© 2000 by Oxford University Press
Journal of the National Cancer Institute, Vol. 92, No. 17, 1388-1402,
September 6, 2000
© 2000 Oxford University Press
REVIEW |
Role of Transforming Growth Factor-ß Signaling in Cancer
Affiliation of authors: Laboratory of Cell Regulation and Carcinogenesis, Division of Basic Sciences, National Cancer Institute, Bethesda, MD.
Correspondence to: Anita B. Roberts, Ph.D., National Institutes of Health, Bldg. 41, Rm. C629, 41 Library Drive, MSC 5055, Bethesda, MD 20892-5055 (e-mail: (Robertsa{at}dce41.nci.nih.gov).
 |
ABSTRACT |
|---|
 Top Abstract IntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Signaling from transforming growth factor-ß (TGF-ß) through its unique transmembrane receptor serine–threonine kinases plays a complex role in carcinogenesis, having both tumor suppressor and oncogenic activities. Tumor cells often escape from the antiproliferative effects of TGF-ß by mutational inactivation or dysregulated expression of components in its signaling pathway. Decreased receptor function and altered ratios of the TGF-ß type I and type II receptors found in many tumor cells compromise the tumor suppressor activities of TGF-ß and enable its oncogenic functions. Recent identification of a family of intracellular mediators, the Smads, has provided new paradigms for understanding mechanisms of subversion of TGF-ß signaling by tumor cells. In addition, several proteins recently have been identified that can modulate the Smad-signaling pathway and may also be targets for mutation in cancer. Other pathways such as various mitogen-activated protein kinase cascades also contribute substantially to TGF-ß signaling. Understanding the interplay between these signaling cascades as well as the complex patterns of cross-talk with other signaling pathways is an important area of investigation that will ultimately contribute to understanding of the bifunctional tumor suppressor/oncogene role of TGF-ß in carcinogenesis.
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Research during the last 20 years has made it clear that tumor cells acquire an advantage over their normal cellular counterparts by a complex set of genetic alterations and that the particular mutational targets involved are representative of classes of genes with certain functional attributes. This paradigm replaces the older concept that mutational activation or repression of any one specific oncogene or tumor suppressor gene, respectively, was a prerequisite for malignant transformation. Rather, the current model focuses on certain essential alterations of cellular behavior that are required for the expression of malignant and invasive phenotypes. These alterations include changes in responses to positive and negative regulators of cellular proliferation, angiogenic factors and apoptotic signals as well as acquisition of the ability to undergo uncontrolled replication and invasion (1). Some of these effects may also result from changes in tissue architecture, altering the complex patterns of communication between the tumor, infiltrating immune cells, and its surrounding stroma (2–4).
Nearly all cells, including not only epithelial and lymphoid cells that are the targets of most human cancers but also the surrounding stromal, immune, and endothelial cells that participate in the formation of tumors, can express both transforming growth factor-ß(TGF-ß) ligands and their receptors. These ligands and receptors play a prominent, although complex, role in carcinogenesis. Negative regulation of cellular proliferation by TGF-ß has been shown to constitute a tumor suppressor pathway (5). However, a reduction in TGF-ß signaling in tumor cells is often accompanied by increased secretion of the ligand, which functions independently through its effects on accessory cells and, in certain instances, also on the tumor cells themselves, to promote tumorigenesis and increase metastasis (6,7). The latter effects are of particular importance for tumor cells in which certain TGF-ß-signaling pathways remain functional even though growth control by TGF-ß may be lost (8–11). Understanding the function of those TGF-ß-signaling pathways that remain operative in tumor cells is of paramount importance, since most tumors with defects in TGF-ß signaling still have functional components of these pathways that may play critical roles in determining the malignant phenotype.
In this review, we will focus primarily on a prominent pathway of signaling from the TGF-ß receptors involving Smad proteins. These substrates of the TGF-ß receptor kinases shuttle directly to the nucleus where they participate in the transcriptional regulation of target gene expression. Analysis of Smad protein mutations found in human cancers has provided insight into their biochemical function, as has the study of an expanding set of cytoplasmic and nuclear-interacting proteins that modulate the activity and degradation of the Smads. Furthermore, emerging patterns of cross-talk between the Smads and other signaling pathways suggest that the function of these molecules in carcinogenesis is highly complex and represents a major challenge for cancer research over the next decade.
|
TGF-ß RECEPTORS AS TUMOR SUPPRESSOR GENES |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
The three mammalian isoforms of TGF-ß—TGF-ß1, TGF-ß2, and TGF-ß3—are members of a large family of secreted ligands, including the activins and bone morphogenetic proteins (BMPs) (12). These ligands initiate downstream signaling events by activation of a heteromeric complex of related transmembrane receptors with intrinsic cytoplasmic serine–threonine kinase domains. TGF-ßs interact with the TGF-ß type II receptor (TßRII), which, in turn, recruits a TGF-ß type I receptor, principally TßRI/ALK5, into a heterotetrameric receptor complex. Other members of the TGF-ß superfamily interact with different combinations of homologous type I and II receptor serine–threonine kinases. The ligand-binding type II receptor kinase is constitutively active and activates type I receptors by phosphorylation of serine and threonine residues in the GS box, a conserved stretch of glycine and serine residues preceding the receptor kinase domain. This phosphorylation event is associated with activation of the type I receptor kinase and downstream signaling (13,14).
