Distinct Oligomeric States of SMAD Proteins in the Transforming Growth Factor-β Pathway

Abstract

Protein interactions are critical for the function of SMADs as mediators of transforming growth factor-β (TGF-β) signals. TGF-β receptor phosphorylation of SMAD2 or SMAD3 causes their association with SMAD4 and accumulation in the nucleus where the SMAD complex binds cofactors that determine the choice of target genes. We provide evidence that in the basal state, SMADs 2, 3, and 4 form separate, strikingly different complexes. SMAD2 is found mostly as monomer, whereas the closely related SMAD3 exists in multiple oligomeric states. This difference is due to a unique structural element in the MH1 domain of SMAD2 that inhibits protein-protein interactions in the basal state. In contrast to SMAD2 and SMAD3, SMAD4 in the basal state is found mostly as a homo-oligomer, most likely a trimer. Upon cell stimulation with TGF-β, SMAD proteins become engaged in a multitude of complexes ranging in size from SMAD2–SMAD4 heterodimers to assemblies of >650 kDa. The latter display the highest DNA binding affinity for the TGF-β-response elements of JUNB and collagen 7. These observations, all validated with endogenous SMAD proteins, modify previous models regarding the assembly and activity of SMAD complexes in the TGF-β pathway. Protein interactions are critical for the function of SMADs as mediators of transforming growth factor-β (TGF-β) signals. TGF-β receptor phosphorylation of SMAD2 or SMAD3 causes their association with SMAD4 and accumulation in the nucleus where the SMAD complex binds cofactors that determine the choice of target genes. We provide evidence that in the basal state, SMADs 2, 3, and 4 form separate, strikingly different complexes. SMAD2 is found mostly as monomer, whereas the closely related SMAD3 exists in multiple oligomeric states. This difference is due to a unique structural element in the MH1 domain of SMAD2 that inhibits protein-protein interactions in the basal state. In contrast to SMAD2 and SMAD3, SMAD4 in the basal state is found mostly as a homo-oligomer, most likely a trimer. Upon cell stimulation with TGF-β, SMAD proteins become engaged in a multitude of complexes ranging in size from SMAD2–SMAD4 heterodimers to assemblies of >650 kDa. The latter display the highest DNA binding affinity for the TGF-β-response elements of JUNB and collagen 7. These observations, all validated with endogenous SMAD proteins, modify previous models regarding the assembly and activity of SMAD complexes in the TGF-β pathway. transforming growth factor-β hemagglutinin polyacrylamide gel electrophoresis Transforming growth factor-β (TGF-β)1 family of secretory polypeptides regulate various important cellular processes such as differentiation, adhesion, tissue repair, and apoptosis of many different cell types. Included in this family are TGF-β, the bone morphogenetic proteins, the activins and the inhibins, the nodals, and some other related factors. Signaling by TGF-β is initiated upon its binding to two cell-surface receptors termed type I (TβRI) and type II (TβRII). Both receptors are serine/threonine kinases, and binding by TGF-β results in phosphorylation of TβRI by TβRII. The only substrates known to date for phosphorylated TβRI are the SMAD proteins. Upon phosphorylation of conserved serine residues at their extreme C terminus, the TGF-β-activated SMADs, SMAD2 and SMAD3 (or SMAD1 in the case of bone morphogenetic proteins), translocate into the nucleus. En route to the nucleus they associate with SMAD4 which is a shared partner (1Massague J. Annu. Rev. Biochem. 1998; 67: 753-791Crossref PubMed Scopus (3975) Google Scholar). Both groups of SMAD proteins share a common structure with a conserved N-terminal (or MH1) and C-terminal (or MH2) domain and a more variable linker region that connects the two (1Massague J. Annu. Rev. Biochem. 1998; 67: 753-791Crossref PubMed Scopus (3975) Google Scholar). Upon entry into the nucleus, the receptor-activated SMAD complex can interact with a number of partner proteins that are cell type-specific and jointly with the associated SMADs provide DNA binding activity specific to particular downstream target genes. The activated SMAD complexes then recruit the transcriptional coactivators CBP/p300 (2Janknecht R. Wells N.J. Hunter T. Genes Dev. 1998; 12: 2114-2119Crossref PubMed Scopus (434) Google Scholar, 3Shen X. Hu P.P. Liberati N.T. Datto M.B. Frederick J.P. Wang X.F. Mol. Biol. Cell. 1998; 9: 3309-3319Crossref PubMed Scopus (184) Google Scholar, 4Feng X.-H. Zhang Y. Wu R.