Phosphorylation of Smad7 at Ser-249 Does Not Interfere with Its Inhibitory Role in Transforming Growth Factor-β-dependent Signaling but Affects Smad7-dependent Transcriptional Activation*

Smad proteins are major components in the intracellular signaling pathway of transforming growth factor-β (TGF-β), and phosphorylation is an important mechanism in regulation of their functions. Smad7 was identified as a potent inhibitor of TGF-β-dependent signaling. We have identified serine 249 in Smad7 as a major phosphorylation site, the phosphorylation of which was not affected by TGF-β1. Abrogation of the phosphorylation by substitution of Ser-249 with alanine or aspartic acid residues did not affect the ability of Smad7 to inhibit TGF-β1 and BMP7 signaling. No differences were found in the stability or in the intracellular distribution of Smad7 mutants compared with the wild-type molecule. However, Smad7 fused to the DNA-binding domain of GAL4 induced transcription from a reporter with mutated TATA minimal promoter in a Ser-249-dependent manner. Moreover, a reporter with the SV40 minimal promoter was inhibited by GAL4-Smad7, and this effect was also dependent on Ser-249 phosphorylation. The amplitude of effects on transcriptional regulation was dependent on cell type. Our results suggest that phosphorylation of Smad7, unlike phosphorylation of the receptor-regulated Smads, does not regulate TGF-β signaling but rather affects TGF-β-independent effects of Smad7 on transcriptional regulation.

Smad proteins are major components in the intracellular signaling pathway of transforming growth factor-␤ (TGF-␤), and phosphorylation is an important mechanism in regulation of their functions. Smad7 was identified as a potent inhibitor of TGF-␤-dependent signaling. We have identified serine 249 in Smad7 as a major phosphorylation site, the phosphorylation of which was not affected by TGF-␤1. Abrogation of the phosphorylation by substitution of Ser-249 with alanine or aspartic acid residues did not affect the ability of Smad7 to inhibit TGF-␤1 and BMP7 signaling. No differences were found in the stability or in the intracellular distribution of Smad7 mutants compared with the wild-type molecule. However, Smad7 fused to the DNA-binding domain of GAL4 induced transcription from a reporter with mutated TATA minimal promoter in a Ser-249-dependent manner. Moreover, a reporter with the SV40 minimal promoter was inhibited by GAL4-Smad7, and this effect was also dependent on Ser-249 phosphorylation. The amplitude of effects on transcriptional regulation was dependent on cell type. Our results suggest that phosphorylation of Smad7, unlike phosphorylation of the receptor-regulated Smads, does not regulate TGF-␤ signaling but rather affects TGF-␤-independent effects of Smad7 on transcriptional regulation.
Discovery of Smad proteins provided insights into the intracellular signaling by TGF-␤ 1 family members. Three groups of Smads have been described. Receptor-regulated Smads (R-Smads) are direct targets of activated receptors and provide the signaling specificity. Common mediator (Co-Smad) Smad4 forms complexes with R-Smads and is involved in signal propagation. Finally, Smad7 and Smad6 were identified as inhibitors (I-Smads) of signaling by members of the TGF-␤ family (1)(2)(3). Smad7 has been suggested to inhibit TGF-␤ signaling by inhibiting the phosphorylation of R-Smads by type I receptors (1)(2)(3)(4). Interestingly, Smad7 was found to occur abundantly in the nuclei of certain cells and to be exported from the nucleus upon TGF-␤ stimulation or a change in cell substrate (5,6). This suggests that Smad7 may also have a function in the nucleus, which may be independent of the inhibition of ligandinduced signaling at the receptor level.
Since expression of Smad7 is up-regulated after TGF-␤ stimulation, it has been suggested that Smad7 is involved in negative feedback of TGF-␤ signaling. Induction of Smad7 has also been described as the pivotal mechanism whereby tumor necrosis factor-␣ and interferon ␥ inhibit TGF-␤ signaling (7,8). Smad7 has been found to be up-regulated in human tumors, but no mutation in human cancers has yet been described (9).
