Distortion of autocrine transforming growth factor b signal ... - Nature

Oncogene (2003) 22, 2309–2321

& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

Distortion of autocrine transforming growth factor b signal accelerates malignant potential by enhancing cell growth as well as PAI-1 and VEGF production in human hepatocellular carcinoma cells Yasushi Sugano1, Koichi Matsuzaki*,1, Yoshiya Tahashi1, Fukiko Furukawa1, Shigeo Mori1, Hideo Yamagata1, Katsunori Yoshida1, Masanori Matsushita1, Mikio Nishizawa2, Junichi Fujisawa3 and Kyoichi Inoue1,4 1 Third Department of Internal Medicine, 10–15 Fumizonocho, Moriguchi, Osaka 570–8507, Japan; 2Department of Medical Chemistry, 10–15 Fumizonocho, Moriguchi, Osaka 570–8507, Japan; 3Department of Microbiology, Kansai Medical University, 10–15 Fumizonocho, Moriguchi, Osaka 570–8507, Japan

Resistance to growth inhibitory effects of transforming growth factor (TGF)-b is a frequent consequence of malignant transformation. On the other hand, serum concentrations of TGF-b, plasminogen activator inhibitor type 1 (PAI-1), and vascular endothelial growth factor (VEGF) are elevated as tumor progresses. The molecular mechanism of autocrine TGF-b signaling and its effects on PAI-1 and VEGF production in human hepatocellular carcinoma (HCC) is unknown. TGF-b signaling involves TGF-b type I receptor-mediated phosphorylation of serine residues within the conserved SSXS motif at the C-terminus of Smad2 and Smad3. To investigate the involvement of autocrine TGF-b signal in cell growth, PAI-1 and VEGF production of HCC, we made stable transfectants of human HCC line (HuH-7 cells) to express a mutant Smad2(3S-A), in which serine residues of SSXS motif were changed to alanine. The transfectants demonstrated an impaired Smad2 signaling. Along with the resistance to growth inhibition by TGF-b, forced expression of Smad2(3S-A) induced endogenous TGF-b secretion. Moreover, this increased TGF-b enhanced liganddependent signaling through the activated Smad3 and Smad4 complex, and transcriptional activities of PAI-1 and VEGF genes. In conclusion, distortion of autocrine TGF-b signals in human HCC accelerates their malignant potential by enhancing cell growth as well as PAI-1 and VEGF production. Oncogene (2003) 22, 2309–2321. doi:10.1038/sj.onc.1206305 Keywords: TGF-b; Smad; HCC; PAI-1; VEGF

Introduction Human hepatocellular carcinoma (HCC) is one of the most frequent tumors, particularly among Asian and *Correspondence: Dr K Matsuzaki; E-mail: [email protected] 4 Deceased Received 20 May 2002; revised 6 December 2002; accepted 6 December 2002

African populations (Yamanaka et al., 1990). A notable feature of most HCC is hypervascularity; therefore, it is likely that HCC produces angiogenic factors. Vascular endothelial growth factor (VEGF) plays an important role in the development of neovascularization in HCC, since VEGF expression is significantly higher in HCC with strong or moderate angiographic staining than in those with weak or no angiographic staining (Mise et al., 1996). Plasminogen activator inhibitor type 1 (PAI-1) is also highly expressed in human HCC (Zhou et al., 2000). PAI-1 is the main inhibitor of the urokinase-type plasminogen activator system, and conducts cancer cells to migration, invasion, and metastasis by blocking cellular adhesion and by promoting basement membrane degradation in a variety of malignancies (Gutierrez et al., 2000). Another characteristic of HCC is increased production of transforming growth factor (TGF)-b, which is a potent growth inhibitor for hepatocytes. This observation implies the resistance to growth inhibitory effects caused by the autocrine TGF-b signal in human HCC (de Caestecker et al., 2000). However, it remains unclear whether the endogenous TGF-b signal and transcriptional regulations of PAI-1 and VEGF genes are linked during hepatocarcinogenesis. Progress over the past several years had elucidated how TGF-b initiates its response. Members of the TGFb family exert their function by binding to and activating specific pairs of receptor serine/threonine kinases. The TGF-b family members interact with the type II receptor, which, in turn, recruits the type I receptor (TbRI) into a heteromeric receptor complex, and then the ligand-binding type II receptor kinase is constitutively active and activates TbRI by phosphorylation of serine and threonine residues in the GS box (Massague´, 1998). The activated TbRI phosphorylates the serine residues within the conserved SSXS motif at the C-terminus of receptor-regulated Smads (R-Smads), such as Smad2 and Smad3 (Macı´ as-Silva et al., 1996). These phosphorylated R-Smads heteromerize with a common partner Smad4, and then accumulate in the nucleus. The activated Smad complex binds to target promoters including PAI-1 and VEGF in association

Autocrine TGF-b signal in human HCC Y Sugano et al

2310

with DNA-binding cofactors, and recruits coactivators to activate transcription (Derynck et al., 1998). Alternatively, the activated Smad complex can also recruit corepressors, which in turn bind histone deacetylases. As a result, Smads can either positively or negatively regulate the transcription of specific genes in response to TGF-b signaling (Wrana, 2000). In addition to these positively acting Smads, Smad7 blocks the phosphorylation of R-Smads by acting as a pseudosubstrate for TbRI. Interestingly, transcription of Smad7 gene is induced by TGF-b, providing a negative feedback mechanism for the regulation of Smad signaling (Nakao et al., 1997). We previously demonstrated that R-Smads activated by autocrine TGF-b signal propagate the negative or positive growth signal in cultured HCC cells, and that they play an important function in the regulation of gene-specific transcriptional responses (Matsuzaki et al., 2000a, b). The activated R-Smads might determine the biological character of human HCC, such as invasive activity and metastatic behavior. On the other hand, mutations have been reported in the signaling molecules, including in Smad2 and Smad4, of some HCC cells (Yakicier et al., 1999). Accordingly, it is important to investigate whether the inactivation of the signaling molecules modulates the character of human HCC by blocking autocrine TGF-b signal pathway. In the present study, we found that the inhibition of Smad2 activation did not shut off the entire TGF-b signaling pathway, but altered the pathway to allow for the preferential usage of Smad3, leading to upregulation of factors that are involved in metastasis and angiogenesis. Thus, such an alteration of Smad signaling would enhance the malignant potential in addition to increasing resistance to the antiproliferative effects of TGF-b.

