Transforming Growth Factor β1 Induces Nuclear Export of Inhibitory Smad7*

Transforming growth factor β (TGF-β) signals from membrane to nucleus through serine/threonine kinase receptors and their downstream effector molecules, termed Smad proteins. Recently, Smad6 and Smad7 were identified, which antagonize TGF-β family signaling by preventing the activation of signal-transducing Smad complexes. Here we report that Smad7, but not Smad6, inhibits TGF-β1-induced growth inhibition and the expression of immediate early response genes, including Smad7. Interestingly, in the absence of ligand, Smad7 was found to be predominantly localized in the nucleus, whereas Smad7 accumulated in the cytoplasm upon TGF-β receptor activation. The latter is in accordance with the physical association of Smad7 with the ligand-activated TGF-β receptor complex in the cell membrane. Whereas the ectopically expressed C-terminal domain of Smad7 was also exported from the nucleus to the cytoplasm upon TGF-β challenge, a Smad7 mutant with a small deletion at the C terminus or only the N-terminal domain of Smad7 was localized mainly in the cytoplasm in the absence or presence of ligand. This suggests that an intact Mad homology 2 domain is important for nuclear localization of Smad7. The nuclear localization of Smad7 suggests a functional role distinct from its antagonistic effect in receptor-mediated Smad activation.

Transforming growth factor ␤ (TGF-␤) 1 family members control a broad spectrum of cellular processes, including proliferation, differentiation, apoptosis, and migration (1,2). Signaling by these pleiotropic cytokines occurs via ligand-induced heteromeric complex formation of distinct type I and type II serine/ threonine kinase receptors (3,4). Type I receptors act downstream of the type II receptors and, upon their phosphorylation and activation by the constitutively active type II receptor kinase, propagate the signal downstream through the activation of Smad proteins. Smads, of which homologous proteins were first identified through genetic screens in Drosophila (5), play a pivotal role in the intracellular signaling of TGF-␤ family members (6,7). The so-called pathway-restricted Smad2 and Smad3 transiently interact with and become phosphorylated by the activated TGF-␤ type I receptor (T␤R-I; Refs. 8 -11), after which they oligomerize with the common mediator Smad4 (12)(13)(14). The hetero-oligomeric Smad complex is then translocated to the nucleus and regulates the transcription of target genes (15,16). Smad proteins share two regions of high sequence similarity, termed Mad homology (MH) 1 and MH2 domains, at the N-and C-terminal regions, respectively. The N-terminal domain has been shown to have direct sequencespecific DNA-binding activity (17)(18)(19)(20), and the C-terminal domain can act as a transcriptional activator (21,22).
Recently, inhibitory Smads, e.g. Smad6 (23) and Smad7 (24,25), have been identified that form stable interactions with the activated T␤R-I, thereby preventing binding to and activation of pathway-restricted Smads. The inhibitory Smads diverge structurally from other Smad family members; whereas they share extensive sequence similarity with other Smads in the MH2 domain, their N-terminal regions share limited sequence similarity with other Smads. Smad7 mRNA expression is potently induced by TGF-␤1, consistent with the possibility that it may act in an autoregulatory negative-feedback loop (25).
In the present investigation, we have further characterized the negative role of Smad7 on TGF-␤-induced responses, e.g. growth inhibition and induction of mRNA levels for direct target genes, using a cell line stably expressing Smad7. Furthermore, Smad7 was found to be localized within the nucleus in the absence of ligand and translocated to the cytoplasm upon TGF-␤1 stimulation. The differential localization of different Smad7 deletion mutants suggests that an intact MH2 domain is important for its nuclear localization and TGF-␤1-induced nuclear export. Differential compartmentalization is an important control mechanism for the regulation of the activity of signal-transducing components. Implications of TGF-␤1-induced nuclear export of Smad7 for Smad7 function are discussed.
