Transforming growth factor-β (TGF-β) inhibits growth and induces apoptosis of colon epithelial cells. Binding of TGF-β to its receptor induces phosphorylation of the Smad proteins Smad2 and Smad3, which then form heteromeric complexes with Smad4, translocate to the nucleus, and activate gene transcription. Smad4 function has been considered an obligate requirement for TGF-β signaling, and Smad4 mutations present in some cancers have been considered sufficient to inactivate TGF-β signaling. In this work, we describe studies with a nontransformed human colon epithelial cell line that is mutant for Smad4 but remains growth-inhibited by TGF-β. The colon cell line VACO-235 has lost one of its Smad4 alleles via a chromosome 18q deletion. The remaining allele bears two missense point mutations located in regions important for Smad4 trimer formation, which is thought necessary for Smad4 function. As expected, pSBE4-BV/Luc, a Smad4-activated transcriptional reporter, was inactive in VACO-235. Nonetheless, VACO-235 demonstrated 80% growth inhibition in response to TGF-β, as well as retention of some TGF-β-mediated activation of the p3TP-Lux transcriptional reporter. Transient transfection of the VACO-235 Smad4mutant allele into a Smad4-null cell line confirmed that this allele is functionally inactive as assayed by both the pSBE4-BV and p3TP-Lux reporters. The simplest explanation of these results is that there is a non-Smad4-dependent pathway for TGF-β-mediated signaling and growth inhibition in VACO-235 cells.
The TGF-β3superfamily of cytokines, comprising TGF-β, activins, and bone morphogenic proteins, plays an important role in a variety of cellular responses such as differentiation, cell cycle arrest, adhesion,migration, and extracellular matrix production (1). Signaling for TGF-β occurs through a transmembrane heteromeric receptor complex of RI and RII serine/threonine kinases. Binding of TGF-β to this receptor complex results in the phosphorylation of RI by RII, which activates the RI kinase domain (2, 3, 4). RI then phosphorylates intracellular mediators called Smads (5, 6). In the case of TGF-β signaling, Smad2 and/or Smad3 become phosphorylated and translocate to the nucleus with Smad4 (also called DPC4), a common Smad for the TGF-β, activin, and bone morphogenic protein signaling pathways (7). Once in the nucleus, this heteromeric Smad complex can bind directly or in cooperation with other transcription factors to DNA to activate the transcription of defined genes (8, 9, 10).
For many epithelial cell types, including colon, TGF-β inhibits growth and/or induces apoptosis (11). In colon epithelium,disruption of the TGF-β signaling cascade is considered an important mechanism by which tumor cells can escape growth suppression (11, 12). In a number of colon cancers, resistance to TGF-β growth inhibition is associated with mutations either in RII or in the signal transducers Smad4 and Smad2. RIImutations have been found to occur in ∼30% of all colon adenocarcinomas, whereas the frequencies of Smad4 and Smad2 mutations are 20% and ∼7%, respectively,thus indicating a role for these genes as tumor suppressors in human colon cancers (12, 13, 14, 15, 16, 17). The function of Smad4 as a tumor suppressor is believed to be the result of its ability to bind to specific DNA sequences (SBE) and transcriptionally activate TGF-βresponsive genes. Evidence has suggested that Smad4 plays a central role in mediating most cellular responses to TGF-β. Experiments with a colorectal cancer cell line in which Smad4 has been deleted through homologous recombination demonstrated the loss of TGF-β responsiveness (18). Furthermore, studies with differing Smad4 mutants derived from a variety of cancers have demonstrated that all mutations tested disrupt the ability of Smad4 to transcriptionally activate a TGF-β-responsive luciferase reporter (19). Finally, experiments in which Smad4 was localized to the nucleus demonstrated growth inhibition in the Smad4-null breast cancer cell line, MDA-MB-468(20). However, in this study we describe a nontransformed colon cell line that retains TGF-β-mediated growth inhibitor responses despite bearing only mutant Smad4 and despite the loss of Smad4-dependent transcriptional activity.
