Regulation of the Tgfb Signalling Pathway by Ubiquitin-Mediated Degradation

Regulation of the TGFb signalling pathway by ubiquitin-mediated degradation

Luisa Izzi2 and Liliana Attisano*,1,2

1Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada M5S 1A8; and 2Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada M5S 1A8

The transforming growth factor-b (TGFb) superfamily disruption or mutations in TGFb pathway components controls a plethora of biological responses, and alterations are associated with several human diseases, including in its signalling pathway are associated with a range of cancer (de Caestecker et al., 2000;Derynck et al., 2001). human diseases, including cancer. TGFb superfamily Approximately 40 distinct TGFb members have been ligands signal through a heteromeric complex of Ser/ identified in vertebrates and they are generally divided Thr kinase receptors that propagate the signal to the into two branches, the TGFb/activin/nodals and the Smad family of intracellular proteins. The ubiquitin- bone morphogenetic proteins (BMPs), each of which mediated proteasomal degradation pathway is an evolu- can activate one of the two intracellular signalling tionary conserved cascade that tightly regulates TGFb cascades (de Caestecker et al., 2000;Moustakas et al., superfamily signalling. Both the size of the Smad pool in 2001;Attisano and Wrana, 2002;Shi and Massague, unstimulated cells and Smad protein levels subsequent to 2003). Ligands initiate signalling by bringing together a the activation of the pathway are controlled by ubiquitina- complex of serine/threonine kinase receptors, consisting tion. E3 ligases are components of the ubiquitin-degrada- of the type I and type II class. Ligand binding triggers tion complex that specifically recognize targeted proteins the phosphorylation of the type I receptor by the type II and the E3 ligases, Smad ubiquitination-related factor 1 receptor kinase, which results in the activation of the (Smurf1), Smurf2 and SCF/Roc1 have been implicated in type I kinase (Figure 1). The signal is then transmitted Smad degradation. The Smurfs are of particular im- from the type I receptor to the Smads, the intracellular portance to TGFb signalling, as Smads also function as effectors of the TGFb and BMP signalling pathways adapters that recruit the Smurfs to various pathway (Moustakas et al., 2001;Attisano and Wrana, 2002;Shi components including the TGFb receptor complex and the and Massague, 2003). The type I receptor kinase directly transcriptional repressor, SnoN, and thereby regulate the phosphorylates two serine residues at the carboxy- degradation of these Smad-associating proteins. Thus, by terminus of the receptor-regulated Smads or R-Smads, controlling the level of Smads as well as positive and Smad1, -2, -3, -5 and -8. Smad1, -5 and -8 are negative regulators of the pathway, Smurfs provide for phosphorylated by BMP type I receptors, while Smad2 complex and fine control of signalling output. Finally, and -3 are activated by the TGFb and activin type I growing evidence demonstrates that ubiquitination and receptors (Moustakas et al., 2001;Attisano and Wrana, proteasomal degradation is also implicated in the turnover 2002;Shi and Massague, 2003). Phosphorylation of of tumor-derived Smad mutants and may thus contribute the R-Smads permits their association with the common to disease progression. Smad or co-Smad, Smad4. Once formed, this R-Smad/ Oncogene (2004) 23, 2071–2078. doi:10.1038/sj.onc.1207412 Smad4 complex translocates to the nucleus and interacts with DNA binding proteins to regulate transcriptional Keywords: smad;smurf;ubiquitination responses (Figure 1). Smad partners include the DNA binding factors FoxH1, Mixer, c-jun, Lef-1/TCF, OAZ, as well as many others all of which are required for the regulation of specific target genes (Attisano and TGFb signalling pathway Wrana, 2000;ten Dijke et al., 2000;Moustakas et al., 2001;Shi and Massague, 2003). Smads can positively The transforming growth factor-b (TGFb) superfamily regulate gene expression by recruiting coactivators of cytokines control a plethora of biological responses such as CBP/p300 or negatively by direct recruitment and have been shown to play a prominent role during of histone deacetylases (HDACs) or corepressors, such embryonic development, regulation of cell growth as c-ski and SnoN, which themselves associate and differentiation and apoptosis (Massague´ , 1998; with HDACs (ten Dijke et al., 2000;Moustakas et al., Whitman, 1998;Derynck et al., 2001). In addition, 2001;Attisano and Wrana, 2002;Shi and Massague, 2003). In addition to R-Smads and co-Smads, the *Correspondence: L Attisano, Department of Biochemistry, Medical I-Smads or inhibitory Smads (Smad6 and -7) are crucial Sciences Building, Room 6336, University of Toronto, Toronto, for appropriate regulation of TGFb and BMP signalling Ontario, Canada M5S 1A8;E-mail: [email protected] pathways. Smad6 and -7 block TGFb and BMP Regulation of the TGFb signalling L Izzi and L Attisano 2072 targets Smads to early endosomes (Tsukazaki et al., 1998). Both the MH1 and MH2 domains mediate interactions with transcription factors, and transcriptional