Introduction

Transforming growth factor- (TGF-) is a family of multifunctional cytokines that regulate the growth, differentiation, apoptosis, and matrix accumulation of wide varieties of cells (Blobe et al., 2000). TGF- belongs to a large superfamily of structurally related proteins termed the TGF- superfamily, which includes activins, Nodal, myostatin, bone morphogenetic proteins (BMPs), anti-Müllerian hormone (AMH), and growth-differentiation factors (GDFs). TGF- acts as a potent growth inhibitor of most types of cells, including epithelial cells, endothelial cells, hematopoietic cells, and lymphocytes. Perturbations of TGF- signaling therefore result in progression of some tumors (Derynck et al., 2001; Wakefield and Roberts, 2002). However, TGF- stimulates growth of certain types of cells, including mesenchymal cells, and induces epithelial-to-mesenchymal transdifferentiation (EMT) of some epithelial cells. Moreover, TGF- stimulates extracellular matrix accumulation, angiogenesis, immunosuppression, and secretion of parathyroid hormone-related peptide (PTHrP). TGF- may thus function as a pro-oncogenic cytokine under certain conditions.

Runx proteins are a family of transcription factors having a Runt domain responsible for DNA binding (Ito, 1999). Runx was originally given three different names, that is, subunits of polyomavirus enhancer binding proteins 2 (PEBP2), acute myeloid leukemia (AML), and core binding factor a (Cbfa). Three Runx proteins are present in mammals, termed Runxl (PEBP2B/AMLl/Cbfa2), Runx2 (PEBP2A/AML3/Cbfal), and Runx3 (PEBP2C/AML2/Cbfa3). The subunit of PEBP2 PEBP2, binds to Runx proteins through the Runt domains, and stabilizes Runx proteins by preventing ubiquitin-dependent degradation. Runxl plays an important role in definitive hematopoiesis during development, and mutations in Runxl have been found in 30% of human acute leukemias. Runx2 plays a pivotal role in osteogenesis, and haploinsufficiency of Runx2 results in the autosomal dominant bone disease cleidocranial dysplasia (CCD). Runx3 is important for thymogenesis and neurogenesis; and a recent study indicated that alterations of expression of the Runx3 gene result in progression of gastric cancer.

Interestingly, some of the functions of the TGF- superfamily cytokines are similar to those of Runx proteins. TGF- acts on B lymphocytes and induces synthesis of IgA, and Runx3 exhibits a similar effect. BMPs induce bone formation, similar to the function of Runx2. Runx3 has been shown to be involved in growth and apoptosis of gastric epithelial cells, which are also regulated by TGF-. We have recently reviewed functional cooperation between Runx and TGF- superfamily signaling, focusing on their roles in progression of cancer (Ito and Miyazono, 2003). In this review, we present the mechanisms of cooperation between TGF- superfamily signaling and Runx, and some intriguing evidence that this cooperation is regulated through multiple mechanisms.

TGF- signaling pathways

Members of the TGF- superfamily bind to two distinct transmembrane serine/threonine kinase receptors, termed types II and I (Heldin et al., 1997; Shi and Massague, 2003). Upon ligand binding to corresponding type II and type I receptors, type II receptor kinases transphosphorylate the juxtamembrane GS domains of type I receptors, leading to activation of type I receptor kinases. The type II and type I receptors form hetero-tetramers composed of two molecules each of type II and type I receptors (Figure 1). Based on the intracellular signals activated by type I receptors, the TGF- superfamily cytokines are sorted into two subgroups: those activating TGF-/activin-like signals and those activating BMP-like signals. In addition to TGF- and activin, Nodal and myostatin belong to the former group, whereas most BMP family proteins and AMH belong to the latter (Miyazawa et al., 2002). However, in endothelial cells, TGF- binds to two distinct type I receptors, and activates not only TGF-/activin-like signals but also BMP-like signals.

Figure 1.