-Y. Derynck R. Genes Dev. 1998; 12: 2153-2163Crossref PubMed Scopus (448) Google Scholar, 5Pouponnot C. Jayaraman L. Massague J. J. Biol. Chem. 1998; 273: 22865-22868Abstract Full Text Full Text PDF PubMed Scopus (290) Google Scholar) or the corepressors TGIF, Ski or Sno (6Wotton D. Lo R.S. Lee S. Massague J. Cell. 1999; 97: 29-39Abstract Full Text Full Text PDF PubMed Scopus (479) Google Scholar, 7Stroschein S.L. Wang W. Zhou Q. Lou K. Science. 1999; 286: 771-774Crossref PubMed Scopus (436) Google Scholar, 8Sun Y. Liu X. Eaton E.N. Lane W.S. Lodish H.F. Weinberg R.A. Mol. Cell. 1999; 4: 499-50911Abstract Full Text Full Text PDF PubMed Scopus (224) Google Scholar). The set of gene responses set in motion by TGF-β in a particular cell type therefore depends on the interaction between the SMADs and various partner proteins that dictate DNA binding specificity and transcriptional activity of the resulting complexes. Whereas the list of proteins that interact with the SMADs is increasing, the oligomeric nature of these complexes or their composition has remained a matter of controversy. There is biochemical evidence to suggest that SMAD2, -3, and -4 proteins are capable of associating both with themselves as well as with each other, in the basal and in the stimulated state. This association occurs chiefly, although not exclusively, through interactions of the MH2 domains of the proteins (9Hata A. Lo R.S. Wotton D. Lagna G. Massague J. Nature. 1997; 388: 82-87Crossref PubMed Scopus (294) Google Scholar, 10Wu R.-Y. Zhang Y. Feng X.-H. Derynck R. Mol. Cell. Biol. 1997; 17: 2521-2528Crossref PubMed Scopus (186) Google Scholar). Determination of the three-dimensional crystal structure of the MH2 domain of SMAD4 identified a homotrimer (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar). Interestingly, the monomer interfaces forming the trimer contain a majority of conserved amino acids. Missense mutations of several of these interface residues have been identified in pancreatic or colon cancer-derived alleles of SMAD2 and SMAD4, providing strong genetic evidence for a role of this interface in vivo. More recently, however, the notion that SMADs may exist as oligomers in the basal state was challenged by the observation that overexpressed SMADs 2, 3, and 4 appeared as monomers (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google Scholar). These results were obtained utilizing transfection of tagged SMAD constructs and therefore may or may not accurately reflect the state of endogenous SMADs. To address this controversy, we have undertaken an analysis of endogenous SMAD2 and -4 proteins so as to better understand their oligomeric composition in the absence of ligand and any changes that occur upon its addition. HaCaT and COS-1 cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, penicillin/streptomycin, and fungizone. For transient transfections COS-1 cells were seeded in 10-cm dishes and transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. In cases where low levels of proteins were expressed, cells were transfected using DEAE-dextran. FLAG-tagged SMAD2 Δ exon 3 was a kind gift from M. Kato. FLAG SMAD3/2 and SMAD2/3 were constructed by polymerase chain reaction amplification of MH1 domain of SMAD3 and linker + MH2 of SMAD2 or vice versa and cloning into the pCS2 expression vector. FLAG SMAD2 Δ 12aa containing a 12-amino acid deletion of residues 88–99 of SMAD2 was constructed by polymerase chain reaction amplification using appropriate primers and cloned into pCS2. Anti-SMAD2 antibodies were raised in rabbits by immunization with the recombinant linker region of human SMAD2 (residues 183–273) (13Kretzschmar M. Doody J. Timokhina I. Massague J. Genes Dev. 1999; 13: 804-816Crossref PubMed Scopus (849) Google Scholar). They were affinity-purified with immobilized SMAD2 prior to use and cross-react with SMAD3. Anti-SMAD3-specific antibodies (14Ulloa L. Massague J. Nature. 1999; 397: 710-713Crossref PubMed Scopus (717) Google Scholar) were raised against synthetic peptides that correspond to residues 192–211 of human SMAD3 (Zymed Laboratories Inc.). Anti-SMAD4-specific antibodies (15Calonge M.J. Massague J. J. Biol. Chem. 1999; 274: 33637-33643Abstract Full Text Full Text PDF PubMed Scopus (130) Google Scholar) were raised in rabbits by immunization with a recombinant human SMAD4 peptide spanning residues 40–333 and were affinity-purified prior to use. The anti-FLAG and anti-HA (12CA5) monoclonal antibodies were from Sigma and Roche Molecular Biochemicals, respectively. COS-1 cells (48 h post-transfection) or untransfected HaCaT cells (untreated or treated with 200 pmol TGF-β) were washed twice with phosphate-buffered saline, scraped, and resuspended in lysis buffer containing 20 mm Tris (pH 8.0), 150 mm NaCl, 0.5% Triton X-100, 1 mmdithiothreitol, and