Arkadia Induces Degradation of SnoN and c-Ski to Enhance Transforming Growth Factor-β Signaling*

Transforming growth factor-β (TGF-β) signaling is controlled by a variety of regulators that target either signaling receptors or activated Smad complexes. Among the negative regulators, Smad7 antagonizes TGF-β signaling mainly through targeting the signaling receptors, whereas SnoN and c-Ski repress signaling at the transcriptional level through inactivation of Smad complexes. We previously found that Arkadia is a positive regulator of TGF-β signaling that induces ubiquitin-dependent degradation of Smad7 through its C-terminal RING domain. We report here that Arkadia induces degradation of SnoN and c-Ski in addition to Smad7. Arkadia interacts with SnoN and c-Ski in their free forms as well as in the forms bound to Smad proteins, and constitutively down-regulates levels of their expression. Arkadia thus appears to effectively enhance TGF-β signaling through simultaneous down-regulation of two distinct types of negative regulators, Smad7 and SnoN/c-Ski, and may play an important role in determining the intensity of TGF-β family signaling in target cells.

Members of the transforming growth factor-␤ (TGF-␤) 2 family have a diverse array of activities, including regulation of growth, motility, extracellular matrix production, differentiation, and apoptosis, in various target cells. TGF-␤ signal is transduced through two different types of serine/threonine kinase receptors, termed type I and type II, on the cell surface (1)(2)(3)(4). Upon binding of TGF-␤ to the type II receptor, the type I receptor is recruited to the ligand-receptor complex and is phosphorylated by the constitutively active type II receptor kinase. Type I receptor is then activated and phosphorylates the receptor-regulated Smads, Smad2 and Smad3. Phosphorylated Smad2 and Smad3 form oligomeric complexes with Smad4, a common Smad, and translocate into the nucleus. The activated Smad complexes then bind to promoter regions of target genes either directly or together with other transcription factors and regulate their transcription in cooperation with transcriptional coactivators and corepressors (5,6). TGF-␤ signaling must be tightly controlled, because its aberration has been reported to cause progression of various diseases, including cancer and fibrosis (7,8).
TGF-␤ signaling is regulated by various proteins that target either signaling receptors or activated Smad complexes. Smad7, an inhibitory Smad, appears to play a central role in down-regulation of the activity of signaling receptors. Smad7 is located in the nucleus and is translocated to the plasma membrane in a Smurf (Smad ubiquitin regulatory factor) 1/2-dependent fashion (9 -12). Smad7 then competitively inhibits phosphorylation of Smad2 and Smad3 through binding to activated type I receptor kinase (13,14). In addition, Smad7 recruits Smurf1/2 to the type I receptor. Because Smurf1 and Smurf2 are HECT-type E3 ubiquitin ligases, they down-regulate the levels of type I receptor proteins through ubiquitylation and proteasomal degradation (9,10). In contrast, SnoN and c-Ski, which are nuclear corepressor proteins of the Ski family, are potent regulators that target the activated Smad complexes (15)(16)(17)(18)(19). They have been shown to recruit histone deacetylases (HDACs) to the Smad complexes and to compete with transcriptional coactivator p300/CBP for binding to the Smad complexes (15,16,19).
Misexpression of these negative molecules has been implicated in various pathological conditions. Reduction of Smad7 protein has been reported in human fibroblasts of patients with scleroderma (20) and in tissues with renal fibrosis in mice (21). Systemic deletion of one copy of the Sno or Ski gene causes increased susceptibility to chemical carcinogens (22,23). Increased expression of Smad7 has been found in inflammatory bowel disease (24) and pancreatic cancer (25). Increased expression of SnoN or c-Ski has been implicated in the progression of esophageal squamous cell carcinomas (26,27), melanomas (28), estrogen receptor-positive breast carcinomas (29), and colorectal carcinomas (30). Maintenance of expression levels of these negative regulators within the appropriate ranges thus appears to be important. Arkadia was originally isolated by gene-trap mutagenesis in mice as a factor required for induction of the mammalian node in the extraembryonic lineages (31), and it was found to induce mesendoderm by enhancing nodal-related signaling (32). Arkadia is a nuclear protein with 989 amino acid residues, with a characteristic RING domain in its C terminus. We previously found that Arkadia is an E3 ubiquitin ligase that enhances TGF-␤ signaling by targeting a negative regulator, Smad7 (33). Another mechanism of enhancement of TGF-␤ signaling involving degradation of phospho-Smad2 and Smad3 has also recently been proposed (34). Arkadia is the first example of an E3 ubiquitin ligase that positively regulates TGF-␤ family signaling.