Induction of osteoblastic differentiation by BMPs through Runx proteins. BMPs induce expression of Runx2 through Dlx5 in osteoprogenitor cells, and this process is sufficient to inhibit their differentiation into myocytes. However, Runx2 alone is not sufficient to induce osteoblastic differentiation; BMP-specific R-Smads interact with Runx2 and possibly other proteins, and induce osteoblastic differentiation. TF, transcription factors

Full figure and legend (161K)

Type I receptors activate various intracellular substrates, among which Smad proteins play central roles in exhibition of biological activities of the TGF- superfamily cytokines. Smad proteins are classified into three subtypes, that is, receptor-regulated Smads (R-Smads), common-partner Smads (Co-Smads), and inhibitory Smads (I-Smads). Smad2 and Smad3 serve as R-Smads transducing TGF-/activin-like signals, whereas Smads 1, 5, and 8 act as R-Smads transducing BMP-like signals. Smad4 is the only Co-Smad in mammals, and is shared by the TGF-/activin and BMP pathways. Smad7 functions as an I-Smad for TGF-/activin and BMP signaling, whereas Smad6 is an I-Smad preferentially inhibiting BMP signaling.

R-Smads are anchored to the plasma membrane through certain molecules, including Smad anchor for receptor activation (SARA) (Qin et al., 2002). The activated type I receptor kinases phosphorylate the last two serine residues at the C-terminal Ser-Ser-X- Ser motif of R-Smads. R-Smads then form hetero-oligomeric complexes with Co-Smad, and translocate into the nucleus (Figure 1). Although the structures of the R-Smad–Co-Smad complexes have not yet been fully determined, hetero-trimers composed of two molecules of R-Smads and one molecule of Co-Smad, and hetero-dimers composed of one molecule each of R-Smad and Co-Smad, may be formed by them.

In the nucleus, the R-Smad–Co-Smad hetero-oligomers interact with various transcription factors and transcriptional coactivators/corepressors, leading to transcriptional regulation of target genes. In addition to these transcription factors, R-Smads and Co-Smads directly bind to specific DNA sequences. A wide range of biological activities of TGF- may result from the ability of Smads to interact with various transcription factors. Thus, more than 30 transcription factors, including FAST1, c-Jun, Spl, TFE3, Mixer, vitamin D receptor, and GATA-3, have been reported to interact with Smads (Miyazawa et al., 2002). By selecting these transcription factors as their interaction partners, Smads may exhibit specific biological effects on target cells.

Cooperative induction of IgA by Runx and TGF--specific R-Smads

Cooperation between TGF- superfamily signaling and Runx proteins was first discovered in the regulation of IgA synthesis in B lymphocytes. TGF- inhibits growth and function of B cells, T cells, and natural killer cells. However, TGF- has been shown to act on B cells and to direct class switching to IgA (Coffman et al., 1989; Sonoda et al., 1989), and a TGF- responsive element (TRE) was identified in the promoter regions of the germline immunoglobulin constant region (Ig C) in mouse and human (Lin and Stavnezer, 1992; Nilsson and Sideras, 1993). Interestingly, the TRE in mouse contains two Smad binding sites (CAGAC motif; Dennler et al., 1998) and three Runx binding sites (Melnikova et al., 1993; Bae et al., 1994) in close vicinity. Runx proteins physically interact with Smad2/3 acting on the TGF- signaling pathway; thus, they cooperatively stimulate the synthesis of IgA through directly inducing the transcription of Ig C (Hanai et al., 1999; Pardali et al., 2000). Although all three Runx proteins interact with Smads, mRNA for Runx3 is predominantly induced by TGF-1 in B lymphocytes (Shi and Stavnezer, 1998), suggesting that Runx3 plays a central role in IgA synthesis.

Both Smad binding sites and Runx binding sites are essential for efficient activation of the Ig C promoter. Mutations of either of these binding sites severely impair transcriptional activation by TGF- and Runx. Smads interact with Runx proteins through multiple domains; both the N-terminal MH1 and C-terminal MH2 domains in Smads are responsible for such interaction, and the C-terminal transactivation domain as well as a region flanking the Runt domain of Runx are involved in their interaction (Hanai et al., 1999; Pardali et al., 2000). Runx–Smad complexes appear to be formed in the cytoplasm without ligand stimulation, but are detected in the nucleus only after TGF- stimulation.

