Ubiquitin-dependent degradation of TGF-β-activated Smad2

Transcription

1 Ubiquitin-dependent degradation of -ativated Roger S. Lo* and Joan Massagué* *Cell Biology Program, Howard Hughes Medial Institute, Memorial Sloan-Kettering Caner Center, New York, New York 10021, USA SMAD proteins are phosphorylated by transforming growth fator-β () reeptors and transloate to the nuleus, where they ontrol transription. Here we investigate the fate of ativated. We show that reeptormediated ativation leads to multi-ubiquitination and subsequent degradation of by the proteasome. Ubiquitination of is a onsequene of its aumulation in the nuleus. If degradation is averted, the phosphorylated remains in the nuleus in an ative state. By targeting for destrution, ensures the irreversible termination of its own signalling funtion. T GF-β and related growth and differentiation fators are ritial regulators of tissue development and homeostasis in metazoan organisms 1 4, and alterations in the ativity of these fators underlie several human developmental defets, hyperproliferative disorders and various forms of aner 5. Progress over the past few years has led to the eluidation of a signal-trandution pathway from membrane reeptors to nulear target genes 5,6. In this pathway, the ligand brings together two reeptor serine kinases on the plasma membrane, leading one kinase ( reeptor-ii, or TβR-II) to phosphorylate and ativate the other (TβR-I), whih then phosphorylates and Smad3 proteins. The phosphorylated SMADs bind the related fator Smad4, move into the nuleus, and, in assoiation with DNA-binding ofators, interat with the promoter elements of target genes, regulating transription. SMADs exert these funtions through a highly onserved amino-terminal domain (termed the MH1 domain) that ontats DNA and a highly onserved arboxy-terminal domain (MH2 domain) that has ontat sites for reeptors, assoiated SMADs, DNA-binding ofators, and transriptional oativators and o-repressors. SMAD phosphorylation by the reeptors ours at C-terminal serine residues, whereas phosphorylation by the mitogen-ativated protein kinase Erk, whih inhibits nulear aumulation of SMADs, ours in the region linking the two onserved domains 5. Probably beause of its key funtion in proesses that need to be tightly ontrolled, the /SMAD pathway is under stringent regulation at every level. Various soluble and membrane-bound proteins regulate the aess of ligands to their reeptors, the proesses of reeptor ativation and SMAD phosphorylation, the aumulation of SMAD omplexes in the nuleus, and the interation of SMAD omplexes with target genes 5,7. However, the eventual fate of reeptor-ativated SMADs, whih are the entral messengers in this pathway, remains unknown. We have investigated this problem here. A plausible hypothesis was that reeptor-ativated SMAD phosphorylations are removed by the ation of a phosphatase, but we found little evidene in support of this possibility. Rather, we provide evidene that stimulation not only ativates but also indues its onjugation with ubiquitin and degradation by the proteasome. This provides a mehanism for the targeted destrution of ativated, ensuring the irreversible termination of SMAD proteins signalling funtion. Results Proteasome inhibitors extend the life of ativated. To follow the fate of endogenous one ativated in response to stimulation, we first studied the amount of C-terminally phosphorylated (-P) as a funtion of the duration of treatment with. The -responsive human keratinoyte ell line HaCaT was inubated for inreasing lengths of time with, and ell lysates were probed with an antibody raised against a phosphopeptide that orresponds to the reeptor-phosphorylated C-terminal segment of (GSPSVRCSS*MS*, where asterisks denote phosphoserines) 8,9. We deteted endogenous -P only when was added (Fig. 1a). With ontinuous treatment, -P aumulation peaked at about 1 h and delined thereafter. We onsidered phosphatase-mediated -P dephosphorylation and proteasome-mediated -P degradation as two possible mehanisms that might lead to this deline. Addition of two ell-permeable, wide-spetrum phosphatase inhibitors sodium vanadate and okadai aid during inubation of HaCaT ells with had only a small effet in slowing down the rate of -P loss after peak aumulation (Fig. 1b). In ontrast, addition of MG-132 (ref. 10) or lataystin 11, inhibitors of the 26S proteasome, bloked this deline (Fig. 1b). The speifiity of this effet was onfirmed by the inability of the ell-permeable protease inhibitors pepstatin, leupeptin and alpain inhibitor II to affet the kinetis of -P deline (data not shown). The ability of MG-132 to prevent the deline in -P aumulation extended to other ell types, inluding COS-1, HeLa, HepG2 (Fig. 1), EpH4 and EpRas ells, when a high onentration was used (data not shown), establishing the generality of this