Cell Lines-Mink lung epithelial (Mv1Lu) cells and COS1 cells were obtained from the American Type Culture Collection. Mv1Lu cells and COS1 cells were cultured in Dulbecco's modified Eagle's medium (Sigma) with 10% fetal bovine serum and antibiotics. Growth Inhibition Assay-Mv1Lu cells were seeded at 1 ϫ 10 4 cells/ well in 24-well plates. Before the addition of TGF-␤1, the Mv1Lu cells were simultaneously treated (or not treated) with 100 M zinc chloride and the indicated concentrations of TGF-␤1 for 20 h to induce Smad7 expression. Before harvesting, the cells were pulsed with 0.2 Ci of [methyl-3 H]thymidine (81.0 Ci/mmol; Amersham Corp.) for 2 h. The cells were fixed with ice-cold 5% trichloroacetic acid for more than 20 min and washed twice with 5% trichloroacetic acid and once with water. Solubilization of the cells was done with 400 l of 0.1 M NaOH for 20 min at room temperature. The 3 H radioactivity incorporated into DNA was determined by liquid scintillation counting.
RNA Isolation and Northern Blotting-Mv1Lu cells were kept in 0.5% fetal bovine serum and induced with 100 M zinc chloride 20 h before stimulation with 10 ng/ml TGF-␤1 for 2 h followed by RNA extraction. Isolation of total RNA and Northern blotting were performed essentially as described previously (27). The intactness and amount of RNA loaded were checked by staining the gel with ethidium bromide. The cDNA probes used in the hybridizations were a 1.8kilobase pair EcoRI/XhoI human Smad7 fragment, a 3-kilobase pair EcoRI fragment of the human plasminogen activator inhibitor 1 (PAI-1) gene, and a 1.5-kilobase pair mouse JunB fragment.
Iodination of Ligands and Affinity Cross-Linking-TGF-␤1 was iodinated using the chloramine T method according to Frolik et al. (28). Cross-linking was performed as described previously (29), except that incubation with 125 I-TGF-␤1 was performed at room temperature for 2 h. The transfected cells lines were stimulated for 24 h with 100 M zinc chloride to induce the expression of Smad7. Complexes of Smad7 and affinity-labeled receptors were immunoprecipitated with antibody directed against the Flag in F-Smad7. To investigate the expression levels of receptors, aliquots of cell lysates were directly analyzed by SDS-polyacrylamide gel electrophoresis.
Transfections, Metabolic Labeling, and Immunoprecipitation-Transient transfections of COS1 cells were performed using the DEAEdextran protocol. Stable transfection of Mv1Lu cells with pMEP4 expression vector was done using the calcium phosphate precipitation method, as described previously (29); selection was performed with 420 units/ml hygromycin B (Calbiochem). Induction with zinc chloride was done with 100 M zinc chloride for 20 h, unless otherwise indicated. Metabolic labeling of cells and immunoprecipitation were performed as described previously (29).
Immunofluorescence-Cells were grown in LAB TEK chambers (Nunc, Naperville, IL) and incubated with Dulbecco's modified Eagle's medium containing 0.3% fetal bovine serum in the absence or presence of 10 ng/ml TGF-␤1 for 2 h. The slides were washed once with phosphate-buffered saline, fixed for 10 min with 4% paraformaldehyde, washed three times with phosphate-buffered saline, subsequently permeabilized with 0.1% Triton X-100 in phosphate-buffered saline for 5 min, and washed again three times with phosphate-buffered saline. Slides were blocked by 10% goat serum for 1 h at room temperature and then incubated with 10% goat serum with anti-Flag antibody (20 g/ml) for 15 h at 4°C. The slides were subsequently washed three times, incubated with tetramethylrhodamine B isothiocyanate-conjugated goat anti-mouse IgG antibody (diluted 1:40), and washed again four times. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (1 g/ ml) for 10 min at room temperature followed by three washes. To visualize the fluorescence, a Zeiss microscope was used. For cell counting, a square lattice mounted in one of the eyepieces was used. In the COS1 cell experiments, 200 cells were counted in five different fields, and the nuclear localization was checked by 4Ј,6-diamidino-2-phenylindole staining.