In addition to Smad and Runx binding sites, an Ets binding site and an ATF/CRE site are located in the vicinity of TRE, and transcription factors of the Ets family (Xie et al., 1999) and CREB (Zhang and Derynck, 2000) cooperatively regulate the transcription of Ig C together with Runx and Smads.

Synergy between BMP and Runx in osteoblastic differentiation

Runx2 is an essential transcription factor required for bone formation. CCD is characterized by patent fontanelles and sutures, hypoplastic clavicles, and certain other skeletal abnormalities, and is induced by heterozygous mutation of the Runx2 gene (Mundlos et al., 1997). Targeted disruption of Runx2 in mice results in the absence of mature osteoblasts and lack of bone formation (Komori et al., 1997; Otto et al., 1997). In fact, cooperation between Runx and Smads has been most clearly documented in bone formation induced by BMPs.

BMP-specific Smads, including Smadl and Smad5, physically interact with Runx2, and synergistically induce the osoteoblast-like phenotype in C2C12 mesenchymal progenitor cells (Lee et al., 2000; Zhang et al., 2000). Interaction of endogenous Smadl and Runx2 was observed only upon BMP stimulation in C2C12 cells. Moreover, a truncation of the C-terminal region of Runx2 found in a CCD patient resulted in loss of interaction with BMP-specific Smads, and dominant interference of differentiation of C2C12 cells into osteoblasts upon BMP stimulation (Zhang et al., 2000). Runx proteins have nuclear matrix-targeting signals (NMTS) in their C-terminal regions. Activated Smad proteins are recruited to specific foci in the nucleus through Runx proteins, and recruitment of Smad-Runx complexes to specific subnuclear foci is essential for transcriptional regulation of their target genes (Zaidi et al., 2002). These findings indicate a critical role of cooperation of Runx2 and BMP signaling in bone formation.

Expression levels of Runx2 in mesenchymal progenitor cells appear to be low. BMPs induce osteoblastic differentiation together with Runx2 through two steps: first, BMPs act on mesenchymal progenitor cells and induce expression of Runx2. Second, BMP-specific R-Smads interact with the induced Runx2 protein and further induce differentiation of cells into osteoblastic cells (Figure 1). Although Runx2 is induced in mesenchymal cells, it may not be a direct target of BMP signaling (Lee et al., 2000). BMPs induce expression of Dlx5, which is specifically expressed in osteogenic cells (Miyama et al., 1999; Lee et al., 2003). Dlx5 then induces the expression of Runx2 in osteoprogenitor cells. In contrast to BMPs, TGF- does not induce expression of Dlx5.

BMPs prevent differentiation of mesenchymal cells into myoblasts and Runx2 appears to play a pivotal role in this process. Runx2 prevents differentiation of C2C12 cells into myoblasts by inducing matrix gene products, suppressing the expression of MyoD and inhibiting myotube formation. However, Runx2 alone is insufficient for induction of osteoblastic differentiation; for this process, Runx2 needs to interact with BMP-specific R-Smads. It is currently not known how BMP-specific R-Smads regulate osteoblastic differentiation together with Runx2. It should be noted that osteoblast differentiation is not induced by TGF--specific R-Smads. Since BMP-specific R-Smads bind to DNA sequences different from the typical Smad-binding sequences (Ishida et al., 2000), they may bind to such sequences in promoters of bone-specific genes together with Runx2. Another important possibility is that Dlx5 induced by BMPs may support the action of BMP-specific R-Smads and Runx2.

In contrast to BMPs, TGF- negatively regulates bone formation in most instances, although the effects of TGF- on osteoblast differentiation depend on the status of differentiation of cells and the extracellular milieu. Although there are some controversial data (Lee et al., 1999, 2000), TGF- has been reported to inhibit the expression of Runx2 in some osteoprogenitor cells, and this occurs independent of new protein synthesis but requires Runx2 itself (Alliston et al., 2001). Smad3 activated by TGF- physically interacts with, and represses the transcriptional activity of, Runx2. This results in inhibition of transcription of the Runx2 gene itself and some osteoblast-specific genes, including osteocalcin. Interestingly, Smad3 does not directly bind to the osteocalcin promoter, but binds indirectly to it through Runx2. Thus, the Smad3–Runx2 complex induced by TGF- plays a critical role in repression of osteoblast differentiation in certain cells.