phenomenon. Interestingly, in the two Smad4-defetive ell lines tested, the oloni arinoma ell line SW480 and the breast arinoma ell line MDA468, there was no signifiant deline in -P levels up to 6 h after addition. Stable introdution of Smad4 expression into SW480 ells did not resue -P degradation, indiating that -P loss may our independently of its funtional partner Smad4 and that these tumour ells appear to ontain a Smad4-independent defet in agonist-indued degradation (data not shown). How does the duration of signal input affet the dynamis of -P level? To address this question, we stimulated HaCaT ells with a pulse of (for up to 10 min) and washed the ells free of ligand. Inubation was then ontinued with no addition, addition of lataystin, or re-addition of. The results (Fig. 1d) show that, without re-addition, the level of -P peaked and started to deline muh earlier than that from ells under ontinuous stimulation, that is, with ligand re-addition after washout. This indiates that the rate of -P loss had quikly overtaken the rate of -P generation after ligand washout, resulting in an earlier peak and deline in -P levels. This earlier deline was also bloked by the presene of lataystin (Fig. 1d). To further investigate the role of proteasome-dependent degradation in the removal of ativated, we determined how a prolonged treatment with might affet the steady-state amount of Mamillan Magazines NATURE Ltd CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om

2 a (min) b (h) P Relative units (h) COS-1 HeLa HepG Time (min) MG-132 Additions after washout d Time (min) Vanadate + Okadai aid MG-132 Washout Pulse Post-Washout period Relative units Relative units Time (h) MG-132 V + OA Control Time (min) Figure 1 Proteasome inhibition bloks the time-dependent loss of endogenous reeptor-phosphorylated. All lysates were analysed by immunoblotting using an antibody speifi to C-terminally-phosphorylated (-P) (a d). a, Upon ontinuous stimulation, the level of -P peaks at about 1 h and delines thereafter. HaCaT ells were either left untreated (time 0 min) or treated with and lysed after inreasing durations of treatment. The results were plotted relative to the 60-min value as a funtion of time. b, The proteasome inhibitors MG- 132 and lataystin, but not the phosphatase inhibitors vanadate (V) and okadai aid (OA), blok the time-dependent loss of -P after stimulation. HaCaT ells were treated with alone or with the indiated additions for the stated lengths of time. The quantified values were plotted relative to the 1.5-h value as a funtion of time., The proteasome inhibitor MG-132 also bloks the time-dependent loss of -P in COS-1 monkey kidney ells, HeLa human ervial arinoma ells and HepG2 human hepatoma ells. d, After a short pulse of stimulation (10 min) and a ligand washout at 4 C to attenuate reeptor ativation, (top panel), nothing (middle panel) or lataystin (bottom panel) was added to the ulture medium. With a short pulse of treatment, in ontrast to ontinuous stimulation (that is, re-addition), the aumulation of -P peaked and started to deline muh earlier; this deline was still bloked by lataystin (ompare middle and bottom panels). The immunoblot signals were quantified and the values relative to the value at 10 min were plotted as a funtion of time. endogenous. Overnight inubation with led to an appreiable derease in the overall level of, and this derease was prevented by addition of MG-132 or lataystin during the inubation (Fig. 2a). The extent of this effet depended on the duration of MG-132 treatment (Fig. 2b). The proteasome inhibitors had no detetable effet on the basal levels of in the absene of TGFβ treatment (Fig. 2a). Furthermore, treatment dereased the metaboli stability of endogenous, and this derease was reversed signifiantly by o-inubation with MG-132 (Fig. 2). No hange in messenger RNA levels was observed during treatment for up to 24 h (data not shown). Taken together, these results indiate that -ativated may be seletively eliminated, and that this elimination is dependent on the proteasome. Prolonged presene of ativated in the nuleus. Next we investigated whether proteasome funtion dereases the pool of that is engaged in nulear signalling funtions. The indued nulear aumulation of endogenous peaked at 1 h of treatment and delined thereafter (Fig. 3a), thus following the kinetis of -P aumulation (Fig. 1a). In COS-1 and HaCaT ells (Fig. 3a), as well as in HeLa and EpH4 mouse mammary epithelial ells (data not shown), treatment with lataystin prevented the time-dependent derease in nulear aumulation of. Thus, when the loss of in TGFβ-treated ells is prevented with a proteasome inhibitor, the resulting aumulation of ours mostly in the nuleus. In the nuleus, Smad4 omplexes, in assoiation with DNA-binding ofators, bind to speifi enhaner elements on target genes 5. In assoiation with the