protease inhibitors. Whole cell extracts were prepared by rocking at 4 °C for 20 min and centrifugation at 10,000 × g for 10 min followed by centrifugation at 50,000 × g for 10 min to pellet cell debris. Cell supernatant was collected, and an aliquot (0.2 ml) was applied to a 200 gel with buffer containing mm Tris (pH 8.0), 200 mm NaCl, and mm The was at and were aliquot of each was on an to with the and using an according to manufacturer's instructions. from the were as in the and for min with and then with the and for an 3 h at 4 The were washed with lysis and the complexes were by in buffer and by were then to with the and as were as and They were then at 4 °C with containing element which the complexes were on and was on and as The of the for the are as and The of antibodies against the SMAD proteins to the oligomeric state of endogenous SMADs. Whole cell extracts from HaCaT a human cell in of TGF-β were to size on a 200 were to with affinity-purified or SMAD2 in to its T. S. M. M.B. L. Cell. Full Text Full Text PDF PubMed Scopus Google Scholar) 1 of SMAD2 in to a of kDa. The low of this not of its In contrast to SMAD2, SMAD4 from the in 1 that to a of from a SMAD4 monomer M. A. S. Science. PubMed Scopus Google Scholar). of SMAD4 as a in The of FLAG-tagged SMAD2 and SMAD4 from extracts of transfected COS-1 cells was to that of the endogenous proteins from HaCaT cells 1 that the size of SMAD4 not from an interaction with a endogenous To the SMAD4 an or as (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google a monomer we the of a on the of this form Cell prepared in lysis buffer containing were on a that was and in buffer containing SMAD2 in the as in low and 1 SMAD3 was these as a monomer the of this observation is SMAD4, on the other in with the in This that SMAD4 oligomers in the basal state is by a is that of to 1 not the of SMAD4 not that the interactions to the basal SMAD4 complex are of the of endogenous HaCaT cell extracts were prepared and in lysis buffer containing were as on an to and with a containing and cells were with FLAG-tagged and The and that both proteins were the results of an with the anti-FLAG The proteins were on an gel and with the anti-HA cells were as with and SMAD4 and prepared in buffer containing The two that both proteins transfected were The two the results of an reaction on cell with the anti-FLAG The proteins to were using the anti-FLAG or anti-HA as cells were transfected with FLAG and SMAD4 and cell prepared h Cell extracts were to with the anti-FLAG or the anti-HA followed by with of the two antibodies as by were and the results are as a of was on obtained upon of the To determine the basal SMAD4 was by the association of multiple SMAD4 we COS-1 cells with two of SMAD4 with an N-terminal FLAG and the other SMAD4 with a C-terminal Cell were on a 200 and groups of were and to anti-FLAG followed by anti-HA to determine the of such interactions were the of with cell extracts of evidence of SMAD4 association that was in SMAD4 into The difference between in the absence and of is not and in SMAD4 monomers not with is that the SMAD4 monomer in the of is and has a its state at low was important to the association that we between two or more SMAD4 was of the oligomeric state of the or a of associating SMAD4 To this COS-1 cells transfected with and were and were to followed by for the of the obtained that of the anti-FLAG or the anti-HA was for the the of was of the by which of transfected was by the anti-FLAG of was using the anti-HA which the of SMAD4, the anti-FLAG of SMAD4 a of SMAD4 with is with SMAD4 with the other that these the of SMAD4 that are as they not into complexes by SMAD4 containing the results therefore suggest that a of SMAD4 is in a complex with The size of the basal SMAD4 complexes was with a or a trimer. to SMAD4 in the using cells with SMAD4 containing different with results not the of the reaction with each of the of SMAD4 was not all to the of low of the or between at the of more SMAD4 in the this and the previous and genetic evidence that SMAD4 can form the endogenous SMAD4 SMAD3 has been to a monomer in the basal state overexpressed in COS-1 cells (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google Scholar). To this to endogenous SMAD3 as HaCaT cell extracts 200 were to using an that SMAD3 not SMAD3 was in many different most of the 3 however, was in to of kDa. not results were obtained with against SMAD2, which with SMAD3. In this SMAD3 can from SMAD2 by its 3 In the SMAD2 and SMAD4 in their 3 of SMAD3 in these using the was variable from to For SMAD3 was not in the in 1 at that This may due to the of SMAD3 several and the of SMAD3. as in endogenous SMAD3 as