanine nucleotide exchange factor (SOS in Drosophila), rus, and ruf, leads to activation of p42ip44 MAP kinases through phosphorylation and activation of a dual function threonineltyrosine kinase, MAP kinase kinase, or MEK (10-22). Therefore, the activation

* Supported by National Cancer Institute Grants R01 CA50645 and CA45783 (to E. F.) and National Cancer Institute Gastrointestinal On- cology Training Grant IT32CA09628 (to S. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “uduertisement” in ac- cordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom all correspondence should be addressed: Memorial Sloan- Kettering Cancer Center, Box 244, 1275 York Ave., New York, N. Y.

extracellular signal-regulated kinase; MAP kinase, mitogen-activated ‘The abbreviations used are: EGF, epidermal growth factor; erk,

protein kinase; MBP, myelin basic protein; PAGE, polyacrylamide gel electrophoresis, TGFp1, transforming growth factor pl; DMEM, Dul- becco’s modified Eagle’s medium; PIPES, 1,4-~iperazinediethanesul- fonic acid.

10021. -1.: 212-639-2837; Fax: 212-717-3053.

of rus proteins conveys mitogenic signals from growth factors to MAP kinases (23-27). When activated by phosphorylation on tyrosine and threonine (28-32) MAP kinases can translocate into the nucleus to phosphorylate and activate various transcription factors leading to cell proliferation (33-34). The TGFp family of hormonal polypeptides, in contrast to mitogenic growth factors like EGF and insulin, strongly inhibits the proliferation of epithelial cells, making TGFp appear as an un- likely initiator of rus activation in epithelial cells. However, rus activation has been reported to mediate a growth-inhibitory signal from TGFPl in two epithelial cell lines, the 4-1 intesti- nal cell line and CCL64 mink lung cells (35). TGFp receptors have recently been cloned by several groups (36-401, but the signal transduction pathway activated by TGFp has not yet been elucidated. We have cloned several epithelial cell lines that display one of two opposite responses to TGFp1, growth inhibition as expected of normal epithelial cells, or an unexpected response, growth stimulation (41, 42). By using these well characterized cell lines, we have asked if ras activation occurs in both signal transduction leading to growth inhibition and in signal transduction leading to growth stimulation. The study was then extended to determine whether either TGFp1- initiated signal transduction pathway activated any MAP kinases. The latter are assayed by their activity on the exogenous substrate, MBP, and so designated as MBP kinases. To mini- mize the possibility that any differences observed in either activation pathway could be due to extraneous genetic changes due to tumor cell heterogeneity, we have chosen lines of colon carcinoma cells that have been derived from the same tumor cell line, have the same mutations in two colon cancer genes, APC (adenomatous polyposis coli) and p53, and have no activating mutations in rus proteins.’ In this communication we show that the two TGFpl-initiated signal transduction systems differ in their activation of ras proteins and MBP kinases.

EXPERIMENTAL PROCEDURES M~teriuls-~~PO,, [y-32P]ATP, [3SSlmethionine, and [CY-~~PIGTP were

obtained from DuPont NEN human platelet TGFPl was from R&D Systems (Minneapolis, MN); EGF was from Collaborative Research, Bedford, MA, antisera Y13-259 was from Oncogene Science Inc. (Manhasset, NY); protein A-Sepharose was from Pharmacia LKB Bio- technology Inc.; polyvinylidene difluoride transfer paper Immobulin-P was from Millipore Corp.; polyethyleneimine cellulose F thin-layer chromatography plates were from EM Separations, Gibbstown, NJ. Rabbit polyclonal anti-rat erkl (R2), sheep polyclonal anti-rat erk3 (p63), and protein tyrosine phosphatase-1B were purchased from UBI, Lake Placid, NY. Rabbit anti-sheep IgG coupled to alkaline phosphatase and mouse monoclonal antibody to erkl and erk2 were purchased from Zymed Labs., Inc. (San Francisco, CA). Rabbit polyclonal antisera 691 was raised against a synthetic peptide corresponding to a sequence in conserved subdomain XI of erkl(27) and was the kind gift of Dr. Mela- nie Cobb, Southwestern Medical School (Dallas, TX). Monoclonal antibody RK2 to the EGF receptor was obtained from Dr. J. Schlessinger, New York University Medical Center.

Z. Yan, S. Winawer, and E. Friedman, manuscript in preparation.

13231

13232 Signal Dansduction Pathways Activated by TGFPl Cell Culture-The U9, HD3, and HP1 human colon carcinoma cell

lines were subcloned from the HT29 cell line and maintained in DMEM containing 7% fetal bovine serum, as described (41). HD3 cells were synchronized in early G1 by culture a t confluent density for two days followed by release by plating a t one-third confluent density on collagen I-coated Petri dishes in serum-free insulin-transferrin-selenous acid- supplemented DMEM medium (42).

Determination of ras GTPIGDP Ratio-Cells were cultured in 100-mm dishes until subconfluent in modified DMEM supplemented as described (41) and then labeled overnight with 200 pCi of 32P-labeled H,PO, in serum-free phosphate-free insulin-transferrin-selenous acid- DMEM (41). Cells in each dish were lysed in 0.5 ml of buffer consisting of 0.5% Nonidet P-40, 50 mM Tris-HC1, pH 7.5, 20 m MgCl,, 150 m~ NaCI, 100 p~ GTP, 100 p~ GDP, 1 m ATP, 1 m~ Na,PO,, pH 7.4, 10 pg/ml aprotinin, 10 pg/ml leupeptin, 10 m benzamidine, io pg/ml soybean trypsin inhibitor, 1 mM phenylmethylsulfonyl fluoride, and 50 pg of affinity-purified monoclonal antibody Y13-259. The lysate was then incubated for 10 min a t room temperature with 50% w/v norit A charcoal to absorb free nucleotides. p21m"-Y13-259 complexes were precipitated by the addition of 20 pg of rabbit anti-rat IgG (Cappel) per 1 mg of total protein and 40 pl of protein A-Sepharose. After washing the immunoprecipitates, GTP and GDP were eluted with 20 pl of 1 M KH,PO,, pH 3.4, with heating for 3 min a t 90 "C. The relative amounts of ras-bound GTP and GDP were determined both by autoradiography and by direct scanning for p emissions after thin-layer chromatography performed on polyethyleneimine cellulose F (EM Separations) developed with 1 M KH,PO,, pH 3.4 (43). For initial rus immunoprecipitations, cells were prelabeled overnight with [3sS]methionine (625 pCi/ml, 1220 Ci/mmol).

