Smad Complex Formation: Structural Perspectives

The interactions through which the Smads complex with each other and with various partner proteins are critical determinants of signalling specificity and we are now beginning to understand more about how these interactions are mediated structurally. Much evidence suggests that oligomeric complexes of Smads are trimers that interact through their MH2 domains (Chacko et a l, 2001; Jayaraman and Massagué, 2000; Kawabata et ah, 1998; Shi et al., 1997). Structural support for this comes from the crystal structure of the Smad4 MH2 domain (Shi et ah, 1997) which is a trimer. Each monomer within this complex contains a core p-sandwich capped at one end by a three- helix bundle (H3, H4 and H5 a-helices) and at the other end by a loop-helix region (LI, L2 and L3 loops and HI a-helix). In the crystals, the homotrimer has 3 identical extensive protein-protein interfaces comprising the three-helix bundle of one subunit packing against the loop-helix bundle of the adjacent subunit (Shi et al., 1997). Phosphorylation of Smad2 also results in the formation of a trimer, the crystal structure of which is very similar to that of the Smad4 trimer (Wu et al., 2001b). This trimer is further strengthened when the phosphorylated C-terminus of one monomer contacts the L3/B8 ‘loop-strand’ pocket of the adjacent monomer. Importantly, although the L3/B8 ‘loop-strand’ pocket in Smad4 is poorly conserved with respect to R-Smads, the four residues that coordinate the phosphate groups of the C-terminal tail are invariant. It is thus supposed that heterotrimers of Smads would also assemble in such a manner (Chacko et al., 2001; Qin et al., 2002). Many of the interface residues that mediate important contacts with the adjacent subunit are also invariant between the R-Smads and the Co-Smads, consistent with the physiological importance of the trimer (Shi et al.,

1997).

By far the most commonly mutated components of the TGF-P pathway in cancer are Smad2 and Smad4. DPC4 (Smad4) is a tumour suppressor gene located on 18q21 which has been demonstrated to be mutated or deleted in approximately 50 % of pancreatic carcinomas, and 15 % of colorectal cancers (Duff and Clarke, 1998). Smad2

is also located on 18q21 (Eppert et al., 1996) and is found to be mutated in a small number of colorectal and lung cancers. Tumour-derived missense mutations of the

Chapter 1_____________________________________________________Introduction

Smads commonly occur in the MH2 domain. Interestingly, nearly half of these mutated residues are found at the trimer interface and map to the loop-helix and three-helix bundles, and would be involved in critical hydrogen bonds and hydrophobic interactions (Shi et ah, 1997). Thus, these mutations prevent homo- and heterotrimer formation of the Smads in response to TGF-p providing a possible explanation for the disruptive effects of these mutations.

There is some biochemical evidence for dimeric Smad2/Smad4 complexes (Jayaraman and Massagué, 2000; Wu et a i, 2001a; Wu et ah, 2001b), suggesting that Smad complexes with different stoichiometries are possible. In support of this, the R- Smads have recently been shown to form hetero-dimers or trimers with Smad4 when complexed with transcription factors on DNA (Inman and Hill, 2002). The exact stoichiometries of these complexes may be determined by the associated DNA-binding transcription factors: when Smad2/Smad4 complexes interact with XFast-1 or XFast-3 they do so as heterotrimers, but Smad3/Smad4 complexes interact with the Smad- binding element of the c-Jun promoter as heterodimers (Inman and Hill, 2002).

1.6.7 Insights into the Functional Role of the Smads

Targeted gene disruption in mice has been invaluable for our understanding of the functions of this family of proteins. Genetic manipulation has highlighted the role of the Smads in multiple biological processes during embryonic development including angiogenesis, gastrulation and organogenesis, as well as tumorigenesis, wound healing and the immune response (Weinstein et aL, 2000).