There is a strong correlation between malignant progression and loss of sensitivity to the antiproliferative effects of TGF-ß which is frequently associated with reduced expression or inactivation of TGF-ß receptors (5,15). There are only sporadic reports of mutations or deletions in TßRI in malignancy (16). In contrast, TßRII is a frequent locus of inactivating mutations (Fig. 1
). These are particularly common in replication error repair-positive colon and gastric carcinomas that harbor a deficient DNA mismatch repair system (5). In these tumors, genomic instability is associated with frameshift mutations in a 10-base-pair polyadenine tract (big polyadenine tract [BAT]) in exon 3 of the TßRII gene (BAT-RII), resulting in truncated receptors that lack the serine–threonine kinase domain (17,18). Other TGF-ß receptor mutations, like the Thr315Met germline mutation of TßRII in a family with hereditary nonpolyposis colorectal cancer, do not interfere with the kinase activity of the type II receptor but enhance the metastatic potential of tumor cells by specifically impeding TGF-ß-mediated growth arrest without affecting induction of extracellular matrix formation (11).
|
Whereas TGF-ß receptor mutations other than the BAT-RII frameshift mutations are rare events in tumorigenesis, repression of TGF-ß receptor expression is a common mechanism that enables tumor cells to escape from negative regulation of growth by TGF-ß. Thus, aberrant receptor trafficking can reduce cell surface receptor expression, for example, in pristane-induced mouse plasmacytomas (19) and a human cutaneous T-cell lymphoma (20), although this more frequently involves transcriptional silencing of the TGF-ß receptor promoters. Several members of the E-twenty-six (ETS) acutely transforming retrovirus of chickens family of transcriptional transactivators, including ERT/ESX/ESE-1/ELF3/jen and Fli-1, are critical for the expression of TßRII and reduced levels of these genes have been shown to correlate with reduced receptor expression in gastric cancers (21,22). In Ewing's sarcoma (EWS), any of several members of the ETS family are fused to the EWS gene as a result of chromosomal translocations. The EWS/Fli-1 fusion protein represses TßRII expression as it retains the DNA-binding activity of Fli-1 but is unable to activate transcription because of deletion of the transactivating amino-terminus of Fli-1 (22). TßRII expression is also negatively regulated by H-Ras (10) and the adenoviral oncoprotein E1A (23). Another mechanism of transcriptional repression observed in some gastric cancer cell lines involves hypermethylation of CpG islands in the TßRI promoter (24). Methylation of gene promoters results in chromatin condensation, limiting accessibility of transcription factors to the DNA. While this is of critical importance for gene silencing in development, the mechanisms underlying abnormal methylation in cancer are unknown (25). Finally, there are mutations in the TßRII promoter that can contribute to reduced receptor expression in tumors by interfering with binding of transcriptional regulators to the mutated promoter (24,26–28).
|
TGF- ß RECEPTOR-INTERACTING PROTEINS |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Several cytoplasmic proteins have been identified that interact with the TGF-ß receptors (Table 1
). Apart from the Smad family of proteins, none of these have been implicated in carcinogenesis, although it remains possible that aberrant expression and/or regulation of these receptor interactors could perturb TGF-ß signaling.
|
A notable exception to this is the pseudoreceptor BAMBI, identified from a complementary DNA expression screen in Xenopus embryos. BAMBI is a naturally occurring truncated type I receptor that lacks a kinase domain. It can interact with TGF-ß in the presence of TßRII, binding TßRI and TßRII and disrupting ligand-induced receptor heteromerization (35). Of interest, nma, the mammalian orthologue of BAMBI, was identified from a differential display analysis in which it was decreased in highly metastatic melanoma (36), suggesting that derepression of TGF-ß signaling following a decrease of nma expression could account for the increased metastatic potential of these cells.
|
TGF- ß SIGNALING BY SMAD PROTEINS |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Ligand-induced heteromerization of TßRI and TßRII results in activation of downstream signaling pathways. An intense period of discovery over the last 4 years has resulted in elucidation of an evolutionarily conserved signal transduction pathway downstream of the TGF-ß receptors mediated by Smads (13,14). These proteins are characterized by two regions of homology to the Drosophila orthologue, Mad, at their N- and C-termini, termed the Mad-homology domain MH1 and MH2, respectively, connected by a less conserved, proline-rich linker domain, and can be divided into three functional groups: the receptor-activated Smads (R-Smads), Smad1, Smad2, Smad3, Smad5, and Smad8; the common mediator Smad (Co-Smad), Smad4; and the inhibitory Smads (I-Smads), Smad6 and Smad7 (Fig. 2
, A and B).
|
Receptor-Activated Smads
Activation of the TGF-ß family of receptors results in phosphorylation of specific R-Smads: Smad1, Smad5, and Smad8, following treatment with BMPs, and Smad2 and Smad3, following treatment with TGF-ß or activin (Fig. 3, A) (14,40,41). For purposes of this review, we will restrict our discussion to the TGF-ß-activated R-Smads, Smad2 and Smad3. Phosphorylation occurs principally on two serine residues within a conserved –SS(M/V)S motif at the extreme C-terminus of the R-Smads, both of which are required for activation of Smad-dependent signaling events (42,43). R-Smads have been shown to interact with cross-linked ligand–receptor complexes (44–46), and in vitro kinase reactions suggest that these proteins are direct substrates of their respective type I receptors (45,46). Receptor specificity is defined by a matching set of two residues within the C-terminal MH2 domain of Smad2 and Smad3 (47). Smad2 and Smad3 are recruited to TßRI through interaction with a membrane-associated, lipid-binding, phenylalanine, tyrosine, valine, and glutamic acid (FYVE) domain protein, Smad-anchor for receptor activation (SARA) (38). Comparison of the three-dimensional structures of the TßRI kinase domain with that of the Smad-binding domain of SARA in complex with the Smad2-MH2 domain (48) suggests that SARA is required to bring Smad2 into close proximity with a cluster of residues within the L45 loop of the TßRI kinase, known to determine the R-Smad specificity of the receptor (39,47).
|
B. Proteasomal degradation of Smads also negatively regulates the pathway, as, for example, in the degradation of nuclear Smad2 by the UbcH5 family of ubiquitin-conjugating enzymes. TNF-Since the Smad proteins do not have enzymatic activity, this signaling pathway is not amplified, and, as such, cellular responses are sensitive to small changes in the level of Smad protein expression. This is illustrated by the resultant embryonic lethal phenotype of compound heterozygous Smad2/Smad3 mutant mice, when each of the individual heterozygotes is viable (49). This suggests that there is a gene dosage effect for the transcriptional targets of the Smads and that the relative level of Smad2 and 3 in a given cell not only affects its ability to respond to TGF-ß but also influences the nature of that response.
