Nội dung

MINIREVIEW Cytoskeleton-modulating effectors of enteropathogenic and enterohaemorrhagic Escherichia coli: Tir, EspFU and actin pedestal assembly Kenneth G. Campellone Department of Molecular and Cell Biology, University of California, Berkeley, CA, USA Keywords actin assembly; Arp2 ⁄ 3 complex; bacterial pathogenesis; cell signaling; EHEC; EPEC; EspF; membrane dynamics; N-WASP; tyrosine kinase Correspondence K. G. Campellone, Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, CA 94720 USA Fax: +1 510 642 8620 Tel: +1 510 642 5525 E-mail: campellone@berkeley.edu (Received 27 October 2009, revised 12 February 2010, accepted 15 March 2010) A variety of microbes manipulate the cytoskeleton of mammalian cells to promote their internalization, motility and ⁄ or spread. Among such bacteria, enteropathogenic Escherichia coli and enterohemorrhagic Escherichia coli are closely related pathogens that adhere to human intestinal cells and reorganize the underlying actin cytoskeleton into ‘pedestals’. The assembly of pedestals is likely to be an important step in colonization, and is triggered by the E. coli virulence factors translocated intimin receptor and E. coli secreted protein F in prophage U, which modulate multiple host signaling cascades that lead to actin polymerization. In recent years, these bacterial effectors have been exploited as powerful experimental tools for investigating actin cytoskeletal and membrane dynamics, and several studies have significantly advanced our understanding of the regulation of actin assembly in mammalian cells and the potential role of pedestal formation in pathogenesis. doi:10.1111/j.1742-4658.2010.07653.x Introduction Numerous microbial pathogens share an ability to trigger localized actin polymerization in host cells. Such pathogens include several bacteria that invade mammalian cells, gain access to the cytoplasm and stimulate actin assembly at their surface to propel them throughout the cell. The actin-based motility of several of these bacteria, including Listeria monocytogenes and Shigella flexneri, has been utilized by many investigators to better understand how mammalian cells control actin dynamics in the cytoplasm [1]. Interestingly, other noninvasive extracellular pathogens have also been shown to stimulate actin polymerization upon binding to the surface of mammalian cells, and have been exploited to study how cells regulate actin assembly beneath the plasma membrane [2]. Two examples of such bacteria are enteropathogenic Eschericia coli (EPEC), which cause diarrhea in children of developing countries, and enterohemorrhagic Eschericia coli (EHEC), which have emerged as a major cause of hemorrhagic colitis and pediatric kidney failure in the USA, UK and Japan [3,4]. One common manifestation of infection with EPEC and EHEC is the formation of intestinal Abbreviations AI, autoinhibitory; Arp2 ⁄ 3, actin-related protein 2 ⁄ 3; B, basic; EHEC, enterohemorrhagic Escherichia coli; EPEC, enteropathogenic Escherichia coli; Esp, E. coli secreted protein; EspFU, E. coli secreted protein F in prophage U; F-actin, filamentous actin; F-BAR, Fes CIP4-Bin-amphiphysin-Rvs167; GBD, GTPase-binding domain; I-BAR, inverse Bin-amphiphysin-Rvs167; LEE, locus of enterocyte effacement; PRD, proline-rich domain; PtdIns3K, phosphatidylinositol 3-kinase; SH2, Src-homology 2; SH3, Src-homology 3; Tir, translocated intimin receptor; WASP, Wiskott–Aldrich syndrome protein; WCA, WASP homology 2 connector acidic; WTF, WASP–Tir–EspFU. 2390 FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS K. G. Campellone lesions that are characterized by tight bacterial attachment to the surface of epithelial cells and the effacement of microvilli [5,6]. The formation of these ‘attaching ⁄ effacing’ lesions is critical for the pathogenesis of EPEC and EHEC, because bacterial mutants that cannot form lesions do not colonize their hosts or cause disease [7,8]. In addition to their tight apposition to the plasma membrane, adherent EPEC and EHEC reorganize the underlying host cytoskeleton into actin-rich pedestals (Fig. 1). These dynamic structures have been shown to promote bacterial motility along the surface of cultured cells [1,2], although their role in pathogenesis in vivo has not been well defined. Importantly, our view of how EPEC and EHEC expertly commandeer the mammalian actin polymerization machinery to drive pedestal formation has come into greater focus in recent years. In this minireview, I first describe the mechanisms by which cells normally regulate actin assembly and discuss how EPEC and EHEC share an ability to form attaching ⁄ effacing lesions. I then examine how one EPEC protein called translocated intimin receptor (Tir) promotes pedestal formation by intercepting mammalian tyrosine kinase signaling cascades that lead to actin polymerization. Finally, I explore how the last several years, in particular, have yielded valuable new insights into the clever and distinct methods that the EHEC proteins Tir and E. coli secreted protein F in prophage U (EspFU) utilize to trigger actin polymerization. These discoveries have implications for understanding both EHEC pathogenesis and the strategies that mammalian cells normally employ to control actin cytoskeletal and membrane dynamics. Regulation of actin assembly by Tir and EspFU Cellular control of actin assembly by the actin-related protein 2 ⁄ 3 complex, neuronal-WASP and its activators To initiate the assembly of actin filaments de novo, cells must convert actin