The Deubiquitinating Enzyme UCH37 Interacts with Smads and Regulates TGF-B Signalling

The deubiquitinating enzyme UCH37 interacts with Smads and regulates TGF-b signalling

Stephen J Wicks1, Katherine Haros1, Marjorie Maillard1, Ling Song2, Robert E Cohen2, Peter ten Dijke3 and Andrew Chantry*,1

1School of Biological Sciences, University of East Anglia, Earlham Road, Norwich Norfolk NR4 7TJ, UK; 2Department of Biochemistry, University of Iowa, Bowen Science Building, 51 Newton Road, Iowa City, IA 52242-1109, USA; 3Department of Molecular Cell Biology, Leiden University Medical Center, Div. 5, Wassenaarseweg 72, 2333 AL Leiden, The Netherlands

Disruption of components in the transforming growth phosphorylated and activated by the TGF-b receptor factor-b (TGF-b) signalling cascade is a common complex. Co-Smad (Smad4) associates with activated occurrence in human cancers. TGF-b pathway activation R-Smads, and these heteromeric complexes are translo- is accomplished via serine/threonine kinase receptors and cated to the nucleus where they regulate gene transcrip- intracellular Smad transcription factors. A key regulatory tion by either association with DNA-binding proteins step involves specific ubiquitination by Smurfs that or direct binding to promoter sequences in target genes. mediate the proteasomal degradation of Smads and/or I-Smads (Smads 6 and 7) inhibit Smad-mediated signal- receptors. Here, we report a novel interaction between ling and transcription. Smads and ubiquitin C-terminal hydrolase UCH37, a Downregulation of TGF-b signalling is affected, in deubiquitinating enzyme that could potentially reverse part, by a feedback mechanism that involves specific Smurf-mediated ubiquitination. In GST pull down experi- protein ubiquitination and proteasomal degradation of ments, UCH37 bound weakly to Smad2 and Smad3, and Smads and receptors. Ubiquitination plays a key role in bound very strongly to Smad7 in a region that is distinct a number of biological processes including signal from the –PY– motif in Smad7 that interacts with Smurf transduction, cell cycle, and gene expression (Glickman ubiquitin ligases. Endogenous Smad7 and UCH37 formed and Ciechanover, 2002). Recently, ubiquitin ligases a stable complex in U4A/JAK1 cells, and FLAG-Smad7 known as Smurfs have been shown to bind to Smads co-immunoprecipitated with HA-UCH37 in transfected and have been implicated in their specific ubiquitination. HEK-293 cells. In addition, we show that UCH37 can Smurf1 can interact selectively with Smad1 (Bone deubiquitinate and stabilize the type I TGF-b receptor. morphogenic protein pathway-specific), and this Furthermore, overexpression of UCH37 upregulates mechanism appears to regulate the abundance of Smad1 TGF-b-dependent transcription, and this effect is reversed in unstimulated cells since it is not affected by receptor in cells subject to RNAi-mediated knockdown of endo- activation (Zhu et al., 1999). Activated Smad2 is genous UCH37. These findings support a new role for targeted for ubiquitin-mediated proteasomal degrada- deubiquitinating enzymes in the control of the TGF-b tion by Smurf2 in the nucleus (Lo and Massague, 1999). signalling pathway, and provide a novel molecular target In all cases, a small region in Smurfs known as a WW for the design of inhibitors with therapeutic potential in domain is responsible for the interaction with a –Pro– cancer. Pro–X–Tyr– sequence motif in Smads. Interestingly, Oncogene (2005) 24, 8080–8084. doi:10.1038/sj.onc.1208944; Smurfs also bind to inhibitory I-Smads. Following published online 18 July 2005 prolonged exposure of cells to TGF-b, the Smad7/ Smurf complex forms, exits the nucleus, and binds to the Keywords: transforming growth factor-b; Smads; ubi- activated TGF-b receptor complex; the associated quitin Smurf then ubiquitinates the receptor, thereby leading to its rapid degradation (Kavsak et al., 2000; Ebisawa et al., 2001; Suzuki et al., 2002). Aberrant TGF-b signalling is responsible for a number The conjugation of ubiquitinated substrates can be of developmental disorders, human cancers, and fibrotic reversed by deubiquitinating enzymes (DUBs) that diseases. The TGF-b superfamily transmits signals via specifically cleave the isopeptide bond at the C-terminus cell surface serine/threonine kinase receptors and intra- of ubiquitin (Glickman and Ciechanover, 2002; Wing, cellular Smad transcription factors (Shi and Massague, 2003). Most DUBs are members of either of two families 2003). Smads are segregated into three functional of cysteine proteases: the ubiquitin-specific processing groups. R-Smads (Smads-1, 2, 3, and 5) are directly proteases (UBPs) and the ubiquitin C-terminal hydro- lases (UCHs). There are emerging roles for DUBs in *Correspondence: A Chantry; E-mail: [email protected] cell growth, oncogenesis, development, memory, and Received 11 March 2005; revised 26 May 2005; accepted 17 June 2005; transcriptional regulation. For example, the human published online 18 July 2005 UBP known as HAUSP/USP7 (herpesvirus-associated The deubiquitinating enzyme UCH37 regulates TGF-b signalling SJ Wicks et al 8081 ubiquitin-specific protease) was found to specifically deubiquitinate and stabilize the p53 tumour suppressor protein (Li et al., 2002). BAP1 (BRCA1-associated protein 1) is a novel UCH enzyme that was found to interact with BRCA1 and enhance BRCA1-mediated growth suppression (Jensen et al., 1998). Here, we demonstrate a unique role for ubiquitin C-terminal hydrolase UCH37, also known as UCHL5 in mouse, in TGF-b signalling. Our results suggest that competing effects of ubiquitin ligases and deubiquitinating enzymes in complex with Smad7, in particular, could play a central role in fine-tuning cellular responses to TGF-bs under various physiological and pathological conditions. To identify proteins that interact with MHI and linker regions of Smad3, we performed a yeast two-hybrid screen using a mouse brain library and Smad31–240 as bait. From 65 positive colonies displaying growth on selection media as well as b-galactosidase activity, we selected mouse UCHL5, also known as UCH37 in humans, for further study. UCH37 was first identified as a component of the 26S proteasome that is thought to be involved in the editing of polyubiquitinated protein substrates (Lam et al., 1997a, b). It is a member of a family of UCH enzymes that have a conserved catalytic Figure 1 Interaction of Smads with UCH37. (a) Schematic domain and, in some instances, have a nonconserved representation of UCH37/UCHL5 compared with other UCH family members. The conserved catalytic domain is shown in dark extended C-terminal tail (Figure 1a). Human UCH37 grey, C-terminal extension in light grey and the BRCA1-interaction and mouse UCHL5 are 99% identical. BRCA1-asso- domain in BAP1 shown in black. For the yeast two-hybrid ciated protein BAP1 and UCH37 are unique in that they screening, a cDNA encoding amino acids 1–240 of Smad3 was have an extended C-terminal tail, and in BAP1 this cloned in frame with the Gal4 DNA binding domain in vector region contains the BRCA1 interaction domain. Proteo- pAS2-1 and used to screen a mouse brain library (Clontech). Mouse UCH37/UCHL5 was identified as a positive and subse- some-associated UCH37 removes ubiquitin sequentially quently obtained as a full-length EST clone (EST clone 1197628), from the distal end of the Lys48-linked polyubiquitin and subcloned into pCMV1-FLAG and pCMV1-HA. (b) Smad- chain and can potentially rescue ubiquitinated sub- GST fusion proteins were prepared and attached to glutathione- strates from proteasomal degradation. Since recent sepharose beads as described previously (Wicks et al., 2000). Immobilized proteins were then incubated with lysates from HEK- reports have elucidated an important role for Smurf 293 cells expressing HA-UCH37 and washed extensively before ubiquitin ligases in the regulation of Smad signalling separation by SDS–10% PAGE and immunoblotting against (Zhu et al., 1999; Lo and Massague, 1999; Kavsak et al., anti-HA antibody. The presence of the GST fusion proteins was 2000; Ebisawa et al., 2001; Suzuki et al., 2002), we confirmed by staining gels with Coomassie Blue (C/Blue). (c) The hypothesized that UCH37 could potentially reverse following GST fusion proteins were prepared: full-length Smad7, Smad7 , full-length Smad7DPY, Smad7 DPY, and Smurf-mediated ubiquitination and play an important 1–260 1–260 Smad7260–426. The DPY mutation involves changes of the proline role in the control of Smad/TGF-b signalling activity. to glycine and tyrosine to serine in the conserved ÀPYÀ motif. Initially, to verify the interaction of UCH37 with GST proteins immobilized to glutathione-sepharose beads were Smad3 as well as other Smads, we performed GST pull then incubated with lysates from HEK-293 cells expressing HA- UCH37, and washed extensively before separation by SDS–10% down experiments using lysates from cells expressing PAGE and immunoblotting against anti-HA antibody. The HA-tagged UCH37. In this instance, UCH37 bound to presence of the GST fusion proteins was confirmed by staining Smad3 and marginally to Smad2, however, we found a gels with C/Blue. The GST fusion proteins are indicated with an very strong interaction with Smad7 (Figure 1b). This arrowhead difference in relative affinity between Smad2 and Smad3 could be explained by the presence of an additional exon in the N-terminal region that makes these R-Smads be present in other regions that is not so apparent from behave differently with respect to MH1 domain func- primary structure. tion. Based on a multialignment of Smad7 and Smad3, The interaction of UCH37 with Smad7 is of particular there are only a few stretches of very weak amino-acid interest since Smad7 has been shown to function as an sequence homology within the N-terminal region. One adaptor that recruits the Smurf E3 ubiquitin ligase to example is a –V–X–R–L– motif (where X is either lysine the TGF-b receptor complex to promote its ubiquitina- or arginine), at positions 12–15 in Smad3 that is also tion and proteasomal degradation. To identify the region present in Smad7 (residues 10–13). However, this might in Smad7 that interacts with UCH37 we performed be insufficient homology for a conserved interaction GST pull down experiments of full-length and trunc- motif and, in any case, it is always difficult to predict ated Smad7. Both full-length Smad7 and Smad71–260, with certainty since homology at structural level might but not Smad7260–426, were able to efficiently pull