We report here that Arkadia targets SnoN and c-Ski in addition to Smad7. Arkadia thus effectively enhances TGF-␤ signaling through simultaneous down-regulation of two distinct types of negative regulators.
cDNA Construction-Expression constructs encoding mouse Arkadia, human SnoN, and human c-Ski were described previously (15,33,35). Mutants of Arkadia and c-Ski were described previously (15,33) or generated using a PCR-based method. SnoN mutants were also generated using a PCR-based approach. Expression constructs encoding a constitutively active mutant of the TGF-␤ type I receptor, ALK-5-TD, ubiquitin, Smad7, Smad4, and Smad2, were described previously (33).
Luciferase Assay-A TGF-␤-responsive reporter, (CAGA) 9 -MLP-Luc (38), was used. Cells were transiently transfected with an appropriate combination of promoter-reporter constructs, pSUPER constructs, and expression plasmids, including pcDEF3-ALK-5-TD-HA, using FuGENE 6 transfection reagent (Roche Diagnostics). Total amounts of transfected DNAs were the same in each experiment. Cells were cultured for 24 h after transfection. Cell lysates were then prepared, and luciferase activities in the lysates were measured by the dual-luciferase reporter system (Promega) using a luminometer (MicroLumat Plus LB96V, Berthold). Values were normalized using Renilla luciferase activity under the control of thymidine kinase promoter.
Antibodies-The antibodies used were as follows: anti-SnoN H-317 (rabbit polyclonal, Santa Cruz Biotechnology) for immunoblotting of endogenous SnoN; anti-Arkadia 65 (see below) for immunoprecipitation of endogenous Arkadia; normal rabbit IgG (Santa Cruz Biotechnology) as a negative control for immunoprecipitation using anti-Arkadia 65; anti-Arkadia 62 (33) for immunoblotting of transfected Arkadia in the ubiquitylation assay; anti-Arkadia 3AP4 (see below) for detection of Arkadia in mouse ES cell lines; anti-tubulin DM 1A (Sigma); anti-FLAG M2 (Sigma); anti-Myc 9E10 (Pharmingen); and anti-hemagglutinin 3F10 (Roche Diagnostics). Anti-Arkadia 65 was prepared by immunizing a rabbit with mouse Arkadia (amino acid residues 854 -936) expressed as a fusion protein with glutathione S-transferase. Anti-Arkadia 3AP4 was prepared by immunizing a rabbit with a peptide derived from mouse Arkadia (amino acid residues 742-756) conjugated to bovine thyroglobulin followed by affinity purification.
Pulse-Chase Analysis-Pulse-chase analysis was performed as described previously (33).

Arkadia May Have Substrates
Other Than Smad7-We previously reported that Arkadia induces ubiquitylation and degradation of Smad7 to augment TGF-␤ signaling (33). To determine whether Smad7 is the major substrate of Arkadia in enhancement of TGF-␤ signaling, we first examined the effect  of Arkadia on TGF-␤ signaling when expression of Smad7 was silenced by RNA interference. HepG2 cells were transfected with Smad7-pSUPER and FLAG-Arkadia and stimulated by cotransfection of ALK-5-TD. TGF-␤ signaling was determined by the TGF-␤-responsive reporter (CAGA) 9 -MLP-Luc. As shown in Fig. 1, silencing of Smad7 resulted in enhancement of TGF-␤ signaling, but Arkadia further enhanced TGF-␤ signaling even after silencing of Smad7, suggesting that Arkadia may have substrates other than Smad7 for enhancement of TGF-␤ signaling.