Smad7 but not Smad6 Inhibits TGF-␤1-induced Growth
Inhibition-To investigate the effect of inhibitory Smads on TGF-␤1-mediated growth inhibition, mouse F-Smad6 and F-Smad7 were stably transfected into Mv1Lu cells; Smad6 and Smad7 were placed under transcriptional control of the human inducible metallothionein IIA promoter using the pMEP4 expression vector. Two forms of Smad6 were tested: a mouse long Smad6 version (Smad6L; Ref. 23), and a human short Smad6 version (Smad6S; Ref. 26). When the expression of Smad6S, Smad6L, and Smad7 was analyzed, we observed that the inhibitory Smads were also expressed in the absence of the inducer zinc chloride. This is the result of leakage of the metallothionein promoter. However, pretreatment of the cells with zinc chloride in each transfectant led to increases in Smad6 and Smad7 expression (Fig. 1). Subsequently, we tested the response of the Smad6 and Smad7 (and empty pMEP4) transfectants to TGF-␤1-induced growth inhibition in the absence or presence of zinc chloride. Multiple independent Smad7-expressing clones were analyzed. We found that Smad7 prevented TGF-␤1-induced growth-inhibitory effects in Mv1Lu (Fig. 2, A-C). In all Smad7 clones, TGF-␤1-induced growth inhibition was blocked to an extent that correlated with the expression levels of Smad7. pMEP4-transfected Mv1Lu cells showed a similar dose-response curve after TGF-␤1 stimulation as nontransfected Mv1Lu cells in the absence or presence of zinc chloride. Ectopic expression of Smad6L or Smad6S did not affect the TGF-␤1induced growth inhibition (Fig. 2, D and E). Thus Smad7, but not Smad6, can antagonize TGF-␤1-induced growth inhibition.
Smad7 Inhibits TGF-␤1-induced Transcriptional Responses-We subsequently examined the effect of ectopic Smad7 expression on TGF-␤1-induced expression of the endogenous early response genes JunB, Smad7, and PAI-1 (Fig. 3). After treating cells with zinc chloride 20 h before 2 h of TGF-␤1 stimulation, the ectopically induced Smad7 was found to partially inhibit TGF-␤1-induced immediate early gene responses at a Smad7 mRNA ectopic expression level similar to that of the endogenous gene after TGF-␤ treatment (Fig. 3B). Quantitative analysis of the changes in gene expression by Smad7 using a PhophorImager showed that basal gene expression levels were slightly reduced, and that among the three genes, PAI-1 expression was most prominently affected by ectopic expression of Smad7. Interestingly, upon treatment with zinc Northern blot analysis on RNA from Mv1Lu cells that were stably transfected with pMEP4-Smad7 with or without pretreatment with zinc chloride on exposure to TGF-␤1 is shown. The endogenous mRNAs for JunB, Smad7, and PAI-1 are indicated by arrows. The asterisk in the Smad7 blot indicates the ectopically expressed F-Smad7 mRNA. The amount and intactness of total RNA loaded were checked by ethidium bromide staining (C and E). For hybridizations with probes for JunB (A) and Smad7 (B), the same blot was used, which was different from the blot used for hybridization with the PAI-1 probe (D).

FIG. 4. Association of Smad7 with TGF-␤ receptors in Mv1Lu cells.
Association of F-Smad7 with endogenous TGF-␤ receptors was tested on Mv1Lu cells that were stably transfected with pMEP4-Smad7 (7-10). Cells were pretreated (where indicated) with zinc chloride, and receptors were covalently affinity-labeled with iodinated TGF-␤1. Cell lysates were subjected to Flag immunoprecipitation and analyzed by SDS-polyacrylamide gel electrophoresis and a FujiX BioImager. As a control, the absence of specific immunoprecipitation with Flag antibody on pMEP4-transfected cells in the absence or presence of zinc chloride is shown. Expression of receptors was determined by analyzing an aliquot of cell lysate by SDS-polyacrylamide gel electrophoresis without immunoprecipitation.

FIG. 2. Smad7, but not Smad6, inhibits TGF-␤1-mediated growth inhibition.
The inhibitory effect of Smad7 on TGF-␤1-induced growth inhibition in three independent clones (A-C) is shown, as well as the lack of effect of Smad6S (D) and Smad6L (E) on TGF-␤1-induced growth inhibition. As a control, the effect of TGF-␤1 on pMEP4-transfected cells in the absence or presence of zinc chloride is shown. The relative growth compared with control is plotted against the concentration of TGF-␤1. Different lots of TGF-␤1 were used in experiments A-C versus D and E, which explains the difference between the ED 50 values on pMEP4-transfected cells in these experiments. All data shown are means Ϯ S.D. chloride for 6 h, we observed no significant effect on the TGF-␤1 stimulation of early response genes in the Smad7-expressing clones, whereas the apparent ectopic Smad7 mRNA levels were higher after 6 h of pretreatment versus 20 h of pretreatment. Possibly, differences in Smad7 mRNA do not correlate exactly with Smad7 protein levels upon zinc chloride treatment, or a particular posttranslational modification occurs with slow kinetics that may be required for interaction of Smad7 with receptor in Mv1Lu cells.