Kinase Assays in MBP-containing Polyacrylamide Gels-The method is adapted from one previously used (44,45). Cell lysates were boiled in Laemmli sample buffer for 2 min, then electrophoresed in a 7.5% SDS- PAGE (0.5 mm thick and 5 cm long) containing 0.5 mg/ml MBP (Sigma). After fixing the gel with four changes of 20% 2-propanol in 50 m Tris-HCl buffer (pH 8.0) for 2 h, SDS was removed by washing the gel for 2 h in several gel volumes of 50 mM Tris-HC1 (pH 8.0) containing 5 mM 2-mercaptoethanol, with frequent changes. The MBP kinases were then redenatured with 6 M guanidine HCl for 2 h and then renatured by 10 washes of 20 min each in several gel volumes of 50 n" Tris-HC1 (pH 8.0) containing 0.04% Tween 40 and 5 m~ 2-mercaptoethanol. After preincubation for 1 h with 5 ml of 40 m~ HEPES (pH 8.0) containing 2 m mercaptoethanol and 10 m MgCI,, phosphorylation of MBP within the gel was carried out by incubating the gel a t room temperature for 1 h in 5 ml of 40 m~ HEPES, pH 8.0, containing 25 pCi of [y32PlATP, 40 p~ ATP, 0.5 m EGTA, and 10 m~ MgCI,, and then washing the gel in 5% (w/v) trichloroacetic acid containing 1% sodium pyrophosphate several times until the radioactivity reached background levels, exactly as described in Refs. 44 and 45.

Cell Permeabilization-The method was adapted from Buday and Downward (46). Cells were cultured 2 days postplating at 4 x 105/cm2 in 10-cm tissue culture dishes to bring cells into log phase. The medium was then changed to serum-free insulin-transferrin-selenous acid- DMEM, and cells were cultured overnight. After 1 x wash with phosphate-buffered saline, cells were placed in 2.4 ml of permeabilization buffer consisting of 150 m KCl, 37.5 m~ NaCl, 6.25 m MgCl,, 0.8 m~ EGTA, 1 mM CaCI,, 1.25 m ATP, 12.5 m PIPES (pH 7.4), and incubated a t 37 "C for 10 min. 0.6 ml of 2 international unitdml streptolysin 0 (Sigma) in permeabilization buffer was then added and the incubation continued for 5 min. 15 pCi of [a-3ZPlGTP (300 Ci/mmol) and 5 ng/ml TGFpl were added and incubated for the times indicated. The buffer was then removed, the cells lysed, and the ras proteins immunoprecipitated as above. Total specific radioactivity associated with immunoprecipitated rus proteins was detected by a Beckman 2000 counter. Samples were assayed in triplicate.

Zmmunodetection-Proteins blotted onto polyvinylidene difluoride membranes were detected using a 1:4000 dilution of anti-erkl and -erk2 (Zymed), 1 pg/ml antiphosphotyrosine monoclonal antibody 4G10, or a U125 dilution of anti-EGF receptor antisera RK2 followed by several washing steps, then either incubation for 30 min a t 25 "C with 1 pCi/ml lZ5I-protein A, followed by washing and autoradiography or a l/2000 dilution of rabbit anti-mouse IgG coupled to alkaline phosphatase, and developed as described (42). Qrosine phosphatase treatment of lysates was performed a t 37 "C for 30 min using the supplier's protocol and removed from the lysates before SDS-PAGE by pelleting the protein tyrosine phosphatase-1B bound to glutathione-agarose beads.

ERKl immunoprecipitations, 1 mg/ml total protein lysates of synchronized HD3 cells, 6 h postrelease (42). were diluted 4.5-fold with distilled water containing protease inhibitors, 100 m NaF, 200 p~

HD3 U9

Y13-259 qq p21 "c

FIG. 1. The HD3 and US colon carcinoma cell lines were prelabeled with [%]methionine, then immunoprecipitated with anti- ras monoclonal antibody Y13-259 (+) or rat IgG (-) in the control, and then analyzed by SDS-PAGE and autoradiography of the dried gel.

sodium orthovanadate, 10 pg/ml aprotinin, 10 pg/ml leupeptin, and 1 m~ phenylmethylsulfonyl fluoride, and SDS was added to 0.15%. Lysate proteins were incubated with 10 pVml of antisera 691 for 1 h a t 4 "C and then incubated with 40 pl of protein A-Sepharose coupled to rabbit anti-mouse IgG for another hour. After centrifugation the bound proteins were washed four times with lysis buffer containing 0.025% SDS, then boiled in Laemmli sample buffer, subjected to SDS-PAGE, and detected by alkaline phosphatase reaction.