Arkadia Binds to SnoN and c-Ski-Because SnoN and c-Ski have been shown to be important negative regulators of TGF-␤ signaling, these proteins could be candidates for novel sub-strates of Arkadia. To test this possibility, we first examined the physical interaction of Arkadia with SnoN and c-Ski. Arkadia-C937A (CA), a mutant lacking ubiquitin ligase activity (33), was used to avoid degradation of proteins bound to Arkadia. When 293T cells were transfected with SnoN and Arkadia, Arkadia was coprecipitated with SnoN ( Fig. 2A). As shown in Fig. 2B, Arkadia also bound to c-Ski. These interactions were observed in both the presence and absence of ALK-5-TD, indicating that they are minimally affected by TGF-␤ signaling (Fig. 2, A and B).
Interaction of endogenous Arkadia and SnoN proteins was also examined in NMuMG mouse mammary epithelial cells, because endogenous Arkadia and SnoN proteins are detectable in this cell line, and expression of SnoN was induced by treat- ment with BMP-4 (data not shown). Cells were treated with MG132 to prevent proteasomal degradation of proteins bound to Arkadia. When Arkadia was immunoprecipitated by anti-Arkadia antibody, coprecipitation of SnoN was detected by immunoblotting (Fig. 2C).
We then examined the subcellular localization of Arkadia and SnoN or c-Ski in HeLa cells transfected with Arkadia-CA and SnoN or c-Ski. As shown in Fig. 2D, Arkadia was colocalized with SnoN (upper panels) and c-Ski (lower panels) in the nucleus of transfected HeLa cells.
Arkadia Promotes Ubiquitylation and Degradation of SnoN and c-Ski-Because Arkadia interacted with SnoN and c-Ski, and was colocalized with them in the nucleus, we next examined whether SnoN and c-Ski serve as substrates of Arkadia. Ubiquitylation of SnoN was observed in the presence of wildtype Arkadia but not in the presence of inactive mutant Arkadia-CA or Arkadia ⌬C (amino acid residues 1-936), which lacks C-terminal RING domain ( Fig. 3A and data not shown). Ubiquitylated SnoN was considerably decreased when cells were not treated with MG132, suggesting that the SnoN ubiquitylated by Arkadia was degraded by proteasomes. Arkadia induced ubiquitylation of Smad7 as reported previously (33) but failed to induce that of Smad4, which does not interact with Arkadia (33). Similar results were obtained for c-Ski (Fig. 3B).
Because Arkadia induced ubiquitylation of SnoN and c-Ski, we next used pulse-chase analysis to examine whether Arkadia promotes degradation of SnoN and c-Ski. Turnover of SnoN protein was accelerated in the presence of Arkadia but not in the presence of Arkadia-CA (Fig. 3C). Similar results were obtained in the case of c-Ski (Fig. 3D). These findings indicate that Arkadia promotes degradation of SnoN and c-Ski through ubiquitylation.

Arkadia Ϫ/Ϫ ES Cells Express Higher Levels of SnoN Protein Than Do Wild-type ES Cells-Because
Arkadia enhanced the degradation of SnoN, we compared the levels of expression of SnoN protein between wild-type and Arkadia Ϫ/Ϫ ES cell lines (31) (Fig. 4). Lack of Arkadia protein as well as mRNA was confirmed by immunoblotting and semi-quantitative RT-PCR (Fig. 4, 2nd and 5th panels). We found that levels of expression of the SnoN protein were higher in Arkadia Ϫ/Ϫ ES cell lines than in the wild-type ES cell lines (Fig. 4, top panel). However, levels of expression of SnoN mRNA in the Arkadia Ϫ/Ϫ cell lines were similar to those in wild-type cells (Fig. 4, 4th panel). These findings suggest that absence of Arkadia results in accumulation of SnoN protein through a post-translational mechanism. We thus conclude that endogenous Arkadia contributes to degradation of SnoN. We also examined the expression levels of c-Ski, but it was not detectable in these cell lines. We have previously reported that knockdown of Arkadia increases Smad7 protein in HaCaT cells (33); however, it was not detectable in the ES cell lines.