Forkhead family member Fast2, ativated omplexes an bind to the ativin-response element (ARE) from the Mix2 gene To determine whether proteasome funtion regulates the amount of available for binding to a DNA target site, we determined the effet of lataystin on ARE binding by omplexes from treated ells. We transfeted COS-1 ells with Fast2 and inubated them with together with or without lataystin. ARE-binding omplexes were olleted from the ell lysates using a biotinylated ARE oligonuleotide 16, and this preipitate was probed with anti- antibodies. Inubation with led to the formation of omplexes that bound to the DNA probe, and addition of lataystin markedly inreased the amount of these omplexes (Fig. 3b). Calpain inhibitor II, whih does not prevent Smad-P degradation (see above), had no effet on the amount of ARE omplexes (Fig. 3b), and the expression levels of Fast2 were not hanged signifiantly. These results show that -ativated averted from proteasome-mediated degradation persists in the nuleus in an ative state. Proteasome-dependent degradation therefore limits the signalling funtion of in the pathway. -indued proteasome-dependent loss of. To further analyse the effet of in stability, we transfeted ells with vetors enoding onstruts of interest. COS-1 ells expressing an epitope-tagged were inubated with and lataystin individually or in ombination for 4 h. Immunopre- NATURE CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om 1999 Mamillan Magazines Ltd 473

3 a MG-132 b MG-132 (h) a (h) HaCat Lata Chase time (h) Lata + + +MG S- 35 S- 35 S- 35 S- (Perentage of time 0) Chase time (h) + +MG COS-1 Lata Figure 2 Proteasome inhibitors protet from a -indued depletion. a, COS-1 ells were either left untreated or treated with a proteasome inhibitor (MG-132 or lataystin) and/or for 15 h. The lysates were analysed by immunoblotting using a /3-speifi antibody; the band orresponding to endogenous is indiated by an arrow in a, b. b, Prolonging MG-132 treatment enhanes protetion from the -indued deline in amounts. COS-1 ells were treated with for 10 h, and MG-132 was added for the indiated times before the end of ligand treatment., dereases the metaboli stability of. COS-1 ells inubated with the indiated additions were subjeted to a pulse hase metaboli labelling (with [ 35 S]methionine/ ysteine), and SMADs were immunopreipitated with an anti-/3 antibody. The speifiity of immunopreipitation was onfirmed by using preimmune serum as well as protein A Sepharose beads alone (data not shown). The lower band orresponds to Smad3. The signals were quantified and the values relative to the value at time 0 were plotted as a funtion of hase time. +Lata b DNAP (ARE) CII My-Fast Figure 3 Proteasome inhibition enhanes -indued nulear aumulation of endogenous and transriptional omplex assembly. a, -indued nulear loalization of /3 peaks around 1 h after treatment and delines thereafter. In the presene of lataystin, robust nulear loalization persists. Before fixation, ells were inubated with for the indiated lengths of time. Some ultures reeived lataystin alone for 3.5 h. Endogenous and Smad3 were then visualized by anti-/3 immunofluoresene. b, dependent reruitment of endogenous to the ativin/-responsive element (ARE) is markedly enhaned by o-treatment with lataystin. COS-1 ells were transfeted with a My-epitope-tagged Fast2 vetor and, as indiated, treated with alone or together with lataystin or a negative ontrol protease inhibitor (alpain inhibitor II, or CII) for 5 h. ARE-binding omplexes in the ell lysates were analysed by ARE-oligonuleotide-mediated preipitation (DNA-affinity preipitation, or DNAP) followed by anti-/3 immunoblotting of the preipitates (top). The expression level of My Fast2 was monitored by anti-my immunoblotting of total lysates (bottom). ipitation of the exogenous followed by immunoblotting with anti--p antibodies showed that lataystin inreased the -dependent aumulation of exogenous -P (Fig. 4a). and/or lataystin treatments did not alter signifiantly the levels of exogenous in COS-1 ells, perhaps as a result of overexpression of. We also studied the response of the onstrut ( ), whih inludes the MH2 domain and a small portion of the linker region. This and other SMAD onstruts laking the MH1 domain are onstitutively ative beause they an move into the nuleus and form transriptional omplexes regardless of whether or not they undergo reeptor-mediated phosphorylation Interestingly, lataystin also inreased the level of reeptor-phosphorylated ( ) after a prolonged treatment (6 h), as determined by anti--p immunoblotting (Fig. 4b). This indiates that the ativation-dependent signal for degradation may reside within this region of. Moreover, when exogenous was expressed at a relatively low level in the mink lung epithelial ell line derivative R-1B/L17, whih laks TβR- I, its metaboli stability was dereased upon oexpression of an ativated TβR-I (Fig. 4). Involvement of the ubiquitination pathway in -P degradation. Conjugation of ubiquitin to protein substrates requires ubiquitin-onjugating enzymes (designated Ubs), also known as E2 enzymes. E2s transfer ativated ubiquitin from an E1 ubiquitinativating enzyme to the substrate, either diretly or through E3 ubiquitin ligases to whih the substrate protein is speifially bound 20,21. To establish the involvement of this enzymati system in Mamillan Magazines NATURE Ltd CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om