RESULTS

ras Is Activated in Both TGFpl Growth-inhibited and EGF- stimulated Epithelial Cells but Not in TGFpl Growth-stimu- luted Epithelial Cells-Two types of TGFpl-responsive cell lines had been subcloned from the HT29 human colon carcinoma line (41). The HD3 subline responds to TGFPl by growth inhibition correlated with a block in phosphorylation of the retinoblastoma gene product, whereas the more invasive and tumorigenic U9 and HP1 lines respond to TGFpl by growth stimulation with an increase in phosphorylation of the retinoblastoma protein (41, 42, 47). Analysis of immunoprecipitates from both HD3 and U9 cells shows that p21" proteins can be readily detected as a double band of equal intensity (Fig. 1). None of the p21"" proteins in these cell lines bind elevated levels of GTP as none contain ras proteins mutated to activated forms.' For example ras proteins in the HD3 line bind less than 1% of the total guanosine nucleotides as GTP compared with over 50% in the SKCOl line, which has a mutated and activated k-ras gene (data not shown). EGF is known to induce mitogenesis by a signal transduction pathway that includes ras (1-7). Because EGF stimulates the growth of HD3 cells we compared ras activation in parallel cultures of HD3 cells 5 min after addition of either a mitogenic concentration of EGF or a growth-inhibitory concentration of TGFp1. Both growth factors activated ras proteins, increasing their binding of GTP (Fig. 2, left panel). A 5-min treatment of HD3 cells with either TGFPl or EGF caused an increase in the ratio of GTP to total guanine nucleotides bound to ras p21 proteins (Fig. 2, one of identically duplicate experiments). Thus ras activation was a downstream event in both growth-inhibitory and growth-stimulatory signals in the same cell. In contrast, when TGFPl was added for 5 min to the TGFpl growth-stimulated line, U9, there was no increase in the percentage of nucleotides bound to p21" that were [32P]GTP (Fig. 2, one of three duplicate experiments; identical data also obtained for a second TGFpl growth-stimulated line, HP1, but not shown). EGF did activate ras proteins in the U9 line (Fig. 2, one of identically duplicate experiments). Thus the ras proteins in U9 epithelial cells were capable of transmitting a growth modulating signal from EGF, but not from TGFp1, while ras proteins in HD3 cells were able to transmit signals from both growth factors, from EGF to stimulate growth, and from TGFPl to inhibit growth.

The observation that ras was activated in epithelial cells only when TGFPl led to growth inhibition, but not when it led to growth stimulation, was confirmed with two time-course experiments, in which TGFpl was added for 5,10, 15, or 20 min. In both cases, the fraction of GTP to the total of GDP plus GTP bound to p21" proteins increased approximately 40-fold within

Signal Dansduction Pathways Activated by TGFpl 13233

HD3 u9 EGF ' - - + ' ' " + '

TGFPl - + - e + - a "

GTP - # 4 FIG. 2. Autoradiogram of thin-layer chromatograms of GTP

and GDP molecules bound to ras proteins immunoprecipitated from s2P0,-prelabeled U9 and HD3 cells. Cells were treated for 5 min with either 5 ng/ml TGFpl, 10 ng/ml EGF, or untreated before extraction for immunoprecipitation.

5 min in the growth-inhibited HD3 cells (Fig. 3, entire thin- layer chromatography shown to demonstrate equal loading). The increase was maximal after 5 min and then slowly declined (Figs. 3 and 4). In contrast, the TGFpl-induced proliferative signal in U9 cells induced no increase in GTP bound to ras proteins (Fig. 4).

Asecond method of measuring ras activation was then used. The intrinsic rate at which ras proteins exchange guanine nucleotides is very low, but can be increased by EGF in permeabilized rat-1 fibroblasts (46). Both HD3 and U9 cells were permeabilized by the bacterial toxin streptolysin 0 and then [CY-~~PIGTP was added to the cells along with TGFp1. After 0-, 5-, lo-, or 15-min aliquots of the cells were lysed with detergent in the presence of excess-unlabeled GTP, and the ras-bound- labeled GTP molecules were precipitated with antibody Y13- 259 (see "Experimental Procedures"). The amount of labeled GTP specifically bound to ras proteins increased over 2-fold when HD3 cells were treated with TGFPl for 5 min in each of two experiments (Fig. 5). The amount of bound GTP decreased slowly with longer incubations. In contrast, no increase in GTP bound to ras proteins was induced in TGFpl growth-stimulated U9 cells (Fig. 5). The low level of GTP associated with ras proteins in permeabilized HD3 cells compared with prelabeled nonpermeabilized cells (Fig. 4) was also seen in the EGF-stimulated rat-1 cells (46) and was probably due to the high level of nucleotide triphosphatase activity in permeabilized cells. Thus two different methods demonstrated that the TGFpl signal transduction pathway in TGFpl growth-inhibited epithelial cells used activated ras proteins. In contrast no activation of ras proteins could be detected in signal transduction initiated by TGFp1, which resulted in epithelial cell proliferation.

Different Patterns of MBP Kinase Activation in TGFpl Growth-stimulated and TGFpl Growth-inhibited Cells-We then compared the MBP kinases activated by TGFpl treatment in TGFPl growth-inhibited HD3 cells and TGFPl growth- stimulated U9 cells. One major effect of TGFpl in HD3 cells was induction of a 57-kDa MBP kinase, which was maximal after 5 min of TGFPl treatment and then declined by 10 min and further by 15 min of treatment (Fig. 6). The activation of the 57-kDa MBP kinase by TGFpl was parallel to the time course of ras activation in these cells by TGFPl (Figs. 3-51. The activities of MBP kinases of 130 and 105 kDa decreased with TGFpl treatment in HD3 cells (Fig. 6, one of duplicate experiments with identical results). The pattern of MBP kinase activation when U9 colon carcinoma cells were growth-stimulated by TGFpl differed from that seen in TGFpl growth-inhibited HD3 cells. Instead of the decrease in 130- and 105-kDa MBP