MATERIALS AND METHODS
VACO-235 was maintained on rat tail collagen-coated plates in MEM media supplemented with 2% fetal bovine serum, insulin, transferrin,selenium, l-glutamine, and hydrocortisone (MEM2+ medium),as described previously (21). VACO-9M was maintained in MEM supplemented with 8% bovine calf serum.
VACO-235 cells were plated at 8 × 104 cells/well on fresh collagen-coated 6-well plates in MEM supplemented with hormones but without fetal bovine serum(MEM+ medium). The next day the medium was removed and fresh MEM+medium containing 10 ng/ml EGF (Sigma Chemical Co., St. Louis, MO) or EGF plus 20 ng/ml TGF-β (R&D Systems, Minneapolis, MN) was added. Cells were then counted after 5 days of growth, and cell numbers were determined by removing the cells from the plates with EDTA, treating them with Pronase (13 units/ml; Sigma) to cause disaggregation, and counting them in a hemacytometer.
RNA from VACO-235 was prepared by extraction with guanidine isothiocyanate (14). RT-PCR amplification of full-length VACO-235 Smad4 was achieved using the forward primer 5′-TACGCGGATCCACCATGGACAATATGTCTATTACGAATAC-3′ and the reverse primer 5′-TACCGGAATTCCGGATAAACAGGATTGTATTTTGTAGTCC-3′. The PCR conditions for amplification were 1 cycle of 95°C for 5 min; 35 cycles of 95°C for 30 s, 58°C for 1.5 min, and 72°C for 1.5 min; and 1 cycle of 72°C for 7 min. The full-length product was then purified on 1.0% agarose gels and digested using the enzymes BamHI and EcoRI, which are incorporated into the forward and reverse RT-PCR primers, respectively (underlined sequences). The digested full-length product was then purified from the smaller digested end fragments using a QIAquick nucleotide removal kit(Qiagen, Valencia, CA) and subsequently cloned into pcDNA3.1(Invitrogen, Carlsbad, CA). Clones were then isolated and sequenced to ensure that no other mutations were present except for the known G→C and A→C mutations at bases 1116 and 1192, respectively. Wild-type Smad4 in pcDNA3.1 was a generous gift from Dr. S. E. Kern (Johns Hopkins University, Baltimore, MD).
Reporter Assay Studies.
Cells were seeded at 5 × 105cells/well for VACO-235 and VACO-9M in 6-well plates 1 day before transfection. The next day the medium was changed, and the cells were then transfected with 3 μg of DNA and 9 μl of FuGENE 6 (Roche,Indianapolis, IN), following the manufacturer’s protocol. For the VACO-235 and VACO-9M studies, p3TP-Lux or pSBE4-BV/Luc reporter (1.5μg), wild-type Smad4- or VACO-235 Smad4-pcDNA3.1 (1.5 μg), and the internal control reporter plasmid pRL-CMV (0.06 μg; Promega, Madison, WI) were cotransfected,and the cells were subsequently incubated for 48 (VACO-9M) or 72 h(VACO-235) with or without 10 ng/ml TGF-β. After the appropriate incubation time, cells were lysed and assayed for reporter activity using a Dual-Luciferase Reporter Kit (Promega) and a MLX Microtiter Plate Luminometer (Dynex Technologies, Chantilly, VA). The pRC-CMV-RII vector was generously provided by Dr. M. Brattain (University of Texas,San Antonio, TX).
Smad4 Mutation in VACO-235.
