All molecular modelling was performed by Drs Paul Bates and Christopher Page from the Biomolecular Modelling Laboratory, Cancer Research UK.
2.7.1 Molecular Modelling for Chapter 5
The Mixer SIM sequence was aligned with the rigid coil region of the SARA SBD and the alignment used to replace the side chains of SARA with those of the Mixer SIM in the Smad2-SARA SBD complex (Wu et al., 2000), using the modelling programme 3D- JIGSAW (Figure 5.IB and C; Bates and Sternberg, 1999). Side chain conformations were allowed to vary on both the SIM and Smad2, keeping the protein backbones of both fixed. To remove the small number of steric clashes, 100 steps of steepest descents energy minimisation (all atoms unrestrained) were run using the programme CHARMM (Brooks et al., 1983). The overall quality of side chain packing and stereochemistry of the final model were checked using QUANTA (Molecular Simulations software, version 3.3); no bad clashes or poor side chain packing at the Smad2-SIM interface were found. The side chain conformations of Smad2 at the Smad2-SIM interface were mainly conserved between the SARA SBD and the SIM; the few exceptions being side chains at the edge of the interface, e.g. K375 and C380 in Smad2. All key side chain conformers on Smad2 at the interface, such as W368 and N381, were conserved, indicating that the backbone of the SIM only need undergo minor adjustments relative to the SARA SBD to maintain a similar binding energy.
2.7.2 Molecular Modelling for Chapter 6
The model of the predicted Smad2/Smad4 MH2 domain heterotrimer (Figures 6.11, 6.13A and 6.14) was generated from the crystal structure of the phosphorylated Smad2 (Wu et al., 2001b). From the PDB entry, a homotrimer was constructed based on the operations z, x, y; 1/2 + y, 1/2 -z, -x; and -y, 1/2 -z, x; (7 2i3 symmetry). A Smad4 subunit, taken from the Smad4 MH2 domain homotrimer (Chacko et al., 2001; Qin et
al., 1999) was directly superimposed on one of the Smad2 subunits using the 3D- JIGSAW software (C„ RMSD: 1.33 À; 45 % identity (Bates and Sternberg, 1999). The programme PASS (Brady and Stouten, 2000) was used to identify potential binding sites, based on the size, shape and extent of the partially-buried volume across the protein surface.
Chapter 3_________________________________________Characterisation o f the SIM
Chapter 3
Characterisation of a Smad2-Interaction Motif
3.1 Introduction
As outlined in Chapter 1, the TGF-P superfamily signal transduction cascade is exquisitely controlled at a number of different levels. One of the most important levels of regulation, in terms of specific activation/repression of TGF-P target genes, is the interaction of the Smads with various transcription factors in the nucleus. One of the best characterised systems for studying signalling specificity is the Xenopus embryo. Here, members of the TGF-P superfamily act as morphogens, specifying different cell types at different ligand concentrations. For example. Nodal-like ligands specify endoderm in the vegetal hemisphere where they are synthesised and are thus at high concentrations, and mesoderm in the marginal zone where they are expressed at lower concentrations (Schier and Shen, 2000). Thus, the same signal can regulate the expression of a number of different genes and consequently induce the specification of different cell types. This cell-type specific response is controlled in part by the presence of other signalling cascades acting in the embryo, such as the Wnt and FGF pathways. In addition, the existence of different Smad-interacting transcription factors with distinct DNA-binding specificities in different cells of the embryo will also be critical (Germain
e ta l, 2000; Hill, 2001).
There are now two classic examples of Smad-interacting transcription factors in
X enopus embryos which possess different DNA-binding specificity. The first Smad-transcription factor complex to be identified bound to the activin-responsive element (ARE) of the Xenopus Mix. 2 promoter and was termed ARE, for Activin Responsive Factor. It is composed of Smad2, Smad4 and XFast-1, a winged helix/forkhead transcription factor (Chen et al., 1996; Chen et al., 1997a) and is stabilised through an interaction between Smad2 and XFast-1; Smad4 is brought into the complex through its interaction with Smad2 and acts to stabilise the Smad2-Fast-1 complex (Chen et al., 1997a). In Xenopus embryos, there is a second promoter element
responsive to activin, the sequence of which is completely unrelated to the ARE of the
Mix. 2 promoter. This is the distal element (DE) of the promoter of the mesoendodermal gene, goosecoid which contains a paired-like homeodomain-binding site (McKendry et al., 1998). One family of paired-like homeodomain transcription factors in Xenopus is the Mix family, of which there are seven members; Mixer (Henry and Melton, 1998), Mix.l and the highly related Mix.2 (Rosa, 1989; Vize, 1996), Bixl, Bix2 (also called Milk), Bix3 and Bix4 (Ecochard et a l, 1998; Tada et al., 1998). Recent work in the Hill lab identified a subset of the Mix family, specifically Mixer and Milk/Bix2, as candidates for the endogenous DE-binding transcription factor (Germain et al., 2000). Mixer was demonstrated to form a ligand-dependent DE-binding complex with endogenous Smad2 and Smad4. Like ARE, this Smad-transcription factor complex was mediated solely through the interaction of Mixer with Smad2, Smad4 again being recruited via its interaction with Smad2 (Germain et al., 2000).
Thus, transcription factors of very distinct binding specificity can interact with the same Smads, thereby recruiting them to different promoter elements. Germain et al.
went on to investigate whether these transcription factors employed a common protein-protein interaction mechanism in order to recruit Smad2. Sequence analysis identified a short 25 amino acid carboxy-terminal sequence in a subset of the Mix family (Mixer, Milk/Bix2 and Bix3) characterised by a completely conserved P-P-N-K-S/T-W core, flanked by other highly conserved residues, which was termed the Smad interaction motif (SIM). Importantly, this motif is also present in Xenopus Fast-1 (FoxHIa) and Fast-3 (FoxHlb), zebrafish Fast-1 (FoxHl), human Fast-1 (FoxHl) and mouse Fast-2 (FoxHl; Germain et al., 2000; Howell et al., 2002; Pogoda et al., 2000), all of which have been shown to interact with Smad2, with the exception of zebrafish FoxHl for which this analysis has not yet been performed. It was demonstrated that the SIM was both necessary and sufficient for Mixer and Milk/Bix2 to interact with the MH2 domain of Smad2 (Smad2C) in vitro, and that this sequence was required for Mixer/Smad complex-mediated TGF-|3-induced transcription. It was also shown that a peptide comprising the 25 amino acid Mixer SIM could compete efficiently with full- length Mixer for binding to Smad2C.
Chapter 3________________________________________ Characterisation o f the SIM
In this Chapter I present the results of a project designed to determine what constitutes a functional SIM. Here I characterise the SIM with respect to the amino acids required for Smad2 interaction by extensive site-directed mutagenesis, I then go on to determine which members of the Xenopus Mix family contain a functional SIM and are therefore capable of interaction with Smad2.