coactivators or corepressors (Attisano and Wrana, 2000;Wotton and Massague´ , 2000;Moustakas et al., 2001). The linker region contains a number of phosphorylation sites crucial for crosstalk with other signalling pathways and a PY motif, which specifically interact with E3 ubiquitin ligases implicated in the regulation of the TGFb and BMP pathways (Attisano and Wrana, 2002). Although abundant evidence demonstrates that Smads are critical for TGFb family signalling, accumulating data suggests that Smad-independent pathways also exist (de Caestecker et al., 2000;Mulder, 2000). For instance, TGFb has been reported to activate mitogen- activated protein kinases including the extracellular- regulated kinases, c-Jun N-terminal kinases and p38 kinases. There is also evidence for TGFb-dependent activation of PKB/Akt through PI3 kinase, and a RhoA-dependent signalling pathway has been suggested to function in TGFb-mediated epithelial–mesenchymal transition (Bakin et al., 2000;Bhowmick et al., 2001). As the molecular details of these signalling pathways are Figure 1 The TGFb superfamily signalling pathway. Binding of not well understood, they will not be considered further TGFb superfamily ligands triggers the formation of a heteromeric in this review. complex of serine/threonine kinase receptors. The type II receptor kinase then transphosphorylates and activates the type I receptor, which in turn phosphorylates and activates R-Smads. Activated R- Signalling-independent regulation of R-Smad protein Smads associate with Co-Smad, Smad4 and translocate to the levels by ubiquitination nucleus where they bind to distinct DNA binding protein and regulate the transcription of specific target genes Smads are critical intracellular mediators of TGFb signalling pathways, and thus alterations in Smad protein levels can profoundly effect signalling. Consis- tent with this, the ubiquitin–proteasomal pathway signalling by competing with R-Smads for association has been shown to regulate the basal level of Smads with type I receptors or by targeting receptors (Figure 2). That the level of the Smad pool is regulated for ubiquitin-mediated degradation (Moustakas et al., in a signalling-independent manner was first described 2001;Attisano and Wrana, 2002;Shi and Massague, for the R-Smad, Smad1. A yeast two-hybrid screen 2003). using Smad1 as bait identified Smad ubiquitination- Smads comprise two conserved globular domains, related factor 1 (Smurf1) as an interacting protein the N-terminal MH1 domain and C-terminal MH2 (Zhu et al., 1999). Smurf1 is an E3 ubiquitin ligase of the domain, which are separated by a proline-rich linker C2-WW-HECT domain class and contains an region (de Caestecker et al., 2000;Moustakas et al., N-terminal C2 domain, two WW domains and a 2001;Attisano and Wrana, 2002;Shi and Massague, carboxy-terminal HECT domain that catalyses the 2003). While the MH1 domain of R-Smads and Smad4 transfer of the ubiquitin moiety to its target substrates are highly conserved, the amino-terminal region of the I- (Figure 3). The WW domains of Smurf1 mediate Smads only shows a weak sequence similarity with the a specific interaction with the PY motif of the MH1 domain of other Smads. The MH2 domain BMP-regulated Smad1 and -5 (Zhu et al., 1999), thereby is highly conserved among all Smads, whereas the linker allowing Smurf1 to target Smad1 for ubiquitination regions are divergent. Functionally, the MH1 domain and proteasomal degradation. Ectopic expression is implicated in nuclear import, and DNA binding, of Xenopus Smurf1 in combination with Smad1 in whereas the MH2 domain is involved in receptor Xenopus animal caps results in the downregulation of interaction and Smad oligomerization (de Caestecker Smad1-induced genes and the development of a et al., 2000;Moustakas et al., 2001;Attisano secondary axis (Zhu et al., 1999). Thus, Smurf1 appears and Wrana, 2002;Shi and Massague, 2003). The MH2 to regulate BMP signalling by targeting non- domain is also important for the cytoplasmic anchoring activated Smad1 and -5 for protein degradation, of Smads. The Smad anchor for receptor activation thereby preventing spurious activation of the pathway. or SARA interacts with the MH2 of Smad2 and -3 This may be particularly important in a developmental and targets the R-Smads to the cell surface (Tsukazaki context as it allows cells to be competent to respond to et al., 1998). Moreover, SARA contains an FYVE BMP stimulation only when appropriate (Zhu et al., domain, which mediates binding to phospholipids and 1999).

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2073 unactivated Smad1, -2 and -3 but not with the co-Smad, Smad4, which lacks a PY motif (Lin et al., 2000; Zhang et al., 2001). Moreover, a decrease in the steady- state level of Smad1 and to a lesser degree Smad2 has been observed upon expression of Smurf2, suggesting that the R-Smad pool might be regulated by Smurf2. However, the association of Smad2 with Smurf2 is dramatically enhanced upon activation of the TGFb signalling pathway (Lin et al., 2000;Bonni et al., 2001), suggesting that Smurf2 may be of particular importance for ligand-dependent regulation of the pathway (see below).