monomers into multimeric configurations that act as templates for subsequent polymerization. This rate-limiting step is catalyzed by cellular nucleation factors. Multiple mammalian nucleators have been shown to promote the formation of linear unbranched actin filaments, whereas only one nucleator, actin-related protein 2 ⁄ 3 (Arp2 ⁄ 3) complex, is known to generate densely branched filamentous actin (F-actin) networks [9,10]. This complex is a stably associated group of seven proteins that act collectively to generate new actin filaments upon binding to the sides of existing filaments. By itself, however, the Arp2 ⁄ 3 complex is inactive. To efficiently nucleate filaments, it interacts with proteins called nucleation-promoting factors from the Wiskott–Aldrich syndrome protein (WASP) family. Mammalian cells express at least eight such factors, including the founding Arp2 ⁄ 3 activators, WASP and neuronal-WASP (N-WASP) [9]. The nucleation-promoting activity of all of these proteins resides in their conserved C-terminal WASP homology 2 connector acidic (WCA) domains, which are comprised of WASP homology 2 motifs that bind actin monomers, plus a connector region and acidic peptide that together bind the Arp2 ⁄ 3 complex. By contrast, the N-terminal sequences of these proteins control their spatial and temporal regulation in cells [9,11]. WASP, which is expressed specifically in hematopoetic cells, and its ubiquitous homolog N-WASP, each Fig. 1. Actin pedestal formation on mammalian cells. Upon infection of mammalian host cells, EPEC and EHEC reorganize the cytoskeleton into actin-rich pedestals [2]. Multiple examples of actin pedestals are shown, including a scanning electron micrograph of EPEC generating a pedestal on a polarized epithelial cell (left) and transmission electron micrographs of EPEC and EHEC (dark ovals) sitting atop the electrondense actin pedestals of non-polarized HeLa cells (middle). EHEC pedestal formation can be recapitulated using biomimetic experimental systems in which HeLa cells expressing a Tir–EspFU fusion protein are treated with non-pathogenic E. coli expressing intimin (E. coli + pIntimin) or Staphylococcus aureus particles coated with anti-Tir antibodies (S. aureus + aTir) as described previously [55] (right; color panels). E. coli and S. aureus are shown in blue, the Tir–EspFU fusion protein is shown in yellow and F-actin pedestals are shown in red. All scale bars are 0.5 lm. FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS 2391 K. G. Campellone Regulation of actin assembly by Tir and EspFU have a well-defined modular domain organization [9,11]. This consists of an N-terminal WASP homology 1 region, central basic (B), Cdc42 ⁄ Rac-interactive binding and autoinhibitory (AI) motifs that comprise the GTPase-binding domain (GBD) and a proline-rich domain (PRD) that lies adjacent to the WCA region (Fig. 2). By itself, N-WASP has little nucleation-promoting activity, because its CA region is sequestered by interactions with the AI portion of the GBD. This inactive conformation is also influenced by WASP homology 1-binding proteins like WIP, an actin-binding factor that forms a stable complex with N-WASP. To stimulate N-WASP, signal transduction pathways generally utilize cellular factors that interact with the GBD or PRD [9,11]. For example, binding of the small GTPase Cdc42, membrane phosphatidylinositol 4,5-bisphosphate or Src-homology 2 (SH2) domains to the GBD activates N-WASP by destabilizing the inhibitory interactions between the AI region and the WCA Fig. 2. Regulation of N-WASP-mediated actin assembly. The actin nucleation-promoting factor N-WASP forms a complex with the protein WIP via its N-terminal WASP homology 1 domain, and is regulated by intramolecular autoinhibitory interactions between its GTPase-binding domain (GBD) and its C-terminal WH2 connector acidic (WCA) region. In normal cells, N-WASP can be activated by binding of phosphatidylinositol (4,5) bisphosphate [PI(4,5)P2] to its basic (B) region, Cdc42 to its Cdc42 ⁄ Rac-interactive-binding (CRIB) motif, SH2 domains to a central phosphotyrosine residue (p) and ⁄ or SH3 domains to its proline-rich domain (PRD). The EHEC effector protein EspFU activates N-WASP by out-competing the WCA domain for binding to the autoinhibitory (AI) portion of the GBD. Active N-WASP binds both to actin monomers (G-actin) and the Arp2 ⁄ 3 complex, but it is unclear if it maintains an association with WIP (?). 2392 domain, thereby freeing the latter segment to activate the Arp2 ⁄ 3 complex. In addition, adaptor proteins with Src homology 3 (SH3) domains, such as Nck1 and Nck2, and Src-family tyrosine kinases, like c-Fyn, can bind to the PRD to activate N-WASP. Although each of these factors can stimulate N-WASP individually, multiple signaling inputs need to be integrated to promote maximal N-WASP activity. Intriguingly, EPEC and EHEC have deciphered the signaling mechanisms that control N-WASP activity [12], and have evolved multiple strategies for triggering N-WASPmediated actin assembly during pedestal formation. The locus of enterocyte effacement-encoded type III secretion system and attaching ⁄ effacing lesions To promote interactions with the intestinal epithelium and generate actin pedestals, EPEC and EHEC utilize specialized secretion systems that translocate bacterial effector proteins into host cells [13–15]. Genes encoding a complete ‘type III’ secretion system are found within a pathogenicity island called the the locus of enterocyte effacement (LEE), which is present in all attaching ⁄ effacing bacteria including EPEC and EHEC. The LEE also encodes transcriptional regulators, chaperones and several substrates that are transported via the type III apparatus. These substrates include the E. coli secreted protein EspA, which is likely important for contacting the host-cell membrane, plus EspB and EspD, which reorganize the brush border and cytoskeleton during effacement and also combine to form a pore in the membrane that allows the delivery of other effectors. Comprehensive deletion analyses of LEE genes in Citrobacter rodentium, a murine pathogen that is related to EPEC and EHEC, indicates that each of the structural components of the apparatus, as well as EspA, EspB and EspD, are essential for virulence in mice [16]. It has been estimated that EHEC may inject more than 30 different effector proteins through this translocation pore into host cells [17]. Many of these effectors have substantial effects on the cytoskeleton [18,19], although fewer than half have been well characterized. Among the LEE-encoded effectors that are delivered into the cytoplasm are EspF, EspG, EspH and Map. These factors modulate various mammalian signal transduction cascades and cytoskeletal elements, and each one influences the location and ⁄ or efficiency of colonization in vivo [16,20]. However, none of these effectors is absolutely essential for colonization in a murine model of C. rodentium infection [16], an infant rabbit model of EHEC infection [20] or a human FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS K. G. Campellone intestine organ culture model of EPEC infection [21]. The only LEE-encoded effector that is essential for colonization in all experimental systems is the translocated intimin receptor, Tir. More than a decade ago, the remarkable phenomenon that EPEC translocated its own receptor into mammalian cells was first observed [22,23]. Shortly thereafter, this landmark discovery was confirmed by the observation that EHEC delivers a homologous Tir molecule into host cells [24]. Each of the Tir proteins inserts into the plasma membrane in a hairpin-loop conformation, featuring an extracellular domain that is flanked by two transmembrane segments and Nand C-terminal cytoplasmic domains. The extracellular portion of Tir binds to intimin, a critical adhesin expressed on the surface of both EPEC and EHEC. Intimin and Tir can each multimerize, and intimin binding results in a higher order clustering of Tir in the host-cell plasma membrane beneath adherent bacteria [25,26]. Not surprisingly, given the essential roles of intimin and Tir in intimate cell attachment, bacterial mutants that lack intimin or Tir do not colonize the intestine and are avirulent in the murine and infant rabbit models of infection [16,27,28]. Importantly, in addition to its role in bacterial adhesion, intimin-mediated clustering of Tir is also the signal that triggers the actin assembly that drives pedestal formation. The dirty dozens: Tir peptides that initiate actin assembly EPEC Tir exploits tyrosine kinase signaling cascades to promote Nck recruitment For canonical EPEC strains, Tir is the only translocated effector protein essential for actin pedestal formation [29]. All of the information required for EPEC-mediated actin assembly lies within the C-terminal cytoplasmic region of Tir, because a membrane-targeted version of Tir lacking its N-terminal cytoplasmic domain is sufficient to trigger actin pedestal formation when expressed in mammalian cells and clustered with particles coated with intimin or anti-Tir IgG [29]. Clustering induces the tyrosine phosphorylation of Tir [29–31], which is necessary for actin pedestal formation by EPEC [32]. The predominant site of tyrosine phosphorylation is residue Y474 [32], although a second tyrosine, Y454, can also be phosphorylated [30]. Interestingly, several studies have begun to uncover the complex mechanisms that control Tir phosphorylation (Fig. 3). Y474 (and Y454) can be phosphorylated by a recombinant version of the Src-family kinase c-Fyn Regulation of actin assembly by Tir and EspFU Fig. 3. Regulation of actin assembly by EPEC Tir. Clustering of EPEC Tir (brown) triggers phosphorylation of residues Y474 and (to a lesser extent) Y454 by the redundant host kinases c-Fyn, Abl, Arg and Etk. Some of these kinases can interact with a proline-rich sequence in the N-terminus of Tir (dotted arrow). A 12-residue peptide encompassing phosphorylated Y454 (p) can interact with the SH2 domain from the p85 subunit of phosphatidylinositol 3-kinase (PI3K), but the role of this interaction in pedestal formation is unclear (?). A 12-residue peptide encompassing phosphorylated Y474 (p) binds the SH2 domains of the Nck adaptor proteins. WIPlike proteins may be involved in the subsequent recruitment of N-WASP to a complex of Tir and Nck (not shown), but the three tandem SH3 domains of Nck1 and Nck2 likely activate N-WASP by directly binding to its PRD. Multimerization of N-WASP can further enhance Arp2 ⁄ 3-mediated actin nucleation and pedestal assembly. in vitro, and a priming-and-challenge experimental technique that employs sequential steps of Tir delivery by intimin-deficient EPEC followed by synchronized Tir clustering by intimin-expressing bacteria indicates that c-Fyn associates transiently with Tir and phosphorylates Y474 in cells [31]. These interactions occur within minutes of Tir clustering, which brings Tir into the proximity of c-Fyn in detergentresistant membrane