Smad7 Associates with the Activated TGF-␤ Receptor Complex-We and others have shown previously that Smad7 efficiently associates with the TGF-␤1-induced receptor complex in transfected COS1 cells; Smad7 may inhibit TGF-␤ signaling by preventing the receptor interaction of pathway-restricted Smads (24,25). In an analogous fashion, we examined whether ectopically expressed F-Smad7 could bind to the endogenous TGF-␤ receptor complex. Upon affinity labeling of receptors with 125 I-TGF-␤1 followed by the immunoprecipitation of receptor complex with Flag antibodies, we detected an interaction between TGF-␤1 receptor complex and Smad7 (Fig. 4).
TGF-␤1-induced Nuclear Export of Smad7 in Transfected COS1 cells-The subcellular localization of F-Smad7 and, for comparison, that of F-Smad2 in transfected COS1 cells was analyzed in the absence or presence of the constitutively active T␤R-I(T204D) by immunofluorescence using the Flag antibody. Ectopically expressed F-Smad7 was located mainly in the cell nucleus but was exported from the nucleus to the cytoplasm in a large proportion of treated cells upon cotransfection with T␤R-I(T204D) (Fig. 5, A and B). In contrast, ectopically expressed F-Smad2 was located mainly in the cytoplasm and accumulated in the nucleus upon cotransfection with T␤R-I(T204D) (Fig. 5, A and B), which is in agreement with earlier reports (8,11). The nuclear localization was checked by 4Ј,6diamidino-2-phenylindole staining.

Differential Localization of Smad7 Deletion Mutants in the
Absence or Presence of TGF-␤1-We analyzed the subcellular localization of F-Smad6 and F-Smad7 in the stable Mv1Lu transfectants (Fig. 1) using the Flag antibody and immunofluorescence. F-Smad7 was localized mainly in the nucleus of cells and was exported from the nucleus in response to TGF-␤1. In contrast, F-Smad6L and F-Smad6S were localized in the cytoplasm, and the localization remained unchanged after TGF-␤1 stimulation ( Fig. 6A; data not shown). To gain more insight into regions in Smad7 that are important for nuclear localization and for TGF-␤1-induced nuclear export, we analyzed the subcellular distribution of different Smad7 deletion mutants. pMEP4 expression constructs for the C-terminal domain of F-Smad7 (7C; amino acids 204 -426), F-Smad7C with a C-tail deletion (7C⌬; amino acids 204 -407), Smad7 with deletion of the C-tail (7NC⌬; amino acids 1-407), and the N-terminal domain of Smad7 (7N; amino acids 1-261) were stably transfected into Mv1Lu cells, and cell lines were characterized for Smad7 expression upon zinc chloride treatment (Fig. 1). All cells showed some degree of leaky expression, but in all cell lines, zinc treatment induced the expression of the Smad protein. Subcellular distribution of the 7C mutant was similar to that of wild-type Smad7 in the absence or presence of TGF-␤1. An intact MH2 domain seemed important for nuclear localiza- tion because F-Smad7C⌬ and F-Smad7NC⌬ mutants were predominantly localized in the cytoplasm in the absence of TGF-␤1. Smad7N was found in the cytoplasm in the absence and presence of TGF-␤1; it had a spotted localization, suggesting association with a particular cell structure (data not shown). None of the Smad7 deletion constructs were able to interfere with TGF-␤1-induced growth inhibition as observed for wildtype Smad7 (data not shown). DISCUSSION In many signal transduction pathways, there is a regulated nuclear entry of specific proteins in response to ligand stimulation (30). Examples include pathway-restricted Smad2 and Smad3 and common mediator Smad4, which relay signals from cell surface TGF-␤ receptors to the nucleus, where they modulate the expression of specific genes (6,7). Smad6 and Smad7 have inhibitory roles in TGF-␤ signaling (23)(24)(25)31) and prevent receptor-dependent activation of signal-transducing Smads. Here we have further characterized the signaling responses of TGF-␤1 that are regulated by inhibitory Smads. Examination of the subcellular distribution of Smad7 revealed a nuclear localization in the absence of ligand; after TGF-␤1stimulation, Smad7 is exported into the cytoplasm. Additional insight into the mechanism for the differential compartmentalization of Smad7 was obtained from analyzing the subcellular distribution of Smad7 deletion mutants.
