Rik Derynck

Smad-dependent and Smad-independent pathways in TGF-|[beta]| family signalling

Transforming growth factor-β (TGF-β) proteins regulate cell function, and have key roles in development and carcinogenesis. The intracellular effectors of TGF-β signalling, the Smad proteins, are activated by receptors and translocate into the nucleus, where they regulate transcription. Although this pathway is inherently simple, combinatorial interactions in the heteromeric receptor and Smad complexes, receptor-interacting and Smad-interacting proteins, and cooperation with sequence-specific transcription factors allow substantial versatility and diversification of TGF-β family responses. Other signalling pathways further regulate Smad activation and function. In addition, TGF-β receptors activate Smad-independent pathways that not only regulate Smad signalling, but also allow Smad-independent TGF-β responses.

Molecular mechanisms of epithelial–mesenchymal transition

The transdifferentiation of epithelial cells into motile mesenchymal cells, a process known as epithelial–mesenchymal transition (EMT), is integral in development, wound healing and stem cell behaviour, and contributes pathologically to fibrosis and cancer progression. This switch in cell differentiation and behaviour is mediated by key transcription factors, including SNAIL, zinc-finger E-box-binding (ZEB) and basic helix–loop–helix transcription factors, the functions of which are finely regulated at the transcriptional, translational and post-translational levels. The reprogramming of gene expression during EMT, as well as non-transcriptional changes, are initiated and controlled by signalling pathways that respond to extracellular cues. Among these, transforming growth factor-β (TGFβ) family signalling has a predominant role; however, the convergence of signalling pathways is essential for EMT.

TGF-|[beta]| signaling in tumor suppression and cancer progression

Epithelial and hematopoietic cells have a high turnover and their progenitor cells divide continuously, making them prime targets for genetic and epigenetic changes that lead to cell transformation and tumorigenesis. The consequent changes in cell behavior and responsiveness result not only from genetic alterations such as activation of oncogenes or inactivation of tumor suppressor genes, but also from altered production of, or responsiveness to, stimulatory or inhibitory growth and differentiation factors. Among these, transforming growth factor β (TGF-β) and its signaling effectors act as key determinants of carcinoma cell behavior. The autocrine and paracrine effects of TGF-β on tumor cells and the tumor micro-environment exert both positive and negative influences on cancer development. Accordingly, the TGF-β signaling pathway has been considered as both a tumor suppressor pathway and a promoter of tumor progression and invasion. Here we evaluate the role of TGF-β in tumor development and attempt to reconcile the positive and negative effects of TGF-β in carcinogenesis.

TGF-β-induced epithelial to mesenchymal transition

During development and in the context of different morphogenetic events, epithelial cells undergo a process called epithelial to mesenchymal transition or transdifferentiation (EMT). In this process, the cells lose their epithelial characteristics, including their polarity and specialized cell-cell contacts, and acquire a migratory behavior, allowing them to move away from their epithelial cell community and to integrate into surrounding tissue, even at remote locations. EMT illustrates the differentiation plasticity during development and is complemented by another process, called mesenchymal to epithelial transition (MET). While being an integral process during development, EMT is also recapitulated under pathological conditions, prominently in fibrosis and in invasion and metastasis of carcinomas. Accordingly, EMT is considered as an important step in tumor progression. TGF-β signaling has been shown to play an important role in EMT. In fact, adding TGF-β to epithelial cells in culture is a convenient way to induce EMT in various epithelial cells. Although much less characterized, epithelial plasticity can also be regulated by TGF-β-related bone morphogenetic proteins (BMPs), and BMPs have been shown to induce EMT or MET depending on the developmental context. In this review, we will discuss the induction of EMT in response to TGF-β, and focus on the underlying signaling and transcription mechanisms.