Changes in the levels of Smad proteins may result from the transcriptional regulation of Smad expression, e.g., the decrease in Smad3 messenger RNA (mRNA) following treatment with TGF-ß in lung epithelial and mesangial cells (50,51). More recent data suggest that the expression of Smads is also dynamically regulated at a post-translational level, through targeted, ubiquitin-mediated proteasomal degradation. In the BMP-signaling pathway, the cytoplasmic E3 homology to E6-AP carboxylterminus (HECT) domain ubiquitin-ligase, SMURF1, has been shown to target Smad1 for proteasomal degradation in a ligand-independent manner through interactions with a specific proline–tyrosine motif in the middle-linker (52). While a TGF-ß R-Smad-specific E3 ligase remains to be identified, nuclear accumulation of phosphorylated Smad2 is targeted for destruction through the UbcH5 family of E2 ubiquitin-conjugating proteins (53), providing an additional level of ligand-dependent feedback regulation of Smad signaling (Fig. 3
, B).
Smad2 and Smad3 have also recently been shown to interact with ß-tubulin and to be distributed along microtubular networks within the cell (54). Chemical disruption of these microtubular networks interferes with Smad2 binding to tubulin and is associated with ligand-independent phosphorylation of Smad2 and activation of TGF-ß-like signals. This raises the intriguing possibility that changes in the cytoskeleton associated with cellular transformation may directly influence TGF-ß signaling.
Common Mediator Smad, Smad4
Following TGF-ß receptor-induced phosphorylation, R-Smads partner with the Co-Smad, Smad4, and translocate to the nucleus where they activate downstream transcriptional responses. Smad4 was first cloned as a candidate tumor suppressor gene, deleted in pancreatic carcinoma locus 4 (DPC4), being lost or functionally inactivated in nearly 40% of pancreatic carcinomas, underscoring the importance of this signaling pathway in carcinogenesis (55). Smad4 lacks the C-terminal –SS(M/V)S motif found in all R-Smads and is not phosphorylated following activation of TGF-ß receptors (56,57). It acts as a convergent node in the Smad pathways downstream of TGF-ß superfamily receptors, complexing not only with TGF-ß/activin-activated Smad2 and Smad3 (56,57) but also with BMP-activated Smad1, Smad5, and Smad8 (Fig. 3, A) (58–60). Hetero-oligomerization of R-Smads and Smad4 is dependent on their MH2 domains, while the amino-terminal MH1 domain of Smad4 provides an auto-inhibitory mechanism by preventing ligand-independent R-Smad/Smad4 interactions (see Fig. 5, A) (61). Smad4 can translocate to the nucleus only when complexed with R-Smads, whereas ligand-activated Smad2 and Smad3 translocate to the nucleus in a Smad4-independent fashion (62). This implies, by default, that the principal function of Smad4 is to regulate transcription rather than to transmit the signal from the cytoplasm to the nucleus.
|
Initial studies (56,62,63) suggested that Smad4 was an essential component in all TGF-ß signaling responses. More extensive studies in Smad4 null cells (64–66) have since identified a number of Smad4-independent TGF-ß responses. On this basis, it is possible that Smad2 and/or Smad3 alone could activate TGF-ß-dependent responses in the absence of Smad4. However, a number of these Smad4-independent responses, including the induction of fibronectin and TGF-ß-induced cell cycle arrest in some Smad4 null cells, depend on TGF-ß-induced activation of Jun N-terminal kinase (JNK) and extracellular signal-related kinases (ERK) 1/2 mitogen-activated protein (MAP) kinase pathways, respectively (64,65). As such, these responses may involve activation of alternative pathways downstream of the TGF-ß receptor that bypass the Smad signaling pathway. Alternatively, the expression of another as yet unidentified Co-Smad could substitute for the absence of Smad4 in Smad4 null cells. Recent identification of a second Co-Smad in Xenopus, XSmad4b (66), suggests that this is a possibility, although the mammalian XSmad4b orthologue remains to be identified.
Inhibitory Smads
Given the central importance of TGF-ß in regulation of cellular homeostasis, it is not surprising that there are also a number of feedback mechanisms regulating this process (Fig. 3, B). Like Smad4, the I-Smads (Smad6 and Smad7) lack a C-terminal –SS(M/V)S motif and are not phosphorylated following receptor activation. I-Smads interfere with receptor-mediated phosphorylation of R-Smads by competing for binding of R-Smads to their respective receptors (67–70). In addition, Smad6 competes with Smad4 for binding to BMP-activated Smad1 (71). Both of these mechanisms may be operative and are likely to depend on the relative levels of the various Smad proteins in the cell. While both of the I-Smads have the capacity to inhibit TGF-ß and BMP-dependent signaling, there is evidence that the principal function of Smad7 is to regulate TGF-ß and of Smad6 is to regulate BMP responses. Therefore, while overexpression of Smad7 inhibits TGF-ß-dependent signaling in most of the assay systems tested (68–70), Smad6 overexpression may either inhibit (67,72) or have no effect on (71,73) these responses. Furthermore, Smad7 expression is induced following treatment with TGF-ß or activin (69,74,75), while BMPs induce Smad6 expression in a variety of cell types (76).
|
TRANSCRIPTIONAL REGULATION OF TARGET GENES BY SMADS |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Identification of the target genes of TGF-ß and understanding their roles in both normal and malignant cells will provide insight into the pathogenetic mechanisms of action of TGF-ß. Effects of TGF-ß on cellular proliferation are known to depend, in part, on transcriptional regulation of several target genes, the aberrant expression of which has been associated with carcinogenesis, including TGF-ß signal transduction components, cell cycle regulators, and transcription factors (Table 2
).
|
This review will focus on the mechanism of regulation of certain of these TGF-ß target genes by the Smad-signaling pathway. Several different modes of transcriptional regulation by Smad proteins have now been described, including direct DNA binding by the R-Smad/Smad4 complex and interaction of Smad proteins with other DNA-binding and non-DNA-binding transcriptional activator and repressor components.
