My laboratory focuses on signalling by members of the transforming growth factor β (TGF-β) superfamily, a group of pleiotropic growth and differentiation factors that include the TGF-βs, Activins, Nodals, BMPs and GDFs. We want to understand the molecular mechanism by which these ligands transmit signals to the nucleus to regulate the transcription of target genes and to ascertain their function in early vertebrate development and in untransformed tissue culture cells. Importantly, these signalling pathways are perturbed in cancer. We want to determine how this occurs and its role in tumour progression.
A mathematical model to describe TGF-β/Smad signalling
The best understood signalling pathway downstream of receptors for TGF-β superfamily members is the Smad pathway. In essence, ligand stimulation induces receptor activation which leads to the phosphorylation and activation of a subset of R-Smads, for example, Smad2 and Smad3 in the case of TGF-β/Activin/Nodal ligands. Activated R-Smads then form homomeric and heteromeric complexes with Smad4 that accumulate in the nucleus, where they are directly involved in regulating transcription of target genes. However, it has become apparent from our experimental work over the past several years that the TGF-β/Smad signal transduction pathway is not a simple linear, unidirectional pathway from receptors to the nucleus. Instead, it is a dynamic network where the Smads constantly shuttle between the cytoplasm and nucleus both in the absence and presence of a signal, the latter being driven by successive rounds of Smad phosphorylation in the cytoplasm and dephosphorylation in the nucleus. In this model, nuclear accumulation of active Smad complexes results from a change in mean residence time of the Smads in nucleus versus the cytoplasm.
To address whether this experimental model can account for the observed kinetics of Smad nuclear accumulation in response to signal and their redistribution to the cytoplasm upon receptor inactivation, we have developed a computational model of the TGF-β/Smad signalling pathway in collaboration with Paul Bates' lab, using four distinct kinetic datasets generated in our lab. The resulting highly-constrained mathematical rate equation model simultaneously fits all the datasets with excellent accuracy and in doing so, verifies the plausibility and mechanistic relevance of the underlying network topology. We have used the model to make predictions about the outcome of fluorescence recovery after photobleaching experiments and the behaviour of a functionally-impaired Smad2 mutant, which we have experimentally verified. Most importantly, our computational model clearly demonstrates a functional role for Smad nucleocytoplasmic shuttling in the dynamic and quantitative interpretation of TGF-β signals, which can explain how TGF-β superfamily ligands act as morphogens (Schmierer et al., 2008, PNAS, 105, 6608-6613) (Figure 1).
E3 ubiquitin ligases in TGF-β superfamily signalling
In the last few years we have used high-throughput siRNA screening to discover new components and regulators of TGF-β/Smad signalling, and have identified two E3 ubiquitin ligases, Arkadia and Ectodermin, that play crucial roles. Following on from our publication in 2007 that Arkadia is an essential component of the Smad3-dependent branch of TGF-β/Activin/Nodal signalling which functions by inducing ligand-dependent degradation of the transcriptional repressor, SnoN, we have focused this year on understanding the relevance of this for cancer progression and for early vertebrate development, using zebrafish as a model system. We have identified a lung tumour cell line which harbours a homozygous nonsense mutation in the Arkadia gene, which generates a non-functional protein. Stable reintroduction of wild-type Arkadia into this cell line restores TGF-β-induced SnoN degradation, Smad3-dependent transcription and substantially reduces the ability of these tumour cells to grow in soft agar. This suggests that Arkadia may be a novel tumour suppressor gene and we are currently investigating this in more detail by generating a conditional mouse knockout of Arkadia and searching for Arkadia mutations in other tumour cell lines. In contrast to Arkadia, Ectodermin acts as a negative regulator of TGF-β superfamily signalling pathways. We have shown that its repressive activity requires its PHD and Bromo domains, and consistent with the ability of such domains to bind modified histones, we have shown that Ectodermin is recruited to promoters of TGF-β target genes in a ligand-dependent manner. We are currently investigating how this is achieved and how Ectodermin functions as a repressor.
Two highly-related PP2A regulatory subunits exert opposite effects on TGF-β/Activin/Nodal signalling
Our high-throughput screening also led us to the identification of two highly-related PP2A regulatory subunits, Bα and Bδ as important modulators of TGF-β/Activin/Nodal signalling. In a collaboration with Laurel Raftery (MGH, Boston, USA) and using a combination of experimental systems (tissue culture cells, Xenopus and Drosophila embryos) we have shown that Bα enhances TGF-β/Activin/Nodal signalling by stabilising basal levels of type I receptors, whereas Bδ negatively modulates these pathways by inhibiting receptor activity (Batut et al., 2008, Development, 135, 2927-2937).
A novel branch of TGF-β/Smad signalling is required for anchorage-independent growth
Contrary to the original view of the TGF-β superfamily signalling whereby TGF-β/Activin/Nodal stimulation leads to activation of Smad2 and 3, whereas BMP/GDF signalling is mediated via Smad1, 5 and 8, we have now discovered that TGF-β also strongly induces phosphorylation of Smad1 and 5 in epithelial cells and fibroblasts. This signalling requires TβRII, ALK5, and additionally the type I receptors ALK2 and/or ALK3. Simultaneous activation of the R-Smads, Smad2/3 and Smad1/5 results in the formation of novel 'mixed' R-Smad complexes containing, for example, activated Smad1 and Smad2, which we propose are responsible for transducing the signal to the nucleus. Finally, we have shown that Smad1/5 activation by TGF-β is not required for growth inhibition, but is specifically required for TGF-β-induced anchorage independent growth, suggesting that this branch of TGF-β/Smad signalling may play an important role in tumour promotion (Daly et al., 2008, Mol. Cell. Biol. 28, 6889-6902).
Figure 1. a) Using our mathematical model of TGF-β/Smad signalling, receptor levels were altered in a step-wise fashion as indicated. The simulation demonstrates that Smad nucleocytoplasmic shuttling couples the concentration of nuclear Smad2¿Smad4 complexes to receptor activity, but with a time delay. b) Because of the time delay, fluctuations in receptor activity are dampened and thus do not cause corresponding fluctuations in the concentration of nuclear Smad2¿Smad4 complexes. Only sustained changes in receptor activity are therefore transmitted to the nucleus.
For a list of refereed research papers, see Publications (in navigation on left).
