Functional consequences of tumorigenic missense mutations ... - Nature

ã

Oncogene (2000) 19, 4396 ± 4404 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc

Functional consequences of tumorigenic missense mutations in the amino-terminal domain of Smad4 Anita MoreÂn1, Susumu Itoh2, Aristidis Moustakas1, Peter ten Dijke2 and Carl-Henrik Heldin*,1 1

Ludwig Institute for Cancer Research, Box 595, S-751 24 Uppsala, Sweden; 2Division of Cellular Biochemistry, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands

Smads, the intracellular eectors of transforming growth factor-b (TGF-b) family members, are somatically mutated at high frequency in particular types of human cancers. Certain of these mutations aect the Smad amino-terminal domain, which, in the case of Smad3 and Smad4, binds DNA. We investigated the functional consequences of four missense mutations in the Smad4 amino-terminal domain found in human tumors. The mutant proteins were found to have impaired abilities to bind DNA although they were fully capable of forming complexes with Smad3. All four Smad4 mutants showed decreased protein stability compared to wild-type Smad4. Two of the Smad4 mutants (G65V and P130S) were translocated to the nucleus and were capable of transactivating a Smad-dependent promoter in a liganddependent manner. In contrast, the L43S and R100T mutants were not translocated eciently to the nucleus and consequently resulted in severely defective transcriptional responses to TGF-b. Moreover, we demonstrate here the critical importance of two basic residues in the b-hairpin loop of Smad3 or Smad4 for DNA binding, consistent with predictions from the Smad3 crystal structure. In addition, our results reveal that in the TGF-b-induced heteromeric signaling complex, loss of DNA binding of Smad4 can be compensated by Smad3, however, both Smad3 and Smad4 are needed for ecient DNA binding and signaling. In conclusion, mutations in the amino-terminal domain of Smad4, that are found in cancer, show loss of multiple functional properties which may contribute to tumorigenesis. Oncogene (2000) 19, 4396 ± 4404. Keywords: mutation; signal transduction; Smad; TGFb-transcription; tumor suppressor gene Introduction Transforming growth factor-b (TGF-b) belongs to a superfamily of structurally and functionally related cytokines, which also includes the activins and bone morphogenetic proteins (BMPs) (MassagueÂ, 1998; Roberts and Sporn, 1990). Members of this family regulate a wide spectrum of cellular responses, including proliferation, dierentiation, migration and apoptosis. They initiate cellular responses by binding to and activating speci®c receptors with intrinsic serine/ threonine kinase activity, whereby downstream eectors, termed Smad proteins, are activated (Attisano and Wrana, 2000; ten Dijke et al., 2000). Smad2 and *Correspondence: C-H Heldin Received 2 May 2000; revised 9 June 2000; accepted 13 July 2000

Smad3 are receptor-regulator Smads (R-Smads) which transiently interact with the activated TGF-b receptor and become phosphorylated at their extreme C-termini on two serine residues (Abdollah et al., 1997; Souchelnytskyi et al., 1997). Subsequently, activated Smad2 and Smad3 form heteromeric complexes with common-partner Smads (Co-Smads) (Lagna et al., 1996), e.g. Smad4, and translocate eciently to the nucleus, where they control the transcription of target genes (Attisano and Wrana, 2000; ten Dijke et al., 2000). R-Smads and Co-Smads share two domains of high sequence similarity at their amino and carboxy termini, termed Mad-homology domain (MH)1 and MH2, respectively. MH1 and MH2, which are separated by a proline-rich linker region of variable length, have distinct functions. The MH2 domain is important for homo- and heteromeric complex formation and for transcriptional activation and repression (Feng et al., 1998; Janknecht et al., 1998; Kawabata et al., 1998; Luo et al., 1999; Wotton et al., 1999). In addition, the MH2 domains of Smad2 and Smad3, but not Smad4, interact with type I receptors (Lo et al., 1998; MacõÂ as-Silva et al., 1996; Zhang et al., 1996). The MH1 domains of Smad3 and Smad4 have intrinsic binding activity to DNA sequences that contain 5'-AGAC-3' sequences, termed Smad-binding elements (SBEs) (Dennler et al., 1998; Jonk et al., 1998; Yingling et al., 1997; Zawel et al., 1998). A protruding b-hairpin loop is responsible for direct DNA contact (Shi et al., 1998). Smad2 is unable to bind directly to DNA because it contains an extra exon just N-terminal of the b-hairpin loop (Dennler et al., 1998; Yagi et al., 1999). SBE-like sequences have been identi®ed in the promoters of multiple TGF-b responsive genes, including plasminogen activator inhibitor-1 (PAI-1) (Dennler et al., 1998; Luo et al., 1999), junB (Jonk et al., 1998), type VII collagen (Vindevoghel et al., 1998) and the germline immunoglobulin Ia region (Hanai et al., 1999; Pardali et al., 2000a; Zhang and Derynck, 2000). Mutation of SBEs attenuate TGF-b inducibility. Multiple repeats provide for a strong enhancer function for TGF-b family members. MH1 and MH2 domains interact with a number of other transcription factors (Attisano and Wrana, 2000; ten Dijke et al., 2000). Moreover, the MH1 and MH2 domains, before activation of RSmads by receptors, repress the activity of each other through an intramolecular interaction (Hata et al., 1997, 1998; Zhang et al., 1996). TGF-b is a potent growth inhibitor for many dierent cell types. Escape from TGF-b-induced growth inhibition is frequently observed in tumors, and suggests a potential role for components in the