Direct Binding of Smads to DNA
The amino-terminal MH1 domains of Smad3 and Smad4 bind a specific DNA sequence, CAGA, called the Smad-binding element (SBE). This was first identified through a polymerase chain reaction-based random oligonucleotide selection process (102) and subsequently found in a number of mammalian promoter elements (88,89,103–107). In addition, a consensus sequence, GCCGnCGC, first identified in the Drosophila vestigial dpp enhancer (108), is required for TGF-ß-induced activation of Smad4-responsive elements in the human type VII collagen and mouse goosecoid promoters (104,109,110). In contrast, Smad2, which is 91% identical to Smad3, does not bind to DNA. This inability of Smad2 to bind DNA has been mapped to a 30 amino-acid insertion within its MH1 domain, which is deleted in a naturally occurring splice variant of Smad2, Smad2 
exon3, restoring its ability to bind to DNA (Fig. 2, B) (111,112). This difference in DNA-binding properties of Smad2 and Smad3 could account for their divergent functions, illustrated by the completely different phenotypes of Smad2- and Smad3-deficient mice (49). These distinct functional features and cellular targets may also explain the observed differences in the frequency of Smad2 and Smad3 mutations seen in human malignancy.
Interaction of Smads With Other DNA-Binding Factors
The SBE frequently overlaps, or is in close proximity with, the binding sites of other transcription factors. This has been demonstrated in several TGF-ß-responsive elements, such as AP-1 sites in collagenase I, type VII collagen, and plasminogen activator inhibitor-1 (PAI-1) promoters (104,106,109,113), a TFE3-binding E box in the PAI-1 promoter (114), an acute myeloid leukemia site in the germline immunoglobulin A1 promoter (115), and an FAST2-binding site in the goosecoid promoter (110). While in certain contexts the SBE is dispensable (106), it is usually required for TGF-ß-dependent transcriptional activation. However, as Smad3 and Smad4 bind DNA relatively weakly (116), and Smad2 does not bind DNA at all, it is likely that Smads are primarily recruited to DNA through their interactions with other DNA-binding transcription factors and that the principal function of Smad/DNA interactions is to stabilize preformed, DNA-binding, transcriptional complexes. A summary of transcription factors known to interact with TGF-ß-dependent Smads is presented in Table 3. These interactions of Smad proteins with transcription factors may occur through either, or both, the MH1 and MH2 domains. A novel Smad-interaction motif, characterized by a core PPNK sequence, has recently been identified in several transcription factors that interact with Smad2, including the winged helix/forkhead proteins FAST1 and FAST2 and the paired-like homeodomain factors Mixer and Milk (118). Identification of this motif should facilitate discovery of novel transcription factors interacting with Smad2.
|
Smad-Dependent Recruitment of Transcriptional Activator and Repressor Complexes
While transcriptional activation of target genes by R-Smads and Smad4 requires their C-terminal MH2 domains (137), the mechanism has been elusive. New insights come from the findings that Smads interact with proteins having histone acetyl transferase and histone deacetylase activities (Table 3), the primary functions of which are to modify transcription by altering chromatin structure so that the underlying DNA sequences are either exposed to, or concealed from, respectively, the transcriptional apparatus (138). These include the closely related transcriptional adaptor proteins, CBP and p300, which not only have histone acetyl transferase activity but also may act directly by recruiting the RNA polymerase II holoenzyme to the promoter (139). CBP and p300 can interact with Smad2, Smad3, and Smad4 and are required for transcriptional activation of a variety of TGF-ß-dependent promoters (140–145). Moreover, Smad4 contains a unique transcriptional activation domain, the Smad4 activation domain, within the middle-linker region that is both necessary and sufficient to activate Smad-dependent transcriptional responses and that interacts with a distinct domain within the amino-terminus of p300 (63,146). Since CBP and p300 interact with many transcription factors and coactivators (139), this mechanism provides an additional level of transcriptional cross-talk with TGF-ß-signaling pathways. For example, MSG1 is a transcriptional coactivator that interacts functionally with Smad4 in signaling downstream of TGF-ß (127) and modifies transcription through its interaction with CBP and p300 (128). In addition, several proteins have been identified that compete for R-Smad/Smad4 binding to CBP and p300, providing a mechanism of competitive transcriptional inhibition. These include the viral oncoprotein EIA (129), the proto-oncogenes c-Ski and SnoN (134), and the DNA-binding homeodomain protein TGIF (110).
Smad proteins can also interact with proteins that recruit repressor components with histone deacetylase activity to the transcriptional complex (Table 3). TGIF not only interferes with Smad2–p300 interactions but also recruits the histone deacetylase HDAC1 into Smad2-containing transcriptional complexes (125,126). This has the net effect of enhancing the relative balance of suppressor versus activator components associated with the Smad complex and results in the repression of TGF-ß-dependent target gene expression. Likewise, the proto-oncogene products Ski and SnoN interact with Smad2, Smad3, and Smad4, compete for binding to CBP and p300 (109), and recruit the transcriptional corepressor N-CoR to the Smad complex (133–136). N-CoR then recruits mSin3A/B, which, in turn, interacts with HDAC1 and/or HDAC2 to form higher molecular weight transcriptional repressor complexes. Of interest, while Ski and SnoN protein levels are rapidly decreased following activation of TGF-ß receptors as a result of nuclear Smad3-dependent, proteasomal-mediated degradation (133,136), SnoN mRNA expression is increased after prolonged TGF-ß stimulation (136), providing an integrated mechanism for the regulation of TGF-ß-dependent transcriptional responses.