microdomains [33], observations consistent with results indicating that membrane cholesterol is important for Tir signaling [34] and that EPEC transiently colocalizes with the signaling phosphoinositide phosphatidylinositol 4,5-bisphosphate early during wild-type infections [35]. The c-Fyn kinase is not a stable component of the pedestal, however, and < 1 h after it phosphorylates Tir, c-Fyn is inactivated [33]. Importantly, kinases other than c-Fyn have also been shown to phosphorylate Tir in cells. These include the Abl-family kinases Abl and Arg, along with the Tec-family kinase Etk [36,37]. Unlike c-Fyn, these kinases appear to be stable pedestal constituents. The enduring localization of Abl- and Tec-family FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS 2393 Regulation of actin assembly by Tir and EspFU K. G. Campellone kinases appears to reside in their ability to associate with an N-terminal proline-rich peptide of Tir using their SH3 domains [36]. Collectively, these studies suggest a model in which c-Fyn phosphorylates Tir immediately upon multimerization and entry into a plasma membrane microdomain to promote a rapid burst of signaling (Fig. 3). Other kinases then help to maintain long-term Tir signaling by ensuring that it is persistently tyrosine phosphorylated. Once clustered and phosphorylated, a 12-residue Tir peptide encompassing Y474 binds to the SH2 domaincontaining adaptor proteins and known N-WASP activators, Nck1 and Nck2 [38,39]. When immobilized on bacterium-sized beads, this Tir phosphopeptide promotes Nck-dependent actin tail formation in cell-free extracts [29], implying that Nck recruitment by this tiny segment of Tir is sufficient to activate the actin assembly machinery. Nck1 and Nck2 have partially overlapping functions in cells, and genetic deletion studies indicate that they are critical for the major pathway of actin pedestal formation triggered by EPEC Tir [39]. As adaptor proteins, Nck1 and Nck2 each possess three SH3 domains in addition to their phosphotyrosine-binding SH2 regions (Fig. 3). The individual SH3 domains can each bind and activate N-WASP, but their native tandem configuration results in cooperative N-WASP activation, leading to high levels of actin polymerization in vitro [40,41] and in cells [42]. Following Tir clustering and Y474 phosphorylation, Nck adaptors are crucial for recruiting N-WASP, because N-WASP localization to EPEC Tir is diminished in the absence of Nck1 and Nck2 [39], and overexpression of the SH2 domain of Nck1 inhibits actin pedestal formation [30]. It is not entirely clear whether N-WASP is recruited to Tir via direct binding of the Nck SH3 domains to the PRD of N-WASP or indirectly through a WIP-like protein. In either case, N-WASP is absolutely essential for actin pedestal formation by EPEC, as demonstrated by a lack of pedestal formation on N-WASP knockout fibroblasts [43]. Presumably, the required role of N-WASP in pedestal formation lies in its ability to stimulate the actin nucleation activity of the Arp2 ⁄ 3 complex, although a role for Arp2 ⁄ 3 in pedestal assembly has not been directly tested. Collectively, these studies highlight the fact that the predominant pathway for actin pedestal formation by EPEC is incredibly simple after Y474 phosphorylation (Fig. 3), because a 12-residue phosphopeptide can trigger a complete signaling cascade that leads to actin assembly. This sequence binds the SH2 domain of the redundant Nck adaptors, which in turn utilize their 2394 SH3 domains to bind and activate N-WASP to drive Arp2 ⁄ 3-mediated actin nucleation. Interestingly, this mechanism of localizing Nck and activating N-WASP is reminiscent of the signaling pathways that enable vaccinia virus actin tail formation [44], as well as nephrin-mediated actin organization in podocytes [45]. In fact, the Nck-binding Tir peptide can functionally replace the analogous region of nephrin [46]. Further dissection of Nck-mediated signaling cascades will surely illuminate additional similarities, and perhaps differences, in how EPEC pedestals and these other actin-rich structures are formed. EPEC Tir possesses Nck-independent activities that influence actin pedestal formation It is important to note that Tir activities other than Nck recruitment may influence the maintenance or architecture of the pedestal. This suggestion is supported by the previously mentioned observation that the N-terminus of Tir can interact with the SH3 domains of tyrosine kinases [36]. In addition, several Tir activities aside from Nck-binding that might modulate actin pedestal formation have also been uncovered. Nearly 25% of adherent EPEC can recruit N-WASP and form pedestals on cells lacking Nck1 and Nck2, suggesting that Tir has subsidiary cellular targets that affect actin dynamics [30]. Much of this Nck-independent signaling requires Y474 phosphorylation, because overexpression of the Y474-binding SH2 domain of Nck1 diminishes this remaining pedestal formation (to < 5%), as does a Y474F mutation [30]. Interestingly, a proteomic screen for phosphopeptidebinding proteins identified the SH2 domain of c-Src as a potential Y474-binding partner [47], raising the possibility that the Tir-associated kinases themselves might activate the actin assembly machinery. However, the abilities of Src-family kinases (and their SH3 domains) to directly drive N-WASP–Arp2 ⁄ 3-mediated actin pedestal formation in the absence of Nck remain to be tested. Although the vast majority of pedestal formation arises from