Missense mutations in Smad4 N-terminal domain A MoreÂn et al

TGF-b signal transduction pathway as tumor suppressors (de Caestecker et al., 2000). Genetic alterations in genes encoding TGF-b signaling components have been found in a wide variety of cancers. TbR-II is often mutated in an inherited form of colon cancer associated with microsatellite instability (Markowitz et al., 1995). Smad2 and Smad4 are found to be mutated in particular tumor subsets; the frequent mutation and homozygous deletion of Smad4 in pancreatic cancer led to its original discovery as the DPC4 tumor suppressor gene (Hahn et al., 1996). Somatic mutations in Smad4 are less frequent in other types of cancers such as colon, breast and lung cancers (Schutte et al., 1996). Additionally, the TGF-b pathway can be subverted in tumor cells by the action of oncogenes such as E1A, Ski, Sno-N or Evi-1 which interact with and block Smad nuclear functions (de Caestecker et al., 2000; ten Dijke et al., 2000). Upon loss of Smad4 expression in colon and breast epithelial cells, cells are no longer growth inhibited by TGF-b, and transfection with Smad4 rescues this response (de Winter et al., 1997; Le Dai et al., 1999; Zhou et al., 1998). Interestingly, Smad4 is not required for TGF-b-mediated induction of ®bronectin in MDAMB 468 breast cancer cells (Hocevar et al., 1999). In addition, embryonic ®broblasts derived from Smad4 knock out mice exhibit normal growth inhibitory and extracellular matrix gene responses to TGF-b, but are de®cient in activin and BMP-speci®c gene responses, indicating that in certain cell types, Smad4 is not necessary for TGF-b family signaling (Sirard et al., 2000). In cancer cells lacking particular Smad proteins, certain responses to TGF-b thus remain and may positively drive tumor progression. TGF-b may at this late stage of tumor progression function as a tumor promoter in a cell-autonomous manner by promoting tumor cell growth, migration and invasion (Cui et al., 1996; Oft et al., 1998). Furthermore, late stage tumor cells often produce high amounts of TGF-b, which, in a paracrine manner, may stimulate a favorable microenvironment for rapid tumor growth and metastasis by suppressing the immune system and stimulating angiogenesis and extracellular matrix formation (Heldin et al., 1997).

The characterization of the altered molecular properties of Smad proteins with missense mutations found in human cancers have been very informative with respect to the elucidation of the relationship between structure and function. The MH2 domain of Smad4 is often a target for point mutations and frame-shift mutations that lead to premature stops. Depending on which amino acid residues in the MH2 domain are mutated, this domain may have a perturbed ability to form stable homo- or heteromeric Smad complexes or may have a disrupted MH2 domain core structure. In the present study, we have determined the functional consequences of missense mutations in the MH1 domain of Smad4 and investigated the requirement of DNA binding of Smad3 and Smad4 in a TGF-binduced transcriptional response.

4397

Results and discussion Missense mutations in the Smad4 amino-terminal domain Genetic alterations in Smad2 and Smad4 genes in cancer cells include deletions, as well as missense, nonsense and frameshift mutations. Most of the point mutations are located in the carboxy-terminal portion of the Smad molecule, and have been shown to functionally inactivate Smad proteins by disrupting their structure, aecting homo- and heteromeric complex formation between Smads, blocking receptor-dependent R-Smad phosphorylation, or by interfering with formation of functional complexes with other proteins (Hata et al., 1997, 1998). The fewer missense mutations identi®ed in the amino-terminal region (also termed MH1 domain), are the subject of the present study (Figure 1). The L43S mutation in Smad4, identi®ed in pancreatic cancer (Johnson et al., 1999), aects a completely buried residue in the middle of an a-helix (Shi et al., 1998). Change of the aliphatic leucine residue into a polar serine residue may thus aect the domain structure. This residue is invariant among R- and Co-Smads and is located in near vicinity of a conserved putative nuclear localization signal sequence (NLS). The G65V mutation in Smad4 was

Figure 1 Mutations in the N-terminal domain of Smad4. The amino acid sequences of the N-terminal domains of Smad3 and Smad4 are aligned. Shared amino acid residues are shown by dots in the Smad3 sequence. Dash symbols indicate spacing introduced to maximize the alignment. The DNA binding loop, as assigned by Shi et al. (1998), and a putative NLS are indicated by a box and a line, respectively. The two residues of the b-hairpin loop that were mutated to abolish DNA binding, and the missense mutations in Smad4 analysed in the present study, are indicated. Numbers indicate amino acid sequence relative to the initiation codon Oncogene