Phosphorylation has been found to be a potent regulatory mechanism of Smad functions (10 -13). Sequential phosphorylation of R-Smads at two C-terminal serine residues by activated receptors is the triggering event in ligand-dependent intracellular signaling (13). Phosphorylation in the linker region, presumably by mitogen-activated protein kinase, may affect nuclear translocation of R-Smads (12). Moreover, phosphorylation of R-Smads by other kinases has also been shown to influence intracellular signaling (14,15). Phosphorylation of mammalian Smad4 has not been reported, but Xenopus Smad4␤ has been identified as a phosphoprotein (16). Among inhibitory Smads, Smad6 is a phosphoprotein (17), whereas phosphorylation of Smad7 has not been characterized.
We show here that Smad7 is phosphorylated, and we identify Ser-249 as a major phosphorylation site. Mutation of Ser-249 did not affect the inhibitory effect of Smad7 on TGF-␤ or BMP7 signaling and did not interfere with nuclear localization of Smad7. However, the TGF-␤-independent transcriptional activity of Smad7 was affected by mutation of Ser-249, suggesting that phosphorylation of Smad7 at Ser-249 is important for its ligand-independent ability to regulate transcription.
Fusion constructs of the DNA-binding domain of GAL4 with wildtype Smad7 and Smad7 mutants were obtained by subcloning Smad7 into the pM expression vector (CLONTECH, Palo Alto, CA). The GAL4binding site-containing luciferase reporter GAL4-TGTA-luc, containing a mutated TATA box from retinoid acid receptor ␤2, and GAL4-TK-luc, containing a herpes simplex virus-thymidine kinase basal promoter, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Supported by a scholarship from Svenska Institutet. were obtained from Johan Ericsson, and GAL4-SV40-luc, containing an SV40 early promoter, was from Cory Abate.
COS-1, Mv1Lu, and NIH-3T3 cell lines were obtained from ATCC, Manassas, VA. DR-26 clone of Mv1Lu cells, lacking TGF-␤ type II receptor, was obtained from Joan Massagué. For stable transfection, wild-type FLAG-Smad7 was subcloned in pMEP4 vector. The construct and empty vector were transfected in DR-26 and NIH-3T3 cells, followed by selection in presence of hygromycin B.
All data in figures are presented as average Ϯ S.E. Standard tests were used to evaluate statistical significance of the obtained results.
Identification of Phosphorylation Sites in Smad7-[ 32 P]Orthophosphate labeling and mapping of tryptic phosphopeptides, as well as phosphoamino acid analysis, were performed as described (13). FLAG-Smad7 protein was immunoprecipitated with M2 anti-FLAG antibody (Sigma). Visualization of phosphopeptides was carried out using the Fujix BAS2000 imaging system.
[ 35 S]Methionine Metabolic Labeling-Detection and quantification of transiently expressed or stably transfected wild-type and mutant FLAG-Smad7 was performed by [ 35 S]methionine metabolic labeling (10 Ci/ml, 4 h), followed by immunoprecipitation with M2 anti-FLAG antibody. To determine protein stability, cells were pulse-labeled with [ 35 S]methionine (100 Ci/ml, 1 h), chased with medium containing unlabeled methionine, and immunoprecipitated with M2 anti-FLAG antibody. The immunoprecipitated material was resolved by SDS-PAGE and quantified using the Fujix BAS2000 imaging system.
Luciferase Reporter Assays-Luciferase reporter constructs were cotransfected into cells together with Smad7 expression plasmids using various transfection techniques (DEAE-dextran, LipofectAMINE, or Fugene6) according to the manufacturer's instructions. Cells were incubated and treated as described in the figure legends. Luciferase activity in stimulated and nonstimulated cells was determined using a luciferase detection kit (Promega, Madison, WI). Luminescence was measured in a Victor multilabel plate reader (EG & G Wallac, Turku, Finland). Cells were cotransfected with a ␤-galactosidase expression vector, and luciferase activity values were compensated for differences in transfection efficiency by a colorimetric ␤-galactosidase assay.
For transcriptional activation studies, GAL4 DNA-binding domain fusion constructs were cotransfected into cells together with GAL4binding site-containing reporter constructs, and the level of luciferase expression was determined as described above.