Taken together, these findings suggest a model in which Smad proteins are initially recruited to specific promoters through a combinatorial interaction between the Smad/DNA and Smad/transcription factor complexes. These interactions stabilize preformed protein–DNA complexes and function to regulate transcription through the recruitment of high-molecular-weight complexes containing dominant transcriptional activator or suppressor activities. The formation of these complexes depends on the abundance of these activator and suppressor components at the promoter of interest. This is determined not only by their recruitment by Smads and other locally sequestrated transcription factors but also by the overall abundance of these factors in a particular cell at a given time.
|
CROSS-TALK WITH OTHER SIGNALING PATHWAYS |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
Signaling downstream of TGF-ß receptors cannot be considered in isolation from other intracellular signaling pathways, since these can interface with the Smad-signaling pathways in a number of different ways. The mechanisms of cross-talk between these pathways include the regulation of R-Smad activity by MAP kinase cascades, interaction of Smads with different transcriptional cofactors that are themselves regulated by different signaling pathways, and regulation of I-Smad expression.
Regulation of R-Smads by MAP Kinase Cascades
MAP kinase cascades play pivotal roles in cellular signaling from a wide variety of stimuli resulting, ultimately, in phosphorylation and activation of transcription factors. Distinct MAP kinase pathways lead to downstream activation of either the ERK1/2 or two stress-activated protein kinases, JNK and p38 MAP kinase. Mutational activation of MAP kinase pathways is frequently found in human cancers, most notably the activation of Ras, an upstream activator of the ERK1/2 pathway, which is an early event in many cancers (7,147,148). Two studies (149,150) suggest that persistent activation of MAP kinase pathways by oncogenic Ras can inactivate Smad signaling by retention of R-Smads in the cytoplasm. This mechanism is also operative in SW480.7 Smad4 null colon carcinoma cells, where expression of an activated Ki-ras oncogene results in cytoplasmic retention of Smad2 and Smad3 such that overexpression of Smad4 is unable to restore TGF-ß signaling (151). Collectively, these studies provide a mechanism for the suppression by hyperactivated Ras of TGF-ß-signaling pathways in epithelial cells.
In contrast, a number of reports also show that activation of MAP kinase pathways can enhance TGF-ß-mediated responses in a cell-specific manner. In this context, it has been shown that treatment of cells with hepatocyte growth factor (HGF) or epidermal growth factor (EGF) activates Smad2/Smad4 signaling via an ERK-dependent pathway in epithelial cells (152) and that Smad2 is activated by MEKK1, an upstream activator of the JNK pathway in endothelial cells (153). TGF-ß itself activates a variety of MAP kinase pathways that may be required for both Smad-dependent and Smad-independent transcriptional responses to TGF-ß (154–156). The particular combinations of MAP kinases activated by TGF-ß are cell type dependent. For example, in mink lung epithelial cells, TGF-ß-induced activation of JNK mediates Smad3 phosphorylation, which is required for the transcriptional activation of Smad3-dependent responses (157), while in rat articular chondrocytes, TGF-ß induces activation of ERK1/2, but not p38 or JNK MAP kinases; this pathway is important in the mitogenic response of these cells to TGF-ß (158). Furthermore, TGF-ß-induced activation of fibronectin in a variety of epithelial cell lines requires activation of JNK and is independent of Smad signaling (64).
Exactly how these MAP kinase pathways are activated by TGF-ß is unknown. TGF-ß-induced activation of JNK signaling may involve activation of Rho-like guanosine triphosphatases (54,157,159), but linkage to the TGF-ß receptor complex is unclear. Some insights into upstream links of MAP kinase pathways to TßRI have been provided with the identification of TAK1 and TAB1, a novel MAP kinase kinase kinase (160,161) and its activator, which have been shown to be important for TGF-ß signaling through p38 MAP kinase (162). TAK1 itself is a substrate for the hematopoietic progenitor kinase 1, which is activated following treatment with TGF-ß (163), but direct linkage between the TAB/TAK1 cascade and TGF-ß receptors is unclear. In the case of another TGF-ß superfamily type I receptor, BMPR1A, a direct physical linkage of BMPR1A-XIAP-TAB1-TAK1, has been shown, where XIAP is a human X-chromosome-linked member of the inhibitor of apoptosis (IAP) family (164). Whether this protein interacts with TßRI remains to be determined.
MAP kinases themselves can modify Smad signaling by phosphorylation-dependent modification of ligand-dependent R-Smad nuclear translocation. While the sites of oncogenic Ras-dependent Smad2 and Smad3 phosphorylation have not been identified, mutation of all the potential ERK or JNK sites in the middle-linker regions of Smad2 and Smad3 can restore their TGF-ß-dependent activation and nuclear translocation, even in the presence of activated Ras (150). In contrast, activation of Smad2 by MEKK1- or HGF-induced activation of ERK1/2 MAP kinases is also associated with phosphorylation of the R-Smad outside the C-terminal –SSMS motif and yet results in enhanced nuclear translocation of Smad2 (152,153). Furthermore, TGF-ß-dependent phosphorylation sites in Smad2 overlap with the Smad2 phosphopeptide maps seen following activation of the ERK1/2 MAP kinase pathway by HGF (152), suggesting that common sites of MAP kinase-induced phosphorylation may also be required for TGF-ß-dependent nuclear translocation of the R-Smad/Smad4 complex. These contrasting findings indicate that activation of MAP kinase pathways may have positive or negative regulatory effects on R-Smads, depending on the nature of MAP kinase activation, which, in turn, may affect both the specificity and multiplicity of MAP kinase-dependent phosphorylation events.