HD3 mini 0 5 10 15 20 I

GDP-

GTP-

Ori- we@. labeled HD3 cells treated with 5 ng/ml TGFpl for 0,5, 10, 15, or

FIG. 3. Ras proteins were immunoprecipitated from '?O,-pre-

20 min, as indicated. The rus-bound nucleotides were then eluted from the immunoprecipitates (see "Experimental Procedures"), and the bound GDP and GTP molecules were separated by thin-layer chromatography and autoradiographed. Entire chromatogram shown to demonstrate equal loading.

n c (5

4-

3-

2-

1-

I

0 1 0 2 0

Minutes After TGFR1 Addition FIG. 4. Time course for activation of ras proteins by TGFpl in

TGFPl growth-inhibited HD3 cells. rus activation was not detected in TGFPl growth-stimulated U9 cells. The ratio of GTP to the total GDP plus GTP bound to immunoprecipitated rus proteins is given, as determined by direct scanning for p emissions of thin-layer chromatograms.

kinases observed during TGFpl growth inhibition, growth stimulation by TGFPl was correlated with increased activity of both the 105- and 130-kDa MBP kinases. Because ras proteins were not detectably activated even after 15-20 min of treatment with TGFpl (Figs. 3-5), activation of the 130-, 105-, and 57-kDa MBP kinases must have occurred by a ras-independent pathway. Common to both TGFpl-mediated responses was an increase in the 57-kDa MBP kinase activity but with a different time course. The increase occurred only after 10 min of TGFpl treatment in the growth-stimulated U9 cells, while an increase in activity peaked at 5 min and then decreased in the growth- inhibited HD3 cells (Fig. 6). Thus in cells with different biological responses to TGFp1, this growth factor induced a different pattern of activation or inhibition of the same three major MBP kinases.

Different MBP Kinases Are Activated by TGFpl and by EGF in Epithelial Cells-The known MAP kinases of 42,44,54, and 57 kDa all have kinase activity on the substrate MBP (23-25, 44, 45) and so can be designated MBP kinases. The MBP kinases of 105 and 130 kDa modulated by TGFPl have not been identified by other investigators, to our knowledge, and may or may not be related structurally to the cloned MAP kinases, p42

13234

- 7 E Y 0

Signal Dansduction Pathways Activated by TGFpl MBP Kinase

EGF I - - + ‘ TGFPl - Kd

+ - a 2 400-

21 5.5 - F h( 105.1 - -g a 69.8 - 0 43.3 - U 25.3 - c ¶ 0 18.1 - P

-4 c. m-

a 200- Q immunoprecipitations (Fig. 2, l e f t ) were treated for 5 min with E either 10 ng/ml EGF or 5 ng/ml TGFpl then analyzed for activa- 0 tion of MBP kinases by the in gel assay (see UExperimental Pro-

FIG. 7. Parallel HD3 cell cultures to those used for rae protein

c.

- Q cedures”). 0

z = loo i

0 1 0 2 0 crease in tyrosine phosphorylation was observed in a 44-kDa

Minutes after TGFBl Addltlon FIG. 5. Time course for activation of ras proteins by TGFf31 in

TGFf3l growth-inhibited HD3 cells. rus activation was not detected in TGFPl growth-stimulated U9 cells. The mean 2 S.E. of the counts! min of [3zP]GTP, which entered permeabilized cells and bound to immunoprecipitated rus proteins in cells treated with TGFPl for the times indicated.

HD3 u9 min 0 5 10 15 0 5 10 15

KD 21 5-

105 - 69.8 -

43.3 - 28.3 -

F’IG. 6. Time course of modulation of M6P kinase activities in HD3 cells and in U9 cells treated for the times indicated (0,5,10, and 15 min) with 5 ng/ml TGFpl before cell lysis and assay.

(erk2) and p44 (erkl). Because we had determined that both TGFPl and EGF activated ras proteins in HD3 cells, we then asked first, whether TGFpl and EGF activated any of the same MBP kinases and second whether any of these MBP kinases could be identified as any of the known MAP kinases. Mitogenic stimulation by EGF induced increases in the activity of MBP kinases of 85 and 44 kDa, with an increase in a major MBP activity at 57 kDa (Fig. 7, parallel cultures to Fig. 2, left, one of three experiments with same results). However, MBP kinase activities of 130 and 105 kDa and a minor activity of 42 kDa were not measurably changed by EGF treatment.

The MAP kinases of p42, p44, p54, and p57 are activated by tyrosine phosphorylation (28-32, 44, 45). We first determined whether there was an increase in phosphotyrosine content of any proteins that co-migrated with any of the MBP kinases activated by TGFPl or by EGF in HD3 cells. The major MBP kinases in these epithelial cells exhibited lower activities in the early G1 phase than at later times in the cell cycle (data not shown), so if any were activated by tyrosine phosphorylation, it would be easiest to detect in cells synchronized in early G1. Synchronized (see “Experimental Procedures”), early G1 HD3 cells were treated with either growth-stimulating levels of EGF for 5, 10, or 20 min or with growth-inhibitory levels of TGFpl for the same times.

In lysates from synchronized EGF-treated HD3 cells, an in-

band (Fig.”, arrow) that-co-migrated with erkl (data not shown) and in a 57-kDa band (Fig. 8 A , arrow). An increase in tyrosine phosphorylation was also seen in bands that co-migrated with the EGF receptor at 170 kDa (Fig. 8C), indicating activation of this receptor, as expected. In contrast, TGFP1- treated HD3 cells exhibited no detectable induction or increase of any tyrosine-phosphorylated band (Fig. EA). Treatment of lysates from EGF-stimulated HD3 cells with tyrosine phosphatase protein tyrosine phosphatase-1B eliminated phosphotyrosine only from bands migrating at 170,57, and 44 kDa (Fig. 8B, arrows), confirming their identification as phosphotyrosine-containing proteins. The apparent tyrosine phosphorylation in bands at 130,63, and 54 kDa present constitutively in all HD3 cell lysates was an artifact (Fig. 8, A and B ) .