Signalling-dependent regulation of R-Smads by ubiquitination In addition to regulating steady-state levels of R-Smads, the ubiquitin–proteasome pathway is involved in the degradation of activated R-Smads (Figure 3). In human keratinocytes, endogenous Smad2 is targeted for ubiquitination and proteasomal degradation in response to TGFb stimulation (Lo and Massague´ , 1999). Destruction of activated Smad2 was shown to occur in Figure 2 Ubiquitin-dependent degradation of Smads. The ubi- the nucleus and was mediated by the E2-conjugating quitin–proteasome pathway regulates both the basal level of Smads as well the turnover of Smads upon the activation of the signalling enzymes, UbcH5b/c and Ubc3 (Lo and Massague´ , pathway. Smad degradation is mediated at least in part by E3 1999). As Smurf2 can associate with activated Smad2 ligases including Smurf1, Smurf2, SCF/Roc1 (Lin et al., 2000;Bonni et al., 2001), Smurf2 is a candidate E3 ligase for activated Smad2 degradation. However, the specific E3 ligase localized to the nucleus that catalyses the ligation of the ubiquitin moieties onto activated Smad2 has not yet been defined. Similar to Smad2, activated Smad3 is also targeted for proteasomal destruction. However, in this case, SCF/ Roc1 E3 ligase complex has been shown to be responsible for triggering the ubiquitination of phosphorylated Smad3 (Fukuchi et al., 2001). Once bound to and ubiquitinated by the SCF/Roc1 complex, nuclear Smad3 is exported from the nucleus and undergoes proteasomal degradation in the cytoplasm. Interest- ingly, recruitment of the transcriptional coactivator Figure 3 Structural features of smurfs. A schematic representa- p300 to activated Smad3 enhances the association tion of human Smurf1 (hSmurf1), human Smurf2 (hSmurf2) and between Smad3 and Roc1, suggesting that ubiquitin- Drosophila Smurf (Dsmurf) is shown. Amino-acid residues dependent degradation of Smad3 may be necessary delineating distinct domains as defined by the ProSite database to terminate Smad3 transcriptional activity (Fukuchi (http://ca.expasy.org) are indicated. All Smurfs contain an N-terminal C2 domain and a C-terminal HECT domain, and et al., 2001). The degradation of activated Smad1 has either two or three WW domains also been reported through a complex containing Smad1, the ornithine decarboxylase antizyme (Az) and the 20S proteasome b subunit, HsN3, which then targets Smad1 for proteasomal degradation (Gruendler et al., A second Smurf1-related E3 HECT domain ligase, 2001). termed Smurf2, has also been described (Kavsak et al., 2000;Lin et al., 2000;Zhang et al., 2001). Alignment Ubiquitination of the common Smad, Smad4 of the amino-acid sequences reveals that Smurf2 exhibits an overall sequence identity of 80% with Smurf1. Although the mechanism of Smad4 ubiquitination is not Smurf2 features an N-terminal C2 domain, WW known, proteasomal degradation of the common Smad domains and a carboxy-terminal HECT ligase domain. has also been described. For instance, ectopic expression Interestingly, while Smurf1 contains two WW domains, of Jab1 (Jun-activating domain binding protein 1), Smurf2 comprises three WW domains, resulting which also induces degradation of p27 and p53, from a 31 amino-acid insertion downstream of the C2 mediated ubiquitination and degradation of Smad4 domain (Figure 2). Overexpression and in vitro (Wan et al., 2002). Further, the overexpression of studies have indicated that Smurf2 can associate with oncogenic Ras has been shown to decrease Smad4

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2074 protein levels, an effect that was partially reversed to recruit HDACs to block transcription (ten Dijke et al., by an inhibitor of the ubiquitin–proteasome pathway 2000;Moustakas et al., 2001;Attisano and Wrana, (Saha et al., 2001). In addition to regulating the 2002;Shi and Massague, 2003). TGF b stimulation proteasomal degradation of Smad4, ubiquitination is has been shown to result in the formation of a Smad2/ also implicated in the modulation of Smad4 function. A Smurf2 complex in which Smad2 then recruits Smurf2 recent report demonstrated that Smad4 is monoubiqui- to the transcriptional corepressor SnoN (Bonni et al., tinated in its MH2 domain at lysine 507 (Moren et al., 2001). Complex formation between Smurf2/Smad2 2003). Unlike polyubiquitination that targets proteins and SnoN then triggers the ubiquitination and degrada- for proteasomal degradation, the role of monoubiqui- tion of SnoN (Bonni et al., 2001). As SnoN is a negative tination is less clear. In the case of Smad4, mutagenesis regulator of TGFb transcriptional responses, its degra- of lysine 507 leads to a weaker affinity for R-Smads and dation may thereby restore TGFb signalling. Smads a lower ability to transactivate target promoters follow- have been shown to function as adaptors for other E3 ing TGFb stimulation (Moren et al., 2003), suggesting ligases that can also function to degrade SnoN. Smad3, that monoubiquitination enhances Smad4 transcrip- and to a lesser extent Smad2, can associate with the tional activity. anaphase-promoting complex (APC) and recruit it to SnoN upon activation of TGFb signalling (Stroschein Sumoylation of Smads et al., 2001;Wan et al., 2001). As with the Smurfs, recruitment of APC to SnoN (Stroschein et al., 2001; Ubiquitination is not the sole modification that appears Wan et al., 2001) results in the ubiquitination and to modulate protein function. Among the ubiquitin- proteasomal degradation of SnoN. The reason why like proteins present in eukaryotic cells, the small Smad2/3 uses two distinct E3 ligase complexes to ubiquitin-related modifier (SUMO) appears to