Transcriptional Activators of TGF-β Responses: Smads

Based on the observations summarized above, the mechanism of transcriptional activation by Smads is most likely defined by the combined requirements for interactions with other transcription factors and with promoter DNA sequences. Consequently, Smad-responsive promoters have a double DNA sequence requirement. One sequence confers the specificity to bind the transcription factors that cooperate with the Smad complex. Another adjacent sequence is required for direct Smad binding and confers Smad selectivity to the first sequence. Thus, only a subset of promoter sequences that bind these cooperating transcription factors are targets for Smad signaling, and this selectivity is provided by flanking or partially overlapping sequences that allow Smad binding. For example, Smad3/4 cooperates with c-Jun to induce transcription from the collagenase I promoter in response to TGF-β; thus, the AP-1-binding site, which binds c-Jun/c-Fos, and the partially overlapping Smad-binding sequence AGAC mediate responsiveness to TGF-β and Smad3/4 (Yingling et al. 1997xYingling, J, Datto, M, Wong, C, Frederick, J, Liberati, N, and Wang, X.-F. Mol. Cell. Biol. 1997; 17: 7019–7028CrossrefSee all ReferencesYingling et al. 1997; Zhang et al. 1998xZhang, Y, Feng, X.-H, and Derynck, R. Nature. 1998; 394: 909–912Crossref | PubMed | Scopus (567)See all ReferencesZhang et al. 1998). However, the AP-1-binding site without a flanking GAC sequence does not allow inducibility by TGF-β or Smad3/4 (Dennler et al. 1998xDennler, S, Itoh, S, Vivien, D, ten Dijke, P, Huet, S, and Gauthier, J.-M. EMBO J. 1998; 11: 3091–3100Crossref | Scopus (1298)See all ReferencesDennler et al. 1998). Similarly, in the PAI-1 promoter, the TFE3-binding E-box sequence and the flanking sequence that binds the Smad3/4 complex are both required for TGF-β-induced transcription, which is mediated by the cooperation of Smad3/4 and TFE3 (Hua et al. 1998xHua, X, Liu, X, Ansari, D.O, and Lodish, H.F. Genes Dev. 1998; 12: 3084–3095CrossrefSee all ReferencesHua et al. 1998). The dual sequence requirement is also exhibited by an activin-response element, which requires adjacent FAST-1- and Smad-binding sequences for full ligand-inducible transcription (Zhou et al. 1998xZhou, S, Zawel, L, Lengauer, C, Kinzler, K.W, and Vogelstein, B. Mol. Cell. 1998; 2: 121–127Abstract | Full Text | Full Text PDFSee all ReferencesZhou et al. 1998). Therefore, at the Mix.2 promoter, Smad2/4 may not only interact with FAST-1 (Chen et al. 1997xChen, X, Weisberg, E, Fridmacher, V, Watanabe, M, Naco, G, and Whitman, M. Nature. 1997; 389: 85–89Crossref | Scopus (446)See all ReferencesChen et al. 1997), but presumably also with an adjacent sequence, most likely through Smad4 (since Smad2 does not bind DNA). Accordingly, Smad2/4 interacts with FAST-2 at the goosecoid promoter and Smad4 interacts with an adjacent GC-rich bipartite sequence. Again, the FAST-2- and Smad4-binding sequences are both required for full inducibility of transcription by ligand or Smad2/4 (Labbe et al. 1998xLabbe, E, Silvestri, C, Hoodless, P.A, Wrana, J.L, and Attisano, L. Mol. Cell. 1998; 2: 109–120Abstract | Full Text | Full Text PDF | PubMedSee all ReferencesLabbe et al. 1998). Finally, this model may also explain why a CRE sequence and a Mad-binding sequence were independently proposed as sequences required for transcriptional induction of the Ubx promoter by Dpp in Drosophila. Mutational analysis suggests that these overlapping sequences are both required for Dpp-mediated induction (Szuts et al. 1998xSzuts, D, Eresh, S, and Bienz, M. Genes Dev. 1998; 12: 2022–2035CrossrefSee all ReferencesSzuts et al. 1998). The required interactions of Smads with other DNA-binding transcription factors explains in retrospect why no consensus sequences for TGF-β and activin-response elements were found in different promoters, and why previously known transcription factors were implicated as essential factors for ligand-induced transcription.Given this cooperativity with other transcription factors, it may be surprising that the Smad3/4-binding elements themselves allow transcription in response to TGF-β and Smad3/4 (Dennler et al. 1998xDennler, S, Itoh, S, Vivien, D, ten Dijke, P, Huet, S, and Gauthier, J.-M. EMBO J. 1998; 11: 3091–3100Crossref | Scopus (1298)See all ReferencesDennler et al. 1998; Jonk et al. 1998xJonk, L, Itoh, S, Heldin, C.-H, ten Dijke, P, and Kruijer, W. J. Biol. Chem. 1998; 273: 21145–21152Crossref | PubMed | Scopus (444)See all ReferencesJonk et al. 1998; Zawel et al. 1998xZawel, L, Dai, J, Buckhaults, P, Zhou, S, Kinzler, K, Vogelstein, B, and Kern, S. Mol. Cell. 1998; 1: 611–617Abstract | Full Text | Full Text PDF | PubMedSee all ReferencesZawel et al. 1998). However, since a single Smad-binding sequence by itself does not allow transcriptional activation by ligand or Smad3/4, perhaps the multimerization of these sequences in the reporter assays allows recruitment of cooperating transcription factors. Alternatively, multimerization of Smad-binding sites may synergistically amplify the low transcriptional activity of Smad3, which is apparent in yeast or in the absence of cooperating transcription factors (Zhang et al. 1998xZhang, Y, Feng, X.-H, and Derynck, R. Nature. 1998; 394: 909–912Crossref | PubMed | Scopus (567)See all ReferencesZhang et al. 1998).This transcriptional cooperativity model also provides a mechanism for integration of two signaling pathways at the promoter sequence. Thus, transcription by c-Jun/c-Fos is induced by mitogenic stimuli and stress or UV irradiation, whereas Smad3/4 activation is induced by TGF-β. The required cooperation of Smad3/4 with c-Jun/c-Fos at select AP-1-binding sites thus illustrates that a convergence of both signaling pathways is at the basis of Smad3/4-induced transcription from these promoters (Zhang et al. 1998xZhang, Y, Feng, X.-H, and Derynck, R. Nature. 1998; 394: 909–912Crossref | PubMed | Scopus (567)See all ReferencesZhang et al. 1998). A similar type of cross-talk is also required for Dpp-induced transcription from the Ubx promoter. Thus, CREB binding to the promoter results from EGF receptor activation, whereas Mad binding is induced by Dpp receptor activation (Szuts et al. 1998xSzuts, D, Eresh, S, and Bienz, M. Genes Dev. 1998; 12: 2022–2035CrossrefSee all ReferencesSzuts et al. 1998), and Mad is likely to cooperate with CREB for Dpp-induced transcription from this promoter. Based on these examples, a variety of Smad responses may depend on activation of other signaling pathways and modifications in the cooperating transcription factors may regulate Smad responses.