4398

identi®ed in colorectal cancer (MacGrogan et al., 1997). Since this residue is completely solvent-exposed and located in a loop region, it may possibly be involved in interactions with other proteins (Shi et al., 1998). The R100T mutation in Smad4 was found in pancreatic cancer (Schutte et al., 1996). An analogous mutation in Smad2, R133C, has been observed in colorectal cancer (Eppert et al., 1996). Mutation of this partially buried arginine residue may have an eect on the local structure, but may leave the global structure of the amino-terminal domain intact (Shi et al., 1998). P130S in Smad4 was identi®ed in colorectal cancer (Thiagalingam et al., 1996). This proline residue is completely buried and found in all R- and Co-Smads; mutation of this residue to a polar serine residue may have a signi®cant eect of amino-terminal domain global structure (Shi et al., 1998; Figure 1). Smad4 with missense mutations have reduced ability to bind DNA The N-terminal domains of Smad3 and Smad4 were found to possess an intrinsic property to bind to speci®c DNA sequences that contain 5'-AGAC-3' sequences, termed Smad binding elements (SBEs) (Dennler et al., 1998; Jonk et al., 1998; Zawel et al., 1998). A protruding b-hairpin loop in the N-terminal domain, which is highly conserved among dierent Rand Co-Smads, is responsible for direct DNA contact (Shi et al., 1998; Figure 1). In order to test the ability of the missense mutants of Smad4 to bind to SBE, we generated GST-fusion proteins of these mutant Smad4 derivatives either with complete Smad4 coding region or GST-Smad4 fusions in which the carboxy-terminal domain was removed. EMSA with radiolabeled SBE probe showed that, in contrast to wild-type Smad4, all mutant proteins containing the complete coding region (data not shown) or lacking the carboxy-terminal domain (Figure 2), were unable to bind DNA. This ®nding is in agreement with Kim et al. (1997), who showed that an equivalent mutation of R100T Smad4 in a carboxy-terminally truncated Drosophila Mad, resulted in a protein unable to bind to DNA. In addition, the lack of DNA binding of R100T Smad4 and P130S Smad4 mutants was recently shown by other research groups (Le Dai et al., 1998; Luo et al., 1999). Interestingly, combination of the R100T or P130S mutants described here with deletions of the carboxy-terminal 38 amino acids of Smad4, led to persistent loss of DNA binding or gain of weak DNA binding, respectively (Le Dai et al., 1998). In the latter case, loss of the C-terminal fragment may result in relief from MH2-mediated transcriptional responses (ten Dijke et al., 2000). Taken together, the data show that missense Smad4 mutations found in human cancer result in proteins with abrogated DNA binding capacity at least in vitro. The data suggest that DNA binding is essential for Smad4 function and that amino acid residues outside the b-hairpin loop in¯uence dramatically the anity of Smad4 for its cognate SBE. Missense mutations in the MH1 domain lead to Smad4 instability When we expressed wild-type Smad4 or the four Smad4 missense mutations in either COS or HEK-

Oncogene

Figure 2 Impaired DNA binding of Smad4 proteins with missense mutations in their N-terminal domains. E. coli expressed GST-Smad4 proteins in which the carboxyterminal MH2 regions were deleted (D), wild-type (wt) or mutant, were incubated with a 32 P-labeled SBE probe. GST protein and no protein addition (7) were included as negative controls. A 500-fold molar excess of unlabeled oligonucleotide competitor was added where indicated (+); such a high excess was used because of the rather low anity of the MH1 domain for DNA binding. Equal amounts of GST fusion proteins were added, as determined by Coomassie staining (shown below the EMSA). Arrows indicate speci®c nucleoprotein complexes

293T cells we noted that wild-type Smad4 was expressed at a higher level than the Smad4 missense mutants, in spite of the fact that identical amounts of DNA were transfected (Figure 3a). In particular, Smad4 R100T was expressed at much lower levels than wild-type Smad4 (Figure 3a) which is consistent with recent observations by Xu and Attisano (2000) who demonstrated that this mutant undergoes ubiquitin-mediated degradation. We could correct for the unequal levels of Smad4 and Smad4 mutant proteins by transfecting proportionally more of the missense Smad4 mutant expression constructs compared to wild-type Smad4 expression plasmid (data not shown), suggesting that the observed instability was dependent on a saturable cellular machinery. Incubation of the transfected cells with the proteasome inhibitor MG132 resulted in a relative increase of the Smad4 missense mutant protein levels (data not shown). The missense mutations may thus have induced a conformational change or misfolding of the Smad protein inducing it to be degraded by a proteasome-mediated mechanism. Previous studies on Smad2 proteins with missense mutations in the C-terminal domain that disrupt the folding of the C-terminal domain, have also shown a decreased stability of the protein compared to wild-type Smad2 (Eppert et al., 1996). The increase in protein stability after incubation with MG-132 was not a property reserved for mutant


proteins only, as also the wild-type Smad4 protein level increased.

Figure 3 Impaired stability but normal heteromeric complex formation of Smad4 with missense mutations in the N-terminal domain. (a) Expression of Smad4 or N-terminal domain missense Smad4 mutants. The arrow indicates the Smad4 proteins. (b) Complex formation of Smad4 or N-terminal domain missense mutants with Smad3. HEK-293T cells transfected with dierentially tagged Smads in the absence or presence of constitutively active TGF-b type I receptor (termed ALK5TD). The cell lysates were sequentially immunoprecipitated with Flag followed by Western blotting with anti-myc. Expression controls are shown below. Arrows indicate the speci®c protein bands

4399

The Smad4 missense mutants do not affect oligomerization with Smad3 but interfere with nuclear translocation To elucidate additional defects of the MH1 missense mutants in Smad4, we tested their ability to form protein complexes with Smad3 (Figure 3b) and to translocate to the nucleus (Figure 4) in response to TGF-b. Wild-type or Smad4 missense mutant proteins were coexpressed transiently together with wild-type Smad3 in HEK-293T cells, as this assay is not sensitive enough to detect interactions in non-overexpressing cells. After immunoprecipitation of Smad4 proteins, the associating Smad3 was monitored by Western blotting. The results demonstrate that Smad3 coprecipitates with all four mutants of Smad4 with the same eciency as with wild-type Smad4 (Figure 3b). Stimulation of the transfected cells with the constitutively-active type I receptor for TGF-b enhanced the co-precipitating complexes signi®cantly, as expected. The variation in complex formation shown in Figure 3b re¯ects the dierences in Smad4 protein expression levels but not their capacity to oligomerize with Smad3. Thus, MH1 domain mutants of Smad4 that confer relative protein instability, are fully capable to form complexes with Smad3. This was anticipated as the critical domain for Smad4 oligomerization with Smad3 lies within the MH2 domain (Lagna et al., 1996; Zhang et al., 1996). In order to test the potential of the Smad4 missense mutants to translocate to the nucleus in response to TGF-b the colon carcinoma cell line SW480.7 was transiently transfected with wild-type or mutant Smad4 constructs (Figure 4). This cell line lacks one allele of Smad4 and expresses very low levels of endogenous Smad4 from the remaining allele (Calonge and MassagueÂ, 1999). All transfected proteins exhibited a uniform cytoplasmic distribution upon immuno¯uorescence microscopy of the SW480.7 cells (Figure 4).