Oncogene The deubiquitinating enzyme UCH37 regulates TGF-b signalling SJ Wicks et al 8082 down UCH37, and mutation of the –PY– motif in both (UCH37DC; lacking residues Trp196–Lys329), was coex- Smad7 and Smad71–260 did not affect this interaction pressed with FLAG-tagged Smad7 proteins. Here we (Figure 1c). Since the –PY– motif is required for the found that Smad7 bound to both full-length UCH37 interaction of Smad7 with Smurf, we conclude that and UCH37DC both in the presence and absence of an UCH37 is likely to bind to the N-terminal half of Smad7 activated type I TGF-b receptor (Figure 2b). To identify in a region that is distinct from the Smurf binding site. whether Smad7 bound to UCH37 can form a complex We then confirmed that Smad7 interacts with UCH37 with the type I TGF-b receptor we performed a endogenously in U4A/JAK1 cells. In the majority of cell co-immunoprecipitation using anti-FLAG antibody types, the basal expression of Smad7 is low, however, its from HEK-293 cells transfected with FLAG-Smad-7, expression can be induced in U4A/JAK1 cells treated HA-UCH37 in the presence and absence of an activated with interferon-g (Ulloa et al., 1999). In cells stimulated type I TGF-b receptor. These data clearly suggest that with interferon-g for 2 h, Smad7 expression is induced UCH37 binds directly to Smad7 in a complex that and it also co-immunoprecipitates with endogenous includes the type I TGF-b receptor (Figure 2c). UCH37 (Figure 2a). No such complex was observed in Next, we examined whether ubiquitination and the parental U4A cell line in which Smad7 is not induced stability of the type I TGF-b receptor were affected by by interferon-g (Figure 2a; Ulloa et al., 1999). Impor- UCH37. We coexpressed the type I TGF-b receptor tantly, these results demonstrate that Smad7 and together with Smad7, Smurf2 and increasing amounts of UCH37 associate at endogenous levels of expression. UCH37 in HEK-293 cells. In the presence of Smurf2, We then performed coexpression and co-immunopreci- the levels of TGF-b type I receptors was significantly pitation studies of epitope-tagged proteins in HEK-293 reduced, based on Western blotting of lysates, confirm- cells. In these experiments, HA-tagged full length ing that Smurf2 can destabilize the TGF-b type I UCH37, or a truncation lacking a C-terminal extension receptor in a Smad7-dependent manner (Figure 2d; left downstream of the N-terminal enzymatic domain panel). However, coexpression of Smad7 and Smurf2