To confirm that erkl (p44 MAP kinase) was activated in HD3 cells by EGF, e rk l was immunoprecipitated from EGF-treated and control HD3 cells and its tyrosine phosphorylation analyzed by Western blotting. Similar amounts of erkl protein (p44) were immunoprecipitated from EGF-treated and untreated cells, as shown by immunoblotting the immunoprecipitates with a different anti-erkl,2 antisera (Fig. 80, lanes 13 and 14). Only the erkl from EGF-treated cells exhibited detectable tyrosine phosphorylation, demonstrating its activation (Fig. 8D, Canes 16 and 17). Thus the 44-kDa MBP kinase activated by EGF in these epithelial cells was very likely to be erkl. The anti-peptide antisera used for the immunoprecipitations had been raised to a conserved domain in erkl also found in erk2 (27), so a low abundant tyrosine-phosphorylated band also observed in the immunoprecipitates at 42 kDa was probably erk2.

EGF treatment of HD3 cells increased tyrosine phosphorylation of a 57-kDa band (Fig. 8 A ) that comigrated with the 57-kDa MBP kinase activated by both EGF and TGFpl (Fig. 7). EGF induced much more activity of this kinase than TGFpl (Fig. 7), so the lack of a detectable increase in tyrosine phosphorylation of p57 after TGFPl treatment (Fig. 8 A ) may simply reflect this quantitative difference in activation. We had previously identified a 57-kDa W M B P kinase in HD3 and U9 cells (44, 45) and have recently purified p57 MAP kinase for microsequencing and cloning. When p57 is cloned, we will be able to determine whether the 57 MBP kinases activated by TGFj3l and by EGF are identical and whether this p57 MBP kinase is the same p57 MBP kinase activated by basic fibroblast growth factor and diolein in U9 cells (44,451. The 57-kDa MBP kinase is not erk3, which migrated at 70 kDa in parallel blots and unexpectedly could only be detected by Western blotting with the polyclonal antisera used after tyrosine phosphatase treatment (data not shown). Thus, following treatment of HD3 cells with the mitogen EGF or with the growth inhibitor

Signal Tkansduction Pathways Activated by TGFpl 13235

A B C D EGF TGFP1 Control PTP-1 B LL aERKl +ERK2 aPTy 5

Min mmw KD KD KD EGF' - + c ' KD - 21 5.5 - 218.3 - 218.3 - 218.3

+

- 71.8 -+

- 100.5

- 69.8 - 71.8 - 71.8

- 43.2 -b - 43.3 - 43.2 " - - - 43.2

- 28.3

- 18.4 - 15.3

- 28.3 - 28.3

- 18.4 - 15.3

- 28.3

- 18.4 - 15.3

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

6 h after release from growth arrest had progressed to early to mid-G1 and then were treated for 5,10, or 20 min with 10 ng/ml EGF ( h n e S 1 3 ) , FIG. 8. A, analysis of t o t a l cell lysates on SDS-PAGE by Western blotting with anti-phosphotyrosine antibody. Density-synchronized HD3 cells

8 ng/ml TGFpl (lanes 4 4 , or left untreated but lysed in parallel with the treated samples (lanes 7-9). Arrows at left margin, proteins showing increased tyrosine phosphorylation. B, lysates from 5 min EGF-treated HD3 cells (parallel to lane 3 ) were treated with tyrosine phosphatase-1B (+, lane 11) or left untreated (lane IO) before SDS-PAGE and immunoblotting with anti-phosphotyrosine antibody and alkaline phosphatase detection. Arrows at left margin, proteins showing increased tyrosine phosphorylation. C, lune 12, parallel lane to lane 1, blotted with anti-EGF receptor antibody, detected by alkaline phosphatase reaction. D, lysates from 5 min EGF-treated HD3 cells (lanes 14,15,17, and 18) or untreated HD3 cells (lanes 13 and 16) were immunoprecipitated wih anti-erkl polyclonal antisera 691 (lanes 13,14,16, and 17) or with control preimmune rabbit serum (panel C, lanes 15 and 18), subjected to SDS-PAGE, then immunoblotted with either anti-erkl.2 antisera (lanes 13-15) or anti-phosphotyrosine antibody (lanes 16-18). Arrows at right of each gel indicate molecular weight markers.

TGFp1, a different set of MBP kinases were effected. TGFpl activated a p57 MBP kinase but inhibited the activity of 105- and 130-kDa MBP kinases. EGF, in contrast, did not modulate the activity of the 105- and 130-kDa MBP kinases but activated three MBP kinases of 85,57, and 44 kDa, the latter being erkl.

DISCUSSION Mammalian ras genes code for 21-kDa membrane-associated

guanine nucleotide-binding proteins. ras proteins, like other GTP-binding proteins, are activated biochemically only when bound to GTP and are inactive when bound to GDP. ras proteins are capable of transmitting mitogenic signals from platelet-derived growth factor, EGF, and serum, as shown by injec- tion of neutralizing antisera to ras (47). More recent biochemical studies have demonstrated a central role for GTP- bound ras proteins in signaling from tyrosine kinase receptors such as the EGF receptor and the insulin receptor (1-9). This common signaling pathway includes guanine nucleotide exchange factors and growth factor receptorlras-linking factors as GRB2, which function upstream of ras proteins, and raf and MAP kinase kinases or MEKs, which transmit signals from activated ras proteins in a protein kinase cascade to MAP kinases (10-22). Activated ras proteins have also been shown to transmit nerve growth factor-initiated mitogenic signals in PC12 cells (10, l l ) and to transmit interleukin 2, interleukin 3, and granulocytdmacrophage colony-stimulating factor mitogenic signals in lymphoid and myeloid cells (48).