Figure 4 Dierential nuclear accumulation of Smad4 missense mutants. Indirect anti-¯ag immuno¯uorescence of SW480.7 cells transiently transfected with wild-type (wt) or missense mutants of Smad4, tagged with the ¯ag epitope. Transfected cells were treated with mock or TGF-b1 (10 ng/ml) for 1 h prior to ®xation. Representative ®elds are shown. The percentage of cells showing nuclear translocation of Smad4 or Smad4 mutants after TGF-b1 stimulation, is shown below the photographs. The bar represents 20 mm Oncogene


4400

Wild-type Smad4 localized in the nuclei of the TGFb1-treated cells in about 50% of the cells. However, whereas the G65V and P130S mutants responded properly to TGF-b stimulation and accumulated in the nucleus, the L43S and R100T mutants did not show any appreciable nuclear translocation (less than 4% of the cells). The phenotype of the G65V and P130S mutant ®ts well with their ability to oligomerize with Smad3 and activate gene transcription. The L43S mutation resides next to a putative NLS (Figure 1). Although the mechanism of Smad4 import to the nucleus is currently unknown, this mutation may aect the integrity of the putative NLS. The R100T mutation, which resides farther away from the putative NLS, exhibits the most unstable protein phenotype (Figure 3a). Thus, arginine 100 may play crucial roles in proper Smad4 folding or recognition by interacting proteins of the nuclear import or proteasome machineries, or both. Alternatively, the L43S and R100T mutants undergo selective cytoplasmic retention by anchoring proteins such as SARA or microtubule associating proteins (Dong et al., 2000; Tsukazaki et al., 1998). All such possibilities deserve further analysis. Impaired transcriptional activity of Smad4 with missense mutations We compared the activities of wild-type Smad4 and its missense mutants in the TGF-b/SBE-mediated transcriptional response assay in MDA-MB 468 cells (Figure 5a). MDA-MB 468 cells have a homozygous deletion of the 18q chromosome and lack both copies of the Smad4 gene thus providing an excellent model system for functional studies of Smad4. The amounts of expression plasmid were adjusted to achieve similar protein expression levels of the wild-type and mutant Smad4 proteins. Transfection of the L43S and R100T Smad4 mutants were unable to restore a TGF-b/SBEmediated response in MDA-MB 468 cells, which is in line with their inability to enter the nucleus (Figure 4). The G65V and P130S Smad4 mutants were as potent as wild-type Smad4 in mediating TGF-b/SBE-driven transcriptional response. This could be due, either to residual weak DNA binding activity in vivo, that could not be recorded in the in vitro assay of Figure 2, or, more likely, to the ability of these mutants to enter the nucleus (Figure 4) and cooperate with other transcription factors. However, when we transfected equal amounts of DNA, their TGF-b-induced SBE reporter activity was at least reduced to half (data not shown). When cotransfected with wild-type Smad3, we found that L43S and R100T Smad4 mutants mediated weak signals in this assay (Figure 5b), in agreement with the ability of all mutants to form receptor-dependent heteromeric complexes with Smad3 (Figure 3b). Our ®ndings on the properties of the R100T Smad4 mutant contrast with those obtained by Hata et al. (1997, 1998), who reported an increase in anity of the amino terminal domain for the carboxy terminal domain in the R100T Smad4 mutant. We found that the R100T Smad4 mutant formed a receptor-dependent heteromeric complex with Smad3 as eciently as wildtype Smad4 (Figure 3b). Our results are consistent with those of Xu and Attisano (2000); they also reported that the R100T Smad4 mutant did not disrupt Smad2/ Smad4 heteromeric complex formation. Furthermore,

Oncogene

Figure 5 Impaired transcriptional activity of Smad4 with missense mutations in the N-terminal domain. (a) Eect of Smad4 or N-terminal domain missense mutants on TGF-bmediated SBE-driven transcriptional reporter activity in MDAMB 468 cells. (b) Eect of Smad4 or N-terminal domain missense mutants in the presence of wild-type Smad3 on TGF-b-mediated SBE-driven transcriptional reporter activity in MDA-MB 468 cells. Relative luciferase activity, normalized against the corresponding b-galactosidase activity, is shown as a bar graph. Standard deviations were derived from triplicate plates on a single experiment. Results from a representative experiment are shown. Open and ®lled bars represent unstimulated cells and cells stimulated with 10 ng/ml TGF-b1, respectively

if there would be a gain of autoinhibition in the R100T Smad4 mutant, one would predict that the lack of DNA binding of the R100T Smad4 complete protein would be unmasked by a R100T Smad4 mutant lacking the carboxy-terminal region. However, we found that both complete and carboxy-terminally truncated R100T Smad4 proteins were unable to bind DNA (Figure 2). Smad3 can complement DNA binding impaired Smad4 mutants To further elucidate the importance of DNA binding of Smad3 or Smad4 in signaling induced by the heteromeric Smad complex, we set out to selectively inactivate the DNA binding ability of these Smads by mutation. Arg74 and Lys81 in Smad3, and Arg81 and Lys88 in Smad4, are located in the b-hairpin loops and are presumed to directly contact DNA. To make the minimal possible change in order to preserve the local and global fold of the Smad protein, we mutated the arginine residues into lysines and the lysine residues into arginines and double mutants were also made (Figure 1). Wild-type and mutated GST-Smad4 full-