Figure 2 UCH37 binds to Smad7 and affects ubiquitination and stability of the type I TGF-b receptor. (a) U4A/JAK1 and parental U4A cells were treated for 2 h with 1000 units/ml interferon-g (Roche Biochemicals). Endogenous Smad7 was immunoprecipitated with rabbit polyclonal anti-Smad7 antibody (Santa Cruz). Samples were separated by SDS–10% PAGE and probed with rabbit polyclonal anti-UCH37 antibody. (b) HEK-293 cells were transfected with FLAG-Smad7 and full-length UCH37 or UCH37DC lacking residues Trp196–Lys329 in the presence/absence of the nontagged constitutively active type I TGF-b receptor as described previously (Wicks et al., 2000). Cell lysates were immunoprecipitated with anti-FLAG antibody, samples were separated by SDS–10% PAGE and then immunoblotted against anti-HA. Western blot expression controls of cell lysates are included to confirm equal expression of all transfected constructs. (c) HEK-293 cells were transfected as indicated and immunoprecipitated with anti-FLAG antibody. Lysates and immunoprecipitated samples were probed by Western blotting against the anti- TGF-b receptor antibody. The position of the type I TGF-b receptor in the immunoprecipitate is indicated (*). (d, left panel) HEK-293 cells were transfected as indicated with different combinations of FLAG-TGF-bRIT204D, FLAG-Smad7 and FLAG-Smurf2 together with either 3 or 10 mg HA-UCH37. Cell lysates were separated by SDS–10% PAGE and probed by Western blotting against anti-FLAG antibody. Expression of HA-UCH37 was confirmed by Western blotting against anti-HA antibody (data not shown). (d, middle panel) HEK-293 cells were transfected as indicated with different combinations of FLAG-TGF-bRIT204D, FLAG-Smurf2 and FLAG-UCH37 in the presence of HA- ubiquitin. Cell lysates were immunoprecipitated with an anti- TGF-bRI antibody covalently attached to agarose beads (Santa Cruz). Samples were separated by SDS–10% PAGE and probed by Western blotting against anti-HA antibody to identify polyubiquitinated species of the type I TGF-b receptor. The migration of the polyubiquitinated TGF-b receptor is indicated (Ub-RI), and all other bands are nonspecific. Expression of FLAG-TGF-bRIT204D, FLAG-Smurf2 and FLAG-UCH37 was confirmed by Western blotting against anti-FLAG antibody (data not shown). (d, right panel) HEK-293 cells were transfected as indicated and labelled with [35S-methionine] as described previously (Chantry, 1995). Pulse-chase was then applied for the indicated times and immunoprecipitations were performed using anti-type I TGF-b receptor antibody. Samples were separated by SDS–10% PAGE, and gels were dried and exposed to X-ray film