In all of these studies ras activation mediated mitogenic signaling. I t was therefore unexpected that the epithelial growth inhibitor, TGFp1, would also activate rus, as shown by Mulder and Moms (35) in each of two epithelial cell lines. In this study we have confirmed Mulder's observations in another TGFpl-inhibited epithelial cell line, HD3, and have extended the observations in two ways. First, we have compared ras activation by TGFpl in two closely related epithelial cells with opposite responses to TGFp1, growth inhibition, or stimulation. Within 5 min of the addition of TGFp1, the signal transduction pathway leading to growth inhibition activated ras proteins, while no ras activation was detected in the TGFpl growth-

stimulated cells for 5-20 min after the addition of TGFp1. The ras proteins in TGFPl growth-stimulated U9 cells were functional as they were capable of transmitting an EGF-initiated signal. Thus ras activation-mediated TGFpl initiated epithelial cell growth inhibition but did not mediate proliferative signals initiated by TGFpl in epithelial cells although these cells contained functional ras proteins.

The second part of our study identified the MBP kinases activated following TGFpl treatment of growth-stimulated and growth-inhibited epithelial cells. MBP kinases were identified by their kinase activity on the MAP kinase substrate MBP in an in gel assay that allowed determination of their molecular weights. MAP kinases occupy a focal point in signal transduction converting tyrosine phosphorylation signals induced by different growth factors and other stimuli into biochemical events modulated by serinelthreonine phosphorylations. MAP kinases of 42,44, 54, 57, 63, and 97 kDa have been identified, and several have been cloned (23-26,44-45,51,52). The MAP kinases of 42, 44, 54, and 57 are activated by phosphorylation on tyrosine and threonine by a dual function kinase, MEK (10-22). MEK can be activated by either of two pathways, one using ras and raf and the other not (53). The alternative pathway in some cells is activated by heterotrimer G protein-linked receptors, which activate MEK in a raflras-independent man- ner, possibly through MEK kinase (53).

The addition of a growth-stimulating concentration of TGFPl did not detectably activate ras proteins within 20 min of addition but within this time frame increased the activity of three MBP kinases of 57,105, and 130 kDa. These proteins may be novel MAP kinases, because they are mitogen-activatedpro- tein kinases. The 105- and 130-kDa species have not been described previously to our knowledge, and their structural rela- tionship, if any, to the well known 42- and 44-kDa MAP kinases, erkl and erk2, is unknown. Antibody made to a conserved sequence in erkl and erk2 (antibody 691, Fig. 8 0 ) did not co-precipitate any molecules with molecular weights around 57, 105, or 130 kDa. The 57-kDa MBP kinase may be the 57-kDa MAPfMBP kinase that we previously identified in

13236 Signal fiansduction Pathways Activated by TGFpl TABLE I

Summary of induction of MBP kinases

Growth factor

EGF TGFpl TGFpl

Cell line

ras-activated

Modulation of MBP kinases

HD3 HD3 u9

Yes Yes No

130 kDa

105 kDa

85 kDa

No change

No change

Increase No change detected No change detected No P-TYR detectedb

Large increase Increase Late increase P-TYR-induced P-TYR not detected P-TYR not assayed

Increase No change detected No change detected P-TYR-induced

Proliferation Growth inhibition Proliferation

Decreased Late increase”

Decreased Late increase

57 kDa

44 kDa

Biological result

a Slow activation times, 10-15 min uersus 5 min, are underlined. P-TYR, phosphotyrosine.

the lines used in this study (44, 45). p57 MAP kinase can be recognized by antisera made to conserved sequences found in erkl, erk2, and erk3 (44, 45) and so is probably structurally related to these known MAP kinases. Also, p57 MAP kinase is activated by diolein and by fibroblast growth factor treatment by an increase in both tyrosine and threonine phosphorylation and thus presumably has the tyrosinelthreonine activation sequence of TEY found in erkl and erk2 (28-32). TGFp1, when acting as a mitogen, did not activate the known MAP kinases of 42 and 44 kDa, erkl and erk2 (Table I). Thus three novel MBP kinases were activated during TGFpl growth stimulation, and all were activated by a ras-independent pathway. The p42lp44 MAP kinase kinase or MEK can be activated by a ras-independent pathway through MEK kinase (53). Possibly the 57, 105, and 130 kDa MBP kinases are activated by a ras-independent TGFpl-initiated signaling pathway through a MEK kinase.

The addition of a growth-inhibiting concentration of TGFpl to a second epithelial cell line activated ras proteins and a 57-kDa MBP kinase within 5 min but unexpectedly decreased the activity of two MBP kinases of 105 and 130 kDa 5-15 min after the addition (Table I). This is the first observation to our knowledge of MBP kinases whose activity was inhibited in parallel with inhibition of proliferation. These two higher molecular weight MBP kinases had the identical mobility on SDS- PAGE to the MBP kinases activated in TGFpl-initiated mitogenesis in a related epithelial cell line and may be the identical species. If this hypothesis is correct, TGFpl inhibits the 105- and 130-kDa MBP kinases when it inhibits epithelial cell growth but stimulates their activity when it acts as a mitogen.

The known MAP kinases were functional in these epithelial cells but were activated by growth factors other than TGFP1. EGF activated ras proteins and concomitantly activated e rk l (p44 MAP/MBP kinase) by increasing its phosphotyrosine content and activated a second MBP kinase of 85 kDa in HD3 cells (Table I). HD3 cells were also growth-inhibited by TGFp1. Thus in the same cell ras proteins were activated following either a mitogenic signal from EGF or a growth-inhibitory signal from TGFp1. In fibroblasts an 87-kDa MBP kinase was activated within 30 min of temperature shift of a temperature-sensitive v-src and remained activated in fully transformed cells (50). Both U9 and HD3 cells are known to contain activated c-src kinase and an 85-kDa MBP kinase that is activated by either

diolein or basic fibroblast growth factor (45). Possibly the 85- kDa MBP kinase activated in epithelial cells with constitutively activated c-src and the 87-kDa MBP kinase activated by v-src in fibroblasts are identical.