4 a HA- b Flag-( ) IP: anti-ha WB: anti--p -P IP: anti-flag WB: anti--p ( )-P WB: anti-ha HA- F-( ) Chase time (h) S- (Perentage of time 0) Chase time (h) d DN E2 His-Construts IP: anti-ha WB: anti--p -P WB: anti-ha HA- HA- UbH 5b UbH Ub3 E2-20K WB: anti-his His-DN E2s Figure 4 indues proteasome-dependent loss of involving the ubiquitination pathway. a, Top, the level of reeptor-phosphorylated exogenous is enhaned by lataystin treatment. Bottom, the haemagglutinin (HA) onstrut was transfeted into COS-1 ells and expressed at a similar level in all samples. IP, immunopreipitation; WB, western blot. b, Top, N-terminallytrunated (( )) is phosphorylated in response to, and aumulation of the C-terminally-phosphorylated form is inreased by lataystin. Bottom, the Flag ( ) (F-( )) onstrut was transfeted into COS-1 ells and expressed at a similar level in all samples., TGFβ signalling dereases the metaboli stability of exogenous. -reeptordefetive R-1B/L17 ells were tranfeted with HA- alone ( ) or together with a vetor enoding ativated reeptor (+), and subjeted to a pulse hase metaboli labelling. HA was immunopreipitated and subjeted to gel eletrophoresis, and the autoradiographi signal was quantified. d, Expression of speifi dominant-negative E2 onstruts preferentially bloks the proteasomedependent degradation of -P. treatment served as a ontrol in its ability to blok the degradation of -P after stimulation. Similar expression of the HA and hexahistidine-tagged dominant-negative (DN) E2 onstruts is shown in the bottom panels. the degradation of ativated, we sought evidene for a role of known E2 enzymes in this proess. We transiently transfeted COS-1 ells with haemagglutinin-epitope-tagged alone or together with dominant-negative, atalyti-site-mutant forms of various E2 enzymes (Fig. 4d). After treatment with, the ells were subjeted to immunopreipitation with an anti-haemagglutinin antibody, and the immunopreipitates were probed with an anti--p antibody. Catalyti-site mutants of UbH5b and its lose homologue, UbH5, were as effetive as lataystin at allowing -P aumulation (Fig. 4d). A atalyti-site Ub3 mutant was less effetive than the UbH5 mutants. In ontrast, a atalytisite mutant of E2-20K, an E2 with a narrow substrate espeifiity, had little effet on the aumulation of -P. These results indiate a role of the ubiquitination system, ating through UbH5b/ and/or other, unharaterized, E2 omponents, in the indued ubiquitination of. Nulear loalization leads to ubiquitination. To determine diretly whether ubiquitin is onjugated to endogenous in vivo, we subjeted denatured lysates from HaCaT ells to anti- immunopreipitation and then analysed the immunopreipitates by immunoblotting using an anti-ubiquitin antibody. addition indued the aumulation of high-moleular-mass speies reognized by the anti-ubiquitin antibody (Fig. 5a), indiating that may be multiply ubiquitinated in response to TGFβ. To determine the requirements for this modifiation, we transfeted ells with vetors enoding wild-type or mutant forms of and ubiquitin in various ombinations. COS-1 ells were transfeted with vetors enoding Flag-epitope-tagged and hexahistidine-tagged ubiquitin and lysed under strongly denaturing onditions to inhibit isopeptidases. Ubiquitin-onjugated proteins were isolated by obalt-helate hromatography through the hexahistidine tag and analysed by anti-flag immunoblotting to reveal ubiquitin-onjugated produts (Fig. 5b). We deteted multi-ubiquitinated produts in -treated ells but not in ells left untreated or treated with lataystin alone (Fig. 5b). Furthermore, the ombination of and lataystin led to a marked aumulation of these high-moleular-mass speies. A mutant form of ubiquitin (lysine 48 arginine (K48R)) that has a hain-terminating, dominant-negative effet 22 signifiantly redued the amount of high-moleular-mass forms, even in the presene of lataystin (Fig. 5b). We also determined the effet of mutating s C-terminal sites for phosphorylation by the reeptor from serine to alanine. This mutant ((3SA)) is not phosphorylated in response to 23,24 and was not ubiquitinated in response to, even in the presene of lataystin (Fig. 5b). Colletively, these results show that -reeptormediated phosphorylation leads to ubiquitination. These results were onsistent with two possibilities, namely, that is ubiquitinated beause the reeptor-phosphorylated region onstitutes a signal for ubiquitin onjugation or beause the NATURE CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om 1999 Mamillan Magazines Ltd 475