Degradation of Runx2 by Smurf1: a novel link between the Runx and Smad pathways

Ubiquitin-dependent protein degradation plays pivotal roles in various biological processes (Hershko and Ciechanover, 1998), including TGF- superfamily signaling. In the ubiquitin–proteasome pathway, E3 ubiquitin ligases play key roles in the recognition of target proteins and their degradation by the 26S proteasomes. Smad ubiquitin regulatory factor (Smurf) 1 was originally identified as a HECT type E3 ubiquitin ligase. Smurf1 induces the ubiquitination and degradation of BMP-specific R-Smads, Smadl, and Smad5, in a ligand-independent manner (Figure 2) (Zhu et al., 1999). Smurf1 has WW domains, which specifically interact with PY motif in R-Smads. Smurf2 is structurally very similar to Smurf1, and also degrades Smad1 (Zhang et al., 2001). Moreover, Smurf2 was shown to interact with activated TGF--specific R-Smad, Smad2, and to induce its ubiquitination and degradation (Lin et al., 2000). In addition to their effects on R-Smads, Smurf1, and Smurf2 interact with I-Smads through their PY motifs in the nucleus and induce their export to the cytoplasm. Smurf-I–Smad complexes are then recruited to the type I receptors for TGF- and BMPs, and enhance their degradation (Figure 2) (Kavsak et al., 2000; Ebisawa et al., 2001; Murakami et al., 2003). Thus, Smurfs negatively regulate TGF- superfamily signaling by targeting R-Smads as well as type I receptors bound to I-Smads for ubiquitin-dependent degradation.

Figure 2.

Degradation of Smads and Runx2 by Smurf proteins. Smurf proteins induce degradation of BMP-specific R-Smads. Through complex formation with I-Smads, they induce degradation of type I receptors and R-Smads. In addition, Smurf1 interacts with Runx2 and induces its degradation

Full figure and legend (126K)

Intriguingly, Smurf1 interacts directly with Runx2, and induces its ubiquitin-dependent degradation (Figure 2) (Zhao et al., 2003). Although the domain responsible for interaction between Smurf1 and Runx2 has yet to be determined, Runx2 has a PY motif in the C-terminal region, which may be important for interaction with Smurf1. Smurf1 reduced the alkaline phosphatase activity and production of osteocalcin in osteoprogenitor cells, whereas a Smurf1 mutant lacking the catalytic activity of E3 ligase induced osoteoblastic differentiation. It is currently unknown whether degradation of Runx2 by Smurf1 depends on BMP signaling. However, since Smurfs interact wjth R-Smads directly, Runx proteins may be efficiently degraded by Smurfs as Runx-R–Smad complexes upon BMP signaling. PEBP2 has been shown to bind to the Runt domain and to prevent Runx proteins from degradation (Huang et al., 2001). It will thus be interesting to determine how PEBP2 regulates degradation of Runx2 by Smurf1.

Regulation of expression of Smurf1 and Smurf2 has not been extensively studied yet; however, a recent study showed that Smurf2 is expressed in patients with esophageal squamous carcinoma, and that high expression of Smurf2 correlates with poor prognosis of esophageal cancer (Fukuchi et al., 2002). It will be interesting to examine whether Smurfs can regulate degradation of the other Runx family proteins, and whether expression of Smurfs is also upregulated in other tumors in which functions of Runx proteins are involved (see below).

Cross-talk between STAT and Smad pathways though Runx

The Smad signaling pathway has been shown to cross-talk with various signaling pathways, including ERK MAP kinase pathways and Wnt--catenin pathways (Derynck et al., 2001). Cross-talk with cytokine signaling has also been reported; interferon- (IFN-) induces expression of Smad7 on some cells through the Jak-STAT signaling pathway, and Smad7 then inhibits TGF- signaling pathways (Ulloa et al., 1999) (Figure 3a). In contrast, leukemia inhibitory factor (LIF) and BMP pathways synergistically induce differentiation of neuroepithelial cells into astrocytes through indirect binding of STAT3 and BMP-specific Smads via p300 (Nakashima et al., 1999). Thus, Jak-STAT signaling pathways regulate Smad signaling pathways both positively and negatively.