monomers in the of that endogenous SMAD3, SMAD2, is found in oligomers in the basal state. a of SMAD3 as a monomer, the majority of the SMAD3 is engaged in complexes of results with endogenous SMAD3 from has been with overexpressed SMAD3 (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google we to this To this we transfected COS-1 cells with a FLAG-tagged SMAD3. a of was the overexpressed in the cell from 200 mostly in the monomer 3 in with previous (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google Scholar). a of was in the transfections to a of the from 200 in to the with in the monomer 3 This complex was different from that of endogenous SMAD3 with and In both however, the resulting multiple SMAD3 complexes of and SMAD3 These results that in the basal state endogenous SMAD3 is engaged in multiple with and of the levels of SMAD3 by expression of the size and composition of these complexes. The structural difference between SMAD2 and SMAD3 is the of a acid in the N-terminal domain (or MH1 of SMAD2 Y. Feng X.-H. Wu R.-Y. Derynck R. Nature. PubMed Scopus Google Scholar). This is the of the SMAD2 form containing the is the form in most T. S. M. M.B. L. Cell. Full Text Full Text PDF PubMed Scopus Google Scholar). This is N-terminal to the conserved element that with DNA in SMADs (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar). In SMAD2 the of this with DNA (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar, J. K. S. A. Nature. 1997; 388: PubMed Scopus (448) Google Scholar, C. L. Mol. Cell. 1998; Full Text Full Text PDF PubMed Scopus Google Scholar). in to a for the different oligomeric state of SMADs and 3, we to on this structural We FLAG-tagged of SMAD2 with the MH1 domain by that of SMAD3 a deletion of the Δ exon K. D. T. S. M. Miyazono K. J. Biol. Chem. 1999; 274: Full Text Full Text PDF PubMed Scopus Google Scholar) or a deletion Δ at low levels in COS-1 cells the of all these were to of with 3 a FLAG-tagged of SMAD3 with the MH1 domain by that of SMAD2 mostly as a monomer the 4 with These results suggest that the of SMAD2 in the basal state is an of the in its MH1 In to interaction of SMAD2 with DNA in the activated state, the of this the to interact with other proteins in the basal state. To that the overexpressed SMAD2 and SMAD3 remained capable of interactions with known their as we their interaction with this is the basal T. L. Cell. 1998; Full Text Full Text PDF PubMed Scopus Google Scholar). FLAG-tagged from transfected COS-1 cells from the 200 at a to which is twice its and were overexpressed and to gel they in in the in which overexpressed and some as an monomer in a The overexpressed in transfections in the as monomers to 1 and 3 with causes SMAD2 and SMAD3 to form complexes. is that these complexes contain multiple SMADs in association with More the results of this suggest that SMAD3 overexpressed 3 as a monomer not of an to to other proteins are for the assembly of complexes these To the of TGF-β on the oligomeric state of endogenous SMADs, HaCaT cells were treated with TGF-β for min prior to of cell and then by gel 3 a of SMAD2 and SMAD4 in the as they in the absence of of the in to a of kDa. This is to SMADs have been to associate with many proteins cofactors and transcriptional coactivators and in to TGF-β J. Wotton D. EMBO J. PubMed Google and This is in contrast to the of complexes using overexpressed proteins (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google which SMAD2 and SMAD4 in their endogenous To containing complexes in the from we and to with the followed by SMAD2 and SMAD4 were associated with each other in all from the Interestingly, complexes were in which is in the the of has that at levels of these proteins, binding of a complex to the in the region of a target gene association with a specific X. M.J. M. Nature. PubMed Scopus Google Scholar, C. Massague J. Genes Dev. 1997; PubMed Scopus Google Scholar). We therefore the association of SMAD2 and -4 into reflect such a from cell their by gel were and for their to The of the human JUNB and type collagen have been to to TGF-β and to contain TGF-β S. W. J. Biol. Chem. 1998; 273: Full Text Full Text PDF PubMed Scopus Google Scholar, L. A. J. A. A. J. Biol. Chem. 1998; 273: Full Text Full Text PDF PubMed Scopus Google Scholar). The HaCaT were with the to the TGF-β elements of JUNB or and the resulting complexes were with of the using was to the binding of SMAD complexes to the containing a in the L. A. J. A. A. J. Biol. Chem. 1998; 273: Full Text Full Text PDF PubMed Scopus Google Scholar) was as a of all the containing only of binding activity and SMAD This that SMAD of target in to TGF-β occurs upon the of complexes with a SMAD2, -3, and -4 have been to capable of (9Hata A. Lo R.S. Wotton D. Lagna G. Massague J. Nature. 