Results Establishment of Smad transfectants from HuH-7 cells Considerable experimental evidence supports a model of TGF-b signaling that involves the phosphorylation of serine residues within the conserved SSXS motif at the C-terminus of R-Smads by the kinase present in TbRI (Macı´ as-Silva et al., 1996). Accordingly, Smad2(3S-A) and Smad3(3S-A), in which the serine residues of SSXS motif are changed to alanine, can act as dominantnegative mutants by blocking the functions of endogenous R-Smad proteins. To investigate the effects of the TGF-b signal on the biological features of human HCC, we selected HuH-7 cells, which are known to be a welldifferentiated HCC line and sensitive to the TGF-b signal (Fan et al., 1996). We generated four types of HuH-7 clones, which had been stably transfected with expression vectors encoding Flag-tagged Smad proteins. Two representative clones were selected from cells transfected with wild-type RSmads (2WT-ST1B3, 3WT-ST1B3). Two others were selected that had been transfected with dominantnegative Smad2 (2SA-ST1B3) or Smad3 (3SA-ST1 Oncogene

B2). Initially, we investigated the expression levels of exogenous Smad proteins in each stable transfectant. After the cells were metabolically labeled with [35S]methionine, the cell lysates were precipitated with anti-Flag antibody (Ab) and visualized by autoradiography (data not shown). We chose the clone showing the highest expression of Smad proteins, and performed further molecular analyses. We next confirmed whether the expressions of exogenous Smads overwhelmed those of endogenous Smads. For this purpose, we compared the mRNA levels of exogenous Smads with those of endogenous Smads by Northern blot hybridization (Figure 1a). mRNAs of 3.4 and 2.9 kb would correspond to endogenous Smad2, and 2.4 kb mRNA would correspond to exogenous Smad2 (lanes 2 and 3). The mRNA of exogenous Smad2(3S-A) in 2SA-ST1 cells was more abundant than that of endogenous Smad2 (lane 3), whereas 2WT-ST1 cells showed relatively low levels of exogenous wild-type Smad2 expression (lane 2). mRNAs of 4.9 and 2.5 kb would correspond to endogenous Smad3, and 2.0 kb mRNA would correspond to exogenous Smad3 (lanes 4 and 5). Expressions of exogenous wild-type Smad3 in 3WT-ST1 cells and Smad3(3S-A) in 3SA-ST1 cells were significantly higher than that of endogenous Smad3. The intensities of mRNA for Smads were normalized to the signals of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Since all the Smad constructs were Flag-tagged, we could monitor protein expression levels of Flag-Smads among the transfectants. Followed by immunoprecipitation of cell lysates of parental HuH-7 cells and stable transfectants with anti-Flag Ab, Flag-Smads proteins were detected by immunoblot using anti-Flag Ab (Figure 1b, upper panel). The results showed that Flag-Smad2WT, Smad2(3S-A), Smad3WT, and Smad3(3S-A) proteins were expressed in the transfectants. We next compared the protein levels of exogenous gene products with those of endogenous R-Smads in the stable transfectants. The ratio of exogenous R-Smads to endogenous R-Smads, in the samples immunoprecipitated with anti-Smad2/3 Ab, was determined by immunoblot using anti-Smad2/3 Ab (Figure 1b, lower panel). Sufficient differences in size between exogenous R-Smads and endogeous ones allowed us to know the expression levels of these R-Smads. Thus, Flag-tagged R-Smads migrated slightly slower than endogenous RSmads. The results showed that Flag-Smad2(3S-A) in 2SA-ST1 cells was more abundant than that of endogenous Smad2 (lane 3), whereas 2WT-ST1 cells showed relatively low levels of Flag-Smad2 expression (lane 2). Expression of Flag-Smad3WT in 3WT-ST1 cells or Flag-Smad3(3S-A) in 3SA-ST1 cells was significantly higher than that of endogenous Smad3. These results were similar to the expression profiles at the mRNA levels among the transfectants. Collectively, the present findings suggest that the expression of exogenous R-Smads(3S-A) was sufficiently high to overwhelm that of endogenous Smads, and that RSmads(3S-A) proteins could selectively block their specific signal through the phosphorylated Smads in these transfectants.


2311

a

a Cell line: HuH-7 2WT 2SA -ST1 -ST1

3SA -ST1

3WT -ST1

3.4 Kb -

Cell line: HuH-7 endogenous Smad2 exogenous Smad2

2.9 Kb 2.4 Kb 4.9 Kb -

2WT 2SA -ST1 -ST1

3WT -ST1

3SA -ST1

TGF-β β1

2.5 Kb -

endogenous Smad3

2.5 Kb -

GAPDH

exogenous Smad3

2.0 Kb -

GAPDH

1

2

3

4

5

lane

b

b 2SA 3WT 3SA Cell line: HuH-7 2WT -ST1 -ST1 -ST1 -ST1 - Flag-Smad2 - Flag-Smad3 - IgH

IP: α Flag IB: α Flag

Flag-Smad2 endogenous Smad2 Flag-Smad3 endogenous Smad3

IP: α Smad2/3 IB: α Smad2/3

1

2

3

4

5

lane

Figure 1 Establishment of stable Smad transfectants from HuH-7 cells. (a) High expression levels of exogenous Smads mRNAs relative to those of endogenous Smads mRNAs in stable transfectants. Northern blot hybridization was performed with poly(A) RNAs (2 mg/lane) extracted from parental HuH-7 cells and the stable transfectants. Blots were hybridized with 32P labeled random-primed cDNAs for Smad2, Smad3, and GAPDH. The lowest panel shows GAPDH mRNA expression as an internal control. (b) Expression of exogenous Smads proteins is sufficiently high to overwhelm that of endogenous Smads proteins in stable transfectants. Protein expression levels of Flag-Smads were monitored among the transfectants (upper panel). Followed by immunoprecipitation (IP) of cell lysates of parental HuH-7 cells and stable transfectants with anti-Flag Ab (aFlag), Flag-Smads proteins were detected by immunoblot (IB) using anti-Flag Ab. Next, the protein levels of exogenous gene products were compared with those of endogenous R-Smads in the stable transfectants (lower panel). The ratio of exogenous R-Smads to endogenous RSmads in the anti-Smad2/3 (aSmad2/3) immunoprecipitated samples was determined by IB using anti-Smad2/3 Ab. Flag-tagged R-Smads migrate slightly slower than endogenous R-Smads

By Northern blot hybridization and ELISA techniques, we further examined TGF-b1 expression and secretion into the culture media in parental HuH-7 cells and stable transfectants. Surprisingly, both 2SA-ST1 and 3SA-ST1 cells displayed significant autoinduction of TGF-b1 at both the protein and mRNA levels (Figure 2a, b). To examine the role of TGF-b signaling in liver carcinogenesis, we investigated the mRNA expression of type I and type II TGF-b receptors in these cells. Both parental HuH-7 cells and stable transfectants expressed the type I and type II receptor mRNAs (data not shown). Since TGF-b1 expression was elevated in all stable transfectants, and the cells expressed TGF-b

TGF-β1 concentration (ng/ml)

2-

1-

0Cell line: HuH-7

2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

Figure 2 High expression levels of TGF-b1 mRNA and protein in stable R-Smads(3S-A) transfectants. (a) Northern blot hybridization of TGF-b1 using parental HuH-7 cells and the stable transfectants was performed as described in Figure 1a. (b) TGFb1 concentration in the culture media. Samples were collected from culture media in which the cells were cultured for 60 h, and measured by enzyme-linked immunoassay as described in the text. The data are from one representative experiment with each point determined in triplicate

receptors and Smads, it is possible that TGF-b1 propagates its autocrine signal in these cells. Autocrine TGF-b signal propagates through the activated Smad3 in 2SA-ST1 cells Mutations in Smad2 gene have been reported in various types of human cancer, and are an important factor in tumor progression (Riggins et al., 1997). However, it remains unclear whether or not the Smad3 signal is inactive in those cancer cells carrying Smad2 gene mutations. Accordingly, we focused our studies on R-Smads activation in 2SA-ST1 cells. First, we monitored the phosphorylated state of Smad2 protein by performing immunoblot analysis using anti-phosphorylated Smad2 Ab in parental HuH-7, 2SA-ST1, and 3SAST1 cells. As a control, treatment of parental HuH-7 cells with TGF-b1 induced phosphorylation of Smad2 Oncogene


