The findings reported in this study identify a previously unrecognized role for the prominent epithelial integrin α3β1 in TGF-β1 signaling, providing evidence that the integrin functions as a sensor of cell contacts to regulate TGF-β1 signaling. The essential function of α3β1 appears to depend on the presence of surface complexes on epithelial cells, which are comprised of a subfraction of the surface pool of α3β1, E-cadherin, and TGF-βR1, which critically influence the signaling response to TGF-β1. The components of this complex by itself reveals the likely elements of its function, as modeled in . Surface complexes of E-cadherin and TGF-βR1 bring two key transcription factors involved in induction of EMT into spatial proximity: β-catenin and receptor Smads (Chilosi et al., 2003
; Kalluri and Neilson, 2003
; Yook et al., 2006
). After TGF-β1 stimulation, the surface complexes are internalized, and β-catenin is phosphorylated on Y-654. The formation of stable complexes between pY654–β-catenin and p-Smad2 and the appearance of these complexes strongly correlate with subsequent EMT-related protein expression (). Both internalization and β-catenin tyrosine phosphorylation require the third component of the complex, α3β1. The unique function of α3β1 in this context appears to reside in its affinity for E-cadherin, without which E-cadherin–TGF-βR1 complex internalization is impaired, and this signaling system does not operate. Conceptually, these findings are similar to prior evidence in another system in which integrin engagement is found to regulate growth factor signaling at least in part by altering pathways of endocytosis that influence the quality and duration of receptor signaling (del Pozo et al., 2004
). However, our findings also reveal a completely new point of intersection between β-catenin and Smad signaling pathways (Lei et al., 2004
), and the linkage of this point of intersection to a dynamic interplay between adhesion receptors and their normal pericellular contacts provides new understanding for how the extracellular environment can regulate the epithelial cell response to TGF-β1 (Masszi et al., 2004
; Thiery and Sleeman, 2006
Model of TGF-β1 signaling. The schematic illustration summarizes the influence of α3β1 on TGF-β1 signaling as a function of cell–cell and cell–matrix contacts. For details see the Discussion.
TGF-β1 has been previously reported to promote tyrosine phosphorylation of β-catenin (Tian and Phillips, 2002
), although the specific site of phosphorylation and its functional significance in TGF-β1 signaling has been unknown. Our finding of integrin-dependent tyrosine phosphorylation of Y654–β-catenin is important because phosphorylation of β-catenin at Y654 is known to promote both dissociation of β-catenin from E-cadherin and stabilization of β-catenin from ubiquitination and degradation (Brembeck et al., 2006
). Therefore, independently of Wnt signaling, our findings indicate that TGF-β1 can promote a pathway of cross talk with β-catenin by generating stable pY654–β-catenin–Smad complexes. The data indicate that only a fraction of the β-catenin is phosphorylated, and presumably, this reflects, at least in part, the pool internalized with E-cadherin and TGF-βR1 after TGF-β1 stimulation. However, internalization alone does not appear to be sufficient, as α3-null cells, even though they display unstable surface E-cadherin, fail to phosphorylate β-catenin in response to TGF-β1. We are uncertain of the tyrosine kinases responsible for Y654 phosphorylation. At least three different tyrosine kinases have been reported to phosphorylate this site: c-src (and possibly other src family members; Roura et al., 1999
), Bcr-Abl (Coluccia et al., 2007
), and hepatocyte growth factor receptor Met (Zeng et al., 2006
). In addition, the TGF-β1 receptor complex itself has recently been shown to have tyrosine kinase activity (Lee et al., 2007
). As src family kinases are known to associate with the cytoplasmic tails of β1 integrins (Hynes, 2002
), the requirement for the integrin likely involves a src family kinase. However, the exact kinase and how this kinase activity is locally regulated by α3β1 after TGF-β1 stimulation remains to be defined in future work.
A recent study described defective TGF-β1 signaling in α3β1-deficient murine keratinocytes and attributed the defect to higher levels of the inhibitory Smad, Smad7 (Reynolds et al., 2008
). Corresponding to the higher Smad7 levels was lower overall levels of TGF-β1 receptors and lower Smad2/3 phosphorylation after TGF-β1 stimulation, which is consistent with the known effects of Smad7 on the TGF-β1 signaling pathway. However, neither in the kidney epithelial cell line studied here nor in α3β1-deficient primary lung epithelial cells were we able to detect altered levels of Smad7. Consistent with our data, overall levels of TGF-β1 receptors were not lower in α3-deficient cells, and Smad2/3 phosphorylation was not lower after TGF-β1 stimulation (). Together, these observations may imply that the impact of α3β1 on TGF-β1 signaling operates through different pathways in different cells but may also suggest that this integrin may critically influence TGF-β1 signaling in many, if not all, epithelial cells.
We have previously reported that stable overexpression of uPAR in kidney epithelial cells promotes an EMT response via interactions of uPAR with α3β1 (Zhang et al., 2003
). The functionally important H245A α3 point mutation used in this study was discovered in a screen of several integrin β propeller mutants for inhibitors of uPAR-dependent EMT. Cells expressing the H245A mutant were found to have disrupted uPAR–α3β1 interactions and attenuated EMT compared with cells expressing wt α3. However, basal epithelial cells express little or no uPAR, and uPAR is a known TGF-β1 target gene (Yue et al., 2004
). Therefore, we asked whether the response of these cells to TGF-β1 involved uPAR. Surprisingly, H245A-expressing cells were found to have little or no response to active TGF-β1, implying a critical role for α3β1 in the early response of these cells even before induction of uPAR and leading to the series of experiments reported in this study. It is likely that uPAR, once induced by TGF-β1, further promotes signaling amplifying EMT, as has been recently reported in other systems (Lester et al., 2007
). However, findings reported in this study indicate that, independently of uPAR, α3β1 has a critical role in the initial TGF-β1 signaling leading to EMT.
One limitation of our observations is that it is unclear how pY654–β-catenin–p-Smad2 complexes operate to promote mesenchymal gene responses to TGF-β1. Although there is a strong correlation between formation of these complexes and initiation of EMT in both primary lung epithelial cells as well as the kidney epithelial cells primarily studied here, it is unclear what the critical promoter targets are for this complex. TGF-β1 does not strongly activate the canonical Lef1/T cell factor reporter, and activation of the canonical Smad-binding CAGA reporters is not different between cells expressing wt and H245A mutant α3β1 (Fig. S2), suggesting that a noncanonical signaling mechanism may be in play in this signaling pathway. Moreover, how can the pro-EMT signaling pathway involving p-Smad2 identified in this study be reconciled with recent findings that Smad2-null keratinocytes spontaneously develop EMT (Hoot et al., 2008
), suggesting the possibility that p-Smad2 actually functions to suppress EMT? We note that total p-Smad2 levels after TGF-β1 stimulation are strongly and comparably induced in both responding α3 wt cells and nonresponsive α3-null and H245A mutant–expressing cells (). It is well known that cell-specific coactivators and repressors critically regulate the repertoire of responses to Smad signaling (Derynck and Zhang, 2003
). This is consistent with the view that p-Smad2 may function to suppress EMT unless and until pY654–β-catenin–p-Smad2 complexes form under the conditions revealed by our experiments. If so, the set of TGF-β1 target genes activated by pY654–β-catenin–p-Smad2 complexes and how this switches p-Smad2 from a suppressor to activator role in EMT are important areas for further investigation.