Evidence that transforming growth factor-β is a hormonally regulated negative growth factor in human breast cancer cells

The hormone-dependent human breast cancer cell line MCF-7 secretes transforming growth factor-beta (TGF-beta), which can be detected in the culture medium in a biologically active form. These polypeptides compete with human platelet-derived TGF-beta for binding to its receptor, are biologically active in TGF-beta-specific growth assays, and are recognized and inactivated by TGF-beta-specific antibodies. Secretion of active TGF-beta is induced 8 to 27-fold under treatment of MCF-7 cells with growth inhibitory concentrations of antiestrogens. Antiestrogen-induced TGF-beta from MCF-7 cells inhibits the growth of an estrogen receptor-negative human breast cancer cell line in coculture experiments; growth inhibition is reversed with anti-TGF-beta antibodies. We conclude that in MCF-7 cells, TGF-beta is a hormonally regulated growth inhibitor with possible autocrine and paracrine functions in breast cancer cells.

Human transforming growth factor-α: Precursor structure and expression in E. coli

Abstract Transforming growth factor-α (TGF-α) is secreted by many human tumors and can induce the reversible transformation of nontransformed cell lines. Using long synthetic deoxyoligonucleotides as hybridization probes we isolated an exon coding for a portion of TGF-α from a human genomic DNA library. Utilizing this exon as a probe, a cell line derived from a human renal cell carcinoma was identified as a source of TGF-α mRNA. A cloned TGF-α cDNA was isolated from a cDNA library prepared using RNA from this cell line, and was found to encode a precursor polypeptide of 160 amino acids. The 50 amino acid mature TGF-α produced by expression of the appropriate coding sequence in E. coli binds to the epidermal growth factor receptor and induces the anchorage independence of normal mammalian cells in culture.

Smad3 and Smad4 cooperate with c-Jun/c-Fos to mediate TGF-Beta-induced transcription

Smad proteins transduce signals for transforming growth factor-β (TGF-β)-related factors. Smad proteins activated by receptors for TGF-β form complexes with Smad4. These complexes are translocated into the nucleus and regulate ligand-indu ced gene transcription. 12- O-tetradecanoyl-13-acetate (TPA)-responsive gene promoter elements (TREs) are involved in the transcriptional responses of several genes to TGF-β (refs 5–8). AP-1 transcription factors, composed of c-Jun and c-Fos, bind to and direct transcription from TREs, which are therefore known as AP1-binding sites. Here we show that Smad3 interacts directly with the TRE and that Smad3 and Smad4 can activate TGF-β-inducible transcription from the TRE in the absence of c-Jun and c-Fos. Smad3 and Smad4 also act together with c-Jun and c-Fos to activate transcription in response to TGF-β, through a TGF-β-inducible association of c-Jun with Smad3 and an interaction of Smad3 and c-Fos. These interactions complement interactions between c-Jun and c-Fos, and between Smad3 and Smad4. This mechanism of transcriptional activation by TGF-β, through functional and physical interactions between Smad3–Smad4 and c-Jun–c-Fos, shows that Smad signalling and MAPK/JNK signalling converge at AP1-binding promoter sites.