Downstream components of MAP kinase-signaling pathways may also interact with the R-Smad/Smad4 complex in the nucleus, providing an additional level of transcriptional cross-talk between these pathways. c-Jun and JunB, both downstream substrates of JNK, are components of the AP-1 complex that are transcriptionally regulated by Smads. These may act by binding directly to an SBE (88,89) or, in the case of the c-Jun promoter, through functional synergy of the Smads with JNK to activate an AP-1 element (Fig. 4) (88,113). In addition, ATF-2, a constitutively expressed member of the c-Jun family and a downstream substrate of both JNK and p38 MAP kinases, is transcriptionally activated as a result of its interaction with Smad3 and Smad4 (117,162). Taken together, these effects suggest a complex integration of signaling pathways resulting in activation of AP-1 via a MAP kinase/Smad-interdependent amplification loop (Fig. 4). This has special significance for oncogenesis, since AP-1 activity is critical both for the auto-induction of TGF-ß1 (77,165) and for the acquisition of a metastatic phenotype (166).
|
Transcriptional Cross-Talk
Activated Smads also interact with a range of other transcriptional components, themselves regulated by diverse signaling pathways (Table 3). This has been discussed in the context of the transcriptional adaptor proteins CBP and p300. In addition, interaction of TGF-ß-activated Smads with a variety of DNA-binding transcription factors may not only affect the transcriptional activation by the Smads but also may influence the activity of the Smad-interacting protein itself. For example, Smad proteins provide an interface between signals from TGF-ß and the steroid hormones. Thus, the liganded glucocorticoid receptor interacts with the MH2 domain of Smad3 in such a way as to repress its transcriptional activating activity (123). In contrast, Smad3 interacts with the vitamin D receptor through its MH1 domain, acting as an inducible coactivator to potentiate the activity of the vitamin D receptor on its response element in a ligand-dependent manner (122). This activation is repressed by Smad7 but not by Smad6, suggesting that pathways affecting Smad7 expression may, in turn, modulate events under control of the vitamin D receptor (167).
Regulation of I-Smad Expression
A number of reports suggest that transcriptional activation of the I-Smad, Smad7, is also an important mechanism of cross-talk between TGF-ß and other signaling pathways. Smad7 is induced by EGF (80), interferon gamma (IFN 
] (168), lipopolysaccharide and the proinflammatory cytokines tumor necrosis factor-
and interleukin 1ß (169). Since the induction of Smad7 by IFN is mediated by Jak1/Stat1 and that of the proinflammatory cytokines is mediated by NF-kB, a framework is emerging for understanding the basis of antagonism between TGF-ß signaling and the wide variety of other signaling path ways that share the ability to activate the Smad7 promoter (Fig. 3, B).
|
IMPLICATIONS FOR CARCINOGENESIS |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
The development of human cancer involves expansion of cell populations with competitive advantage over other cells. This process is initiated when a normal cell undergoes a genetic change that conveys a selective growth advantage, predisposing it to additional mutations that confer further malignant potential. This model of a multistep process of carcinogenesis has been clearly developed in human colorectal cancers (148,170) and in mouse models of skin cancer (7). In these contexts, TGF-ß plays important roles both in the early suppression of malignancy and at later stages in tumor progression (6,7,148,171). Given the importance of Smad proteins in regulating TGF-ß-dependent signals, defects in this signaling pathway are likely to be found in malignant disease.
Smad Mutations in Cancer
Many tumor suppressor genes are inactivated by intragenic mutations or deletions in one allele accompanied by the loss of a chromosomal region containing the other allele, called a loss of heterozygosity (LOH) (172). This may occur spontaneously or may be associated with a syndrome of genetic instability resulting from mutations in DNA proofreading, mismatch repair enzymes (173). Attempts to identify the candidate tumor suppressor gene in chromosome 18q 21.1, which has a particularly high frequency of deletion in pancreatic and colorectal carcinomas, resulted in the identification of DPC4/Smad4 (55,174). This is usually associated with complete loss of the remaining Smad4 allele, although functionally inactivating mutations also occur and have proven valuable in establishing the basic paradigms of Smad/Smad interactions and signal transduction mechanisms (Fig. 5, A and B) (61,175).
The functional significance of Smad4 deletions has been explored in a number of animal model systems. Compound inactivation of one Smad4 allele in Apc716 heterozygous mice accelerates the rate of intestinal tumor formation and invasiveness compared with that of Apc716 heterozygotes and is associated with LOH for both genes (176). Since Apc mutations are thought to arise early in the genesis of colorectal tumors (177), these findings have been taken to suggest that Smad4 mutations play a role in tumor progression rather than in the initiation phases of malignancy. However, recent data suggest that Smad4 mutations may also contribute to the initiation phase of tumorigenesis. Smad4 heterozygous mutant mice develop intestinal polyposis and invasive carcinoma after 6–12 months of age (178,179); in one of these studies (179), loss of the remaining Smad4 wild-type allele was detected only at later stages in tumor progression. This result suggests that, like the tumor suppressor gene p53 (180), haploid insufficiency of Smad4 is sufficient for tumor initiation and that biallelic loss of Smad4 plays a distinct role later in the progression of malignant disease. These findings are supported by the observation that biallelic mutation or deletion of Smad4 usually occurs as a late event in human colorectal tumors (181). Of interest, functionally inactivating germline mutations of the Smad4 locus have been described in families with familial juvenile polyposis (FJP), an autosomal dominant, inherited syndrome associated with hamartomatous polyps and increased risk of gastrointestinal cancer (182,183). Whereas it is not known if the truncated Smad4 associated most commonly with FJP has dominant negative activity on TGF-ß signaling, biallelic inactivation of the Smad4 gene is rare, providing support for the hypothesis that haploid insufficiency of Smad4 may be sufficient for tumor initiation in human disease.