These results provide evidence that the signal transduction pathways from TGFpl and from EGF are not simple linear protein cascades but result in activation, and sometimes inac- tivation, of sets or cohorts of MBP kinases. The well known MAP kinases erkl and erk2 can phosphorylate and thus activate a series of transcription factors and other kinases like S6 (for a review, see Ref. 54). TGFpl, when acting as a mitogen, activates three major MBP kinases of 57,105, and 130 kDa and possibly other minor species we could not detect with the meth- odology employed. Each MBP kinase, in turn, could activate its own set of targets including various transcription factors, leading to a wave of new transcription. In a parallel fashion, growth inhibition by TGFp1, through inhibiting the activities of the 105- and 130-kDa MBP kinases, could lead to decreased activity of target proteins. In this study mitogenesis by EGF did indeed activate p44 MAP kinase (erkl) but also activated two other MBP kinases, 57 and 85 kDa, again spreading the mitogenic signaling pathway to three kinases, each possibly having multiple targets.

The change in signal transduction pathways initiated by TGFpl in epithelial cells occurs during tumor progression. TGFpl-induced proliferation characterizes metastatic and invasive colon carcinoma cells, while the more differentiated carcinoma cells respond to TGFPl by growth inhibition, like normal epithelial cells (41, 49, 55) . TGFPl induces 40-fold increases in c-fos expression in growth-inhibited cells but not in growth-stimulated cells (56). TGFpl-induced growth arrest in G1 is correlated with a block in phosphorylation of the retinoblastoma protein, while an increase in retinoblastoma protein phosphorylation is seen in TGFpl-induced proliferation (42). Thus the two TGFpl-initiated signal-transduction systems found in epithelial cells end in fan-like signaling to many different targets following the activation of cohorts of MBP kinases.

REFERENCES 1.

2.

3.

4.

5. 6. 7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

Egan, S. E., Giddings, B. W., Brooks, M. W., Buday, L., Sizeland, A. M., and Weinbere. R. A. (1993) Nature 363.4541

R u z a k i s ~ A ~ ~ k , ~ M . ; Femely, R., Wade, J., Pawson, T., and Bowtell, D. (1993)

Li, N., Batzer, A,, Daly, R., Yajnik, V., Skolnik, E., Chardin, P., Bar-Sagi, D.,

Gale, N. W., Kaplan, S., Lawenstein, E. J., Schlessinger, J., and Bar-Sagi, D.

Buday, L., and Downward, J. (1993) Cell 73, 611-620 Simon, M. A,, Dodson, G. S., and Rubin, G. M. (1993) Cell 73, 169-177 Olivier, J. P., Raabe, T., Henkemeyer, M., Dickson, B., Mbamalu, G., Margolis,

B., Schlessinger, J., Hafen, E., and Pawson, T. (1993) Cell 73, 179-191 Baltensperger, K., Kozma, L. M., Cherniack,A. D., Klarlund, J. K., Chawla,A.,

Banerjee, U., and Czech, M. P. (1993) Science 260, 195G1952 Skolnik, E. Y., Batzer, A,, Li, N., Lee, C-H., Lowenstein, E., Mohammadi, M.,

Margolis, B., and Schlessinger, J. (1993) Science 260, 1953-1955 Thomas. S. M.. DeMarco. M.. D’Arcangelo, G., Halegoua, S., and Brugge, J . S.

Nature 363,8345

Margolis, B., and Schlessinger, J. (1993) Nature 363, 85-88

(1993) Nature 363, S a 9 2

(1992) Cell Se, 1031-io40 - ~~

Wood, K. W., Sarnecki, C., Roberts, T. M., and Blenis, J. (1992) Cell 68,1041-

Shibuya, E. K., Polverino. A. J., Chang, E., Wigler, M., and Ruderman, J. V. 1050

Robbins, D. J., Cheng, M., Zhen, E., Vanderbilt, C. A,, Feig, L. A,, and Cobb, M. (1992) Proc. Natl. Acad. Sci. U. S. A. 89,9831-9835

de-Vries-Smits, A. M. M., Burgering, B. M. T., Leevers, S. J., Marshall, C. J., H. (1992) Prm. Natl. Acad. S c i . U. S. A. 89,69244928

Williams, N. G., Paradis, H., Aganval, S., Charest, D. L., Pelech, S. L., and and Bos, J. L. (1992) Nature 357,602404

Itoh, T., Kaibuchi, K., Masuda, T., Yamamoto, T., Matsuura, Y., Maeda. A., Roberts, T. M. (1993) Proc. Natl. Acad. Sci. U. S . A. 90,5772-5776

Moodie, S. A,, Willurnsen, S. M., Weber, M. J., and Wolfman, A. (1993) Science Shimizu, K., and Takai, Y. (1993) P m . Natl. Acad. Sci. U. S. A 90,976-979

Alessandrini, A,, Crews, C. M., and Erikson, R. L. (1992) Proc. Natl. Acad. Sci. 260,165%1661

Ahn, N. G., Seger, R., Bratlien, R. L., Diltz, C. D., Tonka, N. K., and Krebs, E. U. S. A. 89,8200-8204

Matsuda, S., Kosako, H., Takenaka, K., Moriyama, K., Sakai, H., Akiyama, T., G. (1991) J. Biol. Chem. 266,422M227

Signal Dansduction Pathways Activated by TGFpl 13237

21. Crews, C. M., and Erikson, R. L. (1992) Proc. Natl. Acad. Sci. 89,820543209 22. Wu, J., Hamson, J. K., Vincent, L. A., Haystead, C., Haystead, T. A. J., Michel,

H., Hunt, D. F., Lynch, K. R., and Sturgill, T. W. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 173-177

Gotoh, Y., and Nishida, E. (1992) EMEO J. 11,973-982

23. Ray, L.B., and Sturgill, T. W. (1988) J. Eiol. Chem. 263, 12721-12727 24. Rossomando, A. J., Payne, D. M., Weber, M. J., and Sturgill, T. W. (1989) Proc.