RESULTS
Smad7 Is a Phosphoprotein-To investigate whether Smad7 is a phosphoprotein, full-length and N-terminal or C-terminal deletion mutants of FLAG-tagged Smad7 were transiently expressed in Mv1Lu and COS-1 cells. The cells were incubated with [ 32 P]orthophosphate, treated or not with 10 ng/ml TGF-␤1 for 2 h, and FLAG-Smad7 proteins were immunoprecipitated with anti-FLAG antibodies; immunoprecipitates were analyzed by SDS-PAGE. We found that full-length and an N-terminal deletion mutant of Smad7 (FLAG-Smad7C) were phosphorylated ( Fig. 1, A and C). No phosphorylation of a mutant with deleted MH2 domain was found (Fig. 1, E and F). The protein expression level was monitored by immunoprecipitation of FLAG-Smad7 from [ 35 S]methionine-labeled cells (Fig. 1, B, D, F, and G), which were transfected in parallel with the cells used for [ 32 P]orthophosphate labeling. Significantly higher phosphorylation of FLAG-Smad7C correlated with the higher expression of the protein; therefore, no difference in its specific phosphorylation compared with full-length FLAG-Smad7 was observed (Fig. 1).
Unlike the case for R-Smads, phosphorylation of Smad7 was not affected by TGF-␤1 in COS-1 cells with cotransfected T␤R-I and T␤R-II (Fig. 1E) or without cotransfection of the receptors (Fig. 1, A and C). We did not observe TGF-␤-dependent differences in Smad7 phosphorylation in NIH-3T3 cells, which express endogenous receptors. Smad7 phosphorylation was not affected in DR-26 cells, which lack T␤R-II, and therefore are nonresponsive to TGF-␤ (Fig. 1G). Moreover, the two-dimensional phosphopeptide maps of Smad7 from cells treated or not with TGF-␤1 showed a similar pattern (data not shown). Thus, TGF-␤ has no effect either on the total phosphorylation level or on the pattern of phosphorylated sites. Therefore, Smad7 is phosphorylated by kinase(s) other than the TGF-␤ receptors.
The fact that the phosphorylation of Smad7 was TGF-␤independent prompted us to attempt to evaluate how this phosphorylation is regulated. We observed that the level of Smad7 phosphorylation was higher if cells were serum-starved and that the phosphorylation was decreased if cells were cultured in the presence of serum. The effect of serum starvation was found to be especially strong for FLAG-Smad7C. No changes in the level of protein expression were detected, suggesting that the differences in phosphorylation were not related to the quantity of Smad7. The phosphopeptide patterns of Smad7 from serum-starved and proliferating cells were also similar, as evaluated by two-dimensional mapping (data not shown). Therefore, the activity of kinase(s) and/or phosphatase(s) regulating Smad7 phosphorylation may be dependent on the serum starvation of cells but is not directly regulated by TGF-␤1.
Smad7 Is Phosphorylated at Serine 249 -To identify phosphorylation sites, Smad7 was immunoprecipitated from [ 32 P]orthophosphate-labeled cells and subjected to tryptic digestion; the generated peptides were separated by two-dimensional mapping. We identified three major phosphopeptides in full-length FLAG-Smad7 and four in FLAG-Smad7C (Fig. 2, A  and C). Spots 1 and 2 were reproducibly detected in more than 20 repeated phosphopeptide maps of full-length Smad7, whereas spot 4 was not always seen. Spot 4 was also weak in maps of FLAG-Smad7C (Fig. 2C). The highly phosphorylated phosphopeptide giving rise to the diffuse spot 3 of FLAG-Smad7C was not found in full-length protein in any of the experiments. The phosphopeptides from spots 1-3 contained only phosphoserine, and neither phosphothreonine nor phosphotyrosine were detected (Fig. 2F). Phosphopeptides were also subjected to radiochemical sequencing. However, the sequencing of spots 1 and 2 did not provide reliable data, probably due to the fact that the phosphoserine residue(s) was located too far from the N termini of the phosphopeptides to be detected in radiochemical sequencing by Edman degradation (see below). Therefore, an alternative strategy to localize the phosphorylated residues was used. Since the phosphopeptides found in full-length FLAG-Smad7 (spots 1 and 2) were found also in FLAG-Smad7C (Fig. 2, A and C), we constructed a series of expression vectors containing serine-to-alanine mutations of each of the serine residues present in FLAG-Smad7C. All these mutants were expressed in COS-1 cells that were subjected to [ 32 P]orthophosphate labeling; Smad7C and full-length Smad7 mutants were then trypsinized and subjected to two-dimensional phosphopeptide analysis. Unequivocally, mutation of Ser-249 to an alanine residue resulted in disappearance of the phosphopeptide corresponding to spot 1 (Fig. 2, B and D). Interestingly, in FLAG-Smad7C, mutation of Ser-249 to an alanine residue also led to disappearance of spot 2 (Fig. 2D), suggesting that in this mutant protein, Ser-249 phosphorylation may also affect phosphorylation at another site. Ser-249 is located at position 36 in the corresponding tryptic peptide (Fig.  2G). This explains the lack of detection by radiochemical sequencing, which gives reliable results only for phosphorylated residues in the N-terminal 20 amino acid residues of each peptide.