signaling via phosphorylated Y474, phosphorylated Y454 appears to be important for much of the remaining  5% of pedestal formation [30]. This latter phosphopeptide was recently shown to bind to the SH2 domain of the tyrosine phosphatase Shp-2 [47] and to the SH2 domain of phosphatidylinositol 3-kinase (PtdIns3K) [35,47]. Whereas the role of Shp-2 in Tir signaling has not yet been explored, the interaction of Tir with PtdIns3K was shown to occur in a Y454-dependent manner [35,47]. Tir may also activate PtdIns3K [35], suggesting that it might transiently FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS K. G. Campellone increase the local concentration of phosphatidylinositol 3,4,5-triphosphate in the plasma membrane. Although phosphatidylinositol 3,4,5-triphosphate can accumulate beneath adherent EPEC [35], the role of PtdIns3K in actin pedestal formation requires further study. Overall, these findings have added an additional layer of complexity to our understanding of the mechanisms by which Tir exploits tyrosine kinase signaling cascades, and have opened new paths of investigation into how EPEC Tir alters the composition of the membrane during pedestal formation. EHEC Tir promotes phosphotyrosine- and Nck-independent actin assembly Unlike EPEC Tir, the Tir molecule from canonical EHEC strains of serotype O157:H7 does not have a residue equivalent to Y474, is not tyrosine phosphorylated and does not bind Nck to initiate actin assembly. Moreover, EHEC Tir does not function for actin pedestal formation when expressed in EPEC [38,48,49], because only  5% of these bacteria generate pedestals. These results provided the first evidence that EHEC encodes a separate effector, that is missing from EPEC, and allows its Tir molecule to generate pedestals in the absence of phosphotyrosines. The initial clue about how EHEC forms pedestals came from the observation that the region of EHEC Tir that is essential for signaling to the actin cytoskeleton is homologous to the Y454-containing region of EPEC Tir [50,51]. In fact, a 12-residue peptide encompassing this region harbors all of the essential signaling activity of the C-terminus of EHEC Tir [51]. Within this sequence, an asparagine–proline–tyrosine (N–P– Y458) tripeptide is critical for EHEC Tir function [52]. These three residues are also crucial for the low levels of pedestal formation that are elicited by the EPEC Tir Y454 region [52], revealing an underlying parallel in the Nck-independent methods by which EPEC and EHEC promote actin assembly. However, not until the effector that mediates the interaction between EHEC Tir and the actin assembly machinery was identified did the mechanism of EHEC pedestal formation come into focus. A WASP–TIR–EspFU (WTF) complex: driving EHEC pedestal formation EHEC EspFU associates with Tir and binds WASP ⁄ N-WASP to promote actin assembly Two independent studies, one that employed a genome-scale loss-of-function approach [53] and another Regulation of actin assembly by Tir and EspFU based on microarray expression profiling [54], identified the second EHEC pedestal effector as EspFU (also termed TccP). EspFU is 25% identical to the LEEencoded effector EspF, but is found within prophage U (hence its name). Deletion of EspFU impairs EHEC pedestal formation, whereas deletion of EspF does not [53,54], implying that the proteins have evolved distinct cellular functions. EspFU contains a conserved N-terminal type III secretion signal sequence and a more divergent C-terminus that includes A B Fig. 4. Regulation of actin assembly by EHEC Tir and EspFU. (A) EHEC Tir binds to the membrane-deforming inverse BAR (I-BAR) domains of the host proteins IRSp53 and IRTKS, which can dimerize (not shown) and may influence membrane shape (dotted arrow). The SH3 domains of IRSp53 or IRTKS interact with proline-rich sequences in the repeat region (R1–R6) of EspFU. Distinct sequences within each repeat bind and activate N-WASP by outcompeting its CA domain for binding to the AI region. Multimerization of N-WASP further enhances Arp2 ⁄ 3-mediated actin nucleation and pedestal assembly. Overall, homo-oligomerization of Tir and IRSp53 ⁄ IRTKS, combined with the multivalency of EspFU, results in a dramatic signal amplification such that two Tir molecules could potentially recruit up to 24 N-WASP molecules. (B) The sequence of a representative repeat peptide of EspFU is shown. The N-WASP-binding region and SH3 domain-binding sequence (both shown in bold and underlined) within each EspFU peptide are distinct. FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS 2395 Regulation of actin assembly by Tir and EspFU K. G. Campellone multiple ( 2 to 7) copies of a 47-residue peptide that contains alpha-helical and proline-rich sequences (Fig. 4A,B). By showing that EspFU associates indirectly with Tir [53], activates N-WASP–Arp2 ⁄ 3-mediated actin assembly in vitro [54] and binds directly to the N-WASP GBD using its C-terminal repeat region [53], these initial reports opened up new avenues of inquiry into EHEC signaling. In particular, they suggested the formation of an N-WASP–Tir–EspFU (WTF) complex and elicited investigations both into the mechanism of N-WASP activation, and into the interactions between Tir and EspFU. Tir and EspFU are the only two EHEC effectors required for pedestal formation, because clustering of Tir in the presence of EspFU (and in the absence of all other effectors) is sufficient to trigger actin pedestal assembly [55]. EspF is also capable of stimulating N-WASP