Mice deficient in Smad2 or Smad4 are embryonic lethal (E7.5-8.5), with both exhibiting severe gastrulation defects. Smad2 mutant mice fail to form an egg cylinder and lack all mesoderm, as evidenced by the complete lack of Brachyury (T) expression (Nomura and Li, 1998; Weinstein et a l, 1998). A proportion of Smad2 heterozygous embryos also exhibit gastrulation defects, and lack mandibles and eyes later in development, indicating that signalling through Smad2 may function in a dose- dependent fashion (Nomura and Li, 1998). Smad2 is therefore required for egg cylinder elongation, gastrulation, and mesoderm induction. Mice deficient in Smad2 are

strikingly similar to those mutant for Nodal ox ActRIB (Conlon et al., 1994; Gu et al.,

1998), suggesting that Smad2 functions in the Nodal pathway. Interestingly, Smad2

mutant mice generated by two further laboratories displayed a transient induction of mesoderm (Heyer et al., 1999; Waldrip et al., 1998), suggesting that Smad2 was dispensable for mesoderm induction. This phenotype is, however, likely due to the expression of a small amount of truncated Smad2 protein generated from an internal methionine, since these mutants were created by disruption of the first coding exon (see Chapter 6). Smad4 mutant embryos exhibit a very similar, although more severe, phenotype to those lacking Smad2 (Sirard et al, 1998; Yang et a l, 1998). In addition to all the defects seen in Smad2 mutant mice, those deficient in Smad4 also reveal a defect in epiblast cell proliferation. Since Smad4 is involved in both the TGF-P and the BMP signal transduction pathways, it is not surprising that Smad4 mutant mice suffer a more dramatic phenotype.

In contrast to Smad2 and Smad4 mutant mice, SmadS mutant mice are viable and make it to term (Datto et a l, 1999), thereby demonstrating distinct functions of these three Smads during vertebrate development. These mice exhibit a number of interesting phenotypes, which are often dependent on the genetic background. The major phenotype of Smad3 mutant mice appears to be one of immune compromise, and post- weaning these mice develop inflammation in the stomach, pancreas and mucosal membranes, consistent with an overactive immune system (Yang et a l, 1999b). In addition, these mice exhibit increased susceptibility to infection, often forming large bacterial subcutaneous and mucosal abscesses. This implicates Smad3 in roles to suppress immune cell growth, and in combating bacterial infection. In some cases, adult SmadS knock-out mice have accelerated wound healing characterised by an increased rate of re-epithelialisation and impaired local inflammatory response (Ashcroft

et a l, 1999), indicating that Smad3 mediates TGF-P signalling pathways in vivo that are inhibitory to wound healing. In addition, studies using Smad3 deficient mouse embryonic fibroblasts (MEFs) demonstrate a reduction in TGF-P-dependent growth arrest, firmly establishing a role for Smad3 in mediating TGF-P’s anti-proliferative effects (Datto et a l, 1999). Some Smad3 knock-out mice develop metastatic colorectal cancer (Zhu et a l, 1998), although this phenotype may be the consequence of chronic

Chapter 1_____________________________________________________Introduction

inflammation. Smad3 deficient mice usually die between 1 and 10 months due to chronic infection of the mucosal membranes (Yang et al., 1999b).

Mice mutant for the two BMP R-Smads, Smadl and SmadS, die at E9.5-10.5 (Tremblay et al., 2001; Yang et al., 1999a). SmadS knock-out mice have defects in the circulatory system, including a lack of blood vessels in the yolk sac and the embryo proper, enlarged vessels in the embryonic tissues, and decreased numbers of smooth muscle cells. They also exhibit cranio-facial defects, and enhanced apoptosis in mesenchymal cells, defects in embryonic turning and abnormal heart and gut development. In contrast, Smadl mutant mice die due to a failure to connect to the placenta, illustrating the unique requirement for Smadl in coordinating the growth of extraembryonic structures that are necessary to support embryonic development. This indicates that Smadl and SmadS have non-redundant roles in vertebrate development.

S m ad6 knock-out mice are viable but exhibit striking cardiovascular abnormalities, including defects in endocardial cushion transformation, septation and cardiac valves, illustrating the possible role of Smad6 in homeostasis of the cardiovascular system (Galvin et al., 2000).