The composite spot 3, consisting of three differently migrating phosphopeptides unique to FLAG-Smad7C and not seen in full-length FLAG-Smad7, was very prominent. The phospho-rylated residue in this group of phosphopeptides was found to be Ser-206, since upon mutation of Ser-206 to an alanine residue, all three phosphopeptides of spot 3 disappeared (Fig. 2E). Interestingly, Ser-206 is neighbored by a PPPPY sequence, which may be involved in regulation of ubiquitin-dependent degradation (20). Therefore, phosphorylation of Ser-206 in the truncated Smad7 may interfere with its degradation, which may explain the higher level of FLAG-Smad7C protein compared with full-length FLAG-Smad7. However, because phosphorylation at Ser-206 was found only in the truncated mutant and not in the full-length protein, we considered it an artifact caused by the truncation. Thus, we have identified Ser-249 as a major phosphorylation site in Smad7.

Smad7-dependent Inhibition of TGF-␤ and BMP Signaling
Is Not Dependent on Smad7 Phosphorylation-To investigate whether phosphorylation of Smad7 affects its inhibitory action on TGF-␤ and BMP signaling, we used luciferase reporter assays, which are dependent on the respective intact R-Smaddependent pathways. CAGA (12) -luc is activated upon treatment of cells with TGF-␤ but not BMP, and its activation requires a receptor-dependent phosphorylation of Smad3 (18). Wild-type Smad7 inhibits the receptor-dependent phosphorylation of R-Smads and blocks TGF-␤-induced activation of the CAGA (12)luc reporter (18). We found that S249A and S249D mutants of Smad7 were as potent inhibitors as wild-type Smad7 (Fig. 3A). Neither was any significant effect of the mutation of Ser-249 found on inhibition of BMP7-dependent activation of the GCCG (12) -luc reporter (Fig. 3B). Stimulation of this reporter is dependent on receptor-induced activation of Smad1 and Smad5, which are specific for the BMP signaling pathway (21). In addition, the S249A and S249D mutants of Smad7 were equally efficient as wild-type Smad7 in inhibition of ligandinduced activation of another reporter responsive to TGF-␤ and BMP, SBE (4) -luc (data not shown).
Because CAGA (12) -luc and GCCG (12) -luc reporters have artificially designed promoters, we also tested how reporters with natural promoters, responsive to the ligand, were affected by interference with Smad7 phosphorylation. When reporters containing 800 base pairs of the PAI-1 promoter (p800-luc), 296 base pairs of collagen ␣2(I) promoter (pH6-luc), promoters of the cyclin-dependent kinase inhibitors p21 and p15 (p21-luc, p15-luc), or an E2F1-luc reporter were analyzed, the inhibition of the TGF-␤-induced activation by Smad7 was not found to be affected by mutations of Ser-249 (Fig. 3, C and D, data not shown). The diversity of molecular mechanisms, by which Smads regulate different promoters used in this study, supported the conclusion that phosphorylation at Ser-249 does not affect the ability of Smad7 to interfere with Smad-mediated TGF-␤ and BMP signaling pathways .