Interaction of Arkadia with Its Substrates Is Mediated through Its C-terminal Region-We next examined which regions in Arkadia are responsible for interaction with its substrates. Arkadia protein was divided into five fragments, and constructs expressing Arkadia fragments were designed (Fig.  5D). The interaction of these fragments with substrates of Arkadia was examined in transfected 293T cells (Fig. 5, A-C).
As shown in Fig. 5A, SnoN interacted with the C-terminal region of Arkadia (amino acid residues 772-936), whereas c-Ski and Smad7 mainly interacted with both the N-and C-terminal regions of Arkadia (amino acid residues 1-291 and 772-936, respectively; Fig. 5, B and C). These findings suggest that the C-terminal region of Arkadia is important for the interaction with its substrates, although the N-terminal region of Arkadia also participates in the interaction with c-Ski and Smad7.
SnoN and c-Ski Interact with Arkadia through Regions Containing the SAND Domain-We next determined Arkadia-interacting regions in SnoN and c-Ski. SnoN (684 amino acid residues) was divided into three fragments (residues 1-262, 263-479, and 480 -684), and interaction of them with Arkadia was determined in transfected 293T cells. As shown in Fig. 6A (upper panels), Arkadia interacted with the central region of SnoN (residues 263-479). Further analysis was performed using two fragments derived from the central region of SnoN (residues 263-355 and 356 -479). Arkadia interacted with the SnoN fragment containing amino acid residues 263-355 (Fig.  6A, lower panels), which includes the SAND domain in SnoN. c-Ski (728 amino acid residues) was also divided into three fragments (residues 1-210, 211-490, and 491-728), each of which corresponds to that of SnoN in Fig. 6A. Similar to SnoN, Arkadia interacted with the central region of c-Ski (amino acids 211-490), which contains the SAND domain in c-Ski (Fig. 6B).
Arkadia Interacts with SnoN and c-Ski in Complexes with Smads-Because the Arkadia-interacting regions in SnoN and c-Ski overlap with the Smad4-interacting regions in them (the SAND domain) (41), we examined whether SnoN or c-Ski interacts with the Smad complex and Arkadia simultaneously. A DNA-affinity precipitation method using 3ϫ CAGA probe (40) was performed to precipitate Smad4, and coprecipitation of Smad2, Arkadia, and SnoN was determined in the presence and absence of TGF-␤ signaling induced by ALK-5-TD. As shown in Fig. 6C, SnoN was precipitated only in the presence of Smads (compare lanes 3 and 4 and 11 and 12), suggesting that the precipitated SnoN interacts with the Smad complex. Arkadia was coprecipitated with a complex containing Smad2, Smad4, and SnoN (Fig. 6C, lanes 7-10), whereas coprecipitation was not observed in the absence of SnoN (lanes 15 and 16), indicating that SnoN bridges the Smad complex with Arkadia. SnoN thus interacts with the Smad complex and Arkadia simultaneously. These findings suggest that Arkadia interacts with SnoN in its free form as well as that bound to the Smad complex. Similar results were obtained for c-Ski (Fig. 6D).
Arkadia Targets Three Major Negative Regulators of TGF-␤ Signaling, Smad7, SnoN, and c-Ski-Finally, we examined whether Arkadia enhances TGF-␤ signaling when expression of SnoN, c-Ski, and Smad7 was silenced. As shown in Fig. 7A, Arkadia enhanced TGF-␤ signaling when only one of these regulators was knocked down. However, enhancement of TGF-␤ signaling by Arkadia was not observed when these three proteins were simultaneously knocked down. These findings suggest that Smad7, SnoN, and c-Ski are important substrates of Arkadia in maximal enhancement of TGF-␤ signaling, and that Arkadia effectively enhances TGF-␤ signaling through targeting these three suppressors, which function in two different fashions (Fig. 7B).

