THE TGF- SUPER-FAMILY, STRUCTURE and ACTIVATION

The TGF- super-family of growth factors includes the various forms of TGF-, bone morphogenic proteins (BMP), nodals, activins, the anti-Mullerian hormone, and many other structurally related factors found in vertebrates, insects, and nematodes (^{Massagué, 1998}). There are three mammalian isoforms of TGF- (TGF-1–3), structurally nearly identical, with a knot motif composed of six cysteine residues joined together by three intrachain disulfide bonds that stabilize -sheet bands. One free cysteine forms an interchain disulfide bond with an identical monomeric chain to generate the mature TGF- dimer.

TGF- are secreted as latent precursor molecules (LTGF-) requiring activation into a mature form for receptor binding and subsequent activation of signal transduction pathways. The LTGF- molecules consist of 390–414 amino acids. They contain an amino-terminal hydrophobic signal peptide region, the latency-associated peptide (LAP) region, of 249 residues, and the C-terminal, potentially bioactive region that contains 112 amino acids per monomer. LTGF- is usually secreted as a large latent complex covalently bound via the LAP region to LTGF--binding protein (LTBP;^{Roberts, 1998}) or as a small latent complex without LTBP. The LAP confers latency to the complex, whereas LTBP serves to bind TGF- to the ECM and to enable its proteolytic activation (^{Nunes et al, 1997}).

Activation of TGF- is a complex process involving conformational changes of LTGF-, induced by either cleavage of the LAP by various proteases such as plasmin, thrombin, plasma transglutaminase, or endoglycosylases, or by physical interactions of the LAP with other proteins, such as thrombospondin-1, leading to the release of bioactive, mature, TGF- (^{Roberts, 1998}).

TGF-/Smad SIGNAL TRANSDUCTION MACHINERY

TGF- receptors

Upon activation, TGF- superfamily members initiate their cellular action by binding to serine/threonine kinase receptors. The TGF- receptor family consists of two structurally similar subfamilies, type I and type II receptors, with small cysteine-rich extracellular regions and intracellular portions consisting mainly of their kinase domains. Type I receptors have a region rich in glycine and serine residues (GS domain) preceding the receptor kinase domain (^{Huse et al, 1999}). To exert their signal, type II and type I receptors act in sequence: TGF- first binds to the type II receptor (TRII), which occurs in the cell membrane in an oligomeric form with intrinsic kinase activity; TGF- type I receptor (TRI) is then recruited and phosphorylated in its GS domain by TRII, leading to activation of its kinase activity and subsequent intracellular signaling (^{Massagué, 1998;Piek et al, 1999}). Betaglycan, a transmembrane proteoglycan also known as TRIII (^{Lopez-Casillas et al, 1991}), allows high-affinity binding of TGF- to TRII but does not itself transduce signal (^{Rodriguez et al, 1995;Brown et al, 1999}).

Smad proteins

Following ligand activation, signaling from TRI to the nucleus occurs predominantly by phosphorylation of cytoplasmic mediators belonging to the Smad family (^{Piek et al, 1999;Massagué and Chen, 2000;Massagué and Wotton, 2000}). Type I receptors specifically recognize and phosphorylate the ligand-specific receptor-activated Smad (R-Smad, Figure 1).

Figure 1.

Structural domains of an R-Smad. R-Smad consist of two conserved globular domains known as MH1 (Mad homology 1) and MH2 domains, linked by a linker region. In the basal state, R-Smad remain in an inactive conformation through an auto-inhibitory MH1/MH2 interaction. Phosphorylation of the C-terminal SSXS motif by activated TRI results in R-Smad activation, heteromerization with Smad4, and subsequent translocation into the cell nucleus. The Smad3 domain in MH1 recognizes the DNA sequence CAGAC; the MH2 domain is involved in protein/protein interactions with co-Smad, transcriptional coactivators and corepressors.