5 a + IP: anti-/3 WB: anti-ubiquitin b His-Ub Flag- WT WT DN WT 3S->A WT Ub n - Ub n -/3 BSA F- Flag- FL NLS-FL MH2 NLS-MH2 C N C N C N C N d Flag- FL NLS-FL MH2 NLS-MH Ub - n Figure 5 Nulear loalization of leads to ubiquitination. a, Endogenous and Smad3 are onjugated to multiple ubiquitin moleules. The immunoblot shows that ubiquitin (Ub n )-onjugated endogenous speies were immunopreipitated with the anti-/3 antibody from HaCaT ells in a dependent manner. b, stimulates the formation of multi-ubiquitinated, the detetion of whih is further enhaned by o-inubation with lataystin. Mutating s C-terminal three serines to alanines, as well as transfeting a hainterminating (dominant-negative, DN) mutant of ubiquitin, dramatially redues multi-ubiquitination. COS-1 ells were transfeted with the indiated onstruts, treated with and/or lataystin, and lysed under strongly denaturing onditions. His-tagged ubiquitinated speies were olleted with obalt helate resins. Multi-ubiquitinated was visualized by anti-flag immunoblotting. Flag expression levels were assessed by diret anti-flag immunoblotting of ell lysates (bottom)., The partition of produts into ytoplasmi (C) and nulear (N) frations was analysed in COS-1 ells transfeted with vetors enoding fulllength (FL), full-length fused to a nulear-loalization sequene (NLS FL), the MH2 domain alone (MH2) or this domain fused to the nulearloalization sequene (NLS MH2). Arrow indiates. d, Nulear loalization auses multiple ubiquitination of. COS-1 ells were transfeted with the indiated onstruts and treated with lataystin and as indiated; multiubiquitinated produts were analysed as desribed above. -P Nuleus TβR-II P P TβR-I Smad4 Fast 2 (Ub) n P Figure 6 A model of ubiquitin-dependent degradation of at the end of the signalling pathway. After being phosphorylated by TβR-I, moving to the nuleus, and assembling into transriptional omplexes with Smad4 and DNAbinding ofators (suh as Fast2), is multiply ubiquitinated and so is seletively targeted for proteasome-dependent degradation. Cirled P, phosphorylation. reeptor-ativated aumulates in the nuleus. To distinguish between these two possibilities, we sought to unouple the nulear loalization and the -reeptor-mediated stimulation of. We fused the SV40 large T antigen nulear-loalization signal (NLS) to the N terminus of full-length (NLS FL) or to the N terminus of a MH2 domain onstrut (NLS MH2). Whereas wild-type was enrihed in the ytosoli fration, the NLS FL onstrut was enrihed in the nulear fration (Fig. 5). Wild-type beame ubiquitinated in response to, whereas NLS FL beame ubiquitinated independently of addition (Fig. 5d), indiating that ubiquitination may be a onsequene of its loalization in the nuleus. We obtained further evidene in support of this proposal by using the MH2-domain onstruts. The unmodified MH2 produt frationated almost evenly between the ytoplasmi and nulear frations (Fig. 5). The MH2 onstrut is partly nulear in the absene of, and its nulear loalization is not inreased by (ref. 17 and our unpublished observations), in ontrast to its phosphorylation (Fig. 4b). This onstrut is onstitutively ubiquitinated, and does not inrease this level of ubiquitination (Fig. 5d). Furthermore, attahing an NLS to the MH2 onstrut inreases both its nulear loalization (Fig. 5) and its ubiquitination (Fig. 5d). Disussion SMAD ativation by reeptor-mediated phosphorylation is a entral event in signal transdution. A omplex set of regulatory mehanisms have been identified that modulate the initiation of Mamillan Magazines NATURE Ltd CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om

6 this proess, the aumulation of ativated SMADs in the nuleus, or the assembly of SMAD transriptional omplexes 5,7. However, terminating SMAD signalling in a timely fashion is probably as important as initiating this proess. Here we have provided evidene that, in the pathway, reeptor-ativated is irreversibly eliminated by seletive ubiquitination and proteasomemediated degradation (Fig. 6). Our results show that the aumulation of reeptor-phosphorylated in various ell types is a transient proess that peaks 1 h after addition (at high onentrations) and delines thereafter. The proteasome inhibitors lataystin and MG- 132 prevent this deline. In ontrast, in the basal state appears to turn over by a different mehansism, beause proteasome inhibitors do not inrease the basal level of in the ell lines examined. Persistent stimulation with takes a toll in the steady-state level of, and proteasome-dependent degradation is an important fator in this loss. The metaboli stability of is dereased