2312

(Figure 3A). In 2SA-ST1 cells, immunoblot with antiphosphorylated Smad2 Ab revealed that Smad2(3S-A) successfully repressed the TGF-b-dependent phosphorylation of Smad2. Similarly, Smad3(3S-A) overexpression in 3SA-ST1 cells resulted in no significant stimulation of Smad2 phosphorylation by TGF-b1 treatment. Next, we investigated the phosphorylated state of Smad3 in 2SA-ST1 cells. Flag-tagged Smad3 proteins were transiently expressed in the cells, which were metabolically labeled with [32P]phosphate and subsequently incubated with or without exogenous TGF-b1. As shown in Figure 3B, exogenous TGF-b1 enhanced the phosphorylation of Smad3. However, neither the phosphorylation of Smad3(3S-A) nor that of Smad2(3S-A) was not induced by TGF-b1 treatment. In addition, the basal level for Smad3WT phosphorylation was reduced by the addition of neutralizing Ab against TGF-b (lane 3 compared with lane 1). These results indicate that ligand-dependent Smad3 phosphorylation occurs by an autocrine mechanism. We further attempted to confirm whether or not endogenous Smad3 was phosphorylated by TGF-b in the cells. For this purpose, we determined the phosphorylated state of Smad3 by immunoprecipitation of the metabolically labeled cell lysates with anti-Smad3 Ab. Liganddependent phosphorylation of endogenous Smad3 was also observed in 2SA-ST1 cells (Figure 3C). Collectively, Smad2(3S-A) expression did not inhibit the phosphorylation of Smad3 in response to TGF-b signal. We further confirmed that ligand-dependent signaling occurred at the downstream steps in the TGF-b-induced activation of R-Smads. Key events in this process include the translocation of R-Smads into the nucleus. To determine whether TGF-b could influence nuclear translocation of R-Smads, we observed subcellular localization by indirect immunofluorescence microscopy

after TGF-b1 treatment. In parental HuH-7 cells used as a control, R-Smads proteins were localized in the cytoplasm in unstimulated culture and the TGF-b signal did translocate R-Smads into the nucleus (Figure 3D, a–d). However, even in the presence of the TGF-b signal, Smad2 protein was predominantly located in the cytoplasm of 2SA-ST1 cells (Figure 3D, E and F), indicating that the mutant Smad2 functioned in a dominant-negative fashion. By contrast, most 2SAST1 cells showed a ligand-dependent accumulation of Smad3 into the nuclei (Figure 3D, G and H). These results indicate that the TGF-b signal can translocate Smad3 protein into the nucleus in a Smad2-independent manner. Similar to the phosphorylation results, Smad3(3S-A) overexpression blocked nuclear translocation of both Smad2 and Smad3 in 3SA-ST1 cells (Figure 3D, i–l). The heteromeric complex formation of R-Smads with Smad4 in response to TGF-b is also critical for TGF-b/ Smad signaling. Accordingly, we investigated whether TGF-b signal had an effect on the complex formation between R-Smads and Smad4 in parental HuH-7, 2SAST1, and 3SA-ST1 cells. After the cells were treated with TGF-b1, cell lysates were subjected to immunoprecipitation with anti-Smad4 Ab, and each of the immunoprecipitates was analysed by immunoblot using antiphosphorylated Smad2 Ab. As shown in Figure 3E, an increase in the associations between the phosphorylated Smad2 and Smad4 was observed in parental HuH-7 cells treated with TGF-b1. However, TGF-b1 treatment did not sufficiently induce the complex formation of the phosphorylated Smad2 with Smad4 in both 2SA-ST1 and 3SA-ST1 cells. Next, to investigate heteromeric complex formation of Smad3 with Smad4, 2SA-ST1 cells were transfected with Flag-Smad3 and Smad4-HA in the absence or presence of TGF-b1. Cell lysates were

" Figure 3 The TGF-b signal is propagated through the activated Smad3–Smad4 complex in stable Smad2(3S-A) transfectants. (A) TGF-b-dependent phosphorylation of Smad2 is impaired in stable R-Smads(3S-A) transfectants. Parental HuH-7, 2SA-ST1, or 3SA-ST1 cells were incubated in the absence or presence of 200 pm TGF-b1. Followed by immunoprecipitation (IP) of cell lysates with anti-Smad2/3 Ab (aSmad2/3), the phosphorylated state of Smad2 protein was monitored by immunoblot (IB) using antiphosphorylated Smad2 Ab (a pSmad2) (upper panel). To determine total protein levels, same samples were subjected IB analyses with anti-Smad2/3 Ab (lower panel). (B) Autocrine TGF-b-dependent phosphorylation of Smad3 in stable Smad2(3S-A) transfectants. After transient transfection with Flag-Smad3WT or Flag-Smad3(3S-A), 2SA-ST1 cells were labeled with [32P]phosphate in the absence or presence of 200 pM TGF-b1 or 5 mg/ml anti-TGF-b Ab. Cell lysates were subjected to IP with anti-Flag Ab (aFlag), separated by SDS–PAGE, and visualized by autoradiography. To determine total protein levels, same samples were subjected to IB analyses with anti-Flag Ab (lower panel). pSmad; phosphorylated Smad. (C) TGF-b-dependent phosphorylation of endogenous Smad3 in stable Smad2(3S-A) transfectants. 2SA-ST1 cells were labeled with [32P]phosphate in the absence or presence of 200 pm TGF-b1. Cell lysates were subjected to IP with anti-Smad3 Ab (aSmad3) and visualized as described above. To determine total protein levels, cell lysates were subjected to IB analyses with anti-Smad3 Ab (lower panel). pSmad3; phosphorylated Smad3. (D) In contrast with the blockage of TGF-b-dependent nuclear translocation of Smad2, the TGF-b signal translocates Smad3 protein into the nuclei in stable Smad2(3S-A) transfectants. Parental HuH-7, 2SA-ST1, or 3SA-ST1 cells were incubated in the absence or presence of 200 pm TGF-b1 for 1 h. Smad2 or Smad3 was detected by immunofluorescence using antibody against Smad2 (aSmad2) or Smad3 (aSmad3) followed by FITC-conjugated secondary Ab, and analysed by confocal microscopy. (E) TGF-b treatment does not sufficiently induce the complex formation of the phosphorylated Smad2 with Smad4 in stable R-Smads(3S-A) transfectants. Parental HuH-7, 2SA-ST1, or 3SA-ST1 cells were incubated in the absence or presence of 200 pm TGF-b1 for 1 h. Cell lysates were subjected to anti-Smad4 (aSmad4) IP and then were IB with anti-phosphorylated Smad2 Ab (apSmad2) (upper panel). The expression of Smad2/3 or Smad4 was monitored by IB using anti-Smad2/3 Ab (aSmad2/3) (middle panel) or anti-Smad4 Ab (lower panel). (F) TGF-b-dependent heteromeric complex of Smad3 with Smad4 in stable Smad2(3S-A) transfectants. 2SA-ST1 cells were transiently transfected with the indicated combinations of Flag-Smad3 and Smad4-HA, followed by incubation in the absence or presence of 200 pm TGF-b1. Cell lysates were subjected to anti-Flag (aFlag) IP and then were IB with anti-HA Ab (aHA) (upper panel). The expression of transfected DNAs was determined by IB of total cell lysates using anti-Flag Ab or anti-HA Ab. (G) Endogenous Smad3 activated by TGF-b associates with endogenous Smad4 in stable Smad2(3S-A) transfectants. 2SA-ST1 cells were incubated in the absence or presence of 200 pm TGF-b1. Cell lysates were subjected to anti-Smad3 (aSmad3) IP and then were IB with anti-Smad4 Ab (aSmad4) (upper panel). The relative levels of endogenous Smad3 and Smad4 were monitored by direct IB of total cell lysates Oncogene