Full figure and legend (10K)

The latter are recruited to activated TRI by a membrane bound cytoplasmic protein called SARA (Smad Anchor for Receptor Activation;^{Tsukazaki et al, 1998}). R-Smad include Smad1, Smad5, and Smad8 downstream of the BMP, and Smad2 and Smad3 downstream of TGF- and activin. They all consist of two conserved Mad-homology (MH) domains that form globular structures separated by a linker region (^{Shi et al, 1997}). The N-terminal MH1 domain has DNA-binding activity, whereas the C-terminal MH2 domain has protein-binding properties. Phosphorylation of R-Smad by type I receptors occurs principally on two serine residues within a conserved –SS(M/V)S–motif at their C-terminus (^{Massagué and Chen, 2000;Massagué and Wotton, 2000}).

Upon phosphorylation by TRI, R-Smad form heteromeric complexes with co-Smad, such as Smad4. Co-Smad act as a convergent node in the Smad pathways downstream of the TGF- superfamily receptors, complexing R-Smad, regardless of the TGF- ligand specificity of the latter. R-Smad/Smad4 complexes are then translocated into the nucleus by a mechanism involving the cytoplasmic protein importin (^{Xiao et al, 2000;Kurisaki et al, 2001}). They may then function as transcription factors, binding DNA either directly or in association with other DNA binding proteins (^{Piek et al, 1999;Massagué and Chen, 2000;Massagué and Wotton, 2000}). Maximal affinity of recombinant Smad3 and Smad4 to DNA is observed with the 5 bp sequence, CAGAC (^{Shi et al, 1998;Zawel et al, 1998}). Smad2, on the other hand, does not bind DNA directly, requiring a nuclear DNA-binding protein of the Fast family to bind DNA, in association with Smad4, and activate transcription in response to TGF- and activin (^{Chen et al, 1997;Labbé et al, 1998;Liu et al, 1999}).

A third group of Smad proteins, the inhibitory Smad (I-Smad), such as Smad6 or Smad7, prevent R-Smad phosphorylation and subsequent nuclear translocation of R-Smad/Smad4 heterocomplexes (^{Imamura et al, 1997;Nakao et al, 1997}).

Following target gene transcription, Smad complexes are released from the chromatin and undergo ubiquitination, followed by proteasomal degradation (^{Zhu et al, 1999}).

A summary of the various steps of TGF- signaling through the Smad cascade is provided in Figure 2.

Figure 2.

Schematic representation of the Smad signaling cascade downstream of the TGF- receptors. Initiation of signaling occurs when TGF- binds to the serine/threonine kinase known as TRII. Ligand-bound TRII complexes with and phosphorylates the transducer receptor, TRI, which becomes activated. Activated TRI then recruits and phosphorylates R-Smad, leading to the association of the latter with a co-Smad, such as Smad4, and translocation of the formed heterocomplex into the nucleus where it acts as a transcriptional regulator of target genes.

Full figure and legend (37K)

Transcriptional cross-talks

Several cross-signaling mechanisms have been described that implicate Smad proteins. For example, Smad3 interacts with the vitamin D receptor to mediate the effect of TGF- on vitamin D3-induced transcription (^{Yanagisawa et al, 1999}). Smad3 also interacts with TFE3 and Sp1 to activate transcription from the PAI-1 and p21^cip promoters, respectively (^{Hua et al, 1998;Moustakas and Kardassis, 1998}). Smad/AP-1 interactions have also been reported, which may lead to either additive (^{Zhang et al, 1998;Liberati et al, 1999;Verrecchia et al, 2001c}) or antagonistic (^{Verrecchia et al, 2000;Verrecchia et al, 2001c}) activities on gene transactivation. The outcome is dependent on the structure of target promoters, as not only the presence of AP-1- and/or Smad-specific cis-elements, but also their respective positions, may influence the transcriptional response (^{Liberati et al, 1999;Verrecchia et al, 2001c}). For example, in a promoter context exhibiting Smad-specific boxes distant from functional AP-1 sites, such as in the PAI-1 and c-jun promoters, transcriptional cooperation between Smad and Jun has been observed, without formation of Smad/Jun heterocomplexes on DNA (^{Liberati et al, 1999;Verrecchia et al, 2001c}). In the context of the COL7A1 promoter that possesses an AP-1 site located within a bipartite Smad-specific TGF--response element, inhibition of Smad-driven transactivation by Jun members was observed, again without formation of Smad/Jun heterocomplexes on DNA. In the latter case, Jun was shown to act as a direct repressor of Smad function, binding to and blocking transcriptional activation domains intrinsic to Smad3, and preventing its binding to DNA (^{Verrecchia et al, 2001c}).