Figure 3.

Cross-talk between Smads and STAT. (a) IFN- induces expression of Smad7, which in turn inhibits TGF- signaling. (b) Latent STAT1 interacts with Runx2, and prevents nuclear localization of Runx2. STAT1 thus inhibits osteoblast differentiation induced by BMP-specific R-Smads and Runx2

Full figure and legend (147K)

Recently, another important interaction between Smad and STAT signaling pathways has been reported (Kim et al., 2003). Type I IFN (IFN-/) transduces signals through STAT1, STAT2, and IRF-9, whereas type II IFN (IFN-) does so through STAT1 homodimer. Both type I and type II IFNs function as negative regulators of osteoclastogenesis (Takayanagi et al., 2000, 2002). However, STAT1-null mice exhibit increased bone mass despite excessive osteoclastogenesis, suggesting that STAT1 inhibits the process of bone formation in vivo. Kim et al. (2003) demonstrated that STAT1 physically interacts with Runx2 in the cytoplasm, and inhibits nuclear localization of Runx2 (Figure 3b). Phosphorylation of tyrosine-701 of STAT1 is not required for the interaction with Runx2, indicating that the interaction between STAT1 and Runx2 occurs independent of IFN signaling. In STAT1^-/- cells, enhanced response of osteoblasts to BMP-2 was observed, whereas some effects of BMP that do not require Runx2 appeared to be unaffected. STAT1 thus functions as a cytoplasmic attenuator of Runx2. These findings suggest the interesting possibility that STAT1 regulates the function of other Runx proteins, and that TGF- superfamily signaling is indirectly regulated by STAT1 through the Runx proteins.

Runx3 and TGF-/Smad signaling in gastric mucosa

Runx3-null mice exhibit several interesting phenotypes including defects in the development of proprioceptive dorsal root ganglion neurons (Inoue et al., 2002; Levanon et al., 2002) and of cytotoxic lineage thymocytes (Taniuchi et al., 2002; Woolf et al., 2003). Interestingly, hyperplasia of gastric mucosa was observed in Runx3-null mice and the gastric epithelial cells of these mice were refractory to the growth inhibition and apoptosis induced by TGF- (Li et al., 2002). Moreover, Li et al. (2002) reported that approximately half of human gastric cancer cells do not express Runx3 due to hemizygous deletion and DNA methylation of the Runx3 promoter.

TGF- inhibits growth of epithelial cells through suppression of c-myc and CDC-25A as well as through induction of Cdk inhibitors p21 and p15 (Massague et al., 2000). Various genes including DAP-kinase (Jang et al., 2002), Daxx (Perlman et al., 2001), and SHIP (Valderrama-Carvajal et al., 2002) have been reported to be involved in induction of apoptosis by TGF-. Since Runx3 plays a critical role in the development of gastric cancer, and perturbations of Runx3 function may affect TGF- signaling, it will be interesting to determine whether these genes involved in cell cycle regulation or apoptosis are regulated in gastric epithelium by cooperative activity of Smads and Runx proteins.

Perspectives

Smads have been shown to interact with many different transcription factors and transcriptional coactivators/corepressors. However, most of these proteins preferentially interact with TGF--specific R-Smads, and only several proteins bind to BMP-specific R-Smads (Miyazawa et al., 2002). Thus, Runx proteins are unique, since they interact with both TGF--specific R-Smads and BMP-specific R-Smads. Cooperation between BMP-specific R-Smads and Runx2 has recently been intensively studied, and some regulatory molecules, including Smurf1 and STAT1, have been identified. In contrast, cooperation between TGF--specific R-Smads and Runx proteins have been studied mostly in the induction of IgA synthesis, and relatively little in other biological responses. Since Runxl is involved in definitive hematopoiesis, and mutants of it have been reported to be involved in human acute leukemias, it will be interesting to examine whether Runxl interacts with TGF--specific R-Smads and regulates growth and differentiation of hematopoietic cells. In addition, Runx3 has been shown to play critical roles in the development of neuronal cells, thymocytes, and gastric mucosa. It will be particulaly interesting to examine whether growth and apoptosis of these cells are regulated by R-Smads and Runx3.