Allelic loss of chromosome 18q21.1 is a relatively common event in a wide range of malignancies (184) and is associated with greater mortality and an increased risk of metastatic spread (185,186). However, biallelic deletion or inactivation of Smad4 is largely restricted to tumors of the pancreas and gastrointestinal tract (55,174), with a much lower frequency in lung, breast, ovarian, and squamous cell carcinomas (184,187–190). This discrepancy could reflect the importance of Smad4 haploid insufficiency in the initiation of carcinogenesis but could also result from the presence of additional tumor suppressor loci within this segment of the chromosome. Smad2 and DCC (deleted in colorectal carcinoma) are potential tumor suppressor genes also located on this chromosome 18q21.1 (191). Given its role in TGF-ß signaling, Smad2 is a candidate tumor suppressor; however, unlike Smad4, the remaining allele is mutated in only a small proportion of colorectal and lung tumors (Fig. 5, B) (191–193). Loss of expression of DCC has been reported in colorectal carcinomas (177), but its role as a tumor suppressor is unclear because mutations in its coding region are rare (194,195). Finally, Smad7 is also located on chromosome 18q21.1 (196) and, as a suppressor of TGF-ß signaling, is a candidate oncogene. Two studies (181,197) have failed to identify activating mutations or amplifications of the Smad7 gene in colorectal or pancreatic carcinomas, although increased expression of Smad7 has been reported in pancreatic cancer (198), suggesting that aberrant expression of this I-Smad may play a role in tumorigenesis.
Taken together, these data indicate that the principal oncogenic function of Smad4 mutations in malignancy could result from tumor initiation associated with Smad4 haploid insufficiency. In addition, while mutational inactivation of Smad4 is also of significance as a late event in tumorigenesis associated with pancreatic and gastrointestinal epithelia, the relative infrequency of LOH at this locus in other malignancies suggests that there may be additional, as yet unidentified, tumor suppressor loci within chromosome 18q21.
A second cluster of Smad genes, Smad3 and Smad6, are located on chromosome 15q21–22 (199), which is a frequent site of allelic loss in breast, colorectal, lung and pancreatic tumors (200–203). However, to date, there is no evidence that either Smad3 or Smad6 are the locus of homozygous deletions, functionally inactivating mutations, or amplifications in human malignancy (181,197,204–207). These findings contrast with the high frequency of metastatic colorectal tumors reported in Smad3 null mice (208). While this effect has been shown to be strain dependent and may reflect differences between murine and human intestinal epithelia, two other groups have not found any evidence of colorectal carcinoma in homozygous Smad3 mutant mice (209,210), despite extensive search and rederivation of the mice in different backgrounds (49). This discrepancy could result from the expression of a hypomorphic Smad3 allele that may have oncogenic activity in those mice developing colorectal tumors. Alternatively, undefined epigenetic factors may have contributed to the variable rates of malignancy. Taken together, these data indicate that Smad3 has a complex role in the regulation of malignant cell phenotypes and suggest that, while Smad4, and possibly Smad2, on balance, have tumor suppressor activities, Smad3 may contribute to the oncogenic activities of TGF-ß.
Additional mutations or amplifications of tumor suppressor or oncogenic loci in different chromosomal segments may also interface with the Smad-signaling pathway and promote functional inactivation of these responses in the context of allelic deletions of Smad2, 3, or 4. While there are no data on the role that Smad2- or 4-interacting proteins may play in tumorigenesis, functional inactivation of Smad3 is associated with amplifications of three known Smad3-interacting proteins. Chromosomal rearrangements resulting in the fusion products AML/Evi-1 (211) or AML/MDS/Evi-1 (212) are frequent in myeloid leukemia and myelodysplasia. Both fusion proteins interact with and functionally inactivate Smad3, interfering with growth inhibition of myeloid cells by TGF-ß and suggesting that Evi-1 may interfere with TGF-ß signaling mediated by Smad3 in these tumors. The oncoprotein c-Ski is also frequently involved in chromosomal translocations associated with non-Hodgkin's lymphoma (213) and pre-B acute lymphoblastic leukemia (214), again potentially suppressing the activity of Smad3 in these hematopoietic malignancies (133–135). Taken together, these findings suggest that, while biallelic mutations of Smad3 have not been found in human malignancy, dysregulated expression of Smad3 inhibitors may contribute to the functional inactivation of Smad3, especially in hematopoietic malignancies.
Regulation of Smad Expression in Cancer
It is predictable that regulation of Smad expression, either through gene silencing, induction, or targeted protein degradation, might play some role in the development of malignancy. Gene silencing has been described in association with Smad4 heterozygosity in pancreatic cancer (215), but this is a relatively rare event, and the underlying mechanisms have not been determined. However, immunohistochemical studies do reveal variations in Smad expression in different tumors. For example, the expression of Smad2 and Smad3 is decreased in epithelial components of human skin and rat prostatic carcinomas (216,217), while in colorectal tumors this seems to be increased (218). Increased expression of Smad6 and Smad7 has also been described in human pancreatic and rat prostate carcinomas (198,216,219). The mechanisms underlying these changes in Smad expression are unknown, although a recent study (220) has shown that certain mutations in Smad2 and Smad4 found in human cancers can selectively target these mutant proteins for ligand-independent ubiquitin-mediated proteasomal degradation through the UbcH5 family of E2 ligases, in effect silencing protein expression.
|
TGF- ß SIGNALING IN CONTEXT |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
It is now generally accepted that the growth-suppressive effects of TGF-ß constitute a tumor-suppressor pathway that includes the ligand, receptors, downstream signal transducers, and their transcriptional targets. Loss of this tumor suppressor pathway in epithelial cells is most commonly found at later stages in carcinogenic progression (6,7,148) and results, principally, from mutational inactivation or transcriptional repression of TßRII (5). However, it is beginning to be appreciated that loss of the high-flux growth-inhibitory signaling pathway does not require complete loss of TGF-ß signaling in the tumor cells. Rather, given the complex role of TGF-ß in carcinogenesis, its signaling pathways are likely to have both tumor suppressor and oncogenic components. Reduction in TßRII levels associated, for example, with H-ras expression (10), or the partial inactivation of downstream signaling components, for example, following allelic loss of Smad4 in colorectal tumors, might then allow the oncogenic pathways to function unopposed, including those dependent on AP-1, and involving induction of TGF-ß1, invasion, and metastasis (7). The observation that colonic carcinomas with BAT-TßRII mutations that completely inactivate TßRII have a better prognosis than tumors in which the TGF-ß-signaling pathway remains partially functional (221) suggests that oncogenic responses mediated by TGF-ß are operative in the latter and that tumor cell selection favors partial, rather than complete, inactivation of this signaling pathway.