Another possibility for Smad to interfere with other transcription factors occurs through direct interactions with transcriptional coactivators such as the Ski-interacting protein (SKIP) or CBP/p300. The latter proteins modify transcription either by altering chromatin structure so that the underlying DNA sequences are exposed to the transcriptional apparatus (^{Workman and Kingston, 1998}), or by directly recruiting the RNA polymerase II holoenzyme to the promoter (^{Snowden and Perkins, 1998}). Smad–CBP/p300 interaction is required for transcriptional activation of several TGF--dependent promoters (^{Feng et al, 1998;Janknecht et al, 1998;Nishihara et al, 1998;Pouponnot et al, 1998;Shen et al, 1998}). SKIP, a nuclear hormone receptor coactivator, enhances TGF--dependent transactivation by interacting with Smad2 and Smad3 proteins (^{Leong et al, 2001}).

In addition, several nuclear proteins, including the viral oncoprotein E1A (^{Nishihara et al, 1999}), the proto-oncogenes c-Ski and SnoN (^{Akiyoshi et al, 1999;Stroschein et al, 1999}), TGIF (^{Wotton et al, 1999}), Snip-1 (^{Kim et al, 2000}), SIP1 (^{Verschueren et al, 1999}), and c-Jun (^{Verrecchia et al, 2000}) compete for R-Smad/Smad4 binding to CBP or p300, thereby interfering with Smad-dependent gene expression.

TGF-/Smad MODULATION OF ECM GENE EXPRESSION

Despite extensive research directed toward the elucidation of the role played by Smad proteins downstream of TGF-, very few direct Smad target genes have been characterized. Indeed, it was shown that transactivation of the fibronectin gene downstream of TGF- is a JNK-specific, Smad-independent, mechanism (^{Hocevar et al, 1999}), suggesting that pathways that alternate to the Smad cascade also play an important part in the modulation of ECM gene expression by TGF-.

Recently, using a combined cDNA microarray/promoter transactivation approach, we have identified several new Smad gene targets among which are COL1A1, COL3A1, COL5A2, COL6A1, COL6A3, and TIMP-1. Most notably, these data indicate that the Smad signaling pathway is crucial for simultaneous activation of several fibrillar collagen genes by TGF-. About 60 other ECM-related genes were also identified as immediate-early genes targets downstream of TGF- (^{Verrecchia et al, 2001a}).

Our group identified the first described genuine Smad binding sequence (SBS) within the human COL7A1 promoter (^{Vindevoghel et al, 1998a, b}). Specifically, TGF- upregulation of COL7A1 gene expression is mediated by rapid and transient binding of a Smad-containing complex to the region -496/-444 of the COL7A1 promoter, a bipartite element consisting of a CAGA tandem repeat in its 5' end, and a Medea-like (Drosophila Smad4 homolog) binding site in 3', both framing a potential AP-1 binding site whose integrity is not necessary for either Smad-driven transactivation or Smad/DNA complex formation (^{Verrecchia et al, 2001c}). A few other ECM-related genes whose expression is modulated by TGF- have also been identified as potential Smad targets, and CAGA-like elements have been identified within their promoter regions, that bind Smad complexes. These genes include PAI-1, COL1A2, and the 5 integrin gene (INTB5) (^{Chen et al, 1998;Dennler et al, 1998;Lai et al, 2000}). In these three cases, TGF--induced promoter activation may also involve functional interactions between Smad and Sp1, a mechanism that has also been shown to be critical for activation of p21/WAF1/CIP1 (^{Pardali et al, 2000}) and p15/Ink4B (^{Feng et al, 2000}) by TGF- in the control of cell proliferation.