Phosphorylation of Smad7 Does Not Affect Its Stability and Nuclear Localization-Since regulation of Smad biosynthesis and degradation has been shown to be an important mechanism of modulation of TGF-␤ signaling (20), we investigated whether phosphorylation of Smad7 affects its stability. Transfection of similar quantities of cDNAs of the wild-type and the S249A and S249D mutants of Smad7 resulted in comparable protein expression levels (data not shown). Moreover, pulsechase experiments showed that interference with phosphorylation at Ser-249 did not affect the half-life of Smad7 (Fig. 4). In the presence of the protein synthesis inhibitor cycloheximide, the half-life of FLAG-Smad7 was ϳ1 h, and no significant differences were found for the mutants. In absence of cycloheximide the apparent half-life was estimated at ϳ4 h both for wild-type Smad7 and for Smad7 mutants (data not shown). Thus, the phosphorylation of Smad7 at Ser-249 appears not to affect its stability and half-life.
In two recent reports, the nuclear localization of Smad7 has FIG. 2. Identification of phosphorylation sites in Smad7. Twodimensional phosphopeptide mapping of wild-type (A) and S249A mutant (B) of the full-length FLAG-Smad7, and of wild-type (C), S249A (D), and S206A (E) mutants of the N-terminal deleted construct FLAG-Smad7C. The proteins were expressed in [ 32 P]orthophosphate-labeled COS-1 cells. The arrows show positions of major phosphopeptides numbered from 1-4, and arrowheads show application points. F, two-dimensional phosphoamino acid analysis of major phosphopeptides. The nomenclature corresponds to the designations of spots in A-E. Migration positions of different phosphoamino acids were determined using nonradioactive internal standards. pY, pT, and pS correspond to phosphotyrosine, phosphothreonine, and phosphoserine, respectively. G, location of the identified phosphorylatable serine residues 249 and 206 in the sequence of mouse Smad7.

FIG. 3. TGF-␤ and BMP signaling is not affected by phosphorylation of Smad7
. TGF-␤-responsive CAGA (12) -luc (A) or BMP-responsive GCCG (12) -luc (B) reporter constructs were transfected in Mv1Lu cells together with wild-type and mutants of Smad7. Cells were stimulated with 10 ng/ml of TGF-␤1 (A, C, and D) or 100 ng/ml of BMP7 (B); luciferase activity was measured after 20 h and corrected for transfection efficiency. Natural promoter-derived p800-luc (C) and p21p-luc (D) reporter constructs, susceptible to TGF-␤ stimulation, were cotransfected with Smad7 constructs and cells were treated with TGF-␤1, as described for A and B. In all panels, open bars correspond to unstimulated cells and filled bars to cells stimulated with ligand. Expression levels were normalized by measuring ␤-galactosidase activity. Data are presented as mean from n Ͼ3 experiments, and error bars represent S.D. been described (5,6). These observations also suggested that the nuclear localization is a regulated process. Therefore, we tested whether phosphorylation at Ser-249 affects the intracellular distribution of Smad7. Fig. 5 shows that mutation of Ser-249 did not lead to significant differences in intracellular localization, since S249A and S249D mutants and wild-type of FLAG-Smad7 all localized to the cell nucleus to a similar extent. Thus, the intracellular distribution of Smad7 is not affected by its phosphorylation at Ser-249.
Transcriptional Activity of Smad7 Is Dependent on Phosphorylation at Ser-249 -The findings that Smad7 is a nuclear protein, together with the observation that a GAL4-Smad7 fusion protein showed transcriptional activity in PC3U cells, 2 prompted us to investigate whether the phosphorylation of Smad7 affects its transcription-regulating activity. We tested the effect of full-length Smad7, fused to the DNA-binding domain of GAL4, on the response of various reporters containing different minimal promoters downstream of the GAL4-binding elements (GAL4-TGTA-luc, GAL4-SV40-luc, and GAL4-TKluc). We found that, in NIH-3T3 cells, wild-type Smad7 upregulated the GAL4-TGTA-luc reporter containing a point-mutated TATA box to decrease the background activity (Fig. 6A). Abrogation of Smad7 phosphorylation at Ser-249 by its substitution to an alanine residue significantly inhibited this stimulation (p Ͻ 0.01). Substitution of Ser-249 to aspartic acid residue resulted in an effect similar to the wild-type Smad7, probably due to partial mimicking of phosphorylation by introduction of a negative charge (Fig. 6A). Similar results were obtained for the wild-type and S249A mutant when this reporter was used in COS-1 (Fig. 6C) and Mv1Lu cells (data not shown). For Mv1Lu cells, however, the effect was weaker than for COS-1 cells. Interestingly, using the reporter containing an SV40 minimal promoter instead of TGTA, wild-type Smad7 was found to repress luciferase expression; in this case the S249D mutant of Smad7 was mimicking the wild-type Smad7 effect on transcription, and the S249A mutant even induced it (Fig. 6D). Results obtained with the GAL4-TK-luc reporter were similar to the data for the GAL4-SV40-luc reporter (data not shown). Thus, Smad7 can act as a transcriptional activator or as a repressor depending on the type of promoter. Our data suggest that the phosphorylation at Ser-249 regulates the transcriptional activity of Smad7, probably through regulation of its interaction with other components in transcriptional complexes.