DISCUSSION
Arkadia was originally identified as a protein required for induction of the mammalian node (31) and was shown to enhance nodal-related signaling (32). As noted above, we previously found that Arkadia, an E3 ubiquitin ligase, enhanced TGF-␤ signaling by targeting Smad7, a negative regulator that acts on signaling receptors (33), although another mechanism of enhancement of TGF-␤ signaling involving the degradation of phospho-Smad2/3 has also recently been proposed (34). We report here that Arkadia also targets SnoN and c-Ski, which are potent negative regulators that inhibit activated Smad complexes. Arkadia thus enhances TGF-␤ family signaling through down-regulation of two distinct types of negative regulators (Fig. 7B).
SnoN and c-Ski are members of the Ski family of oncoproteins (6,42). Ski was originally identified as the transforming protein (v-Ski) of the avian retrovirus that induces oncogenic transformation of chicken embryo cells (43). Human c-Ski was found thereafter as the cellular counterpart of v-Ski. During the isolation of human c-Ski, the Ski-related novel gene (Sno) was identified (44). SnoN and c-Ski have a highly conserved region in their N-terminal portions (45). The Ski family of nuclear oncoproteins represses TGF-␤ signaling principally through  interaction with Smad proteins (41). c-Ski has been reported as a component of the HDAC complex and has also been reported to repress transcription by recruiting HDACs (46). In addition, c-Ski interferes with recruitment of the transcriptional coactivator, p300/CBP, to the Smad complex (15,41). We recently found that SnoN and c-Ski enhance the binding of the Smad complex to DNA, and we suggested the possibility that inactive Smad complex is stabilized on the promoter regions of target genes by SnoN and c-Ski (47).
Genetic evidence in Caenorhabditis elegans indicates that Ski family proteins are important downstream regulators of Smad signaling (48,49). Inhibition of TGF-␤ signaling is considered to be a part of the mechanism of oncogenesis induced by SnoN and c-Ski (50). Functions of SnoN and c-Ski thus appear to be intimately related to Smad signaling, although Smad-independent functions of c-Ski and SnoN have been suggested (39,46,51,52).
SnoN has been reported to be ubiquitylated and degraded by Smurf2 and APC CDH1 (53)(54)(55), whereas ubiquitin ligases that target c-Ski have not been identified yet. To address the possibility that Arkadia recruits these ubiquitin ligases to SnoN, we examined the interaction between Arkadia and Smurf2 or APC CDH1 . Arkadia interacted neither with Smurf2 nor CDH1 (data not shown), although further analyses are needed to determine whether Arkadia promotes SnoN degradation in a manner dependent on Smurf2 or APC CDH1 . Ubiquitylation of SnoN by Smurf2 and APC CDH1 has been reported to be increased by Smad2 or Smad3. Smurf2, a HECT type ubiquitin ligase, does not directly interact with SnoN but does indirectly interact with it through Smad2 (35,53). APC CDH1 , a RING type ubiquitin ligase complex, has also been reported to ubiquitylate SnoN through binding to Smad3. Because the interaction of SnoN with Smad2 or Smad3 is dependent on TGF-␤ signaling, the degradation of SnoN by Smurf2 or APC CDH1 appears to be conditional upon activation of TGF-␤ signaling. Consistent with this, SnoN has been reported to be rapidly degraded in response to TGF-␤ signaling (17,18). In contrast, Arkadia interacts with SnoN and c-Ski irrespective of activation of TGF-␤ signaling. Arkadia associates with SnoN and c-Ski in their free forms as well as when they are bound to Smads. These findings suggest that Arkadia induces constitutive degradation of SnoN and c-Ski. The level of expression of Arkadia protein thus appears to determine the intensity of TGF-␤ signaling that is permitted in target cells. The mechanisms of regulation of Arkadia expression are an important issue for future study.