be a mediate SnoN degradation is yet unclear, but it may versatile modifier of a growing number of proteins reflect differences in cell types or perhaps the timing of implicated in various signalling pathways (Desterro SnoN degradation. Smad3 has also been shown to et al., 2000;Melchior, 2000;Muller et al., 2001; promote the ubiquitination and degradation of the Cas Schwartz and Hochstrasser, 2003). Unlike ubiquitina- family member HEF1, which functions as an adaptor tion, sumoylation does not target substrates for protein in a number of signalling pathways (Liu et al., proteasomal degradation. Instead, SUMO conjugation 2000). Whether Smad3 mediates HEF1 degradation by appears to regulate protein–protein interaction, sub- recruiting an E3 ligase is still unclear. nuclear localization, protein–DNA interaction, enzy- matic activity and ubiquitin-dependent degradation I-Smads as adaptors for Smurfs (Melchior, 2000;Muller et al., 2001;Schwartz and Hochstrasser, 2003). Recent studies demonstrate The regulation of cellular processes requires the activa- that the SUMO E3 ligase PIASy, protein inhibitor tion of specific signalling pathways. However, equally of activated STAT (signal transducers and activators of important is the downregulation of the signal. The transcription), interacts and catalyses the sumoylation I-Smads play a pivotal role in this regard as they bind of Smad4 (Lee et al., 2003;Lin et al., 2003a, b; directly to Ser/Thr kinase receptors and thereby block Long et al., 2003). In addition to promoting the nuclear R-Smad access to the receptor (Moustakas et al., 2001; accumulation of Smad4, sumoylation appears to Attisano and Wrana, 2002;Shi and Massague, 2003). enhance Smad4 stability and Smad4-dependent The demonstration that I-Smads bind to the Smurf transcriptional activity (Lee et al., 2003;Lin et al., family of E3 ligases has revealed an additional mechan- 2003a, b). Moreover, PIASy also modulates TGFb ism whereby I-Smads can interfere with TGFb signal- signalling by sumoylating Smad3 (Imoto et al., 2003). ling. Specifically, I-Smads can function as adaptors to The functional consequence of Smad3 sumoylation is recruit Smurfs to the receptor complex and thereby not well understood. Possibly, the conjugation of mediate receptor degradation and downregulation of SUMO to Smad3 may hinder the association of Smad3 TGFb signalling. with Smad4 or prevent Smad3 binding to target The expression of Smad7 is regulated by a number promoters. Further studies are clearly needed to of extracellular signals and both TGFb and BMPs delineate the role of sumoylation in the regulation of have been shown to increase Smad7 protein levels, TGFb signalling. particularly in the nucleus where Smad7 is preferentially localized (Itoh et al., 1998;Moustakas et al., 2001; R-Smads as adaptors for E3 ligases Attisano and Wrana, 2002;Shi and Massague, 2003). Smurf2 resides in the nucleus in unstimulated cells and While several studies have indicated that association thus the increase in Smad7 protein levels results in of Smads with Smurf family ligases results in the association of Smad7 with Smurf2 (Kavsak et al., Smad degradation, accumulating evidence indicates that 2000). This interaction is mediated by the PY motif in Smads also function as adaptors that recruit Smurfs to Smad7 and the WW2/WW3 domains of Smurf2. The target proteins, and thereby control the level of Smad- Smad7/Smurf2 complex is then exported from the associating proteins. Ski and SnoN are two closely nucleus to the cytoplasm where Smad7 then recruits related nuclear proto-oncogenes that bind Smads and Smurf2 to the TGFb receptor complex at the cell antagonize TGFb signalling in part through their ability surface. Once bound to the receptor complex, the

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2075 Smurfs ubiquitinate Smad7 and cause degradation of The ubiquitin–proteasome pathway also plays an both Smad7 and the receptor complex (Figure 4) important role in controlling TGFb receptor trafficking (Kavsak et al., 2000). Smurf1 can also interact with (Di Guglielmo et al., 2003). TGFb receptors can be Smad7 and be targeted to the TGFb receptor to mediate internalized via two distinct routes, a clathrin- and a the turnover of the receptor complex (Ebisawa et al., raft/caveolar-dependent pathway each of which has 2001). In this case, it has been shown that chromosomal distinct outcomes. Receptors that are internalized via region maintenance 1, an importin b-related nuclear the clathrin pathway enter the EEA1-positive early transport receptor, physically interacts with a nuclear endosome where SARA is also enriched. Receptors that export signal present within HECT domain of Smurf1 enter this compartment are competent to transduce and thereby mediates the nuclear export of the Smurf1/ TGFb signals (Ehrlich et al., 2001;Hayes et al., 2002; Smad7 complex (Suzuki et al., 2002;Tajima et al., 2003). Penheiter et al., 2002;Di Guglielmo et al., 2003). In Both of the I-Smads, Smad6 and -7, have also been contrast, the receptor complex that is found in lipid-rich shown to function as adaptors that recruit Smurf1 to raft domains of the plasma membrane associates with BMP receptor complexes and mediate BMP receptor the Smad7/Smurf2 complex (Di Guglielmo et al., 2003). complex turnover (Murakami et al., 2003). The physio- Here, the receptor complex is internalized into a logical specificity of Smad6 and -7 association with caveolin-positive compartment that leads to the degra- Smurf1 and/or Smurf2 to target either TGFb or BMP dation of the receptor complex (Figure 4). It is receptor complex turnover has not been resolved, but