Previous studies karyotyping the nontransformed cell line VACO-235,which was derived from an adenomatous polyp of the colon, showed that one copy of chromosome 18 contained a deletion of q12.3-q22, which would force deletion of one copy of Smad4, located at 18q21.1 (16, 21, 22, 23). Because Smad4 frequently is mutated in pancreatic cancers, and is also mutated at low frequency in some colon cancers (16, 23), we proceeded to analyze the remaining Smad4 copy in VACO-235 for possible inactivating mutations. Sequencing demonstrated that VACO-235 expresses a single Smad4 transcript that contains two missense point mutations, one at codon 330 [GAA to CAA (Glu to Gln)], the second at codon 355 [GAC to GCC (Asp to Ala; data not shown)]. Both mutations occur at amino acids that are highly conserved throughout the Smad family of proteins. A comparison of these mutations with the crystal structure of the 234 amino acids at the COOH terminus of Smad4(residues 319–552) obtained by Shi et al. (24)indicate that the Asp355Ala mutation occurs in what is called the“loop/helix” region of Smad4. The loop/helix region of Smad4 is believed to be important in trimer formation and, therefore, for the function of the Smad4 protein. Mutations in this region have been shown to disrupt Smad4 trimer formation and to abolish Smad4-dependent transcriptional activation (19, 24). Furthermore,mutations that are localized in this loop-helix region have been found in Smad4 in ovarian and colon cancers, in Smad2in colon cancer (16, 17, 25), and in three colon cancer cell lines from our laboratory.4The VACO-235 Asp355Ala Smad4 mutation induces a charge loss in this critical domain that is predicted to both disrupt trimer formation and abolish Smad4 transcriptional activity. Comparison with the crystal structure similarly shows that the VACO-235 Smad4 Glu330Gln mutation is located in the edge of a β-sandwich structural motif and results in a charge shift that abolishes a hydrogen bond to Asn369 and thus should destabilize the rigid structure of the loop/helix region surrounding Asn369 (24).
Growth Suppression of VACO-235 by TGF-β.
Because the Glu330Gln and Asp355Ala mutations appear to be in regions important for Smad4 structure and function, we presumed that one or both of the mutations would disrupt the function of Smad4 in VACO-235. If so, it would be expected that VACO-235 would be resistant to growth inhibition by TGF-β. However, previous studies by our group had demonstrated that the growth of VACO-235 is, in fact, strongly inhibited by TGF-β (22). Accordingly, we reprised studies of TGF-β growth inhibition of VACO-235 using the same culture of cells in which we had demonstrated Smad4 mutations. These cells again showed that VACO-235 growth is inhibited 76% by 20 ng/ml TGF-β (Fig. 1), even in the presence of 10 ng/ml EGF, a growth stimulator(22).
Loss of SBE4-mediated Transcriptional Responses in VACO-235.
To confirm that the VACO-235 Smad4 mutation indeed inactivated Smad4 function, we tested the ability of TGF-β to activate the pSBE4-BV/Luc reporter, which contains four repeats of an eight-base palindromic SBE, driving a luciferase reporter(26). As shown in Fig. 2,A, in VACO-235, TGF-β was unable to activate transcription from the pSBE4-BV/Luc Smad4 reporter. To demonstrate that mutational inactivation of the VACO-235 Smad4 accounted for the loss of pSBE4-BV/Luc transcriptional responses, we cotransfected pSBE4-BV/Luc together with an expression vector expressing either wild-type Smad4 or the VACO-235 Smad4 mutant into VACO-235. Cotransfection of the pSBE4-BV/Luc reporter with wild-type Smad4 led to a 35-fold induction of pSBE4-BV/Luc reporter activity in TGF-β-treated versus untreated cells (Fig. 2,A). A similar 35-fold induction of pSBE4-BV/Luc reporter activity was noted for TGF-β-treated cells transfected with a wild-type Smad4 compared with cells transfected with an empty vector control (Fig. 2 A). In contrast, cells transfected with the expression vector encoding the VACO-235 Smad4 mutant showed no TGF-β-mediated induction of pSBE4/BV-Luc activity. Thus, wild-type Smad4 restored pSBE4-BV-Luc transcriptional responses to VACO-235; whereas even overexpression of the VACO-235 Smad4 mutant did not. This suggests that the VACO-235 Smad4 mutation indeed accounts for the loss of pSBE4-BV/Luc transcriptional activity and that the VACO-235 Smad4 does not contain even an attenuated transcriptional activation function.
