SFK-mediated changes in cell-cell adhesion

Our identification of Fyn and Yes as positive regulators of AJs in the zebrafish embryo provides new insights into how SFKs control cell-cell adhesion under physiological conditions. It is important to note that nearly all the widely accepted evidence stating that SFK activity disrupts AJ function was gathered in experiments using oncogene-transformed cells (see Introduction, paragraph 1.7). In contrast, work from our laboratory and other in vivo studies challenge this view. Concretely, depletion of SFK levels/activity in the zebrafish embryo -either by Fyn/Yes knockdown or indirectly by downregulation of PrP-1- led to abnormal AJ protein levels and localization, whereas elevated SFK activity in PrP OE embryos correlated with increased adhesiveness of embryonic cells (this study and aggregation assays from Málaga-Trillo et al, 2009). Accordingly, in Drosophila, SFK member DSrc42A promotes the localization of DE-cadherin at the cell surface during embryonic morphogenesis (Takahashi et al, 2005) and at contact sites between eye photoreceptor cells (Takahashi et al, 1996). Also in mice, Fyn and Src positively regulate the maintenance of keratinocyte cell-cell adhesion (Calautti et al, 1998). Notably, even in cancer cells SFK activity does not always correlate with diminished cell adhesion. For instance, inhibition of physiological Src activation levels in MCF-7 cells by expression of dominant negative Src or PP2-treatment was shown to perturb E-cadherin localization at cell contacts (McLachlan et al, 2007). Similarly, our present work with MCF-7 cells argues for a positive correlation

between SFK activity and the presence of E-cadherin at cell contact sites. Altogether, the functional data collected in different cells and organisms lead to the paradoxical conclusion that the influence of SFKs on E-cadherin cell adhesion is positive in some cases and negative in others. These seemingly contradictory findings in different models can be best explained by a “bimodal” model, in which SFK activity promotes cell adhesion at low, physiological levels but suppress it at high, oncogenic levels. This hypothesis was derived from the observation that endogenous activation levels of Src or low levels of exogenously expressed CA Src in MCF-7 cells support/enhance E-cadherin contact formation, while high CA Src expression levels weaken adhesion (McLachlan et al, 2007). The results presented in my dissertation further support the physiological “arm” of the bimodal model by showing that Fyn and Yes promote E-cadherin cell adhesion during embryonic development.

A key question that emerges from these findings is how exactly SFKs control E-cadherin adhesive function. As described in paragraph 1.7 of the Introduction, the function of of E-cadherin and β-catenin can be regulated via various mechanisms, from which phosphorylation and endocytosis/degradation are the most relevant to my work. Through biochemical analyses in PrP-1 morphants and PrP OE embryos, we confirmed a positive correlation between E-cadherin adhesion, SFK activity and the levels of β-catenin phosphorylation at Tyr142. Fittingly, this residue was previously identified as a direct substrate of Fyn and an indirect target of Yes activity (Piedra et al, 2003). However, in contrast to our findings, this phosphorylation event disrupts β-catenin binding to α-catenin in cell-free systems, suggesting that it would destabilize AJs by decoupling them from actin (Aberle et al, 1996; Pokutta & Weis, 2000). That increased β-catenin phosphorylation at Tyr142 does not impact negatively on embryonic cell adhesion was particularly evident in our PrP OE experiments with zebrafish: despite exhibiting abnormally high levels of Tyr142-phosphorylated β-catenin, these embryos accumulated E-cadherin and β-catenin at the plasma membrane, and their cells were more adhesive than those of control embryos (this study and Málaga-Trillo et al, 2009). A similar phenomenon was observed in mouse keratinocytes, in which elevated β-catenin phosphorylation concurred with increased cell-cell adhesion, required for their differentiation (Calautti et al, 1998). Accordingly, keratinocytes of fyn/src deficient mice showed significantly reduced levels of β-catenin phosphorylation and could not properly adhere to each other, causing structural defects in the skin (Calautti et al, 1998). Taken together, these data indicate that, in vivo, SFK-induced phosphorylation of β -catenin is not the dominant event controlling AJ assembly but rather a minor modulator of AJ stability, or perhaps, part of a mechanism that regulates an AJ-independent function of β -catenin. These scenarios would also be in line with our hypothesis that E-cadherin -and not β-catenin- is the decisive factor that destabilizes AJs in the absence of PrP-1.

If the above is true, how could SFKs influence E-cadherin localization at the plasma membrane in a β-catenin independent manner? In MDCK cells, expression of v-Src -a virally encoded, constitutively active form of Src- has been suggested to directly phosphorylate E-cadherin, inducing the ubiquitination of the latter at the plasma membrane and its subsequent endocytosis and degradation (Fujita 2002). However, like most other studies performed in oncogene-transformed cells, this one also supports the notion of Src being a negative regulator of E-cadherin-based adhesion. In order to conciliate these and our findings, it will be necessary to determine if such direct regulation of E-cadherin via tyrosine phosphorylation follows the bimodal model and thus has a positive regulatory counterpart under physiological conditions.

Another important route by which SFKs influence E-cadherin adhesion complexes is via the regulation of the actin cytoskeleton (see introduction, paragraph 1.7). In fact, our group reported abnormalities in the distribution of F-actin along the plasma membrane of PrP-1 knockdown zebrafish embryos (Málaga-Trillo et al, 2009). This observation and the finding that exogenous PrP expression triggers actin recruitment to S2 cell-cell contacts strongly suggest a positive influence of PrP on cortical actin (Málaga-Trillo et al, 2009). Interestingly, Cortactin -a known substrate of Src-, becomes localized to cell contacts upon E-cadherin homophilic binding, where it stabilizes AJs by promoting actin assembly (Helwani et al, 2004; Wu & Parsons, 1993; Wu et al, 1991). When phosphorylated by Src, Cortactin triggers actin polymerization by recruiting the actin-nucleating Arp2/3 complex to cortical actin (Tehrani et al, 2007; Uruno et al, 2001; Weaver et al, 2002; Weaver et al, 2001). In addition, Cortactin binds and stabilizes nascent actin filament branches produced by Arp2/3 (Weaver et al, 2001). Remarkably, Cortactin has also been reported to interact with dynamin2 as a component of clathrin-coated pits and influence endocytosis, a process also proposed to depend on Src activity (Cao et al, 2003). Thus, it would be interesting to examine whether altered Cortactin phosphorylation contributes to the zebrafish PrP-1 knockdown phenotype by affecting actin dynamics or the endocytosis of AJ components. Further evidence for a link between PrP-1 and actin dynamics is provided by the finding that zebrafish Fyn and Yes are upstream regulators of the small GTPase RhoA, a known modulator of actin polymerization (Heasman & Ridley, 2008; Jopling & den Hertog, 2005). Although the embryonic requirement of the Fyn/Yes/RhoA pathway was studied in the context of convergence and extension cell movements, which take place after the onset of our PrP-1 phenotype, it is conceivable that earlier stages of epiboly are equally affected by this signaling cascade.