length fusion proteins were produced and tested for DNA binding in an electrophoretic mobility shift assay (EMSA) with radiolabeled SBE probe. Whereas wildtype Smad4 was able to bind DNA, single and double mutations in the b-hairpin loop abrogated Smad4 DNA binding (Figure 6a). Binding of the wild-type Smad proteins was speci®c as shown by the competition with an excess unlabeled wild-type SBE probe. These results are consistent with the crystal structure of Smad3 amino-terminal domain in which the b-hairpin loop was implicated in DNA binding (Shi et al., 1998). The Smad3 b-hairpin mutants were not tested on the EMSA as it is known that full-length protein cannot directly associate with DNA (Jonk et al., 1998; Zawel et al., 1998). The conservation of the mutated residues between Smad3 and Smad4 (Figure 1) implies that the b-hairpin mutants of Smad3 would also abrogate DNA binding much like in Smad4. Upon TGF-b receptor activation, a heteromeric complex of R-Smad and Co-Smad is formed in which both Smad3 and Smad4 have the ability to bind to DNA. To investigate the need for direct DNA binding of Smad3 and/or Smad4 in TGF-b-induced transcriptional responses that are mediated through SBEs in target promoters, we transfected the Smad3 or Smad4 b-hairpin mutants, alone or together with wild-type Smads, in Smad-de®cient cells and analysed for TGFb-induced SBE-reporter activity. Spontaneously immortalized mouse embryo ®broblasts (MEFs) lacking Smad3, derived from embryos with homozygous deletion of Smad3 (Yang et al., 1999), were used. The insensitivity to TGF-b-induced SBE-mediated transcriptional response of these cells could be rescued by cotransfection of wild-type Smad3; a substantial background was seen also in the absence of stimulation with TGF-b (Figure 6b) as noted before in Smad3 knock-out MEFs and other cell systems (Dennler et al., 1998; Yang et al., 1999; Zawel et al., 1998). The R74K Smad3, K81R Smad3 and double R74K/K81R Smad3 (D3) b-hairpin mutants were unable to rescue the Smad3 de®ciency in these cells with respect to SBEreporter activity. These results are in agreement with the notion that these mutants are defective in DNA binding. As expected from their DNA binding properties, transfection of wild-type Smad4, but not the R81K Smad4, K88R Smad4 and double R81K/K88R Smad4 (D4) b-hairpin mutants were able to confer a TGF-binduced SBE-luciferase response in these cells (Figure 6c). Subsequently, we tested combinations of wild-type Smad3, wild-type Smad4 and their double b-hairpin mutant derivatives in TGF-b/SBE-induced transcription in MDA-MB 468 cells. Interestingly, we found that the lack of DNA binding of Smad3 or Smad4 could be partially compensated by overexpression of wild-type Smad4 or wild-type Smad3, respectively. Transfection of b-hairpin mutants of both Smad3 (D3) and Smad4 (D4) in these cells showed no TGFb-induced transcriptional response (Figure 6d). The bhairpin mutants were found to make Smad heteromeric complexes as ecient as wild-type (Figure 6f), as was expected (Lagna et al., 1996). Similar results were obtained when natural TGF-b-responsive promoters, i.e. the PAI-1 promoter which includes multiple SBE's and is induced by TGF-b in an SBE-dependent manner, was used (Figure 6e). In contrast, when the

p21 cell cycle inhibitor promoter, which lacks SBEs but is also induced by TGF-b in a Smad-dependent manner, was used, the same b-hairpin mutants were indistinguishable from wild-type proteins (Pardali et al., 2000b). The latter is in agreement with the conclusion that the two mutated b-hairpin residues are involved in direct contact with SBEs in DNA, but not in protein-protein interaction between Smad members, or between Smads and cooperating transcription factors. Taken together, these results indicate that DNA binding of either Smad3 or Smad4 in the Smad3-Smad4 heteromeric complex is sucient for a response; however, for ecient DNA binding and signaling involving SBE-directed responses, both Smad3 and Smad4 are required. Our results are consistent with observations made by Johnson et al. (1999) who found that Smad3 and Smad4 could transcomplement each other's eects on an SBEcontaining promoter, however, our ®ndings address for the ®rst time this issue in the context of the bhairpin loop.

4401

Concluding remarks Smad4 is a central component in the TGF-b superfamily signaling pathways (Attisano and Wrana, 2000; ten Dijke et al., 2000), however, a number of fundamental questions about its role in signaling pathways remain unanswered, such as the mechanism of activation of Smad4 in response to ligand stimulation, the mechanism of heteromeric R-Smad/Co-Smad complex formation and the individual roles of Smad components in such complexes and the mechanism of nuclear import of Smads. The analysis of four Smad4 mutations seen in human tumors revealed that amino acid residues in dierent parts of the MH1 domain alter the Smad4 conformation in ways that prevent completely its DNA binding ability for SBEs; their phenotypes are identical to that of Smad4 mutated in the b-hairpin loop, which directly disrupt contact with the SBE DNA. Our results are fully consistent with those of Jones and Kern (2000), who recently demonstrated that mutation of conserved residues in the Smad4 MH1 domain from L43 to R135, including those residues that were investigated by us, had a dramatic eect on their ability to bind to SBE DNA. In addition, the mutations analysed by us aected, to dierent extents, the stability of Smad4, suggesting that the MH1 domain contributes to proteasome-regulated degradation. Finally, two of the mutants, L43S and R100T Smad4, did not translocate to the nucleus in response to ligand stimulation. The mechanism for the defective import to the nucleus is not known, but may involve perturbed interactions with components involved in the translocation mechanism. The present study demonstrates that both Smad3 and Smad4 contribute to SBE-dependent transcriptional activation, by binding to SBEs and providing transactivating functions. Importantly, each Smad component compensates for the lack of the other for both transcriptional activities. However, the abilities of the DNA binding defective G65V and P130S Smad4 mutants to activate SBE-dependent transcription as eciently as wild-type Smad4, indicates that Smad3 may play a primary DNA binding role in the SBEs. In Oncogene