by stimulation, whereas mrna levels remain unaffeted. Moreover, stimulation with indues ubiquitination. The loss of -P an be prevented not only with proteasome inhibitors but also with dominant-negative mutant forms of ubiquitin-onjugating enzymes. Our results indiate that the pool of ativated in a treated ell may be subjet to onstant ulling by the ubiquitin/proteasome pathway. The generation of multi-ubiquitinated in response to requires reeptor-mediated phosphorylation of the C-terminal serines, whih auses aumulation of in the nuleus. However, our results suggest that reeptor-mediated phosphorylation is neessary for ubiquitination to the extent that it is required for nulear aumulation of. Using Smad onstruts that onstitutively aumulate in the nuleus, either beause they lak the MH1 and linker domain or beause they are fused to a onstitutive nulear-loalization signal, we have shown that ubiquitination is a result of aumulation in the nuleus and not a result of phosphorylation per se. An MH2-domain onstrut is partly loalized in the nuleus and ubiquitinated in a onstitutive manner. Its phosphorylation in response to, whih presumably affets the MH2 population loated in the ytoplasm at the moment of stimulation, does not inrease the nulear aumulation or ubiquitination of this produt. This provides more evidene that ubiquitination of an be dissoiated from its phosphorylation by the reeptor. theless, it is still oneivable that a reeptor-phosphorylated onstitutes a better substrate for the ubiquitin/proteasome system. Proteasome-mediated elimination has a marked impat on the pool of ativated that is available in the nuleus for formation of transriptional omplexes. In the presene of proteasome inhibitor, the ativated stays in the nuleus and is apable of assembling transriptional omplexes. This result also indiates that proteasomal degradation of may not require the export of ativated from the nuleus. This idea is also supported by the observation that leptomyin B, an inhibitor of Crm1-dependent nulear export, does not protet -P from degradation (our unpublished observations). The speifiity of substrate reognition by E2 enzymes in the ubiquitin system is ahieved by different E3 enzymes apable of interating with speifi substrates 20,21. Our results obtained using dominant-negative versions of various E2 enzymes raise the possibility that UbH5b/ and, to a lesser extent, Ub3 are involved in the degradation of ativated. UbH5-related enzymes funtion in onert with HECT-domain-ontaining E3 proteins but may also funtion in the ontext of the SCF E3 omplex, as in the ase of IκB-α degradation 20,25,26. Ub3/Cd34 typially funtions in onert with the SCF omplex and atalyses substrate ubiquitination with the assistane of F-box proteins 27. The HECT-domain protein Smurf1 mediates the ubiquitination of Smad1 (ref. 28), whih is a mediator of bone morphogeneti protein (BMP) signals 5,6. Smurf1 may ontrol ell ompetene to respond to BMPs by ontrolling the basal level of Smad1, but does not funtion downstream of ativated SMADs to turn off BMP signals 28. Smurf1 reognizes Smad1 independently of reeptor-mediated phosphorylation, and does not interat with (ref. 28). It will be important to identify E3 proteins that speifially reognize ativated in the nuleus. Many mediators of reeptor kinase signals are inativated by phosphatase ation and reyled for subsequent rounds of signal transdution. Our evidene does not exlude the possibility that phosphatase ation plays a part in the reversal of phosphorylation. However, the role of ubiquitin-dependent degradation unovered here is interesting beause this mehanism is often involved in proesses that must undergo swift, irreversible transitions. and related fators regulate very dynami physiologial proesses. Ubiquitin-dependent degradation of their ativated mediators may ensure a thorough elimination of their signal. Methods Transfetion, immunopreipitation and immunoblotting. HaCaT, COS-1, HeLa and HepG2 ells were either left untreated or treated with 1 (200 pm) and/ or the indiated agents: 10 µm lataystin, 50 µm MG-132, 100 nm okadai aid (all from Calbiohem), 1 mm sodium vanadate, 10 µm pepstatin, 10 µm leupeptin or 50 µm alpain inhibitor II (all from Sigma) for the indiated lengths of time under redued serum onditions (0.2% FBS). The ells were then sraped in PBS, pelleted, resuspended in Laemmli sample buffer, and needle-sheared; the resultant lysates were subjeted to SDS polyarylamide gel eletrophoresis, transferred to poly(vinylidene) fluoride (PVDF) membranes (Millipore), and blotted for using an affinity-purified anti-/ 3 antibody 29 or for -P using a phospho-(s465/467)-speifi antibody (Upstate Biotehnology). Proteins were deteted using enhaned hemiluminesene (ECL; Amersham). onstruts were expressed from either pcmv5 or pcs2 and N-terminally Flag- or haemagglutinin-tagged. The