Toward a molecular understanding of skeletal development

Much of our knowledge about cartilage and bone has come from descriptive anatomy, endocrinology, and cellular studies of bone turnover. Recent approaches have led to the identification of local factors that regulate skeletal morphogenesis. Molecular and biochemical studies of bone and cartilage cells in vitro, gene inactivation in mice, and the identification of genes responsible for mouse and human skeletal abnormalities have documented the importance of specific growth and differentiation factors, extracellular matrix proteins, signaling mediators, and transcription factors in bone and cartilage development. The successful convergence of mouse and human genetics in skeletal biology is illustrated in this issue of Cell with two papers that show that mutations in collagen type Xl cause chondrodysplasia both in cho/cho mice as well as in patients with Stickler syndrome (Li et al., 1995; Vikkula et al., 1995). In general, recent results emphasize the need to view skeletal development at various integrated levels of organization and illustrate how single gene products affect development at these different levels. Pattern information determines not only the body plan of the early skeleton but also the shape of each individual skeletal element. In addition, the sequence of events during bone growth and development must be temporally and spatially controlled to ensure correct proportions of bony elements. Positional information must also regulate the establishment of bone internal structure throughout growth, while local homeostatic mechanisms must maintain bone integrity throughout adult life. Lastly, a complex extracellular matrix must generate skeletal tissues with specific biomechanical properties. Ultimately, the morphogenesis of the skeleton derives from the regulated differentiation, function, and interactions of its component cell types. Three major cell types contribute to the skeleton: chondrocytes, which form cartilage; osteoblasts, which deposit bone matrix; and osteoclasts, which resorb bone. Chondrocytes and osteoblasts are of mesenchymal origin, whereas osteoclasts derive from the hematopoietic system. Once embedded in bone matrix, osteoblasts mature into terminally differentiated osteocytes. The activity and differentiation of osteoblasts and osteoclasts are closely coordinated during development as bone is formed and during growth and adult life as bone undergoes continuous remodeling. More specifically, the formation of internal bone structures and bone remodeling result from coupling bone resorption by activated osteoclasts with subsequent deposition of new matrix by osteoblasts (Figure 1). Bone remodeling also links bone turnover to the endocrine homeostasis of calcium and phosphorus, since the mineralized bone matrix serves as the major repository for these ions in the body. Descriptive embryology and anatomy distinguish two types of bone development: intramembranous and endochondral. Intramembranous ossification occurs when mesenchymal precursor cells differentiate directly into bone-forming osteoblasts, a process employed in generating the flat bones of the skull as well as in adding new bone to the outer surfaces of long bones. In contrast, endochondral bone formation entails the conversion of an initial cartilage template into bone and is responsible for generating most bones of the skeleton. Cartilage templates originally form during embryogenesis when mesenchymal cells condense and then differentiate into chondrocytes. These cells subsequently undergo a program of hypertrophy, calcification, and cell death. Concomitant neovascularization occurs, and osteoclasts and osteoblasts are recruited to replace the cartilage scaffold gradually with bone matrix and to excavate the bone marrow cavity. Longitudinal bone growth takes place through a similar pattern of endochondral ossification in the growth plates located at the epiphyses (ends) of long bones. In these epiphyseal plates, the calcified, hypertrophic cartilage provides a scaffold for the formation of new trabecular bone. Ultimately, all remaining cartilage is replaced by bone except at the articular surfaces of the joints (Figure 2). Skeletal Patterning Classical embryology has shown that three distinct embryonic lineages contribute to the early skeleton. The neural crest gives rise to the branchial arch derivatives of the craniofacial skeleton, the sclerotome generates most of the axial skeleton, and the lateral plate mesoderm forms the appendicular skeleton. Transplantation studies have indicated that information regarding the number and ana-

TGF-β signaling in cancer — a double-edged sword

Transforming growth factor (TGF) β1 is a potent growth inhibitor, with tumor-suppressing activity. Cancers are often refractile to this growth inhibition either because of genetic loss of TGF-β signaling components or, more commonly, because of downstream perturbation of the signaling pathway, such as by Ras activation. Carcinomas often secrete excess TGF-β1 and respond to it by enhanced invasion and metastasis. Therapeutic approaches should aim to inhibit the TGF-β-induced invasive phenotype, but also to retain its growth-inhibitory and apoptosis-inducing effects.