1997; 388: 82-87Crossref PubMed Scopus (294) Google Scholar, 10Wu R.-Y. Zhang Y. Feng X.-H. Derynck R. Mol. Cell. Biol. 1997; 17: 2521-2528Crossref PubMed Scopus (186) Google Scholar). This has been to a function of the MH2 with some from the MH1 domain (9Hata A. Lo R.S. Wotton D. Lagna G. Massague J. Nature. 1997; 388: 82-87Crossref PubMed Scopus (294) Google Scholar, 10Wu R.-Y. Zhang Y. Feng X.-H. Derynck R. Mol. Cell. Biol. 1997; 17: 2521-2528Crossref PubMed Scopus (186) Google Scholar). these results were obtained utilizing with overexpressed tagged SMAD proteins. More recently, (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google Scholar) using transfected tagged SMAD proteins that SMAD proteins are as monomers in the absence of ligand In an to this we have the oligomeric of endogenous SMAD2, -3, and In where SMADs were was to in as closely as from the that whereas SMAD2 in to its of SMAD4 in that correspond to a SMAD3 on the other is strikingly different from both SMAD2 and -4 and to a of with endogenous and SMAD4 its as a in the basal state. with tagged SMAD4 that containing SMAD4 can contain two in a these oligomers a of the SMAD4 that the in the size of is to they are or on size of the structure of the MH2 domain of SMAD4 that is capable of forming in (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar). In is genetic evidence to suggest that some of the mutations in SMAD4 to the conserved interfaces of the trimer (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar). the of may an important of vivo. we have found that to at 1 not SMAD4 that SMAD4 interactions are of results therefore the previous that SMAD4 exists as a monomer (12Kawabata M. Inoue H. Hanyu A. Immamura T. Miyazono K. EMBO J. 1998; 17: 4056-4065Crossref PubMed Scopus (246) Google Scholar). The previous that SMAD4 is a monomer that was on the that a SMAD4 upon gel that this interface causes of the only in the of 1 (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google and the may well to SMAD4 at a low and overexpressed SMAD2 as although a is SMAD2 not to form complexes in the basal state, at not many that are SMAD2 is by such as T. L. Cell. 1998; Full Text Full Text PDF PubMed Scopus Google or to C. X. Feng X.-H. Mol. Cell. Full Text Full Text PDF PubMed Scopus Google that in of kDa. is however, that most of these interactions are such that SMAD2 is in a between and and the of cell and gel may to these complexes SMAD3, in contrast to SMAD2, is in multiple oligomeric complexes. and low expression levels of SMAD3 have not SMAD3 to endogenous or the of SMAD3 in the cell has strong on the complexes gel therefore that SMAD3 has a affinity SMAD2 for cellular is that two proteins as as SMAD2 and SMAD3 so in their The difference between the two is the in the MH1 domain of SMAD2, of a of amino that to the residues that to DNA binding of the SMADs (11Shi Y. Hata A. Lo R.S. Massague J. Pavletich N.P. Nature. 1997; 388: 87-93Crossref PubMed Scopus (375) Google Scholar). In the activated state, this of SMAD2 with that in the basal state this acid inhibits interactions with cellular proteins and is for the difference in between SMAD2 and SMAD3. This may by or by a that of proteins to their binding on Interestingly, this has been conserved through that an important is important to in that the between SMAD2 and proteins the cell may not as as is in Both proteins may interact with the cellular that the affinity of SMAD2 for these may the that the of gel results the that some of the between SMAD2 and SMAD3 Y. Feng X.-H. Wu R.-Y. Derynck R. Nature. PubMed Scopus Google Scholar, X. M. M. T. C. A. Cell Biol. 1999; PubMed Scopus Google Scholar) due to the of these two proteins to interact with other in a by the MH1 of TGF-β of SMAD is that complexes to a that of a can obtained results are with the that a SMAD4 homotrimer (or in the basal state is to a in to may well that these are the complexes that form upon TGF-β and as the upon which SMAD transcriptional complexes are some of the SMAD4 containing complexes in binding These contain and SMAD4 and therefore to The of binding to may reflect binding affinity for different observation that SMAD complexes of are the with the highest affinity for binding to the TGF-β elements in JUNB and a has been that binding to a TGF-β element only SMADs and a SMAD cofactors identified to date are proteins of kDa. results therefore suggest that the association of with such is with their association with likely to the or SMAD complexes may therefore to DNA as transcriptional We R. and J. for and C. is for FLAG-tagged