Retention of Partial 3TP-Lux Transcriptional Responses in VACO-235.
Because VACO-235 demonstrated TGF-β-mediated growth inhibition despite the loss of Smad4 activity, we assayed VACO-235 cells for retention of other TGF-β-mediated transcriptional responses using the luciferase reporter construct p3TP-Lux. p3TP-Lux contains TGF-βresponse elements from the TGF-β-regulated PAI-1 promoter plus three 12-O-tetradecanoylphorbol-13-acetate response elements(3) and has been used extensively in the study of TGF-βresponsiveness in a variety of cell lines (3, 12, 27, 28, 29). As shown in Fig. 2,B, VACO-235 demonstrated a 3-fold induction of p3TP-Lux activity when treated with 10 ng/ml TGF-β. When cotransfected into VACO-235 cells together with a wild-type Smad4 expression vector, p3TP-Lux showed a larger, 7-fold induction in response to TGF-β. In contrast, cotransfection of p3TP-Lux together with the VACO-235 Smad4 mutant did not increase the responsiveness of p3TP-Lux to TGF-β compared with cells transfected with an empty expression vector (Fig. 2 B). Thus VACO-235 appears to retain the ability to partially activate p3TP-Lux transcription in response to TGF-β. This activity can be augmented by reexpression of wild-type Smad4, but it does not appear to be the result of residual activity of the VACO-235 Smad4 mutant.
VACO-235 Smad4 Does Not Activate p3TP-Lux or pSBE4-BV/Luc Reporters in a Smad4-null Line.
To further confirm the absence of transcriptional activity in the VACO-235 Smad4 mutant, this allele was transiently expressed in a Smad4-null cell line, VACO-9M, together with the TGF-β-regulated reporter constructs, p3TP-Lux and pSBE4-BV/Luc. VACO-9M does not express Smad4 mRNA, as determined by both RT-PCR and Northern blot.4 As shown in Fig. 3, treatment with 10 ng/ml TGF-β did not significantly induce luciferase activity in VACO-9M transfected with either the p3TP-Lux or pSBE4-BV/luc reporters that were accompanied by control empty expression vectors. Transfection of wild-type Smad4 into TGF-β-treated VACO-9M cells increased the activity of p3TP-Lux 4-fold and the activity of pSBE4-BV/Luc 8.5-fold, compared with transfection with the empty vector control (Fig. 3). In contrast, transfection of the VACO-235 Smad4 into VACO-9M gave results essentially identical to transfection of an empty vector control (Fig. 3). Thus,the VACO-235 Smad4 mutant allele was functionally inactive when tested for the ability to induce TGF-β-dependent transcription of the pSBE4/BV-Luc or the p3TP-Lux reporter.
Our findings demonstrate the preservation of significant TGF-β-mediated growth-inhibitory responses in human colon epithelial cells that bear a mutant Smad4 allele. Additionally, these cells retain at least partial TGF-β-stimulated transcriptional responses as determined by the TGF-β-mediated increase in p3TP-Lux activity. To the extent that we can assay, the TGF-β-induced VACO-235 transcriptional responses appear unlikely to be mediated by residual activity of the mutant VACO-235 Smad4 allele because introduction of this allele into a Smad4-null cell line does not confer TGF-β-mediated induction of either the p3TP-Lux or the pSBE4/BV-Luc reporter. These experiments thus suggest that a non-Smad4-dependent pathway can transduce TGF-β-mediated growth-inhibitory responses and at least some TGF-β-mediated transcriptional responses. Such an interpretation is consistent with a previous observation from our laboratory of a colon cancer in which Smad4 and RII had both been targeted for mutational inactivation(12).