in vitro [56], but this activity does not play any detectable role in pedestal formation [53,54]. The only critical role of EHEC Tir in actin assembly is to recruit EspFU, because a hybrid effector protein in which the C-terminus of Tir is replaced by the C-terminal repeat region of EspFU is fully functional for pedestal formation when clustered by intiminexpressing bacteria or anti-Tir-coated particles in a biomimetic experimental system [55] (Fig. 1). Consistent with recruitment by EspFU, N-WASP localizes to EHEC pedestals in GBD-dependent manner [12], and the GBD by itself is a dominant negative inhibitor of EHEC pedestal formation when overexpressed in cells [55], presumably because of an ability to bind and sequester the C-terminal repeats of EspFU. The length of this EspFU repeat region is somewhat variable among clinical isolates, although all EspFU proteins appear to contain at least two repeats [57], implying that the presence of multiple repeats is related to EspFU function. Importantly, recent biochemical and structural studies have provided insight into the role of the tandem repeat configuration in EspFU signaling. BAR hopping: membrane-deforming proteins and pedestal formation EspFU repeats mimic the N-WASP connector region to relieve autoinhibition The first hint that the EspFU repeats cooperate during actin pedestal formation came from a report in which an EHEC strain engineered to express a two-repeat EspFU derivative was observed to generate pedestals, whereas a strain expressing a one-repeat variant was not [58]. Subsequently, by clustering EspFU truncations at the plasma membrane, two additional studies demonstrated that a single repeat can trigger actin assembly in cells, but the presence of additional 2396 repeats correlates positively with the efficiency of polymerization [55,59]. In vitro, a single repeat can bind and activate recombinant WASP or N-WASP derivatives at much higher levels than its endogenous activators, including Cdc42 [59,60], and a single repeat can also activate the N-WASP ⁄ WIP complex [55]. Nevertheless, quantitative comparisons of the activity of EspFU variants containing different numbers of repeats indicate that the repeats exhibit cooperativity in activating N-WASP and multiple repeat derivatives have remarkably higher stimulatory activity [55,59]. This repeat synergy enhances N-WASP binding to the Arp2 ⁄ 3 complex [55], because N-WASP dimerization increases its affinity for the Arp2 ⁄ 3 complex by > 100-fold [61]. In addition, recent structural analyses have revealed the precise mechanism by which these sequences activate N-WASP. The first half of each EspFU repeat forms an a helix that mimics the connector portion of the WCA domain of N-WASP, which is normally involved in binding to the AI region of the GBD [59,60]. Point mutants of hydrophobic EspFU residues within this helix prevent binding to the GBD [59,60], block WASP activation in vitro [60] and impair the ability of N-WASP to trigger pedestal formation in cells [60]. Thus, given both their high affinity for the AI region and their inherent multivalency, the EspFU repeats are able to outcompete the C motif for binding to the GBD (Fig. 4A). This results in a relief of autoinhibition and physical displacement of the WCA domain that frees it to activate the Arp2 ⁄ 3 complex. These results explicitly defined how structural mimicry and multivalency make the EspFU repeat region such a potent N-WASP activator. Importantly, however, the role of the proline-rich half of each repeat in actin pedestal formation and the mechanisms by which EspFU interacts with Tir in the WTF complex did not become clear until proteome-wide searches for EspFU-binding partners were undertaken. The inverse Bin-amphiphysin-Rvs167 (I-BAR) protein family links Tir to EspFU during actin pedestal formation The findings that Tir and EspFU are the only effectors required for EHEC pedestal formation [55] and these proteins do not bind to one another directly [53], led to the surprising conclusion that the factors that mediate the Tir–EspFU association in the WTF complex must be derived from the host cell. Two recent studies FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS K. G. Campellone have now independently identified factors that bridge the interaction between EHEC Tir and EspFU. An unbiased screen for EspFU-binding proteins in tissue extracts [62] and a targeted screen for EspFU-binding SH3 domains [63] uncovered IRSp53 and IRTKS, respectively. These two factors are members of the inverse Bin-amphiphysin-Rvs167 (I-BAR) family, a group of proteins that possess an I-BAR domain that binds membranes to induce plasma membrane protrusion, and also contain an SH3 domain that interacts with proline-rich peptides (Fig. 4A). IRSp53 and IRTKS bind to the Y458-containing region of EHEC Tir using their N-terminal I-BAR domain and bind to the proline-rich portions of the EspFU repeats with their C-terminal SH3 domains [62,63], thereby physically linking the two EHEC effectors. Presumably, the reason why actin pedestal formation relies on EspFU rather than EspF is that the distinct proline-rich motifs in EspF are not recognized by the SH3 domains of the I-BAR proteins. IRSp53 and IRTKS are critical for actin assembly during EHEC infection, because genetic deletion of IRSp53, RNAi-mediated silencing of IRTKS or overexpression of dominant negative truncations of either of these proteins inhibits EspFU recruitment and pedestal formation in cells [62,63]. This unique