it likely that it is the balance of receptor partitioning may be that different I-Smad/Smurf combinations serve between these two pathways that underlies the variable similar functions in different cell types or tissues. An requirement for clathrin in TGFb signalling (Lu et al., additional level of complexity in Smad/Smurf-depen- 2002;Di Guglielmo et al., 2003). What causes receptors dent regulation of TGFb signalling was recently to segregate between these two routes is unclear since revealed with the demonstration that Smad7 is acety- TGFb ligand does not appear to strongly affect lated by the transcriptional activator p300 (Gronroos partitioning and endocytosis into either intracellular et al., 2002). This acetyltransferase targets two lysine compartment. residues in the amino-terminus of Smad7, which serves to protect Smad7 from Smurf1-mediated ubiquitination. Biological role of Smurfs Acetylation did not appear to alter Smad7 nuclear export and thus may not interfere with subsequent While the biochemical activity of Smurfs has been targeting of the Smad7/Smurf1 complex to the TGFb extensively examined, much less is known of their receptor complex (Gronroos et al., 2002). Instead, biological role. A recent report has revealed that the acetylation may serve to protect Smad7 from premature overexpression of Smurf1 blocks BMP-induced osteo- degradation by Smurfs. It will be interesting to genic conversion in C2C12 myoblasts and can facilitate determine whether other Smads are also similarly myogenic differentiation by inducing the degradation of stabilized. Smad5 (Ying et al., 2003). Smurf1 has also been reported to induce the proteasomal destruction of Cbfa, a transcription factor of the runt-domain gene family implicated in osteoblast differentiation and bone development (Zhao et al., 2003). Interestingly, the overexpression of a catalytic mutant of Smurf1, unable to target Cbfa1 nor Smad1 for ubiquitin-dependent proteasomal degradation, enhanced osteoblast differentiation in C2C12 cells (Zhao et al., 2003). In addition, a recent study has shown that Smurf1 can regulate cell polarity and the formation of cellular protrusions through its ability to target the GTPase, RhoA for degradation (Wang et al., 2003). To date mice null for Smurf1 or Smurf2 have not yet been generated; however, insights into the in vivo role of Smurfs have been provided by studies in Drosophila. Podos et al. (2001) identified Drosophila Smurf (DSmurf, Figure 3) and found that it acts as a negative regulator of decapentaplegic (Dpp) signalling in the fly. Dpp, the ortholog of BMP2/4 in Drosophila, confers positional Figure 4 I-Smads and Smurfs regulate receptor turnover. Smurfs interact with I-Smads in the nucleus. Upon ligand binding, the information in both the wing disc and the embryonic Smurf/I–Smad complex is exported and the I-Smad then recruits ectoderm. Dpp mRNA is produced uniformly across the Smurfs to the receptor complex. I-Smads acts as adaptors and dorsal region of the embryo (Arora et al., 1994); bridge Smurfs to the receptor complex. Smurfs then ubiquitinate however, Dpp gradient activity is required to establish and thereby mediate the degradation of the I-Smad/receptor complex. In the case of Smad7 and Smurf2, the ubiquitin- proper dorso-ventral (D-V) patterning in the embryonic dependent degradation of the TGFb receptor complex was shown ectoderm. While a number of molecules function to occur through a raft–caveolar pathway to establish and maintain a Dpp gradient extracellularly,

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2076 DSmurf appears to function intracellularly to regulate increase in the degradation of Smad2 (Xu and Attisano, Dpp signalling. DSmurf mutant embryos do not 2000). A nonsense mutation at position 515 in the present any overt disturbance in D-V patterning of MH2 domain of Smad4, found in pancreatic adenocar- their cuticle;however, spatial expansion of Dpp signal- cinomas, which results in the truncation of the last ling is observed (Podos et al., 2001). DSmurf mutant 38 amino acids, targets Smad4 for ubiquitin-dependent embryos show increased staining of phosphorylated proteasomal degradation in addition to preventing MAD, a Smad protein, along the D-V axis and it from associating with Smad2 and binding DNA increased expression of the Dpp target genes zerknu¨llt (Maurice et al., 2001). A missense mutation found (zen), Race and u-shaped (ush) (Podos et al., 2001). in human hepatocellular carcinoma in which Smad2 Therefore, it appears that one function of DSmurf is to harbors a glutamine-to-arginine mutation at position regulate the level of Dpp signalling along the D-V axis, 407 in the MH2 and a colorectal cancer-associated thereby contributing to the establishment of an activity mutation in leucine 369 to arginine have also proved gradient. Mutation of the DSmurf locus also results in to be highly unstable and, in the case of Q407R, the appearance of a posterior hole in the cuticle of has been shown to result in rapid ubiquitin-dependent the embryo that results from a hindgut defect, which degradation of the mutant Smad (Eppert et al., likely interferes with dorsal closure of the cuticle. 