DISCUSSION
Here we report the identification of Ser-249 as a major phosphorylation site in Smad7. Unlike the case for R-Smads, the phosphorylation of Smad7 was found not to be dependent on TGF-␤ treatment of cells. We show that phosphorylation at Ser-249 does not affect the stability or subcellular localization of Smad7, nor does it affect the ability of Smad7 to inhibit TGF-␤ signaling. Interestingly, however, the presence of the phosphorylatable Ser-249 in Smad7 was found to be important for a novel function ascribed to Smad7, i.e. its ability to regulate transcription.
Ser-249 is one of the two major phosphorylation sites in Smad7 (Fig. 2); the second major site is currently under investigation. Ser-249 is located in the C-terminal part of the region corresponding to the linker in other Smads. Interestingly, a Smad activation domain (SAD) has been identified in the corresponding part of Smad4; SAD interacts with MSG-1, a transcriptional coactivator of Smad4, and has an important role for the transcriptional activity of Smad4 (22, 23). We have not observed any homology between SAD and the corresponding 2 M. Landström, unpublished data. region of Smad7; however, given the importance of SAD in Smad4 transcriptional regulation, it is interesting that phosphorylation in the corresponding region of Smad7 affects its transcriptional activity.
The nuclear localization (Fig. 5) is consistent with the possibility that Smad7 is involved in regulation of gene expression. R-Smads and Smad4 are known as potent transcriptional regulators, and they are translocated to the nucleus upon ligand addition. Their function is dependent on interactions with coactivators and corepressors and on the ability of Smads to bind DNA directly (1)(2)(3)24). Recently, Bai et al. (25) reported that inhibitory Smad6 also can act as a transcriptional regulator through interaction with the homeobox (Hox)c-8 protein, extending the suggestion of transcriptional regulator function to inhibitory Smads. In line with previous observations (26), we could detect only a weak effect of GAL4-FLAG-Smad7 on the activity of GAL4-binding sequence-containing reporters in Mv1Lu cells of epithelial origin (data not shown). However, in cells of mesenchymal origin, GAL4-FLAG-Smad7 significantly induced the GAL4-TGTA-luc reporter. Interestingly, a GAL4binding sequence-containing reporter with an SV40 or TK minimal promoter was inhibited by the wild-type GAL4-FLAG-Smad7 fusion. Similar opposite effects on gene expression were also described for Smad3; Smad3 was found to potently induce CAGA (12) -luc (18) but inhibited stimulation of the goosecoid reporter (27). In similarity to Smad3, the stimulatory or inhibitory effects of Smad7 on the regulation of transcription may depend on the interaction between Smad7 and other proteins. Our data suggest that such interactions may be regulated by Smad7 phosphorylation, as the phosphorylation-deficient S249A mutant of Smad7 shows impaired effects when compared with the wild-type protein (Fig. 6). The S249D mutant of Smad7, partially mimicking the presence of a phosphoryl group by introducing a negative charge, behaved similar to wild-type Smad7.
Our findings suggest that Smad7 is not only involved in control of signaling pathways of TGF-␤ family members, but may also have direct effects in the nucleus, possibly regulated by other regulatory pathway(s). To explore this possibility, it will be important to identify the kinase phosphorylating Smad7, components interacting with Smad7 in the nucleus, as well as target genes for Smad7 as a transcription factor. These studies are in progress.