4402

Figure 6 Mutation of residues in the b-hairpin loop of Smad3 or Smad4 interfere with DNA binding and TGF-b/SBE-driven transcriptional response. (a) E. coli expressed GST-Smad4 wild-type (wt) or mutant (R81K, K88R, D4) proteins were incubated with a 32P-labeled SBE probe. GST protein and no protein addition (7) were included as negative controls. A 500-fold molar excess of unlabeled oligonucleotide competitor was added where indicated (+). Equal amounts of GST fusion proteins were added, as determined by Coomassie Blue staining (shown below the EMSA result). The arrow indicates the speci®c nucleoprotein complex. (b) Eect of Smad3 and b-hairpin mutants on TGF-b-mediated SBE-driven reporter activity in mouse embryonic ®broblasts that lack Smad3. (c) Eect of Smad4 and b-hairpin mutants on TGF-b-mediated SBE-driven reporter activity in MDA-MB 468 cells. (d,e) Eect of combination of Smad3 and Smad4 and their b-hairpin mutant derivatives on TGF-b-mediated SBE-driven reporter activity (d) or on the PAI-I-derived p800 promoter activity (e) in MDA-MB 468 cells. Results in panels b ± c are shown as in Figure 5; open and ®lled bars refer to unstimulated cells and cells stimulated with 10 ng/ml TGF-b, respectively. (f) Heteromeric complex formation between wild-type (wt) Smad3 and Smad4 and their b-hairpin mutant derivatives. HEK-293T cells transfected with dierentially tagged Smads in the absence or presence of constitutively active TGF-b type I receptor (termed ALK5TD). The cell lysates were sequentially immunoprecipitated with Flag antibodies followed by Western blotting with myc antibodies. Expression controls are shown in the lower part of the ®gure Oncogene


part this may be dictated by a 2 : 1 stoichiometry between Smad3 and Smad4, respectively, that appear to preferentially be formed in the heteromeric signaling complex (Kawabata et al., 1998). The necessity for Smad4 therefore implies a role as a transcriptional activator or possibly yet undiscovered molecular roles.

Materials and methods Materials [g32P]ATP was from Amersham (Buckinghamshire, UK). All reagents for cell culture and the lipofectamine reagent were purchased from Life Technology (Gaithersburg, MD, USA). Glutathione Sepharose 4B was purchased from Pharmacia (Uppsala, Sweden). The Mutagenesis kit was obtained from Stratagene (La Jolla, CA, USA). Modifying and restriction enzymes were purchased from Fermentas (Amhurst, NY, USA). The luciferase assay system were purchased from Promega (Madison, WI, USA). The oligonucleotides were synthesized at Cybergene (Stockholm, Sweden). Flag (M5) and Myc (9E10) antibodies were purchased from Sigma (St Louis, MO, USA), and Santa Cruz Biotechnology (Santa Cruz, CA, USA), respectively. The proteasome inhibitor MG-132 was purchased from CalBiochem (La Jolla, CA, USA). Cell culture The human breast carcinoma cell line MDA-MB 468 was purchased from ATCC. Mouse embryonic ®broblasts (MEFs) lacking Smad3 were generously provided by Drs E Piek and A Roberts. MDA-MB 468, monkey kidney COS cells, HEK293T cells, human colon carcinoma SW480.7 cells and MEFs were cultured in Dulbecco's Modi®ed Eagles Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), L-glutamine and penicillin/streptomycin at 378C, in a 5% CO2 : 95% air atmosphere. Plasmids Human Smad4 was obtained from Dr S Kern. The 6xmyc Smad3 in pCNDA3 was a kind gift from Dr K Miyazono. All mutants were made with Stratagene Quick change mutagenesis kit. The sequences of the mutants were con®rmed using an ABI 310 DNA sequencer. Each mutant was ligated to the Flag pCDNA3 vector (Kawabata et al., 1998). The (CAGA)12 luc vector (Dennler et al., 1998) and p800neoluc (Abe et al., 1994) were used for luciferase assay. The b-galactosidase expression plasmid CH110 was from Pharmacia (Uppsala, Sweden). The GST fusion vector pGEX4T-1 (Pharmacia Biotechnology, Uppsala, Sweden) was used for the GST fusion constructs. Transient transfection and luciferase assays Transient transfections were performed using the calcium phosphate method. Flag pCDNA3 plasmids (0.5 ± 5 mg) were added to 2.56105 MDA-MB-468 cells or 1.256105 MEFs. In all transfections, a b-galactosidase expression plasmid was coexpressed and used for normalization of the luciferase activity (Jonk et al., 1998). The cells were stimulated with 10 ng/ml TGF-b for 16 h. Luciferase activities were measured by a Victor2 plate reader from Wallac. Immunoprecipitation and Western blotting HEK-293T cells were transfected with the calcium phosphate method, with 5 mg of Smad3, 10 mg of Smad4 and 5 mg of receptor plasmid. Twenty-four hours after transfection cells

were treated with 10 ng/ml of TGF-b1 for another 16 h. The cell extracts were immunoprecipitated with Flag M5 antibody and blotted with Myc antibody as described (Howell et al., 1999). To control for transfected protein expression levels, the cell extracts were also analysed by Western blotting. For the protein stability studies, the proteasome inhibitor MG132 was added to the transient transfections at a concentration of 20 mM. After 16 h incubation with the inhibitor, a second dose was added at the same concentration and incubation was continued for another 2 h. The cells were lysed and Western blot was performed as described above.