pcmv-β-galatosidase expression vetor was used to keep the amount of transfeted DNA onstant (ontrol DNA). For immunopreipitation, COS-1 ells were transfeted by the DEAE dextran method as desribed h after transfetion, ells were either left untreated or treated with and/or proteasome inhibitors for 4 6 h under redued serum and lysed in TNE buffer (10 mm Tris-HCl, ph 7.8, 150 mm NaCl, 1 mm EDTA, 1% NP-40) supplemented with both protease inhibitors and phosphatase inhibitors. The lysates were then subjeted to immunopreipitation with anti-flag M2 (Sigma) or anti-haemagglutinin 12CA5 (Boehringer Mannheim) monolonal antibodies. To determine the subellular loalization of wild-type and NLS onstruts, we transfeted COS-1 ells by the DEAE dextran method; h after transfetion, ytoplasmi and nulear frations were prepared using NE-PER (Piere) reagents aording to the manufaturer s instrutions. After SDS PAGE size-frationation, the immunopreipitates were transferred onto PVDF membranes and probed with the indiated antibodies. To show o-immunopreipitation of ubiquitin with endogenous /3, we treated 25-m 2 plates of HaCaT ells as indiated for 6 h. The ells were subsequently lysed in 3% SDS-Tris buffer (3% SDS, 20 mm Tris-HCl, ph 6.8, 150 mm NaCl) supplemented with ubiquitin aldehyde (0.25 µg ml 1 ) (Calbiohem) and protease inhibitors, and the lysates were needle-sheared and diluted tenfold into PBS ontaining 0.5% NP-40 and protease inhibitors. The lysates were then subjeted to anti-/3 immunopreipitation. Co-immunopreipitates were subjeted to SDS PAGE, transferred onto PVDF membranes, western-blotted using an anti-ubiquitin antibody (Santa Cruz Biotehnology) and visualized with ECL Plus (Amersham). Pulse hase experiments. To study the metaboli stability of endogenous proteins, we starved COS-1 ells for 1 h in methionine/ysteine-defiient medium, and then pulsed the ells for 15 min with 500 µci ml 1 [ 35 S]methionine/ysteine alone or together with or plus MG-132. The ells were washed extensively with Dulbeo s modified Eagle s medium (DMEM) and inubated for the indiated lengths of time in DMEM/0.2% fetal alf serum alone or together with or plus MG-132. The ells were lysed in TNE buffer ontaining protease inhibitors, and the lysates pre-leared with protein-a Sepharose beads and subjeted to anti-/3 immunopreipitation. The immunopreipitates were subjeted to SDS PAGE, fixed and enhaned with 1 M sodium saliylate, and visualized by autoradiography. All experiments in whih we examined the metaboli stability of exogenous proteins were essentially the same, exept that plasmids expressing Flag-tagged were transfeted into R-1B/L17 ells with a onstrut expressing either β-galatosidase or an ativated form of the type I reeptor (TβR-I(T204D)) 30 and that immunopreipitation was arried out with an anti-flag antibody. Metal-affinity preipitation. COS-1 ells were transfeted using lipofetamine (Gibo) aording to the manufaturer s instrutions. 30 h after transfetion and before lysis in GTN buffer (6 M guanidinium-hcl, 20 mm Tris-HCl, ph 8.0, 200 mm NaCl, 20 mm imidazole, 0.1% Triton-X100), the ells were either left untreated or treated with and/or proteasome inhibitors for 4 6 h. Histidine-tagged ubiquitin olleted from the lysates was purified by obalt helate resins (Clonteh) previously inubated with 5% bovine serum albumin. The preipitates were then washed in GTN without guanidinium-hcl, resolved on SDS PAGE, and subjeted to immunoblotting using anti-flag antibody. Expression ontrols were derived from parallel transfetions and treatments followed by lysis in TNE buffer rather than GTN buffer. DNA-affinity preipitation. COS-1 ells transfeted with Fast2 by lipofetamine and treated as indiated for 4 6 h were resuspended NATURE CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om 1999 Mamillan Magazines Ltd 477

7 in a solution ontaining 50 mm HEPES, ph 7.4, 50 mm NaCl, 0.1% Tween-20 and 10% glyerol, supplemented with protease and phosphatase inhibitors and lysed by soniation. The lysates were preleared with ImmunoPure streptavidin agarose (Piere), inubated with 200 ng biotinylated doublestranded ARE oligonuleotides and 2 µg poly(di-dc).poly(di-dc). Endogenous bound to the ARE probe was then aptured by streptavidin agarose, size-frationated on SDS PAGE, transferred onto PVDF membrane, blotted with the anti-/3 antibody, and visualized using ECL Plus. The sequene of the biotinylated ARE oligonuleotide has been desribed 16. Immunofluoresene. HaCat, COS-1, HeLa and EpH4 ells were treated as indiated and fixed. Endogenous Smad 2/3 proteins were visualized with affinity-purified anti-/3 antibodies, biotin-onjugated anti-rabbit antibody, and fluoresein-isothioyante-onjugated streptavidin (Jakson ImmunoResearh). All slides were ounterstained with 4,6-diamidino-2-phenylindole (DAPI) to visualize nulei (data not shown). RECEIVED 16 JULY 1999; REVISED SEPTEMBER 1999; ACCEPTED 4 OCTOBER 1999; PUBLISHED 28 OCTOBER Roberts, A. B. & Sporn, M. B. in Peptide Growth Fators and their Reeptors (eds Sporn, M. B. & Roberts, A. B.) (Springer, Heidelberg, 1990). 