Phosphorylation of Smad proteins by TGF-β receptors has been directly demonstrated to induce nuclear migration of activated Smad complexes and activation of a number of TGF-β-regulated gene promoters,including PAI-1 (30, 31, 32) and Mix.2 (33, 34),to which activated Smad complexes bind directly. Our findings of TGF-β-induced responses in Smad4-mutant VACO-235 cells are, however, consistent with observations that the growth of Smad4-null MEF cells remains 50% inhibited by TGF-β and that these cells can also transcriptionally induce fibronectin and PAI-1 (28). The investigators in this study speculated that their findings could in part reflect differences between fibroblasts and epithelial cells or between murine and human cells. We found that the growth of Smad4-mutant human colon epithelial cells is in fact even more profoundly inhibited by TGF-β (76%) than are Smad4-null MEF cells. The Smad4-mutant human colon epithelial cells also differ from the Smad4-null murine fibroblasts in maintaining TGF-β-mediated induction of the p3TP-Lux reporter. Our findings are also consistent with observations of a Smad4-null pancreatic cancer cell line that retains partial sensitivity to growth inhibition by TGF-β (35). Thus, studies in multiple different systems now support a model in which growth-inhibitory responses to TGF-β appear to be transduced at least in part by a Smad4-independent pathway.
A number of alternative signaling pathways have been suggested to potentially participate in TGF-β-induced responses. TGF-βactivation of Ras activity has been suggested to participate in TGF-βsignaling (36) and in TGF-β-mediated growth inhibition(35). Ras mutations have also been suggested to participate in induction of resistance to TGF-β growth inhibition(27). However, VACO-235 cells bear a K-ras codon 12 mutation (22), suggesting that TGF-β-induced growth inhibition in these cells is mediated by a Ras-independent mechanism. TGF-β-mediated activation of SAPK/JNK signaling has been demonstrated to have a Smad4-independent component (37) and to participate in the TGF-β induction of fibronectin (38),but thus far this pathway has not been successfully demonstrated to mediate TGF-β-induced growth inhibition (37, 38). Moreover, the potential for other Smads to act independently of or in place of Smad4 is raised by findings that Smad3 can to bind to other signaling molecules, such as the vitamin D receptor (39),and by the identification of the Smad4-related Smad4β gene in Xenopus (40, 41, 42),which could potentially have an as yet unidentified human orthologue.
In overview, analogous to growth factor receptor tyrosine kinases that activate multiple signaling pathways, we suggest a model in which TGF-β signaling also potentially engages multiple pathways that act to mutually reinforce a common response. An alternative interpretation of our findings would be that the Smad4 missense mutation present in VACO-235 is an unusual hypomorphic allele, which although apparently inactive in transcriptional assays, can nonetheless mediate a TGF-β growth-inhibitory response. Although the observations above disfavor this interpretation, we cannot completely exclude it. However,if correct, this interpretation would suggest that the growth-inhibitory activity of Smad4 is distinguishable from its ability to activate the consensus SBE4, and hence proceeds via an alternative mechanism. In either interpretation, the VACO-235 cell model described here should be of clear utility in elucidating key elements of the TGF-β signaling pathway that specifically participate in the TGF-βgrowth-inhibitory response.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by Public Health Service Grants RO1 CA67409, RO1 CA72160, and P30 CA43703. S. M. is an associate investigator of the Howard Hughes Medical Institute. J. M. and B. V. are investigators of the Howard Hughes Medical Institute.
The abbreviations used are: TGF-β,transforming growth factor-β; RI and RII, TGF-β receptors type I and type II; SBE, Smad binding element; EGF, epidermal growth factor;RT-PCR, reverse transcription-PCR; PAI-1, plasminogen activator inhibitor-1.
S. P. Fink, unpublished data.
We thank Ronda Brady for expert technical assistance.