Cortactin and Crk cooperate to trigger actin polymerization during Shigella invasion of epithelial cells

Shigella flexneri, a gram-negative bacterium, causes bacillary dysentery in humans by invading the colonic epithelium and eliciting an intense inflammatory reaction that leads to destruction of this epithelium (Labrec et al., 1964; Sansonetti, 1998). Bacterial colonization of the mucosa depends on its ability to invade epithelial cells and to spread from cell to cell (Suzuki and Sasakawa, 2001). The ability of enteroinvasive pathogens such as Shigella to induce its internalization into cells that are normally nonphagocytic is a key event to its virulence. The “zipper”-like process used by Yersinia and Listeria is based on a high affinity interaction between a bacterial surface ligand and a host cell surface receptor, and implicates moderate cytoskeletal rearrangements (Gruenheid and Finlay, 2003). In contrast, the “triggering” process used by Salmonella and Shigella involves several bacterial factors and important rearrangements of the actin cytoskeleton leading to the formation of cell extensions that rise several micrometers above the cell surface and engulf the bacterium in a large vacuole (Galan, 2001). During the triggering process, a specialized type III secretion system determines the integration bacterial components into host cell membranes to form a so-called “translocon,” which allows the subsequent delivery of type III effectors into the cell cytosol (Hueck, 1998). In the case of Salmonella, cytoskeleton reorganization is mediated by translocated type III effectors, the SopE and SptP proteins that act as a guanosine exchange factor and GTPase-activating protein toward the Rho GTPases Cdc42 and Rac, respectively (Galan, 2001). Although phenotypically similar, Shigella invasion implicates a different mechanism, as IpaC, a component of the translocon, triggers signals that lead to actin polymerization (Tran Van Nhieu et al., 1999). During the initial phase of Shigella entry, actin polymerization depends on Rho GTPases and Src tyrosine kinase activation, but the molecular links between these two pathways have not been identified (Duménil et al., 1998; Mounier et al., 1999). The Abl tyrosine kinase family was recently shown to be required for Shigella invasion and may provide such a link because Abl kinases are activated by Src and have been involved in F-actin dynamics (Plattner et al., 1999; Woodring et al., 2003). During Shigella invasion, Abl may activate Rho GTPases through phosphorylation of the Crk adaptor protein (Burton et al., 2003).

Another candidate that could link Rho GTPases and Src pathways is cortactin, an actin-binding protein recruited in Shigella entry foci, which is tyrosine phosphorylated in an Src kinase–dependent manner during Shigella entry (Wu et al., 1991; Dehio et al., 1995; Duménil et al., 1998). Cortactin distributes to sites of dynamic actin assembly, including lamellipodia, podosomes, and invadopodia (Wu and Parsons, 1993; Bowden et al., 1999).

Cortactin possesses a unique multidomain structure consisting of an acidic domain at the amino terminus (NTA) followed by 6.5 tandem repeats containing the F-actin binding site, a proline-rich region containing the three tyrosine residues (421, 466, and 482) that are phosphorylated by Src-related kinases (Huang et al., 1998), and a carboxy-terminal Src homology (SH) 3 domain. Cortactin binds to the Arp2/3 complex and stimulates its actin nucleation activity in vitro (Mullins et al., 1998; Weed et al., 2000; Uruno et al., 2001). Stimulation of cortactin-mediated actin nucleation is enhanced by its binding to F-actin (Uruno et al., 2001; Weaver et al., 2001) and to WIP, a protein involved in filopodia formation (Kinley et al., 2003). Binding to the Arp2/3 complex occurs through the cortactin tripeptide motif “DDW” located in its amino-terminal NTA domain, a motif that is conserved in many Arp2/3-binding/activating proteins such as HS1, Abp1, and WASP family proteins (Olazabal and Machesky, 2001). Cortactin-dependent actin nucleation may be relevant for the genesis of filopodial extensions such as dendritic spine (Hering and Sheng, 2003). Cortactin was also reported to prevent the debranching of Arp2/3–F-actin networks (Weaver et al., 2001).

To inhibit cortactin expression, HeLa cells were transfected with an siRNA duplex corresponding to the 5′ sequence of the cortactin mRNA (dsRNA1). Immunofluorescence staining of transfected cells indicated that inhibition of cortactin expression was maximal 24 h after transfection of dsRNA1, with 74% of cells showing no detectable cortactin. Consistently, anti-cortactin Western blot analysis indicated a 70% decrease in cortactin levels in cells transfected with dsRNA1, compared with control samples (Fig. 1 a, cortactin), whereas no difference in the levels of expression of actin or Arp3 could be detected (Fig. 1 a, arp3 and actin). F-actin staining indicated that cells transfected with dsRNA1 showed little modification of their actin cytoskeleton (unpublished data). The effects of cortactin inhibition on Shigella uptake were analyzed by infecting dsRNA1-treated cells with Shigella. As shown in Fig. 1 b, bacterial uptake in cells treated with dsRNA1 was decreased by fivefold compared with control cells. This decrease was still observed even after 40 min of bacterial challenge, showing that the invasion process was strongly inhibited in dsRNA1-treated cells and not simply delayed (Fig. 1 b). In dsRNA1-transfected cells, only weak actin polymerization was detected at the site of bacterial contact (Fig. 1 c), whereas as expected, massive actin polymerization was observed at entry sites in untransfected cells (unpublished data) or in scrambled dsRNA1-transfected cells (Fig. 1 c). The number of entry structures showing actin-rich extensions was decreased 10 times in dsRNA1-treated cells compared with control cells, as 90% of Shigella-induced structures in dsRNA1-treated cells appeared abortive (Fig. 1, c and d).

Conversely, overexpression of Flag-tagged full-length (FL) cortactin led to a twofold increase in bacterial uptake compared with mock cells (Fig. S1 and Fig. 2 a;

We thank Javier Pizarro-Cerda for technical help with sucrose gradients. We thank Thomas Meyer for helpful suggestions with siRNA experiments.

L. Bougnères received a fellowship from the French Ministry of Higher Education and Research and from the Fondation pour la Recherche Médicale. P.J. Sansonetti is a Howard Hughes Medical Institute Scholar.