Recent studies have shown that substrates of Arkadia, Smad7 and SnoN/c-Ski, appear to have Smad-independent functions. Smad7 has been shown to regulate various non-Smad signaling pathways (56). Smad7 and c-Ski have been reported to enhance myogenic differentiation through acting on MyoD (51,57). In contrast, sumoylated SnoN potently suppresses the gene expression of myogenin, a master regulator of myogenesis (58). SnoN and c-Ski have also been reported to repress transcription in cooperation with Mad, thyroid hormone receptor, and MeCP2 (46,59). Arkadia may thus play roles in this type of transcriptional regulation through inducing the degradation of Smad7, SnoN, or c-Ski and affect signal transduction pathways other than TGF-␤ signaling. Similarly, there may be other substrates of Arkadia that are not involved in Smad signaling, although we identified Smad7, SnoN, and c-Ski as important substrates of Arkadia in enhancement of Smad signaling.
For efficient enhancement of TGF-␤ signaling, degradation of Smad7, SnoN, or c-Ski alone does not appear sufficient. Because Smad7 and SnoN are negative feedback regulators (14,17), if Arkadia induces degradation of Smad7 alone, enhanced TGF-␤ signaling would induce SnoN, and transcriptional inhibition of TGF-␤ signaling by SnoN would subsequently occur. Similarly, if Arkadia induces degradation only of SnoN alone, FIGURE 7. Smad7, SnoN, and c-Ski are major targets of Arkadia. A, Arkadia fails to enhance TGF-␤ signaling when Smad7, SnoN, and c-Ski are simultaneously knocked down. Effect of Arkadia on transactivation of a TGF-␤-responsive luciferase reporter, (CAGA) 9 -MLP-Luc, was examined in HepG2 cells. Cells were stimulated by cotransfection of a constitutively active TGF-␤ type I receptor (ALK-5-TD). Expression of endogenous Smad7, c-Ski, and SnoN was silenced by cotransfection of Smad7-pSUPER, c-Ski-pSUPER, and SnoN-pSUPER constructs, respectively. NC1-pSUPER and empty-pSUPER were used as negative controls. Values are means Ϯ S.D. B, model of enhancement of TGF-␤ signaling by Arkadia. Arkadia ubiquitylates both Smad7 and SnoN/c-Ski. The former targets TGF-␤ type I receptor, whereas the latter functions at transcriptional level in the nucleus, to suppress TGF-␤ signaling. By inducing proteasomal degradation of these proteins via different mechanisms of action, Arkadia effectively enhances TGF-␤ signaling.
enhanced TGF-␤ signaling would induce Smad7, and then TGF-␤ signaling would be down-regulated by it. Thus, maximum augmentation of TGF-␤ signaling may not be obtained with degradation of SnoN or Smad7 alone. Denissova and Liu (60) reported that c-Ski transcriptionally represses Smad7 expression. SnoN has also been reported to repress transcription of Smad7 by binding to the promoter region (61). Degradation of SnoN or c-Ski would reactivate transcription of Smad7 and then induce the expression of Smad7 protein. TGF-␤ signaling would thus be down-regulated if Arkadia degraded only SnoN or c-Ski. Arkadia therefore fails to enhance TGF-␤ signaling efficiently unless it induces degradation of all of the three negative regulators of TGF-␤ signaling, Smad7, SnoN, and c-Ski.
In conclusion, we have shown that Arkadia enhances TGF-␤ signaling maximally by inducing ubiquitylation and degradation of SnoN, c-Ski, and Smad7, which are negative regulators of TGF-␤ signaling that act in different ways. Our findings demonstrate an effective mode of action of Arkadia in enhancement of TGF-␤ signaling that is controlled through multiple negative regulators.