Oncogene The deubiquitinating enzyme UCH37 regulates TGF-b signalling SJ Wicks et al 8083 together with increasing amounts of UCH37 led to a substantial stabilization of the TGF-b type I receptor (Figure 2d; left panel). We then examined whether the TGF-b type I receptor is a direct substrate for UCH37 in vivo by cotransfection of HEK-293 cells with an activated TGF-b type I receptor, in the presence or absence of Smurf2 and UCH37 and together with HA- tagged ubiquitin as well as Smad7. The degree of TGF-b type I receptor ubiquitination was then monitored by specific immunoprecipitation of the TGF-b type I receptor and probing of Western blots with anti-HA antibody. The amount of HA-ubiquitin incorporated into the TGF-b type I receptor was substantially increased in the presence of Smurf2, and this was reduced almost back to basal levels following coexpression of UCH37 (Figure 2d; middle panel). The effect that deubiquitination of the type I TGF-b receptor by UCH37 has on receptor turnover was then analyzed by pulse-chase experiments. Acceleration of receptor turnover by Smurf2 and Smad7 was efficiently reversed in the presence of UCH37 (Figure 2d; right panel). Since UCH37 can stabilize the TGF-b type I receptor, we examined its effect on TGF-b-dependent transcriptional responses. In HEK-293 cells, the activated TGF-b type I receptor increased the activity of the Smad-specific CAGA12-luc transcriptional reporter (Figure 3a). This effect was increased 2–3 fold in the presence of UCH37 and significantly attenuated by Figure 3 UCH37 upregulates TGF-b-dependent transcription. (a) HEK-293 were transiently transfected with 1 mg CAGA-12 Smurf2; moreover, coexpression with UCH37 reversed luciferase reporter plasmid together with various combinations of the Smurf2-mediated inhibition almost back to levels FLAG-TGF-bRIT204D, FLAG-Smurf2 and FLAG-UCH37 as seen in the presence of UCH37 alone (Figure 3a). To indicated. Cells were starved in medium containing 0.5% fetal calf further confirm the role of UCH37 in regulating TGF-b serum and approximately 48 h after transfection luciferase activity signalling, we examined the effect of an RNAi plasmid was measured using the Luciferase Reporter Assay (Promega). (b) HEK-293 cells were transfected with 20 mg of a UCH37 RNAi directed against endogenous UCH37. Transfection of plasmid obtained from the Nederlands Kanker Instituut (NKI) and HEK-293 cells with increasing amounts of the UCH37 Cancer Research UK. The RNAi construct is cloned in the RNAi plasmid reduced the expression of UCH37 by pRETROSUPER vector and consists of a mixture of three RNAi’s about 70% based on the densitometric scanning of against UCH37 (Genbank accession number NM_015984). Cell lysates were immunoblotted against rabbit polyclonal anti-UCH37 Western blots relative to an internal b-actin control antibody or anti-actin antibody. (c) In a separate experiment, (Figure 3b), and also abrogated TGF-b-dependent HEK-293 cells were transfected with 1 mg CAGA-12 luciferase CAGA12 transcriptional activity to a similar extent reporter plasmid together with 8 mg RNAi-UCH37 in the presence/ (Figure 3c). These functional effects on TGF-b sig- absence of FLAG-TGF-bRIT204D as indicated. After 48 h, lucifer- nalling are not due to aberrant Smad7 subcellular ase activity was measured as described above localization since immunocytochemical experiments demonstrated that UCH37 did not compromise TGF- b-dependent Smad7 nuclear export (data not shown). intracellular mechanism that may function to stabilize In summary, we have identified a novel complex or reactivate TGF-b signalling that has been repressed between Smad7 and the deubiquitinating enzyme by the Smad7/Smurf complex. Alternatively, it may UCH37. Subsequently, we have shown that this can provide a mechanism for sustaining TGF-b signalling lead to deubiquitination of the associated type I TGF-b immediately following receptor activation. It will now receptor complex and thereby promote TGF-b signal- be interesting to elucidate the mechanisms that deter- ling. Since Smad7 is required for both Smurf- and mine the relative recruitment of Smurfs and UCH37 to UCH37-mediated effects on the TGF-b receptor, it is the TGF-b receptor complex. Upregulation of TGF-b difficult to assess the absolute dependence for Smad7 on signalling by UCH37 could also be partly explained by UCH37 action. However, data presented throughout the deubiquitination and stabilization of Smad3, and supports the notion that Smad7 and UCH37 act future experiments aim to clarify the physiological together to impact on type I TGF-b receptor function, importance of the UCH37/Smad3 complex. We will also and competes directly with the downregulating function address the importance of other deubiquitinating enzymes of the Smurf family of E3 ubiquitin ligases. To our in the regulation of Smad signalling, particularly, in knowledge, this is also the first report of a role for the context of specificity relative to ubiquitination by UCH37 in the specific control of a cellular signalling distinct E3 ubiquitin ligases. In addition, it will be of pathway. Our present study now defines a novel great interest to develop in vitro assays to analyze the