1996;Dumont et al., 2003). Taken together, A prolonged phospho-MAD staining is observed in these observations suggest that certain oncogenic muta- the hindgut and this correlates with the ectopic tions in Smads inactivate the proteins by enhancing expression of Dpp target genes, such as zen (Podos their polyubiquitination and proteasomal destruc- et al., 2001). Therefore, the morphological defect in the tion, and may thus cause abnormal TGFb signalling, cuticle is likely due to a temporal deregulation of Dpp deregulated cell proliferation and promote tumori- signalling in the hindgut (Podos et al., 2001). Thus, genesis. DSmurf appears to be important for both establishment Since E3 ubiquitin ligases control the localization, of Dpp gradient and for temporal control of Dpp activity and stability of numerous components of TGFb signalling. signalling, dysregulated expression or function of E3 ubiquitin ligases may profoundly affect the proper Ubiquitination and degradation of oncogenic Smad transmission of TGFb signals and thus may contribute mutants to the development of human diseases. Consistent with this possibility, a recent study by Fukuchi et al. (2002) By regulating the steady-state levels of nonactivated demonstrated that high-level expression of Smurf2 is Smads, ubiquitin-dependent protein degradation has associated with oesophageal squamous carcinoma and a proven to be essential in maintaining the basal state poor prognosis for patients with the disease. Although of the TGFb and BMP signalling pathways. Further- TGFb functions as a tumor suppressor in early stages more, by targeting the activated receptor complex of cancer progression, abundant evidence indicates and phosphorylated R-Smads for destruction, it is that increased TGFb signalling can promote late-stage also an important mechanism for turning off TGFb tumors (de Caestecker et al., 2000;Derynck et al., 2001). signalling following the completion of the transcrip- Thus, a decrease in Smurf levels with a concomitant tional response. A growing body of work indicates increase in TGFb signalling might also promote that ubiquitination and proteasomal degradation is tumorigenesis. RNF11, a RING-H2 protein that implicated in the turnover of tumor-derived Smad is highly expressed in invasive breast cancer, mutants. Several inactivating mutations in Smad2 was recently shown to interact with Smurf2 (Subrama- and -4 have been identiﬁed in a variety of human niam et al., 2003). Ectopic expression of RNF11 appears tumors, including pancreatic, colorectal and lung to block Smurf2-dependent inhibition of TGFb signal- carcinomas (de Caestecker et al., 2000;Derynck et al., ling possibly by promoting Smurf2 ubiquitination 2001). Although the majority of missense and nonsense and degradation via the ubiquitin–proteasomal path- mutations identiﬁed in Smads are localized in the MH2 way. By blocking Smurf2 activity, RNF11 may thus domains, a number of missense mutations have also enhance TGFb signalling and its tumor-promoting been described in the MH1 domains. Interestingly, activity. some of these oncogenic mutations in Smads do not affect their ability to be phosphorylated, oligomer- Conclusions ize, translocate or carry out their transcriptional functions, rather they appear to affect the steady-state The ubiquitin-mediated proteasomal degradation path- levels of the protein (Moren et al., 2000;Xu and way is an evolutionary conserved cascade involved in Attisano, 2000;Moren et al., 2003). Missense mutations the regulation of key cellular processes, such as signal in the MH1 domain of tumor-derived Smad4, including transduction, cell cycle progression, transcriptional L43S, G65V, R100T and P130S, lead to increased regulation and endocytosis (Hershko and Ciechanover, polyubiquitination and proteasomal degradation of 1998). Like other signalling pathways, the TGFb Smad4 when compared to the wild-type protein (Moren and BMP pathways are tightly regulated by the et al., 2000;Xu and Attisano, 2000;Moren et al., 2003). ubiquitin–proteasomal degradation system and this Similarly, an arginine-to-cysteine mutation at position occurs through a variety of mechanisms. The ubiquitin 133 in the MH1 domain of Smad2 also causes an pathway can regulate the basal level of Smads and as

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2077 Smads are privotal for transducing TGFb signals, the pathway, SnoN. Thus, ubiquitin-mediated alterations in Smad proteins can profoundly affect the degradation of a Smad-associated protein would be capacity of cells to respond to incoming signals. predicted to result in enhanced activation of SnoN Smads are also targeted for ubiquitination and degrada- targeted genes. Undoubtedly, current and future tion in response to TGFb stimulation and this research endeavors will shed light as to how degradation then contributes to the termination of TGFb signals. of TGFb pathway components by Smurfs and other The E3 ligases, including Roc1, Smurf1 and Smurf2, E3 ligases contributes to the complexity of biological have been implicated in Smad degradation, but the responses to TGFb superfamily ligands. Several Smurf family of C2-WW-HECT domain ligases has studies have shown that turnover of Smads harboring gained particular prominence as they also control the tumor-associated mutations are regulated by ubiquiti- degradation of other components of the TGFb nation and degradation (Moren et al., 2000;Xu and signalling pathway. For instance, the I-Smads recruit Attisano, 2000;Moren et al., 2003). Furthermore, a Smurfs to the TGFb receptor complex where they correlation between Smurf2 expression and poor prog- contribute to receptor compartmentalization and nosis in oesophageal squamous carcinoma has been promote receptor degradation, thereby downregulating observed (Fukuchi et al., 2002). Thus, one intriguing TGFb signalling. The R-Smads can also recruit E3 possibility is that the Smurf family of E3 ligases may ligases including Smurf2 and APC to mediate the represent suitable targets for anticancer drug develop- degradation of a negative transcriptional regulator of ment.