References

1. Alliston T, Choy L, Ducy P, Karsenty G and Derynck R. (2001). EMBO J., 20, 2254–2272. | Article | PubMed | ISI | ChemPort |
2. Bae SC, Ogawa E, Maruyama M, Oka H, Satake M, Shigesada K, Jenkins NA, Gilbert DJ, Copeland NG and Ito Y. (1994). Mol. Cell. Biol., 14, 3242–3252. | PubMed | ISI | ChemPort |
3. Blobe GC, Schiemann WP and Lodish HF. (2000). N. Engl. J. Med., 342, 1350–1358. | Article | PubMed | ISI | ChemPort |
4. Coffman RL, Lebman DA and Shrader B. (1989). J. Exp. Med., 170, 1039–1044. | Article | PubMed | ISI | ChemPort |
5. Dennler S, Itoh S, Vivien D, ten Dijke P, Huet S and Gauthier JM. (1998). EMBO J., 17, 3091–3100. | Article | PubMed | ISI | ChemPort |
6. Derynck R, Akhurst RJ and Balmain A. (2001). Nat. Genet., 29, 117–129. | Article | PubMed | ISI | ChemPort |
7. Ebisawa T, Fukuchi M, Murakami G, Chiba T, Tanaka K, Imamura T and Miyazono K. (2001). J. Biol. Chem., 276, 12477–12480. | Article | PubMed | ISI | ChemPort |
8. Fukuchi M, Fukai Y, Masuda N, Miyazaki T, Nakajima M, Sohda M, Manda R, Tsukada K, Kato H and Kuwano H. (2002). Cancer Res., 62, 7162–7165. | PubMed | ISI | ChemPort |
9. Hanai J-i, Chen LF, Kanno T, Fujita N, Kim WY, Imamura T, Ishidou Y, Fukuchi M, Shi M-J, Stavnezer J, Kawabata M, Miyazono K and Ito Y. (1999). J. Biol Chem., 274, 31577–31582. | Article | PubMed | ISI | ChemPort |
10. Heldin C-H, Miyazono K and ten Dijke P. (1997). Nature, 390, 465–471. | Article | PubMed | ISI | ChemPort |
11. Hershko A and Ciechanover A. (1998). Annu. Rev. Biochem., 67, 425–479. | Article | PubMed | ISI | ChemPort |
12. Huang G, Shigesada K, Ito K, Wee HJ, Yokomizo T and Ito Y. (2001). EMBO J., 20, 723–733. | Article | PubMed | ISI | ChemPort |
13. Inoue K, Ozaki S, Shiga T, Ito K, Masuda T, Okado N, Iseda T, Kawaguchi S, Ogawa M, Bae SC, Yamashita N, Itohara S, Kudo N and Ito Y. (2002). Nat. Neurosci., 5, 946–954. | Article | PubMed | ISI | ChemPort |
14. Ishida W, Hamamoto T, Kusanagi K, Yagi K, Kawabata M, Takehara K, Sampath TK, Kato M and Miyazono K. (2000). J. Biol. Chem., 275, 6075–6079. | Article | PubMed | ISI | ChemPort |
15. Ito Y. (1999). Genes Cells, 4, 685–696. | Article | PubMed | ISI | ChemPort |
16. Ito Y and Miyazono K. (2003). Curr. Opin. Genet. Dev., 13, 43–47. | Article | PubMed | ISI | ChemPort |
17. Jang CW, Chen CH, Chen CC, Chen JY, Su YH and Chen RH. (2002). Nat. Cell Biol., 4, 51–58. | Article | PubMed | ISI | ChemPort |
18. Kavsak P, Rasmussen RK, Causing GG, Bonni S, Zhu H, Thomsen GH and Wrana JL. (2000). Mol. Cell, 6, 1365–1375. | Article | PubMed | ISI | ChemPort |
19. Kim S, Koga T, Isobe M, Kern BE, Yokochi T, Chin YE, Karsenty G, Taniguchi T and Takayanagi H. (2003). Genes Dev., 17, 1979–1991. | Article | PubMed | ChemPort |
20. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao YH, Inada M, Sato M, Okamoto R, Kitamura Y, Yoshiki S and Kishimoto P. (1997). Cell, 89, 755–764. | Article | PubMed | ISI | ChemPort |