alternation of bacterial and host components illuminates the complex dialogue between EHEC and mammalian cells during infection. These results have also likely defined the final components necessary for recapitulating EHEC-mediated actin assembly in vitro. In one model of such a minimized system, the clustered Y458-containing Tir peptide first recruits IRSp53 and ⁄ or IRTKS via their I-BAR domains (Fig. 4A). The SH3 domains of IRSp53 or IRTKS then bind to the proline-rich motifs in EspFU, which uses its adjacent N-WASP-binding helices (Fig. 4B) to activate N-WASP–Arp2 ⁄ 3-mediated actin polymerization. Because IRSp53 and IRTKS can homodimerize, and EspFU contains multiple repeats that activate N-WASP, this signaling pathway exhibits a remarkable degree of amplification at each step after Tir clustering. The primary roles of IRSp53 and IRTKS during pedestal formation are apparently to act as adaptors between Tir and EspFU, because the C-terminus of Tir can be functionally replaced with the SH3 domain of either IRSp53 or IRTKS [63], or with the repeat region of EspFU [55]. However, these intermediates might have additional functions in actin pedestal formation. For example, their I-BAR domains may be important for deforming the membrane during pedestal protrusion. Moreover, it seems plausible that Regulation of actin assembly by Tir and EspFU IRSp53 and IRTKS can directly activate N-WASP using their SH3 domains. These putative contributions to pedestal formation may also be conserved during EPEC infections, because the sequences around EHEC Tir Y458 and EPEC Tir Y454 are so similar. It is important to note that these activities of IRSp53 and IRTKS are also likely to be influenced by other accessory factors that are recruited by Tir and EspFU to modulate the dynamics of the plasma membrane and ⁄ or the architecture of the pedestal. Fes ⁄ CIP4-Bin-amphiphysin-Rvs167 (F-BAR) proteins and other regulators of actin dynamics influence pedestal formation Interestingly, in the screen that identified IRSp53 as an intermediate in the Tir–EspFU complex, several other putative EspFU-binding partners were identified, including WIP and VASP [62]. Based on their known biochemical properties, each of these proteins could contribute to pedestal dynamics. As mentioned previously, WIP is a G- and F-actin-binding protein that forms a complex with N-WASP and can modulate its activity, so it might be important for actin assembly in the pedestal. VASP facilitates the elongation of unbranched filaments, so it might affect the architecture of F-actin in the pedestal. EspFU has also recently been reported to interact with CIP4 [62] and Toca-1 (K. G. Campellone, A. D. Siripala, J. M. Leong & M. D. Welch, unpublished results). These two proteins each contain an F-BAR domain that can bind and deform membranes to promote invagination, in contrast to I-BAR domains that facilitate protrusion. They also possess C-terminal SH3 domains. Interestingly, Toca-1 can bind to EspFU and activate NWASP using its SH3 domain, and also contributes to the efficiency of actin pedestal formation (K. G. Campellone, A. D. Siripala, J. M. Leong & M. D. Welch, unpublished results). Although the role of BAR domains in deforming membranes during EHEC infection has yet to be examined, it seems likely that competition or collaboration among the I-BAR and F-BAR proteins will have important consequences for the morphology of pedestals. In addition to the SH3 domains of multiple BAR proteins, EspFU may also bind to the SH3 domain of cortactin, an actin nucleation-promoting factor that is structurally and functionally distinct from WASP and N-WASP [64]. Cortactin is a weak Arp2 ⁄ 3 activator, but is able to bind to F-actin and stabilize Arp2 ⁄ 3bound branch junctions. Along with binding to EspFU, cortactin appears to bind to the N-terminus of EHEC FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS 2397 Regulation of actin assembly by Tir and EspFU K. G. Campellone Tir [64], although the latter interaction does not appear to be sufficient to promote cortactin recruitment in cells [65]. It is interesting to note, however, that the N-terminus of EHEC Tir is known to influence the length of pedestals, because removal of this domain results in abnormally long pseudopods [51]. Thus, it seems plausible that the absence of Tir–cortactin interactions results in an increase in pedestal length because of a decrease in the density of branched filament networks. Clearly, much remains to be learned about how Tir and EspFU cooperate to promote actin assembly, and how their signaling activities resemble the pathways that normally regulate actin assembly beneath the plasma membrane of uninfected mammalian cells. Linking actin pedestal formation and pathogenesis Actin pedestals and colonization It is well established that Tir is essential for intestinal colonization and the formation of attaching and effacing lesions in animals [16,27,28], but the major virulence defects exhibited by Tir-deficient mutants in vivo can likely be attributed to a lack of intimate adherence to host cells. Thus, a specific pathogenic role for the subsequent assembly of actin pedestals has been enigmatic for many years. A C. rodentium derivative expressing a Tir Y474F mutant is quantitatively indistinguishable from a strain expressing wild-type Tir in the colonization of mice, and is also capable of forming attaching ⁄ effacing lesions in the intestine [27]. Moreover, EPEC strains expressing either a Tir Y474F mutant or a Y454F ⁄ Y474F