2313

subjected to immunoprecipitation with anti-Flag Ab and the immunoprecipitates were analysed by immunoblot using anti-HA Ab. As shown in Figure 3F, an increase in the association between Smad3 and Smad4 was observed in the cells treated with TGF-b1. Liganddependent association of endogenous Smad3 with endogenous Smad4 was also confirmed (Figure 3G). The results indicate that TGF-b signaling induces the formation of heteromeric complex composed of Smad3 HuH-7

A

-

and Smad4 in 2SA-ST1 cells. Thus, the results obtained in this study are because of specific perturbation of the Smad2 signals by Smad2(3S-A) overexpression. Smad2 inactivation leads to upregulations of PAI-1 and VEGF by the activated Smad3 PAI-1 is involved in tumor growth, metastasis, and angiogenesis (Gutierrez et al., 2000). Clinically, serum 3SA-ST1

2SA-ST1

-

Tβ1

-

Tβ1

Tβ1

IP: α Smad2/3 IB: α pSmad2

- pSmad2



B Transient transfection

2SA-ST1 Flag-Smad3WT Flag-Smad3(3S-A)

-

Tβ 1

Ab

-

Tβ1

Ab

- pSmad2 - pSmad3

32 P labeling

IP: α Flag

-Smad2 -Smad3 -IgH

IP: α Flag IB: α Flag 1

2

3

4

lane

6

5

2SA-ST1

C

Tβ1

32 P labeling

-pSmad3

IP: αSmad3 α Cell lysate

-Smad3

IB: α Smad3

D

HuH-7 α Smad3

α Smad2

α Smad3

α Smad2

α Smad3

−

α Smad2

3SA-ST1

2SA-ST1

c

e

g

i

k

b

d

f

h

j

l

TGF-β β1

a

Oncogene


2314

E

HuH-7

2SA-ST1

3SA-ST1

- Tβ1

- Tβ1

- Tβ1 Smad4-bound pSmad2

IP: α Smad4 IB: α pSmad2

IgH



Cell lysate IB: α Smad4

Smad4

F

2SA-ST1

Flag-Smad3 Smad4-HA

− +

+

+

+

−

−

+ Tβ1 -

IP: α Flag IB: α HA

- IgH

Cell lysate IB: α HA

- Smad4

Cell lysate IB: α Flag

- Smad3

G

2SA-ST1

− IP: α Smad3 IB: α Smad4

Figure 3

Tβ1

Smad3-bound - Smad4

- IgH


- Smad3


- Smad4

(continued )

concentrations of both PAI-1 and TGF-b are elevated as HCC progresses (Shirai et al., 1994; Zhou et al., 2000). To examine the response of the autocrine TGF-b signal to PAI-1 transcription in these transfectants, we investigated PAI-1 expression and secretion into the culture media by Northern blot hybridization and ELISA. The PAI-1 expression was significantly increased at both the protein and mRNA levels in 3WTST1 and 2SA-ST1 cells (Figure 4a, b). To obtain the Oncogene

Smad3-bound Smad4

evidence that R-Smads phosphorylation by endogenous TGF-b is involved in TGF-b signaling, we investigated the effects of TGF-b on PAI-1 promoter activity. For these assays, we used PF1-Luc, which contains a promoter segment from –794 to –532 in the PAI-1 gene sufficient for the TGF-b-dependent induction (Hua et al., 1998). The cells transfected with PF1-Luc were incubated overnight in the absence or presence of TGFb1, and the relative luciferase activity was measured in


2315

b

a Cell line: HuH-7 2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

3.2 Kb 2.2 Kb -

PAI-1

GAPDH

− TGF-β + TGF-β

420-

Cell line: HuH-7

50 -

0Cell line: HuH-7 2WT -ST1

d

8 - PF1-Luc 6-

100 -

Relative Luciferase Activity

Relative Luciferase Activity

c

PAI-1 concentration (ng/ ml)

150 -

2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

3-

2SA -ST1

3WT -ST1

3SA -ST1

PF1-Luc / 2SA-ST1

2-

1-

0Smad

-

3WT 3(3S-A) 4WT

4∆C

Figure 4 The activated Smad3–Smad4 complex mediates PAI-1 transcriptional response in stable Smad2(3S-A) transfectants. (a) High expression levels of PAI-1 mRNAs in stable Smad2(3S-A) and wild-type Smad3 transfectants. Northern blot hybridization of PAI-1 was performed as described in the text. (b) PAI-1 concentration in the culture media. Samples were collected and measured by enzyme-linked immunoassay, as described in the text. The data are from one representative experiment with each point determined in triplicate. (c) Stable Smad2(3S-A) transfectants reveal TGF-b-dependent PAI-1 transcriptional response. After transfection with PF1-Luc, the cells were incubated overnight in the absence or presence of 200 pm TGF-b1, and the relative luciferase activity was measured in cell lysates. The luciferase activity was normalized to the Renilla luciferase activity and is expressed as the mean7s.d. (n ¼ 4) from a representative experiment. (d) PAI-1 transcriptional response is mediated by Smad3–Smad4 complex in stable Smad2(3S-A) transfectants. 2SA-ST1 cells were transfected with PF1-Luc alone or together with Smad3WT, Smad3(3S-A), Smad4WT, or C-terminal truncated Smad4DC. In all cases, the relative luciferase activity was measured in the cell lysates

the cell lysates. In parental HuH-7 cells as a control, significant activation of PF1 promoter was achieved by treatment of TGF-b1 (Figure 4c). This activation was more potentiated in the cells expressing wild-type Smad3 than in those expressing wild-type Smad2, and was lost in the cells expressing Smad3(3S-A), suggesting that TGF-b signaling is mainly propagated through Smad3. As expected, transfection of PF1-Luc alone into 2SAST1 cells resulted in the elevated basal levels of transcription in unstimulated cells, and this transcription was significantly induced by stimulation with exogenous TGF-b1. These results suggest that autocrine TGF-b pathway is involved in the transcriptional activation of the PAI-1 gene. To prove more direct involvement of the increased Smad3 and Smad4 in PF1 promoter activity, we cotransfected each TGF-b signal component with PF1-Luc. Cotransfection with

wild-type Smad3 or Smad4 increased the sensitivity of the cells (Figure 4d). In contrast, cotransfection with Smad3(3S-A) or C-terminal truncated Smad4 (Smad4DC) led to suppression of the PF1 promoter activity. These results demonstrate direct evidence that autocrine TGF-b signals, despite the Smad2 signal blockage, are propagated through Smad3–Smad4 complex in 2SA-ST1 cells. Since VEGF contributes to tumor angiogenesis in HCC (Mise et al., 1996; Ferrara, 1999), we next focused our analysis on the VEGF gene as a target of the autocrine TGF-b pathway. Similar to the profile of PAI1 expression among the transfectants, VEGF expression increased in 3WT-ST1 and 2SA-ST1 cells (Figure 5a). A large amount of VEGF secretion into the culture media was also observed in these cells (Figure 5b). We further studied the exogenous TGF-b1 effects on VEGF Oncogene