References

Arora K, Levine M and O’Connor MB. (1994). Genes Dev., 8, Hershko A and Ciechanover A. (1998). Annu. Rev. Biochem., 2588–2601. 67, 425–479. Attisano L and Wrana JL. (2000). Curr. Opin. Cell. Biol., 12, Imoto S, Sugiyama K, Muromoto R, Sato N, Yamamoto T 235–243. and Matsuda T. (2003). J. Biol. Chem., 278, 34253–34258. Attisano L and Wrana JL. (2002). Science, 296, 1646–1647. Itoh S, Landstrom M, Hermansson A, Itoh F, Heldin C-H, Bakin AV, Tomlinson AK, Bhowmick NA, Moses HL and Heldin N-E and ten Dijke P. (1998). J. Biol. Chem., 273, Arteaga CL. (2000). J. Biol. Chem., 275, 36803–36810. 29195–29201. Bhowmick NA, Ghiassim M, Bakin A, Aakre M, Lundquist Kavsak P, Rasmussen RK, Causing CG, Bonni S, Zhu H, CA, Engel ME, Arteaga CL and Moses HL. (2001). Mol. Thomsen GH and Wrana JL. (2000). Mol. Cell, 6, Cell. Biol., 12, 27–36. 1365–1375. Bonni S, Wang H-R, Causing CG, Kavsak P, Stroschein SL, Lee PS, Chang C, Liu D and Derynck R. (2003). J. Biol. Luo K and Wrana JL. (2001). Nat. Cell Biol., 3, 587–595. Chem., 278, 27853–27863. de Caestecker MP, Piek E and Roberts AB. (2000). J. Natl. Lin X, Liang M and Feng X-H. (2000). J. Biol. Chem., 275, Cancer Inst., 92, 1388–1402. 36818–36822. Derynck R, Akhurst RJ and Balmain A. (2001). Nat. Genet., Lin X, Liang M, Liang YY, Brunicardi FC and Feng XH. 29, 117–129. (2003a). J. Biol. Chem., 278, 31043–31048. Desterro JM, Rodriguez MS and Hay RT. (2000). Cell. Mol. Lin X, Liang M, Liang YY, Brunicardi FC, Melchior F and Life Sci., 57, 1207–1219. Feng XH. (2003b). J. Biol. Chem., 278, 18714–18719. Di Guglielmo GM, Le Roy C, Goodfellow AF and Wrana JL. Liu X, Elia AE, Law SF, Golemis EA, Farley J and Wang T. (2003). Nat. Cell Biol., 5, 410–421. (2000). EMBO J., 19, 6759–6769. Dumont E, Lallemand F, Prunier C, Ferrand N, Guillouzo A, Lo RS and Massague´ J. (1999). Nat. Cell. Biol., 1, 472–478. Clement B, Atﬁ A and Theret N. (2003). J. Biol. Chem., 278, Long J, Matsuura I, He D, Wang G, Shuai K and Liu F. 24881–24887. (2003). Proc. Natl. Acad. Sci. USA, 100, 9791–9796. Ebisawa T, Fukuchi M, Murakami G, Chiba T, Tanaka K, Lu Z, Murray JT, Luo W, Li H, Wu X, Xu H, Backer JM and Imamura T and Miyazono K. (2001). J. Biol. Chem., 276, Chen YG. (2002). J. Biol. Chem., 277, 29363–29368. 12477–12480. Massague´ J. (1998). Annu. Rev. Biochem., 67, 753–791. Ehrlich M, Shmuely A and Henis YI. (2001). J. Cell Sci., 114, Maurice D, Pierreux CE, Howell M, Wilentz RE, 1777–1786. Owen MJ and Hill CS. (2001). J. Biol. Chem., 276, Eppert K, Scherer SW, Ozcelik H, Pirone R, Hoodless P, 43175–43181. Kim H, Tsui L-C, Bapat B, Gallinger S, Andrulis I, Melchior F. (2000). Annu. Rev. Cell Dev. Biol., 16, 591–626. Thomsen G, Wrana JL and Attisano L. (1996). Cell, 86, Moren A, Hellman U, Inada Y, Imamura T, Heldin CH and 543–552. Moustakas A. (2003). J. Biol. Chem., 278, 