21. Lee KS, Kim HJ, Li QL, Chi XZ, Ueta C, Komori T, Wozney JM, Kim EG, Choi JY, Ryoo HM and Bae SC. (2000). Mol. Cell. Biol., 20, 8783–8792. | Article | PubMed | ISI | ChemPort |
22. Lee MH, Javed A, Kim HJ, Shin HI, Gutierrez S, Choi JY, Rosen V, Stein JL, van Wijnen AJ, Stein GS, Lian JB and Ryoo HM. (1999). J Cell Biochem., 73, 114–125. | Article | PubMed | ISI | ChemPort |
23. Lee M-H, Kim Y-J, Kim H-J, Park H-D, Kang A-R, Kyung H-M, Sung J-H, Wozney JM, Kim H-J and Ryoo H-M. (2003). J. Biol. Chem., 278, 34387–34394. | PubMed |
24. Levanon D, Bettoun D, Harris-Cerruti C, Woolf E, Negreanu V, Eilam R, Bernstein Y, Goldenberg D, Xiao C, Fliegauf M, Kremer E, Otto F, Brenner O, Lev-Tov A and Groner Y. (2002). EMBO J., 21, 3454–3463. | Article | PubMed | ISI | ChemPort |
25. Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ, Lee KY, Nomura S, Lee CW, Han SB, Kim HM, Kim WJ, Yamamoto H, Yamashita N, Yano T, Ikeda T, Itohara S, Inazawa J, Abe T, Hagiwara A, Yamagishi H, Ooe A, Kaneda A, Sugimura T, Ushijima T, Bae SC and Ito Y. (2002). Cell, 109, 113–124. | Article | PubMed | ISI | ChemPort |
26. Lin X, Liang M and Feng XH. (2000). J. Biol. Chem., 275, 36818–36822. | Article | PubMed | ISI | ChemPort |
27. Lin YC and Stavnezer J. (1992). J. Immunol., 149, 2914–2925. | PubMed | ISI | ChemPort |
28. Massague J, Blain SW and Lo RS. (2000). Cell, 103, 295–309. | Article | PubMed | ISI | ChemPort |
29. Melnikova IN, Crute BE, Wang S and Speck NA. (1993). J. Virol., 67, 2408–2411. | PubMed | ISI | ChemPort |
30. Miyama K, Yamada G, Yamamoto TS, Takagi C, Miyado K, Sakai M, Ueno N and Shibuya H. (1999). Dev. Biol., 208, 123–133. | PubMed |
31. Miyazawa K, Shinozaki M, Hara T, Furuya T and Miyazono K. (2002). Genes Cells, 7, 1191–1204. | Article | PubMed | ISI | ChemPort |
32. Mundlos S, Otto F, Mundlos C, Mulliken JB, Aylsworth AS, Albright S, Lindhout D, Cole WG, Henn W, Knoll JH, Owen MJ, Mertelsmann R, Zabel BU and Olsen BR. (1997). Cell, 89, 773–779. | Article | PubMed | ISI | ChemPort |
33. Murakami G, Watabe T, Takaoka K, Miyazono K and Imamura T. (2003). Mol. Biol. Cell, 14, 2809–2817. | Article | PubMed | ISI | ChemPort |
34. Nakashima K, Yanagisawa M, Arakawa H, Kimura N, Hisatsune T, Kawabata M, Miyazono K and Taga T. (1999). Science, 284, 479–482. | Article | PubMed | ISI | ChemPort |
35. Nilsson L and Sideras P. (1993). Int. Immunol., 5, 271–282. | PubMed |
36. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, Rosewell IR, Stamp GW, Beddington RS, Mundlos S, Olsen BR, Selby PB and Owen MJ. (1997). Cell, 89, 765–771. | Article | PubMed | ISI | ChemPort |
37. Pardali E, Xie XQ, Tsapogas P, Itoh S, Arvanitidis K, Heldin C-H, ten Dijke P, Grundstrom T and Sideras P. (2000). J. Biol. Chem., 275, 3552–3560. | Article | PubMed | ISI | ChemPort |
38. Perlman R, Schiemann WP, Brooks MW, Lodish HF and Weinberg RA. (2001). Nat Cell Biol., 3, 708–714. | Article | PubMed | ISI | ChemPort |