2316

a Cell line: HuH-7

2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

3.9 Kb -

VEGF

GAPDH

b VEGF concentration (x 10 3 pg/ml)

10 -

5-

0Cell line: HuH-7

2WT -ST1

VEGF concentration (x 10 3 pg/ml)

c

2SA -ST1

3WT -ST1

3SA -ST1

2SA-ST1

10 -

5-

0-

-

Tβ β1

Ab

Figure 5 Smad3 activated by autocrine TGF-b signal stimulates VEGF production in stable Smad2(3S-A) transfectants. (a) High expression levels of VEGF mRNAs in stable Smad2(3S-A) and wild-type Smad3 transfectants. Northern blot hybridization of VEGF was performed as described in the text. (b) VEGF concentration in the culture media. Samples were collected and measured by enzyme-linked immunoassay, as described in the text. The data are from one representative experiment with each point determined in triplicate. (c) Autocrine TGF-b signal enhances VEGF production in stable Smad2(3S-A) transfectants. After 2SAST1 cells were cultured in the absence or presence of 200 pm TGFb1 or 5 mg/ml anti-TGF-b Ab, the samples were collected and measured as described in (b)

production in 2SA-ST1 cells. VEGF production was elevated by exogenous TGF-b1 treatment (Figure 5c). If the secreted TGF-b stimulates VEGF production, antibody-induced blockage of the endogenous TGF-b may Oncogene

indirectly reduce VEGF production. In fact, addition of anti-TGF-b Ab to 2SA-ST1 cells led to 40% loss in VEGF production after Ab treatment. Taken together, these results suggest that autocrine TGF-b signaling promotes VEGF production in 2SA-ST1 cells. Effects of autocrine TGF-b signal on DNA synthesis and AFP production The discrepancy between the increase of TGF-b in HCC and the highly proliferating cell rate of this tumor suggests that these cells acquire resistance to the growth inhibitory effects of tumor-derived TGF-b. Tumor cells often escape from the antiproliferative effects of TGF-b by various mechanisms, including the mutational inactivation of Smad proteins (Riggins et al., 1997). When challenged with TGF-b1, HuH-7 cells normally undergo growth arrest prior to the G1/S transition and cease to incorporate [3H]thymidine. Measurement of [3H]thymidine incorporation thus provides a sensitive gauge of the effect of TGF-b treatment on cellular proliferation. To assess whether the antiproliferative signal is mediated through R-Smads, [3H]thymidine incorporation in these transfectants was compared to that in the parental HuH-7 cells (Figure 6a). The reduction in [3H]thymidine incorporation into the cells by TGF-b1 treatment is shown as a percentage in Figure 6b. Parental HuH-7 cells reduced their incorporation of [3H]thymidine in response to TGF-b. A significant suppression of [3H]thymidine incorporation was achieved by TGF-b1 treatment in the cells carrying wild-type Smad2. This suppression was more potentiated in 3WT-ST1 cells than in 2WT-ST1 cells, and such suppression was lost in 3SA-ST1 cells, suggesting that the activation of Smad3 is essential for the growth inhibitory signaling in HuH-7 cells. Similar resistance to its growth inhibitory effects was observed in 2SA-ST1 cells. Taken together the findings that autocrine TGF-b signals were mainly propagated through the activated Smad3–Smad4 complex in 2SA-ST1 cells, the data indicate that an achievement of this negative cell growth effect requires Smad2 phosphorylation in HuH-7 cells. Highly expressed in HCC tissues, a-fetoprotein (AFP) is widely used for the diagnosis of HCC and monitoring of HCC patients, and increased AFP production in human HCC patients reflects the DNA synthesis of tumor cells (Abelev, 1971). Therefore, we finally wanted to confirm whether autocrine TGF-b signal affects AFP production of HCC. Among all of the transfectants, 2SA-ST1 cells produced the largest amount of AFP (Figure 7), indicating that the TGF-b signal is involved in the regulation of AFP production in HCC.

Discussion Variations in TGF-b-mediated Smad activation and heteromeric complex formation were recently reported in studies of TGF-b signal transduction (Piek and Roberts, 2001). It is true that the 3S-A derivatives of Smad2 and Smad3 act as dominant-negative mutants by


2317

4-

AFP concentration (x 10 4 ng/ml)

[3H] thymidine incorporation ( x 10 -5 / 10 -5 cells)

a − TGF-β β + TGF- β

3210-

Cell line: HuH-7

% Reduction in DNA synthesis

b

2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

100 -

* 50 -

3WT -ST1

6-

4-

2-

0Cell line: HuH-7

2WT -ST1

2SA -ST1

3WT -ST1

3SA -ST1

Figure 7 High expression levels of AFP in stable Smad2(3S-A) transfectants. AFP concentration in the culture media. Samples were collected from the culture media and measured by enzymelinked immunoassay, as described in the text. The data are from one representative experiment with each point determined in triplicate

*

02WT 2SA Cell line: HuH-7 -ST1 -ST1

8-

3SA -ST1

Figure 6 Growth resistance to TGF-b by overexpression of R-Smads(3S-A). (a) Growth resistance to TGF-b in stable R-Smads(3S-A) transfectants. Parental HuH-7 cells or the stable transfectants were treated with 40 pm TGF-b1 for 20 h followed by 4 h incubation with 1 mCi of [3H]thymidine, and the acid-insoluble [3H]thymidine was counted. (b) Percentage reduction of [3H]thymidine incorporation of parental HuH-7 cells and the stable transfectants. Parental HuH-7 cells and their derivatives were treated with TGF-b1 and were processed as in (a). The acidinsoluble [3H]thymidine was counted and compared to that incorporated by the cells that had not been exposed to TGF-b1 treatment. Each bar represents the mean7s.e.m. values from one representative experiment (n ¼ 5) of a total of four separate experiments. *Po0.05, 2WT-ST1 vs 3WT-ST1 values

blocking their functions of endogenous R-Smads proteins. However, it remains unknown whether each R-Smad selectively blocks its own signal, or blocks its counterpart simultaneously. By the following criteria, we conclude that Smad2(3S-A) transfectants show autocrine TGF-b signal through the activated Smad3: (i) These cells displayed significant expression of TGFb1 at both the protein and mRNA levels in the mutants compared to those in the parental HuH-7 cells; (ii) TGF-b signal phosphorylated Smad3; (iii) TGF-b signal translocated Smad3 protein into the nuclei; and (iv) TGF-b signal induced the formation of Smad3–Smad4