33571–33582. Fukuchi M, Fukai Y, Masuda N, Miyazaki T, Nakajima M, Moren A, Itoh S, Moustakas A, Dijke P and Heldin CH. Sohda M, Manda R, Tsukada K, Kato H and Kuwano H. (2000). Oncogene, 19, 4396–4404. (2002). Cancer Res., 62, 7162–7165. Moustakas A, Souchelnytskyi S and Heldin C-H. (2001). Fukuchi M, Imamura T, Chiba T, Ebisawa T, Kawabata M, J. Cell. Sci., 114, 4359–4369. Tanaka K and Miyazono K. (2001). Mol. Cell. Biol., 12, Mulder KM. (2000). Cytokine Growth Factor Rev., 11, 23–35. 1430–1443. Muller S, Hoege C, Pyrowolakis G and Jentsch S. (2001). Nat. Gronroos E, Hellman U, Heldin CH and Ericsson J. (2002). Rev. Mol. Cell. Biol., 2, 202–210. Mol. Cell, 10, 483–493. Murakami G, Watabe T, Takaoka K, Miyazono K and Gruendler C, Lin Y, Farley J and Wang T. (2001). J. Biol. Imamura T. (2003). Mol. Cell. Biol., 14, 2809–2817. Chem., 276, 46533–46543. Penheiter SG, Mitchell H, Garamszegi N, Edens M, Hayes S, Chawla A and Corvera S. (2002). J. Cell Biol., 158, Dore Jr JJ and Leof EB. (2002). Mol. Cell. Biol., 22, 1239–1249. 4750–4759.

Oncogene Regulation of the TGFb signalling L Izzi and L Attisano 2078 Podos SD, Hanson KK, Wang YC and Ferguson EL. (2001). Tsukazaki T, Chiang TA, Davison AF, Attisano L and Wrana Dev. Cell, 1, 567–578. JL. (1998). Cell, 95, 779–791. Saha D, Datta PK and Beauchamp RD. (2001). J. Biol. Chem., Wan M, Cao X, Wu Y, Bai S, Wu L, Shi X, Wang N and Cao 276, 29531–29537. X. (2002). EMBO Rep., 3, 171–176. Schwartz DC and Hochstrasser M. (2003). Trends Biochem. Wan Y, Liu X and Kirschner MW. (2001). Mol. Cell, 8, Sci., 28, 321–328. 1027–1039. Shi Y and Massague J. (2003). Cell, 113, 685–700. Wang H-R, Zhang Y, Ozdamar B, Ogunjimi AO, Alexandrova E, Stroschein SL, Bonni S, Wrana JL and Luo K. (2001). Genes Thomsen GH and Wrana JL. (2003). Science, 302, 1775–1779. Dev., 15, 2822–2836. Whitman M. (1998). Genes Dev., 12, 2445–2462. Subramaniam V, Li HX, Wong M, Kitching R, Attisano L, Wotton D and Massague´ J. (2000). EMBO J., 19, 745–754. Wrana JL, Zubovits J, Burger A and Seth A. (2003). Br. J. Xu J and Attisano L. (2000). Proc. Natl. Acad. Sci. USA, 97, Cancer, 39, 1538–1544. 4820–4825. Suzuki C, Murakami G, Fukuchi M, Shimanuki T, Shikauchi Ying SX, Hussain ZJ and Zhang YE. (2003). J. Biol. Chem., Y, Imamura T and Miyazono K. (2002). J. Biol. Chem., 277, 278, 39029–39036. 39919–39925. Zhang Y, Chang C, Gehling DJ, Hemmati-Brivanlou A and Tajima Y, Goto K, Yoshida M, Shinomiya K, Sekimoto T, Derynck R. (2001). Proc. Natl. Acad. Sci. USA, 98, 974–979. Yoneda Y, Miyazono K and Imamura T. (2003). J. Biol. Zhao M, Qiao M, Oyajobi BO, Mundy GR and Chen D. Chem., 278, 10716–10721. (2003). J. Biol. Chem., 278, 27939–27944. ten Dijke P, Miyazono K and Heldin C-H. (2000). Trends Zhu H, Kavsak P, Abdollah S, Wrana JL and Thomsen GH. Biochem. Sci., 25, 64–70. (1999). Nature, 400, 687–693.

Oncogene