Regarding COL1A2, initial observations demonstrated that a 135 bp region of the COL1A2 promoter within 330 bp of the transcription start site confers responsiveness to TGF- (^{Inagaki et al, 1994}). The minimal TGF--response element was further refined to the region between nucleotides -271 and -235 (^{Chung et al, 1996}). The latter contains potential overlapping cis-element for Smad, AP-1, and NF-B (^{Chung et al, 1996;Chen et al, 1998;Kouba et al, 1999}). Cooperation between Smad3 and Sp1, the latter binding upstream elements, to transactivate the COL1A2 promoter has been described (^{Zhang et al, 2000;Inagaki et al, 2001}). The role of this cooperation in mediating the TGF- effect is unclear, as elimination of the corresponding Sp1 sites upstream of the Smad-specific CAGACA motif does not prevent TGF--driven promoter transactivation (^{Chung et al, 1996}). Also, it has been shown that Smad–p300/CBP interactions are a key event in TGF--driven collagen gene transactivation (^{Ghosh et al, 2000}), and that interferon--driven antagonistic activity against TGF- effect is integrated at the level of p300/CBP, the latter being sequestered by interferon-induced Stat1 (^{Ghosh et al, 2001}).

Besides playing a part in the regulation of the expression of ECM components, Smad have recently been identified as capable of mediating the inhibitory activity of TGF- on interstitial collagenase (matrix metalloproteinase-1) gene activation by pro-inflammatory cytokines (^{Yuan and Varga, 2001}). In that study, the authors demonstrate that Smad3 and Smad4, but not Smad1 or Smad2, overexpression block IL-1-driven transactivation of the matrix metalloproteinase-1 promoter, both by and via p300 sequestration.

Smad signaling in wound healing and fibrosis

Numerous studies have documented the ability of exogenous TGF- to improve wound healing (^{Roberts and Sporn, 1993}). Recent work involving inactivation of the Smad3 gene in mice has allowed to explore specifically the contribution of the Smad pathway to tissue repair (^{Ashcroft et al, 1999}). In contrast to predictions made on the basis of the ability of exogenous TGF- to improve wound healing, Smad3-null (Smad3^ex8/ex8) mice paradoxically show accelerated cutaneous wound healing compared with wild-type mice, characterized by an increased rate of re-epithelialization and significantly reduced local infiltration of monocytes. Smad3^ex8/ex8 keratinocytes show altered patterns of growth and migration, and Smad3^ex8/ex8 monocytes exhibit a selectively blunted chemotactic response to TGF-. From these data, it may therefore be concluded that Smad3 may mediate in vivo signaling pathways that are inhibitory to wound healing. These data suggest that "natural" wound healing may involve suppression of the Smad3 level, and that complete loss of this signaling intermediate further accelerates the wound healing process. In the same study, it was shown that reduced deposition of ECM in Smad3^–/– mice could be reversed with the addition of exogenous TGF-1, which implies that ECM gene regulation by TGF- in fibroblasts in vivo may also occur via Smad-independent pathways. Also, this report indicates that disruption of the Smad3 pathway in vivo, coupled with exogenous TGF- signaling through intact alternate pathways, may be of therapeutic benefit in accelerating all aspects of impaired wound healing.