heteromer. These observations were in complete contrast to the finding that Smad2 activation mediated by the TGF-b signal was severely impaired in Smad2(3S-A) transfectants. Different binding sites on each Smad2 and Smad3 and/or their relative affinities for TbRI might cause the constitutive Smad3 activation, despite the impaired Smad2 signal observed in the transfectants. On the other hand, Smad2 and Smad3 exert different roles in TGF-b signal transduction (Massague´, 1998). The specificity of TGF-b signaling is, in part, determined by the combination of different Smad complexes (Zhang and Derynck, 1999). Accordingly, the ratio of the activated Smad2 to Smad3 can affect the transcriptional regulations of the proximal target genes. PAI-1 expression is tightly regulated by hormones and cytokines including TGF-b (Healy and Gelehrter, 1994). Besides acting as the main inhibitor of the urokinase-type plasminogen activator system, PAI-1 has been shown to be an important regulator of extracellular matrix interactions (Gutierrez et al., 2000). In particular, observations that neoplastic cells including HCCs express elevated levels of PAI-1 suggest that this protein is involved in tumor migration, invasion, and metastasis by blocking cellular adhesion and by promoting the degradation of basement membrane (Dan et al., 1985; Bajou et al., 1998; Zhou et al., 2000). Confirming the importance of fibrinolytic components and the cancer phenotype, clinical studies have demonstrated that a high PAI-1 level is associated with a poorer prognostic outcome (Pedersen et al., 1994a, b). It is also well known that the development of tumor cells requires oxygen and nutrients, which are supplied through neovascularization. Angiogenic potential is, therefore, a prerequisite for tumor growth. Several factors participating in the development of microvasculature have been identified. Among them, VEGF is the most intriguing factor in Oncogene


2318

regard to tumor angiogenesis (Dvorak et al., 1991). A correlation exists between the degree of vascularization of the malignancy and VEGF expression in human HCC. Also, elevations in VEGF levels have been detected in the serum of HCC patients (Mise et al., 1996). Immunohistochemical studies have localized the VEGF protein not only to the tumor cells, but also to the vasculature associated with such tumors (Mise et al., 1996). This localization indicates that tumor-secreted VEGF accumulates in the target cells, confirming that VEGF is a potent neovascularization agent in human HCC. Our current results demonstrated that treatment of HuH-7 cells with TGF-b resulted in inductions of PAI-1 and VEGF mRNAs and release of these proteins in the culture media, thus providing a possible network among the TGF-b signal, extracellular protease activity, and angiogenesis. Interestingly, both PAI-1 and VEGF were significantly induced in Smad2(3S-A) transfectants. Importantly, forced expression of Smad3(3S-A) blocked TGF-b-induced transcriptional activation, and such transcriptional activation could be stimulated by overexpression of wild-type Smad3. These results prove direct evidence that the autocrine TGF-b signal enhances PAI-1 and VEGF production through Smad3– Smad4 complex. In addition, TGF-b-dependent Smad3 activation caused by blockage of Smad2-mediated signal, enhances PAI-1 and VEGF production, and leads to the development of HCC (Figure 8). Our findings also shed light on the mechanism why HCC is resistant to the growth inhibition induced by the TGF-b signal. Obtaining the resistance to TGF-b in cancer cells has been thought to be one of the important factors in malignant transformation, since TGF-b is a potent natural antiproliferative cytokine for normal epithelial cells. Tumor cells often escape the antiproliferative effects of TGF-b by mutational inactivation or dysregulated expression of components such as TGF-b receptors and/or Smad proteins in its signaling pathway (de Caestecker et al., 2000; Miyazono et al., 2001). Our findings demonstrated that stable transfectants of either Smad2(3S-A) or Smad3(3S-A) acquired resistance to this antiproliferative effect, and implied that the antiproliferative signal was mediated through the heterotrimer comprised of Smad2, Smad3, and Smad4 in HuH-7 cells. These results are consistent with a report that the TGF-b-induced growth inhibitory signal was mediated by the heterotrimeric complex of Smad2, Smad3, and Smad4 (Feng et al., 2000). Accordingly, the functional deficiency of Smad2 or Smad3 would cause the cells to resist growth inhibition by TGF-b. TGF-b1 mediates its autoinduction by Smad signaling. In our studies, an autoinduction of TGF-b1 was displayed in both Smad2(3S-A) and Smad3(3S-A) transfectants. This interesting phenomenon suggests the possibility that TGF-b-mediated autoinduction does not necessarily require the heterotrimer comprised of Smad2, Smad3, and Smad4. Since Smad2, in contrast to Smad3, cannot bind to DNA directly, Smad2 and Smad3 may have a different subset of target genes and thus regulate distinct gene responses (Massague´, 1998). Therefore, a different balance of the TGF-b signal Oncogene

TGF-β1

Autoinduction P

Smad3 P

Smad3

Smad2(3S-A)

Smad4

Progress in Tumor Development

P

Smad3

Smad4

PAI-1 gene

Smad4

VEGF gene

P

Smad3

Nucleus Figure 8 Distortion of the autocrine TGF-b signal enhances HCC development. 2SA-ST1 cells display growth resistance to TGF-b and the blockage of Smad2-mediated TGF-b signal. In a state of upregulated TGF-b production, the cells still show TGF-bdependent Smad3 activation such as phosphorylation, nuclear translocation, and Smad3–Smad4 heteromeric formation. The autocrine TGF-b signal propagates through the Smad3–Smad4 complex, and stimulates the transcription of the metastatic and angiogenic factors including PAI-1 and VEGF genes. Thus, this distortion of autocrine TGF-b signal in HuH-7 cells accelerates their malignant potential by enhancing cell growth as well as PAI-1 and VEGF production

through the Smad2 and Smad3 pathway may affect TGF-b1 transcription. Otherwise, the transcriptional regulation of the TGF-b gene can be reflected by mitogen-activated protein kinase pathways through the activator protein-1 complex (Kim et al., 1990). To date, the TGF-b pathway has been found to be inactivated in human tumor cells through a variety of mechanisms, including mutational inactivation of Smad2 and Smad4 proteins (Riggins et al., 1997). Moreover, numerous reports have explained the mechanisms relating mutational inactivation of Smads to malignant transformation (Miyaki et al., 1999; Schwarte-Waldhoff et al., 2000; Prunier et al., 2001). On the other hand, resistance to growth inhibition by TGF-b does not necessarily mean a complete loss of TGF-b signaling, but rather suggests that tumor cells acquire an advantage by selective loss of the tumor suppressor activities of TGF-b while retaining its tumor-promoting activities including tumor invasion, metastasis, and angiogenesis (Engel et al., 1998). These phenomena might occur not only in our stable transfectant models but also in Smad knockout/


2319

transgenic animals and in spontaneous human cancer cells bearing mutational inactivation in Smads. Based on our data, the signaling pathway from other Smad components, except for the mutationally inactivated Smad, can remain intact in these animal models or human cancer tissues. Further studies will be required to elucidate whether the altered TGF-b signal caused by Smad mutational inactivation leads to modification of biological features in such cases.