On the basis of these observations, we propose that Smad3, which has not been found to be mutationally inactivated in human malignancies, mediates key oncogenic functions of TGF-ß. These functions of Smad3 may include synergism with the Ras-ERK-AP-1-, TAK1-p38-ATF-2-, or Rho-JNK-c-Jun/ATF-2-signaling pathways in activation of AP-1 sites, important in invasion and metastasis as well as in transduction of signals mediating the autoinduction of TGF-ß. These oncogenic functions of TGF-ß may be less critical in certain hematopoietic malignancies where Smad3 function is blocked by fusion proteins containing the Evi-1 or Ski proto-oncogenes, suggesting that the inhibition of Smad3 may provide a specific selection advantage in these tumors. Furthermore, these data favor a model in which production of TGF-ß by a tumor cell serves not only to activate surrounding stromal elements in a paracrine fashion but also to activate oncogenic pathways via autocrine effects on the tumor cell itself.
|
REFERENCES |
|---|
|
TopAbstractIntroductionTGF-{beta} Receptors as Tumor...TGF- {beta} Receptor-Interacting...TGF- {beta} Signaling by...Transcriptional Regulation of...Cross-Talk With Other Signaling...Implications for CarcinogenesisTGF- {beta} Signaling in...References |
|---|
1 Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000;100:57–70.[CrossRef][ISI][Medline]cancerlit;20112457
2 Bissell MJ, Weaver VM, Lelievre SA, Wang F, Petersen OW, Schmeichel KL. Tissue structure, nuclear organization, and gene expression in normal and malignant breast. Cancer Res 1999;59(7 Supp):1757–1763s; discussion, 1763s.
3 Rowley DR. What might a stromal response mean to prostate cancer progression? Cancer Metastasis Rev 1998–99;17:411–9.[CrossRef][ISI][Medline]
4 de Visser KE, Kast WM. Effects of TGF-beta on the immune system: implications for cancer immunotherapy. Leukemia 1999;13:1188–99.[CrossRef][ISI][Medline]cancerlit;99377955
5 Markowitz SD, Roberts AB. Tumor suppressor activity of the TGF-beta pathway in human cancers. Cytokine Growth Factor Rev 1996;7:93–102.[CrossRef][Medline]cancerlit;97017740
6 Reiss M. TGF-beta and cancer. Microbes Infect 1999;1:1327–47.[CrossRef][ISI][Medline]cancerlit;20079459
7 Akhurst RJ, Balmain A. Genetic events and the role of TGF beta in epithelial tumour progression. J Pathol 1999;187:82–90.[CrossRef][ISI][Medline]cancerlit;99273254
8
Chen RH, Ebner R, Derynck R. Inactivation of the type II receptor reveals two receptor pathways for the diverse TGF-beta activities. Science 1993;260:1335–8.
9
Laiho M, Ronnstrand L, Heino J, DeCaprio JA, Ludlow JW, Livingston DM, et al. Control of junB and extracellular matrix protein expression by transforming growth factor-beta 1 is independent of simian virus 40 T antigen-sensitive growth-sensitive growth-inhibitory events. Mol Cell Biol 1991;11:972–8.
10
Zhao J, Buick RN. Regulation of transforming growth factor beta receptors in H-ras oncogene-transformed rat intestinal epithelial cells. Cancer Res 1995;55:6181–8.
11 Lu SL, Kawabata M, Imamura T, Miyazono K, Yuasa Y. Two divergent signaling pathways for TGF-beta separated by a mutation of its type II receptor gene. Biochem Biophys Res Commun 1999;259:385–90.[CrossRef][ISI][Medline]cancerlit;99293068
12
Kingsley DM. The TGF-beta superfamily: new members, new receptors, and new genetic tests of function in different organisms. Genes Dev 1994;8:133–46.
13 Massague J. TGF-beta signal transduction. Annu Rev Biochem 1998;67:753–91.[CrossRef][ISI][Medline]cancerlit;98431667
14
Piek E, Heldin CH, Ten Dijke P. Specificity, diversity, and regulation in TGF-beta superfamily signaling. FASEB J 1999;13:2105–24.
15 Kim SJ, Im YH, Markowitz SD, Bang YJ. Molecular mechanisms of inactivation of TGF-beta receptors during carcinogenesis. Cytokine Growth Factor Rev 2000;11:159–68.[CrossRef][ISI][Medline]cancerlit;20175836
16
Goggins M, Shekher M, Turnacioglu K, Yeo CJ, Hruban RH, Kern SE. Genetic alterations of the transforming growth factor beta receptor genes in pancreatic and biliary adenocarcinomas. Cancer Res 1998;58:5329–32.
17
Markowitz S, Wang J, Myeroff L, Parsons R, Sun L, Lutterbaugh J, et al. Inactivation of the type II TGF-beta receptor in colon cancer cells with microsatellite instability. Science 1995;268:1336–8.
18
Grady WM, Myeroff LL, Swinler SE, Rajput A, Thiagalingam S, Lutterbaugh JD, et al. Mutational inactivation of transforming growth factor beta receptor type II in microsatellite stable colon cancers. Cancer Res 1999;59:320–4.
19
Amoroso SR, Huang N, Roberts AB, Potter M, Letterio JJ. Consistent loss of functional transforming growth factor beta receptor expression in murine plasmacytomas. Proc Natl Acad Sci U S A 1998;95:189–94.
20 Knaus PI, Lindemann D, DeCoteau JF, Perlman R, Yankelev H, Hille M, et al. A dominant inhibitory mutant of the type II transforming growth factor beta receptor in the malignant progression of a cutaneous T-cell lymphoma. Mol Cell Biol 1996;16:34