Type I collagen and ECM accumulation is one of the hallmarks of fibrotic conditions. For example, affected skin areas from patients with systemic sclerosis exhibit abnormal accumulation of various ECM components, predominantly type I and type III collagens, but also types V and VII, as well as various proteoglycans. Accumulation of ECM proteins in fibrotic conditions is accompanied by elevated mRNA steady-state levels of fibrillar collagens (^{Kähäri et al, 1984;Ohta and Uitto, 1987}). It is generally accepted that transcriptional activation of ECM-related genes, and in particular that of collagen genes, is a critical event underlying the progressive nature of tissue fibrosis, as is observed in cutaneous lesions from systemic sclerosis or keloid scars (reviewed in^{Jimenez and Saitta, 1999;Uitto and Kouba, 2000}). An essential factor contributing to the metabolic modulation of ECM genes expression is TGF-, although other factors such as connective tissue growth factor, whose expression is controlled by TGF- in a Smad-dependent manner (^{Holmes et al, 2001;Leask et al, 2001}), which may contribute to some of the TGF- effects on extracellular matrix gene expression and subsequent development of tissue fibrosis.

Interfering with fibrotic processes

It is reasonable to hypothesize that a better understanding of the mechanisms of TGF--mediated upregulation of ECM gene expression in fibrotic conditions will provide promising and novel approaches to the therapy of these incurable diseases. In this context, it should be noted that the transcription factor Sp1 is able to interact physically with Smad proteins, leading for example to the activation of the p21, p15, and COL1A2 genes in response to TGF- (^{Moustakas and Kardassis, 1998;Feng et al, 2000;Inagaki et al, 2001}). Sp1 is also important for the basal expression of various collagens, as well as for the expression of TGF- itself (^{Geiser et al, 1993}). Altering the function of Sp1 may therefore represent an interesting alternative to interfere with the pro-fibrotic activities of TGF-. In this context, we have recently demonstrated that Sp1 gene targeting in fibroblasts broadly inhibits ECM gene expression, as measured using differential hybridization of cDNA microarrays, as well as various promoter transactivation assays (^{Verrecchia et al, 2001b}). In addition, using a transgenic mouse model carrying a COL1A2 promoter/luciferase reporter transgene, we have shown that decoy Sp1-binding oligonucleotides inhibit COL1A2 promoter activity in vivo. Such an approach, once optimized, may represent an interesting therapeutic alternative toward the treatment of fibrotic disorders.

Altering the Smad pathway may also be achieved at the level of transcriptional coactivators, such as p300. For example, it has been shown that RelA, cJun, and Stat1 are able to inhibit Smad-driven promoter transactivation by competing with Smad for limiting amounts of cellular p300 (^{Bitzer et al, 2000;Ghosh et al, 2000;Verrecchia et al, 2000}). These mechanisms may explain, at least in part, the antagonistic activity of tumor necrosis factor- and interferon- on TGF--driven COL1A2 expression (^{Chung et al, 1996;Verrecchia et al, 2000;Ghosh et al, 2001}).

Other approaches may include local overexpression of the inhibitory Smad, Smad7, targeting of connective tissue growth factor in situations where it is proven to be the mediator of the TGF- effect on ECM deposition, Several options have been discussed in a recent review (^{Branton and Kopp, 1999}).

Conclusions and perspectives

Tremendous progress has been accomplished over the past several years in the understanding of the initial steps of TGF- intracellular signaling. The identification of Smad proteins as direct links between the cell surface and the nucleus has allowed for the elucidation of critical events leading to gene activation by TGF-. Several hurdles remain before the TGF-/Smad pathway can be targeted directly in situations such as tissue fibrosis or impaired wound healing. The cell-type specificity of Smad3 effects, together with the identification of alternate signaling pathways for TGF- remain of critical importance. For example, even though fibrillar collagens have been identified as direct Smad3/4 target genes, it has been demonstrated that Smad3-deficient mice heal faster than their wild-type counterparts, due to accelerated re-epithelialization and reduced monocytic influx at the site of injury. In these mice, exogenously added TGF- can still activate type I collagen gene expression. These data suggest that, in vivo, Smad3 may have effects that are inhibitory to wound healing. Also, they indicate that disruption of the Smad3 pathway in vivo, coupled with exogenous TGF- signaling through intact alternate pathways, may be of therapeutic benefit in accelerating all aspects of impaired wound healing.

References

References