Materials and methods

or HA-pcDNA3. Mutants of Smad2(3S-A) and Smad3(3S-A) were produced by PCR-based mutagenesis as previously described (Matsuzaki et al., 2000a). The cDNA for dominant-negative Smad4DC (truncation of the C-terminal 39 amino acids) was amplified by PCR (forward, 50 GGGAATTCGGACAATATGTCTATTACGA-30 ; reverse, 50 -GCGTCGACTGGGTAATCCGGTCCCCAGC-30 ) and subcloned into Flag-tagged pcDNA3. The PF1 element, which comprises base pairs from –794 to –532 of human PAI-1 promoter, was amplified by PCR (forward, 50 -TATCTCGAGTACCATGGTAACCCCTGG-30 ; reverse, 50 -GGAAGCTTAACAGCCACAGGCATGCA-30 ) and subcloned into pGL3-Basic (Promega, Madison, WI, USA). The integrity of the constructs was confirmed by sequencing.

Cell culture procedure HuH-7 cells and stable transfectants were grown as subconfluent monolayer cultures in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal calf serum (FCS). All experiments, except the measurement of proteins secreted from the cells in the culture media, were carried out in the log phase of growth after the cells had been plated for 24 h. Northern blot hybridization analysis mRNAs were isolated and hybridized with labeled cDNA probes for TGF-b1, Smad2, Smad3, PAI-1, and human GAPDH as described (Date et al., 1998). To determine human VEGF mRNA levels, 0.52-kb PCR (forward, 50 -GACGCCCTCCGAAACCATGAAC-30 ; reverse, 50 -CAGCCTGGCTCACCGCCTTG-30 ) fragment of mouse VEGF was used as the cDNA probe. Measurement of TGF-b1, PAI-1, VEGF, and AFP by ELISA For measurement of the concentrations, the cells were plated in flasks and incubated for 24 h in regular growth medium. After having reached 60–80% confluency, the monolayers were washed twice and replaced by serum-free DMEM. The culture media were collected 60 h later. To determine the concentrations of TGF-b1, VEGF, PAI-1, and AFP in the media, we used the enzyme-linked immunoassay kits for human TGF-b1 (R&D System, Minneapolis, MN, USA), VEGF (R&D System), PAI-1 (Biopool, Umea˚, Sweden), and AFP (DAINABOT, Tokyo, Japan). [3H]thymidine incorporation To measure [3H]thymidine incorporation, 3 104 cells were plated in six-well plates, and incubated for 24 h in regular growth medium. The cells were changed to serum-free medium in the absence or presence of 40 pm recombinant human TGFb1 (R&D System), and incubated for 20 h. DNA synthesis was measured by the incorporation of 1 mCi/ml [3H]thymidine (Amersham Pharmacia Biotech, Buckinghamshire, UK) into 5% trichloroacetic acid-precipitable material after a 4 h pulse, as previously described (Matsuzaki et al., 1990). Experimental and control values were compared using the unpaired Student’s t-test and ANOVA; Po0.05 was considered to be significant. Constructs and reagents Mammalian expression vectors with an N- or C-terminal tag (Flag or HA) were constructed by inserting oligonucleotides encoding for epitope-tag sequences into pcDNA3 (Invitrogen, San Diego, CA, USA). The coding regions of Smad2, Smad3, and Smad4 were amplified by PCR and subcloned into Flag-

Transient transfection, immunoprecipitation, and immunoblot The conditions for cell culture and transfection have been described elsewhere (Matsuzaki et al., 2000a). The cells were analysed at 48 h after transfection with the indicated combinations of constructs by LipofectAMINE (GIBCO/BRL, Rockville, MD, USA). The cells were incubated in the absence or presence of 200 pm TGF-b1 for 1 h. Alternatively, the cells were incubated with 5 mg/ml anti-TGF-b neutralizing Ab (R&D System) for 24 h. Subsequently, they were dissolved in the lysis buffer (10 mm Tris-HCl (pH 7.5), 150 mm NaCl, 1% NP40, 1 mm EDTA, 10 mg/ml aprotinin, and 0.3 mm phenylmethylsulfonyl fluoride). To monitor the protein levels of endogenous Smads and exogenous gene products, lysates were subjected to immunoprecipitation with anti-Smad2/3 Ab (Transduction Laboratories, Lexington, KY, USA) followed by absorption to protein G Sepharose (Pharmacia Biotech, Wikstro¨ms, Sweden). For the heteromeric complex experiments, lysates were immunoprecipitated with anti-Smad4 Ab (Abcam, Cambridge, UK), anti-Flag M2 Ab (Kodak, New Haven, CT, USA), or antiSmad3 (I-20) Ab (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The bound protein was eluted by heating in SDS–PAGE sample buffer containing DTT. Samples were subjected to SDS– PAGE and then were transferred to nitrocellulose membranes (Hybond-ECL RPN303D; Amersham). Immunoblot was performed using the following primary antibodies: anti-phosphoSmad2 (Ser 465/467) Ab (Upstate Biotechnology, Lake Placid, NY, USA); anti-Smad2/3 Ab; anti-Smad3 (I-20) Ab; antiSmad4 (B-8) Ab (Santa Cruz Biotechnology); anti-Flag M2 Ab; anti-HA Ab (BAbCO, Richmond, CA, USA). The blots were incubated with primary Ab for 1 h at room temperature and then were washed three times. The appropriate secondary Ab was added for 1 h at room temperature. After washing, the immunoreactive proteins were visualized by ECL (RPN2106; Amersham) and autoradiography. Metabolic labeling Conditions for transfection and metabolic labeling have been described (Matsuzaki et al., 2000a). Briefly, the cells were incubated for 12 h in serum-free DMEM in the absence or presence of 5 mg/ml anti-TGF-b Ab. The cells were preincubated with phosphate-free DMEM (GIBCO/BRL) for 3 h, incubated with the same phosphate-free media containing 300 mCi/ml [32P]phosphate (Amersham) for 2 h at 371C, and then stimulated with 200 pm TGF-b1 for 45 min. Subsequently, metabolic-labeled cells were solubilized in the lysis buffer at 41C. The lysates were precipitated with anti-Flag M2 Ab, resolved by SDS–PAGE, and visualized by autoradiography. To determine the total protein levels, the samples after radioactive decay were subjected to immunoblot with antiFlag M2 Ab and chemiluminescence (ECL; Amersham) as Oncogene


2320 described above. In the experiments for the phosphorylation of endogenous Smad3, the cell lysates were precipitated with antiSmad3 (I-20) Ab. For determination of the total protein levels, aliquots of their lysates before immunoprecipitation were used for immunoblot with anti-Smad3 (I-20) Ab after radioactive decay.

Mounting Medium (Shandon Lipshaw, Pittsburgh, PA, USA), the cells were observed by a fluorescence microscope. Transcriptional response assay

Immunofluorescence study

The cells were seeded at 1 105 cells per well into six-well clusters, and then were incubated for 24 h in regular growth medium. They were subjected to transfection with LipofectAMINE, 0.4 mg of reporter plasmid, and the indicated constructs or with an empty vector alone, and incubated for 4 h. After wash with the medium, the cells were cultured for 24 h in 0.5% FCS/DMEM, and then were further incubated for 24 h in the absence or presence of 200 pm TGF-b1. The cells were lysed, and the luciferase activity of the cell extracts was measured by a luminometer (Berthold, Bad Wildbad, Germany) using Dual-LuciferaseTM Reporter Assay System (Promega). The luciferase activities were normalized based on the Renilla luciferase activity.