2. Massagué, J. The transforming growth fator-β family. Annu. Rev. Cell Biol. 6, (1990). 3. Hogan, B. L. M. Bone morphogeneti proteins: multifuntional regulators of vertebrate development. Genes Dev. 10, (1996). 4. Whitman, M. Smads and early developmental signaling by the TGFβ superfamily. Genes Dev. 12, (1998). 5. Massagué, J. TGFβ signal transdution. Annu. Rev. Biohem. 67, (1998). 6. Heldin, C.-H., Miyazono, K. & ten Dijke, P. signalling from ell membrane to nuleus through SMAD proteins. Nature 390, (1997). 7. Zhang, Y. & Derynk, R. Regulation of Smad signaling by protein assoiations and signaling rosstalk. Trends Cell Biol. 9, (1999). 8. Abdollah, S. et al. TβRI phosphorylation of on Ser465 and Ser467 is required for - Smad4 omplex formation and signaling. J. Biol. Chem. 272, (1997). 9. Souhelnytskyi, S. et al. Phosphorylation of Ser465 and Ser467 in the C terminus of mediates interation with Smad4 and is required for transforming growth fator-β signaling. J. Biol. Chem. 272, (1997). 10. Rok, K. L. et al. Inhibitors of the proteasome blok the degradation of most ell proteins and the generation of peptides presented on MHC lass I moleules. Cell 78, (1994). 11. Fenteany, G. et al. Inhibition of proteasome ativities and subunit-speifi amino-terminal threonine modifiation by lataystin. Siene 268, (1995). 12. Chen, X. et al. Smad4 and FAST-1 in the assembly of ativin-responsive fator. Nature 389, (1997). 13. Liu, F., Pouponnot, C. & Massagué, J. Dual role of the Smad4/DPC4 tumor suppressor in TGFβinduible transriptional responses. Genes Dev. 11, (1997). 14. Labbé, E., Silvestri, C., Hoodless, P. A., Wrana, J. L. & Attisano, L. and Smad3 positively and negatively regulate TGFβ-dependent transription through the forkhead DNA-binding protein FAST2. Mol. Cell 2, (1998). 15. Liu, B., Dou, C., Prabhu, L. & Lai, E. Fast2 is a mammalian winged helix protein that mediates TGFβ signals. Mol. Cell Biol. 19, (1999). 16. Wotton, D., Lo, R. S., Lee, S. & Massagué, J. A Smad transriptional orepressor. Cell 97, (1999). 17. Baker, J. & Harland, R. M. A novel mesoderm induer, mmadr-2, funtions in the ativin signal transdution pathway. Genes Dev. 10, (1996). 18. Liu, F. et al. A human Mad protein ating as a BMP-regulated transriptional ativator. Nature 381, (1996). 19. Hata, A., Lo, R. S., Wotton, D., Lagna, M. & Massagué, J. Mutations inreasing autoinhibition inativate the tumour suppressors and Smad4. Nature 388, (1997). 20. Ciehanover, A. The ubiquitin-proteasome pathway: on protein death and ell life. EMBO J. 17, (1998). 21. Laney, J. D. & Hohstrasser, M. Substrate targeting in the ubiquitin system. Cell 97, (1999). 22. Chau, V. et al. A multiubiquitin hain is onfined to speifi lysine in a targeted short-lived protein. Siene 243, (1989). 23. Maias-Silva, M. et al. MADR2 is a substrate of the TGFβ reeptor and phosphorylation is required for nulear aumulation and signaling. Cell 87, (1996). 24. Kretzshmar, M., Liu, F., Hata, A., Doody, J. & Massagué, J. The mediator Smad1 is diretly phosphorylated and funtionally ativated by the BMP reeptor kinase. Genes Dev. 11, (1997). 25. Spener, E., Jiang, J. and Chen, Z. J. Signal-indued ubiquitination of IκBα by the F-box protein Slimb/β-TrCP. Genes Dev. 13, (1999). 26. Gonen, H. et al. Identifiation of the ubiquitin arrier proteins, E2s, involved in signal-indued onjugation of subsequent degradation of IκBα. J. Biol. Chem. 274, (1999) 27. Maniatis, T. A ubiquitin ligase omplex essential for the NF-κB, Wnt/Wingless, and Hedgehog signaling pathways. Genes Dev. 13, (1999). 28. Zhu, H. et al. A Smad ubiquitin ligase targets the BMP pathway and affets embryoni pattern formation. Nature 400, (1999). 29. Kretzshmar, M., Doody, J., Timokhina, I. & Massagué, J. A mehanism of repression of TGFβ/Smad signaling by ongeni ras. Genes Dev. 13, (1999). 30. Lo, R. S., Chen, Y. G., Shi, Y. G., Pavletih, N. & Massagué, J. The L3 loop: a strutural motif determining speifi interations between SMAD proteins and reeptors. EMBO J 17, (1998). ACKNOWLEDGEMENTS We thank R. R. Kopito for ubiquitin expression onstruts; K. Iwai and A. Ciehanover for E2 onstruts; and E. Lai for Fast2. R.S.L. thanks Y-G. Chen, J. Seoane, C. Pouponnot, J. Doody, D. Wotton and S. H. Roan for help. This work was supported by an NIH grant to J.M. (CA34619) and to the Memorial Sloan- Kettering Caner Center and by an NIH Medial Sientist Training Program (MSTP) grant to R.S.L. J.M. is an investigator of the Howard Hughes Medial Institute. Correspondene and requests for materials should be addressed to J.M Mamillan Magazines NATURE Ltd CELL BIOLOGY VOL 1 DECEMBER 1999 ellbio.nature.om