Using stable cell lines expressing Smad6 and Smad7 under the control of an inducible promoter, we showed that Smad7, but not Smad6 (Smad6L and Smad6S), was able to abrogate TGF-␤1-induced growth inhibition. Both Smad6 and Smad7 have been shown to bind the activated T␤R-I and inhibit certain TGF-␤ responses (23)(24)(25)(26). From our data, however, Smad7 appears to be a more potent inhibitor of TGF-␤1 signaling than Smad6. These observations are consistent with studies in Xenopus, which showed that upon overexpression, Smad7 effectively inhibits both TGF-␤/activin and bone morphogenetic protein-mediated responses, whereas Smad6 preferentially inhibits bone morphogenetic protein pathways (25,29,31).
Ectopic expression of Smad7 was also effective in decreasing the TGF-␤1-induced increase of mRNA for early response genes. A comparison of the expression of Smad7 mRNA that was achieved upon induction of the heterologous promoter with the level of the endogenous Smad7 mRNA expression revealed that ectopic expression was similar to the endogenous TGF-␤1induced Smad7 mRNA expression. Thus, the inhibitory action of Smad7 seems to be exerted at a physiological level of Smad7 mRNA.
In accordance with previous findings (24,25), Smad7 was found to interact with the activated TGF-␤ receptor complex. No dramatic increase in interaction was observed upon zinc chloride treatment. However, in this particular clone, there is a considerable leaky expression, which does not increase much after treatment with zinc chloride.
In the absence of ligand, Smad7 was localized in the nucleus and transported to the cytoplasm upon TGF-␤1 treatment. This is contrast to the observed translocation of Smad2 from the cytoplasm to the nucleus upon cotransfection with the constitutively active T␤R-I in COS1 cells. Moreover, Smad6, ectopically expressed in Mv1Lu cells, was not localized in the nucleus, lending support to the notion that the observed TGF-␤1induced nuclear export of Smad7 is of functional importance.
The analysis of the subcellular distribution of Smad7 deletion mutants revealed that an intact MH2 is required for nuclear localization; the C-terminal domain of Smad7 (7C) was localized in the nucleus, whereas Smad7 proteins with deletions of the C-tail (7NC⌬ and 7C⌬) and the N-terminal domain only (7N) were found to reside constitutively in the cytoplasm.
The fact that 7C, but not 7C⌬, accumulated in the cytoplasm upon TGF-␤1 treatment indicates that the nuclear export requires an intact C-terminal domain. We have previously shown that the C-terminal domain of Smad7 interacts with the activated TGF-␤ receptor complex in a manner similar to that of wild-type Smad7 and inhibited TGF-␤1-induced transcriptional responses only slightly less efficiently than wild-type Smad7 (29). Smad7 does not interact with T␤R-I unless T␤R-I is phosphorylated by the T␤R-II kinase (24,25). Thus, in the absence of ligand, T␤R-I and Smad7 do not interact, and as shown in the present study, Smad7 is considered predominantly in the nucleus. The mechanism for the nuclear accumulation is unclear, as is the mechanism for the ligand-induced nuclear export of Smad7; the latter could possibly involve the association of Smad7 (C-terminal domain) with the activated T␤R-I, shifting the equilibrium toward a cytoplasmic localization for Smad7.
A possible nuclear role for Smad7 may be in transcriptional regulation. Fusing Smad7 with a heterologous DNA-binding domain revealed that the Smad7 MH2 domain has a potential transcriptional activation domain (data not shown), although less efficient than that of the MH2 domains of Smad1 and Smad4 (21,22). The N-terminal region of Smad7 shows little similarity to MH1 domains, which are capable of binding DNA directly. It is unclear whether the N-terminal domain of Smad7 has DNA-binding activity. However, Smad7 may contact DNA indirectly, similar to Smad2, which has been shown to contribute to transcription by complex formation with the DNA-binding factor FAST-1 (15). In future studies, we will explore the nuclear function of Smad7. Which responses, in addition to the inhibitory effect on TGF-␤ signaling, are induced upon ectopical expression of Smad7? The direct or indirect effect of Smad7 on the expression of genes involved in the TGF-␤ signal transduction pathway as well as other signaling pathways will be investigated.