4403

Electrophoretic mobility shift assay Gel electrophoretic mobility shift assay (EMSA) was performed as described previously (Dennler et al., 1998). Oligonucleotide 4XWT (gtaccCAGACAGTCAGACAGTTCAGACAGTCAGACAGTc) from the JunB promoter was used as a probe (Jonk et al., 1998). It was labeled with T4 PNK and [g-32P]ATP. The binding reaction contained 100 ng of puri®ed GST fusion proteins, 30 mM KCl, 0.1 mM EDTA, 4 mM MgCl2, 20% glycerol, 0.8 mM Na phosphate, 20 mM HEPES pH 7.9, 4 mM spermidine, 0.3 mg/ml poly (dI-dC), and approximately 10 fmol of labeled probe. As a control for speci®city, a 500-fold molar excess of cold competitor oligonucleotide was added. Protein binding was allowed to proceed for 30 min at room temperature. Protein-DNA complexes were analysed in 5% non-denaturing polyacrylamide gels run in 45 mM Tris borate, 45 mM boric acid, 1 mM EDTA, pH 8.0. Bacterial expression of proteins The GST fusion proteins were expressed in E. coli strain TG1 and puri®ed according to Jonk et al. (1998). The integrity of the puri®ed fusion proteins was analysed by SDS-polyacrylamide gel electrophoresis and Coomassie blue staining. Indirect immunofluorescence SW480.7 cells were transfected using lipofectamine according to the manufacturer's protocol. Forty-eight hours posttransfection, the cells were treated with 10 ng/ml TGF-b1 for 1 h. Cells were ®xed, permeabilized and processed for anti-¯ag immuno¯uorescence as previously described (Piek et al., 1999).

Abbreviations DMEM, Dulbecco's modi®ed Eagle medium; EMSA, elecrophoretic mobility shift assay; GST, glutathione Stransferase; MEF, mouse embryo ®broblast; MH, Mad homology; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate buered saline; PMSF, phenylmethylsulphonyl ¯uoride; SBE, Smad binding element; Smad, Smad and Mad-related protein; TGF-b, transforming growth factor-b Acknowledgments We thank Napoleone Ferrara for TGF-b, Scott Kern for Smad4 cDNA, Rik Derynck for Smad3 cDNA, Ester Piek and Anita Roberts for MEFs lacking Smad3, and Kohei Miyazono for 6myc tagged Smad3 cDNA expression plasmid. We thank Serhiy Souchelnytskyi for valuable discussion and IngegaÈrd Schiller for preparation of this manuscript. This work was supported in part by the Dutch Cancer Society (NKI 2000-2217). Oncogene


4404

References Abdollah S, MacõÂ as-Silva M, Tsukazaki T, Hayashi H, Attisano L and Wrana JL. (1997). J. Biol. Chem., 272, 27678 ± 27685. Abe M, Harpel JG, Metz CN, Nunes I, Loskuto DJ and Rifkin DB. (1994). Anal. Biochem., 216, 276 ± 284. Attisano L and Wrana JL. (2000). Curr. Opin. Cell Biol., 12, 235 ± 243. Calonge MJ and MassagueÂ J. (1999). J. Biol. Chem., 274, 33637 ± 33643. Cui W, Fowlis DJ, Bryson S, Due E, Ireland H, Balmain A and Akhurst RJ. (1996). Cell, 86, 531 ± 542. de Caestecker MP, Piek E and Roberts AB. (2000). J. Natl. Cancer Inst., in press. de Winter JP, Roelen BAJ, ten Dijke P, van der Burg B and van den Eijnden-van Raaij AJM. (1997). Oncogene, 14, 1891 ± 1899. Dennler S, Itoh S, Vivien D, ten Dijke P, Huet S and Gauthier J-M. (1998). EMBO J., 17, 3091 ± 3100. Dong C, Li Z, Alvarez Jr R, Feng X-H and GoldschmidtClermont PJ. (2000). Mol. Cell, 5, 27 ± 34. Eppert K, Scherer SW, Ozcelik H, Pirone R, Hoodless P, Kim H, Tsui L-C, Bapat B, Gallinger S, Andrulis IL, Thomsen GH, Wrana JL and Attisano L. (1996). Cell, 86, 543 ± 552. Feng XH, Zhang Y, Wu RY and Derynck R. (1998). Genes Dev., 12, 2153 ± 2163. Hahn SA, Schutte M, Hoque ATMS, Moskaluk CA, da Costa LT, Rozenblum E, Weinstein CL, Fischer A, Yeo CJ, Hruban RH and Kern SE. (1996). Science, 271, 350 ± 353. Hanai J-i, Chen LF, Kanno T, Ohtani-Fujita N, Kim WY, Guo W-H, Imamura T, Ishidou Y, Fukuchi M, Shi M-J, Stavnezer J, Kawabata M, Miyazono K and Ito Y. (1999). J. Biol. Chem., 274, 31577 ± 31582. Hata A, Lo RS, Wotton D, Lagna G and MassagueÂ J. (1997). Nature, 388, 82 ± 87. Hata A, Shi Y and MassagueÂ J. (1998). Mol. Med. Today, 4, 257 ± 262. Heldin C-H, Miyazono K and ten Dijke P. (1997). Nature, 390, 465 ± 471. Hocevar BA, Brown TL and Howe PH. (1999). EMBO J., 18, 1345 ± 1356. Howell M, Itoh F, Pierreux CE, ValgeirsdoÂttir S, Itoh S, ten Dijke P and Hill CS. (1999). Dev. Biol., 214, 354 ± 369. Janknecht R, Wells NJ and Hunter T. (1998). Genes Dev., 12, 2114 ± 2119. Johnson K, Kirkpatrick H, Comer A, Homann FM and Laughon A. (1999). J. Biol. Chem., 274, 20709 ± 20716. Jones JB and Kern SE. (2000). Nucleic Acids Res., 28, 2363 ± 2368. Jonk LJC, Itoh S, Heldin C-H, ten Dijke P and Kruijer W. (1998). J. Biol. Chem., 273, 21145 ± 21152. Kawabata M, Inoue H, Hanyu A, Imamura T and Miyazono K. (1998). EMBO J., 17, 4056 ± 4065. Kim J, Johnson K, Chen HJ, Carroll S and Laughon A. (1997). Nature, 388, 304 ± 308. Lagna G, Hata A, Hemmati-Brivanlou A and MassagueÂ J. (1996). Nature, 383, 832 ± 836. Le Dai J, Schutte M, Bansal RK, Wilentz RE, Sugar AY and Kern SE. (1999). Mol. Carcinog., 26, 37 ± 43. Le Dai J, Turnacioglu KK, Schutte M, Sugar AY and Kern SE. (1998). Cancer Res., 58, 4592 ± 4597. Lo RS, Chen YG, Shi Y, Pavletich NP and MassagueÂ J. (1998). EMBO J., 17, 996 ± 1005. Luo K, Stroschein SL, Wang W, Chen D, Martens E, Zhou S and Zhou Q. (1999). Genes Dev., 13, 2196 ± 2206.