Oncogene The deubiquitinating enzyme UCH37 regulates TGF-b signalling SJ Wicks et al 8084 precise kinetics of TGF-b receptor deubiquitination. Acknowledgements Overall, it is likely that the balance between Smurf- This study was supported by research grants from the BBSRC mediated ubiquitination and UCH37-mediated deubi- (AC), Wellcome Trust, UK (AC), NIH Grant R01 GM37666 (REC) and the Dutch Cancer Society (PTD). We thank quitination is critical for ﬁne-tuning responses, and that Caroline Hill, Takeshi Imamura, Ian Kerr and Ana Costa- alterations in this homeostatic mechanism could con- Pereira for vectors and cell lines. We also thank the Neder- tribute to aberrant TGF-b signalling seen in tumorigen- lands Kanker Instituut (NKI) and Cancer Research UK for esis, ﬁbrosis and many other human diseases. allowing us to use the RNAi-UCH37 construct.

References

Chantry A. (1995). J. Biol. Chem., 270, 3068–3073. Lam YA, Xu W, DeMartino GN and Cohen RE. (1997b). Ebisawa T, Fukuchi M, Murakami G, Chiba T, Tanaka K, Nature, 385, 737–740. Imamura T and Miyazono K. (2001). J. Biol. Chem., 276, Li M, Chen D, Shiloh A, Luo J, Nikolaev AY, Qin J and 12477–12480. Gu W. (2002). Nature, 416, 648–653. Glickman MH and Ciechanover A. (2002). Physiol. Rev., 82, Lo RS and Massague J. (1999). Nat. Cell Biol., 1, 472–478. 373–428. Shi Y and Massague J. (2003). Cell, 113, 685–700. Jensen DE, Proctor M, Marquis ST, Gardner HP, Ha SI, Suzuki C, Murakami G, Fukuchi M, Shimanuki T, Shikauchi Chodosh LA, Ishov AM, Tommerup N, Vissing H, Sekido Y, Imamura T and Miyazono K. (2002). J. Biol. Chem., 277, Y, Minna J, Borodovsky A, Schultz DC, Wilkinson KD, 39919–39925. Maul GG, Barlev N, Berger SL, Prendergast GC and Ulloa L, Doody J and Massague J. (1999). Nature, 397, Rauscher III FJ. (1998). Oncogene, 16, 1097–1112. 710–713. Kavsak P, Rasmussen RK, Causing CG, Bonni S, Zhu H, Wicks SJ, Lui S, Abdel-Wahab N, Mason RM and Chantry A. Thomsen GH and Wrana JL. (2000). Mol. Cell., 6, (2000). Mol. Cell. Biol., 20, 8103–8111. 1365–1375. Wing SS. (2003). Int. J. Biochem. Cell Biol., 35, 590–605. Lam YA, DeMartino GN, Pickart CM and Cohen RE. Zhu H, Kavsak P, Abdollah S, Wrana JL and Thomsen GH. (1997a). J. Biol. Chem., 272, 28438–28446. (1999). Nature, 400, 687–693.

Oncogene