25. Kyriakis, J. M., and Avruch, J. (1990) J. Eiol. Chem. 266, 17355-17363 26. Boulton, T. G., Yancopoulos, G. D., Gregory, J. S., Slaughter, C., Moomaw, C.,

27. Boulton, T. G., Nye, S. H., Robhins, D. J., Ip, N. Y., Radziejewska, E., Morgen- Hsu, J., and Cobb, M. H. (1990) Science 249,64-67

besser, S. D., DePinho, R. A,, Panayotatos, N., Cohh, M. H., and Yancopou- los, G. D. (1991) Cell 66,663475

28. Anderson, N. G., Maller, J. L., Tonks, N. K., and Sturgill, T. W. (1990) Nature 343,651-653

29. Seger, R., Ahn, N. G., Boulton, T. G., Yancopoulos, G. D., Panayotatos, N., Radziejewska, E., Ericson, L., Bratlien, R. L., Cobb, M. H., and Krebs, E. G .

30. Wu, J., Rossomando, A. J., Her, J-H., DelVecchio, R., Weber, M. J., and Sturgill, (1991) Proc. Natl. Acad. Sci. U. S. A. 88, 6142-6146

31. Rossomando,A., Wu, J., Weber, M., and Sturgill, T. W. (1992) Pmc. Natl. Acad. T. W. (1991) Prw. Natl. Acad. Sei. U. S. A. 88,9508-9512

32. Dent, P., Haer, W., Haystead, T. A. J., Vincent, L. A., Roherts, T. M., and Sci. U. S. A. 89,5221-5225

33. Gille, H., Sharoocks, A. D., and Shaw, P. E. (1992) Nature 368,414416 Sturgill, T. W. (1992) Science 257, 1404-1407

34. Pulverer, B. J., Kyriakis, J. M.,Avruch, J., Nikolakaki, E., and Woodgett, J. R.

35. Mulder, K. M., and Moms, S. L. (1992) J. Eiol. Chem. 267,5029-5031 36. Lopez-Casillas, F., Cheifetz, S., Doods J., Andres, J. L., Lane, W. S., and

37. Wang, X-F., Lin, H. L., Ng-Eaton, E., Downward, J., Lodish, H. F., and Wein-

Natl. Acad. Sci. U. S. A. 86,6940-6943

(1991) 363, 6 7 M 7 4

Massague, J. (1991) Cell 67, 785-795

berg, R. A. (1991) Cell 67, 797405

38. Lin, H. Y., Wang, X-F., Ng-Eaton, E., Weinberg, R. A,, and Lodish, H. F. (1992)

39. Wrana, J. L., Attisano, L., Carcamo, J., Zentella, A., Doody, J., Laiho, M., Wang,

40. Chen, R-H., Ehner, R., and Derynck, R. (1993) Science 260, 1335-1338 41. Hafez, M. M., Infante, D., Winawer, S . J., and Friedman, E. (1990) Cell Growth

42. Yan, Z., Hsu, S., Winawer, S., and Friedman, E. (1992) Oncogene 7, 801405 43. Gihbs, J. B., Pompliano, D. L., Mosser, S. D., Rands, E., Lingham, R. B., Singh,

S. B., Scolnick, E. M., Kohl, N. E., and Oliff, A. (1993) J. Eiol. Chem. 268,

44. Lee, H., Ghose-Dastidar, J., Winawer, S., and Friedman, E. (1993) J. Eiol. 7617-7620

45. Lee, H., Hsu, S., Winawer, S., and Friedman, E. (1993) J. Eiol. Chem. 268, Chem. 268,52554263

46. Buday, L., and Downward, J. (1993) Mol. Cell. Biol. 13, 1903-1910 81814187

47. Smith, M. R., DeGudicihus, S. J., and Stacey, D. W. (1986) Nature 320,540-543 48. Satoh, T., Nakafuka, M., Miyajima, A,, and Kaziro, Y. (1991) Proc. Natl. Acad.

49. Hafez, M. M., Hsu, S., Yan, Z., Winawer, S., and Friedman, E. (1992) Cell

50. Wang, H-C., and Erikson, R. L. (1992) Mol. Eiol. Cell 3, 1329-1337 51. Gonzalez, F. A,, Raden, D. L., Righy, M. R., and Davis, R. J. (1992) FEES Letts.

52. Linnekin, D., Evans, G., Michiel, D., and Farrar, W. L. (1992) J. B i d . Chem.

53. Lenge-Carter, C. A,, Peiman, C. M., Gardner, A. M., Blumer, K J., and John-

54. Thomas, G. (1993) Cell 68,s 55. Schroy, P., Riflcin, J., Coffey, R. J., Winawer, S., and Friedman, E. (1990)

56. Hsu, S., Huang, F., Hafez, M., Winawer, S., and Friedman, E. (1994) Cell

Cell 68,775-785

X-F., and Massague, J. (1992) Cell 71,1003-1014

& Diff: 1,617-626

Sci. U. S. A. 88, 3314-3317

Growth & Diff: 3,753-762

304, 17C178

267,23993-23998

son, G. L. (1993) Science 260,315-319

Cancer Res. 60, 261-265

Growth &Differ., in press

Documents

Two Different Signal Transduction Pathways Can Be Activated by