Oncogene

MacGrogan D, Pegram M, Slamon D and Bookstein R. (1997). Oncogene, 15, 1111 ± 1114. MacõÂ as-Silva M, Abdollah S, Hoodless PA, Pirone R, Attisano L and Wrana JL. (1996). Cell, 87, 1215 ± 1224. Markowitz S, Wang J, Myero L, Parsons R, Sun L, Lutterbaugh J, Fan RS, Zborowska E, Kinzler KW, Vogelstein B, Brattain M and Willson JKV. (1995). Science, 268, 1336 ± 1338. MassagueÂ J. (1998). Annu. Rev. Biochem., 67, 753 ± 791. Oft M, Heider KH and Beug H. (1998). Curr. Biol., 8, 1243 ± 1252. Pardali E, Xie X-Q, Tsapogas P, Itoh S, Avrantidis K, Heldin C-H, ten Dijke P, GrundstroÈm T and Sideras P. (2000a). J. Biol. Chem., 275, 3552 ± 3560. Pardali K, Kurisaki A, MoreÂn A, ten Dijke P, Kardassis D and Moustakas A. (2000b). J. Biol. Chem., in press. Piek E, Moustakas A, Kurisaki A, Heldin C-H and ten Dijke P. (1999). J. Cell Sci., 112, 4557 ± 4568. Roberts AB and Sporn MB. (1990). Peptide Growth Factors and Their Receptors, Part I, Vol 95. Sporn MB and Roberts AB. (eds). Springer-Verlag: Berlin, pp. 419 ± 472. Schutte M, Hruban RH, Hedrick L, Cho KR, Nadasdy GM, Weinstein CL, Bova GS, Isaacs WB, Cairns P, Nawroz H, Sidransky D, Casero Jr RA, Meltzer PS, Hahn SA and Kern SE. (1996). Cancer Res., 56, 2527 ± 2530. Shi YG, Wang YF, Jayaraman L, Yang HJ, MassagueÂ J and Pavletich NP. (1998). Cell, 94, 585 ± 594. Sirard C, Kim S, Mirtsos C, Tadich P, Hoodless PA, Itie A, Maxson R, Wrana JL and Mak TW. (2000). J. Biol. Chem., 275, 2063 ± 2070. Souchelnytskyi S, Tamaki K, EngstroÈm U, Wernstedt C, ten Dijke P and Heldin C-H. (1997). J. Biol. Chem., 272, 28107 ± 28115. ten Dijke P, Miyazono K and Heldin C-H. (2000). Trends. Biol. Sci., 25, 64 ± 70. Thiagalingam S, Lengauer C, Leach FS, Schutte M, Hahn SA, Overhauser J, Willson JKV, Markowitz S, Hamilton SR, Kern SE, Kinzler KW and Vogelstein B. (1996). Nature Genet., 13, 343 ± 346. Tsukazaki T, Chiang TA, Davison AF, Attisano L and Wrana JL. (1998). Cell, 95, 779 ± 791. Vindevoghel L, Lechleider RJ, Kon A, de Caestecker MP, Uitto J, Roberts AB and Mauviel A. (1998). Proc. Natl. Acad. Sci. USA, 95, 14769 ± 14774. Wotton D, Lo RS, Lee S and MassagueÂ J. (1999). Cell, 97, 29 ± 39. Xu J and Attisano L. (2000). Proc. Natl. Acad. Sci. USA, 97, 4820 ± 4825. Yagi K, Goto D, Hamamoto T, Takenoshita S, Kato M and Miyazono K. (1999). J. Biol. Chem., 274, 703 ± 709. Yang X, Letterio JJ, Lechleider RJ, Chen L, Hayman R, Gu H, Roberts AB and Deng CX. (1999). EMBO J., 18, 1280 ± 1291. Yingling JM, Datto MB, Wong C, Frederick JP, Liberati NT and Wang XF. (1997). Mol. Cell. Biol., 17, 7019 ± 7028. Zawel L, Dai JL, Buckhaults P, Zhou S, Kinzler KW, Vogelstein B and Kern SE. (1998). Mol. Cell, 1, 611 ± 617. Zhang Y and Derynck R. (2000). J. Biol. Chem., 275, 16979 ± 16985. Zhang Y, Feng X-H, Wu R-Y and Derynck R. (1996). Nature, 383, 168 ± 172. Zhou SB, Buckhaults P, Zawel L, Bunz F, Riggins G, Dai JL, Kern SE, Kinzler KW and Vogelstein B. (1998). Proc. Natl. Acad. Sci. USA, 95, 2412 ± 2416.