39. Qin BY, Lam SS, Correia JJ and Lin K. (2002). Genes Dev., 16, 1950–1963. | Article | PubMed | ISI | ChemPort |
40. Shi Y and Massague J. (2003). Cell, 113, 685–700. | Article | PubMed | ISI | ChemPort |
41. Shi MJ and Stavnezer J. (1998). J. Immunol., 161, 6751–6760. | PubMed | ISI | ChemPort |
42. Sonoda E, Matsumoto R, Hitoshi Y, Ishii T, Sugimoto M, Araki S, Tominaga A, Yamaguchi N and Takatsu K. (1989). J. Exp. Med., 170, 1415–1420. | Article | PubMed | ISI | ChemPort |
43. Takayanagi H, Kim S, Matsuo K, Suzuki H, Suzuki T, Sato K, Yokochi T, Oda H, Nakamura K, Ida N, Wagner EF and Taniguchi T. (2002). Nature, 416, 744–749. | Article | PubMed | ISI | ChemPort |
44. Takayanagi H, Ogasawara K, Hida S, Chiba T, Murata S, Sato K, Takaoka A, Yokochi T, Oda H, Tanaka K, Nakamura K and Taniguchi T. (2000). Nature, 408, 600–605. | Article | PubMed | ISI | ChemPort |
45. Taniuchi I, Osato M, Egawa T, Sunshine MJ, Bae SC, Komori T, Ito Y and Littman DR. (2002). Cell, 111, 621–633. | Article | PubMed | ISI | ChemPort |
46. Ulloa L, Doody J and Massague J. (1999). Nature, 397, 710–713. | Article | PubMed | ISI | ChemPort |
47. Valderrama-Carvajal H, Cocolakis E, Lacerte A, Lee EH, Krystal G, Ali S and Lebrun JJ. (2002). Nat. Cell Biol., 4, 963–969. | Article | PubMed | ISI | ChemPort |
48. Wakefield LM and Roberts AB. (2002). Curr. Opin. Genet. Dev., 12, 22–29. | Article | PubMed | ISI | ChemPort |
49. Woolf E, Xiao C, Fainaru O, Lotem J, Rosen D, Negreanu V, Bernstein Y, Goldenberg D, Brenner O, Berke G, Levanon D and Groner Y. (2003). Proc. Natl. Acad. Sci.USA, 100, 7731–7736.
50. Xie XQ, Pardali E, Holm M, Sideras P and Grundstrom T. (1999). Eur. J. Immunol., 29, 488–498. | Article | PubMed | ISI | ChemPort |
51. Zaidi SK, Sullivan AJ, van Wijnen AJ, Stein JL, Stein GS and Lian JB. (2002). Proc. Natl Acad. Sci. USA, 99, 8048–8053. | Article | PubMed | ChemPort |
52. Zhang Y and Derynck R. (2000). J. Biol. Chem., 275, 16979–16985. | Article | PubMed | ISI | ChemPort |
53. Zhang Y-W, Yasui N, Huang G, Fujii M, Hanai J-i, Nogami H, Ochi T, Miyazono K and Ito Y. (2000). Proc. Natl Acad. Sci. USA, 97, 10549–10554. | Article | PubMed | ChemPort |
54. Zhang Y, Chang C, Gehling DJ, Hemmati-Brivanlou A and Derynck R. (2001). Proc. Natl. Acad. Sci. USA, 98, 974–979. | Article | PubMed | ChemPort |
55. Zhao M, Qiao M, Oyajobi BO, Mundy GR and Chen D. (2003). J. Biol. Chem., 278, 27939–27944. | Article | PubMed | ISI | ChemPort |
56. Zhu H, Kavsak P, Abdollah S, Wrana JL and Thomsen GH. (1999). Nature, 400, 687–693. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We thank Drs Tadatsugu Taniguchi and Hiroshi Takayanagi for sharing unpublished observations. We also thank Keiji Miyazawa, Hiromi Fukuda, and Masao Saitoh for the valuable discussion. This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, Technology, Sports and Culture of Japan.