Subcellular localization of Smad2 was determined as previously described (Matsuzaki et al., 2000a). The cells were seeded onto slides and incubated for 24 h. The cells were then incubated in the absence or presence of 200 pm TGF-b1 for 1 h. After fixation with 4% paraformaldehyde, slides were incubated with anti-Smad2 Ab or anti-Smad3 (I-20) Ab at 41C for 16 h. Then, Alexa Fluort labeled goat anti-mouse IgG or rabbit anti-goat IgG (COSMO BIO, Tokyo, Japan) was added. After mounting the slides with Perma Fluort Aqueous

Acknowledgements We thank Dr SW Qian (National Cancer Institute, Bethesda, MD, USA), Dr R Derynck (University of California at San Francisco, San Francisco, CA, USA), Drs CH Heldin, and K Miyazono (Ludwig Institute for Cancer Research) for providing us with cDNAs of rat TGF-b1, human Smad2, Smad3, Smad4, and human TbRI in this study.

Stable transfection of Smads The conditions for cell culture and transfection have been described (Matsuzaki et al., 2000b). For stable transfectants, HuH-7 cells were transfected with the indicated constructs by LipofectAMINE. Selection was initiated with 800 mg/ml of G418 (Life Technologies, Rockville, MD, USA) and continued for several weeks until drug-resistant colonies emerged.

References Abelev GI. (1971). Adv. Cancer Res., 14, 295–358. Bajou K, Noe¨l A, Gerard RD, Masson V, Brunner N, HolstHansen C, Skobe M, Fusenig NE, Carmeliet P, Collen D and Foidart JM. (1998). Nat. Med., 4, 923–928. Dan K, Andreasen PA, Grndahl-Hansen J, Kristensen P, Nielsen LS and Skriver L. (1985). Adv. Cancer Res., 44, 139–266. Date M, Matsuzaki K, Matsushita M, Sakitani K, Shibano K, Okajima A, Yamamoto C, Ogata N, Okumura T, Seki T, Kubota Y, Kan M, McKeehan WL and Inoue K. (1998). J. Hepatol., 28, 572–581. de Caestecker MP, Piek E and Roberts AB. (2000). J. Natl. Cancer Inst., 92, 1388–1402. Derynck R, Zhang Y and Feng X-H. (1998). Cell, 95, 737–740. Dvorak HF, Sioussat TM, Brown LF, Berse B, Nagy JA, Sotrel A, Manseau EJ, Van de Water L and Senger DR. (1991). J. Exp. Med., 174, 1275–1278. Engel ME, Datta PK and Moses HL. (1998). J. Cell. Biochem. Suppl., 30/31, 111–122. Fan G, Ma X, Kren BT and Steer CJ. (1996). Oncogene, 12, 1909–1919. Feng X-H, Lin X and Derynck R. (2000). EMBO J., 19, 5178–5193. Ferrara N. (1999). Curr. Top. Microbiol. Immunol., 237, 1–30. Gutierrez LS, Schulman A, Brito-Robinson T, Noria F, Ploplis VA and Castellino FJ. (2000). Cancer Res., 60, 5839–5847. Healy AM and Gelehrter TD. (1994). J. Biol. Chem., 269, 19095–19100. Hua X, Liu X, Ansari DO and Lodish HF. (1998). Genes Dev., 12, 3084–3095. Kim S-J, Angel P, Lafyatis R, Hattori K, Kim KY, Sporn MB, Karin M and Roberts AB. (1990). Mol. Cell. Biol., 10, 1492–1497. Oncogene

Macı´ as-Silva M, Abdollah S, Hoodless PA, Pirone R, Attisano L and Wrana JL. (1996). Cell, 87, 1215–1224. Massague´ J. (1998). Annu. Rev. Biochem., 67, 753–791. Matsuzaki K, Date M, Furukawa F, Tahashi Y, Matsushita M, Sakitani K, Yamashiki N, Seki T, Saito H, Nishizawa M, Fujisawa J and Inoue K. (2000a). Cancer Res., 60, 1394–1402. Matsuzaki K, Date M, Furukawa F, Tahashi Y, Matsushita M, Sugano Y, Yamashiki N, Nakagawa T, Seki T, Nishizawa M, Fujisawa J and Inoue K. (2000b). Hepatology, 32, 218–227. Matsuzaki K, Yoshitake Y, Miyagiwa M, Minemura M, Tanaka M, Sasaki H and Nishikawa K. (1990). Jpn. J. Cancer Res., 81, 345–354. Mise M, Arii S, Higashituji H, Furutani M, Niwano M, Harada T, Ishigami S, Toda Y, Nakayama H, Fukumoto M, Fujita J and Imamura M. (1996). Hepatology, 23, 455–464. Miyaki M, Iijima T, Konishi M, Sakai K, Ishii A, Yasuno M, Hishima T, Koike M, Shitara N, Iwama T, Utsunomiya J, Kuroki T and Mori T. (1999). Oncogene, 18, 3098–3103. Miyazono K, Kusanagi K and Inoue H. (2001). J. Cell Physiol., 187, 265–276. Nakao A, Afrakhte M, Moreń A, Nakayama T, Christian JL, Heuchel R, Itoh S, Kawabata M, Heldin N-E, Heldin C-H and ten Dijke P. (1997). Nature, 389, 631–635. Pedersen H, Bru¨nner N, Francis D, Østerlind K, Ronne E, Hansen HH, Dan K and Grndahl-Hansen J. (1994a). Cancer Res., 54, 4671–4675. Pedersen H, Grndahl-Hansen J, Francis D, Østerlind K, Hansen HH, Dan K and Bru¨nner N. (1994b). Cancer Res., 54, 120–123. Piek E and Roberts AB. (2001). Adv. Cancer Res., 83, 1–54. Prunier C, Ferrand N, Frottier B, Pessah M and Atfi A. (2001). Mol. Cell. Biol., 21, 3302–3313.


2321 Riggins GJ, Kinzler KW, Vogelstein B and Thiagalingam S. (1997). Cancer Res., 57, 2578–2580. Schwarte-Waldhoff I, Volpert OV, Bouck NP, Sipos B, Hahn SA, Klein-Scory S, Lu¨ttges J, Ko¨lppel G, Graeven U, EilertMicus C, Hintelmann A and Schmiegel W. (2000). Proc. Natl. Acad. Sci. USA, 97, 9624–9629. Shirai Y, Kawata S, Tamura S, Ito N, Tsusima H, Takaishi K, Kiso S and Matsuzawa Y. (1994). Cancer, 73, 2275–2279. Wrana JL. (2000). Cell, 100, 189–192.

Yakicier MC, Irmak MB, Romano A, Kew M and Ozturk M. (1999). Oncogene, 18, 4879–4883. Yamanaka N, Okamoto E, Toyosaka A, Mitunobu M, Fujihara S, Kato T, Fujimoto J, Oriyama T, Furukawa K and Kawamura E. (1990). Cancer, 65, 1104–1110. Zhang Y and Derynck R. (1999). Trends Cell Biol., 9, 274–279. Zhou L, Hayashi Y, Itoh T, Wang W, an Rui J and Itoh H. (2000). Pathol. Int., 50, 392–397.

Oncogene