double mutant successfully infect the duodenum and generate attaching ⁄ effacing lesions in a human intestinal culture model [66]. Although it is difficult to quantify actin pedestal formation versus attaching ⁄ effacing lesion formation in animals, these results are consistent with the hypothesis that the manner in which pedestals are formed in cells in vitro may not correlate with pedestal formation in vivo [5]. It has been suggested that the EPEC and EHEC versions of Tir diverged from an ancestral Tir molecule which used an N–P–Y sequence to stimulate actin assembly [52]. According to this model, EHEC acquired EspFU to amplify its Tir N–P–Y458 pathway, whereas EPEC incorporated Y474 into its Tir molecule to initiate a Nck-dependent signaling cascade that augments its N–P–Y454 signaling module. Intriguingly, some pathogenic E. coli isolates appear to lack both EspFU and Y474-like phosphotyrosine signaling 2398 elements [67], whereas others can actually use EspFU and Y474-mediated actin assembly mechanisms simultaneously during pedestal formation [68]. Although the role of actin pedestal formation in EPEC and C. rodentium pathogenesis remains obscure, the role of EspFU and pedestal formation in colonization by EHEC is now beginning to be elucidated. EspFU does not play a measurable role in intestinal colonization in calf and lamb reservoir models of EHEC infection [69]. However, in other animal models for examining EHEC pathogenesis, EspFU seems to be important for bacterial expansion beyond the original sites of EHEC infection [70]. After initial colonization of the cecum and mid-colon in infant rabbits, an EHEC EspFU mutant fails to proliferate, unlike wildtype EHEC which continues to increase in number [70]. Moreover, qualitative examinations of infections by an EspFU-deficient EPEC isolate engineered to express EspFU from a plasmid suggest that EspFU may improve the colonization efficiency of the terminal ileum in human intestinal organ cultures [67]. Finally, wild-type EHEC colonizes a larger area of the cecum than an EspFU mutant in gnotobiotic piglets [70]. In this experimental model, wild-type EHEC appeared to form electron-dense actin pedestals more frequently than its EspFU mutant counterpart [70], suggesting for the first time a link between actin pedestal formation and colonization. Potential pedestal purposes Although the role of actin pedestal formation in pathogenesis is beginning to come into focus, the biological purpose of the pedestal is still very much a mystery. Even so, several plausible models for its function can be envisioned. For example, because actin assembly mediates EPEC and EHEC motility on the surfaces of cultured cells in vitro [71], the impaired expansion and proliferation of the EspFU mutant in vivo might be explained by a defect in actin-based motility. It is also possible that pedestal-based adherence to the epithelium makes the bacteria more resistant to flow-mediated detachment during diarrhea. However, actin pedestal formation might only be a visual byproduct of an important underlying process. In this scenario, perhaps Tir-mediated actin assembly enhances the translocation of other factors into the host cell. This hypothesis is supported by observations using EPEC, where Tir is the first effector injected into the host cell and has the highest steady-state levels [72], and where Tir secretion is important for the efficient secretion of other effectors [73]. Nevertheless, the hierarchy of effector secretion during EHEC infection FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS K. G. Campellone and the influence of actin assembly on translocation remain to be determined. Interestingly, a growing body of evidence is consistent with the hypothesis that actin pedestal formation is an antiphagocytic mechanism. First, both EPEC and EHEC recruit a remarkable number of host factors that are known to play a role in phagocytosis or endocytosis. These include PtdIns3K, which remodels the plasma membrane during phagocytosis, the membrane invaginating proteins CIP4 and Toca-1, plus N-WASP, cortactin and the Arp2 ⁄ 3 complex, which are thought to provide a burst of actin polymerization during endocytic vesicle scission. EPEC and ⁄ or EHEC have also been shown to recruit clathrin [74], which forms a coat on invaginating vesicles, and dynamin 1 or dynamin 2 [75], which promote vesicle scission. Notably, all of these factors may interact with EspFU [62], suggesting that they are not mere bystanders during EHEC infection of host cells. Thus, the ability of EspFU to alter the activity of these proteins or upset the timing or degree of N-WASP activation may short-circuit normal endocytic processes. These and other hypotheses that might describe the function of the actin pedestal await rigorous experimental testing. Concluding remarks The last several years have resulted in a fine-tuning of our models for how EPEC Tir interacts with tyrosine kinases and the ways in which its phosphotyrosines recruit adaptor proteins and other factors to activate the N-WASP–Arp2 ⁄ 3 actin assembly machinery. As a consequence, the relationship between EPEC Tir signaling and endogenous mammalian tyrosine kinase cascades has come into greater focus. In addition, the last 2 years in particular have yielded major advances in our understanding of the seemingly crafty mechanisms by which EHEC EspFU triggers pedestal assembly. These most recent studies have provided biochemical and structural insights into N-WASP activation, suggested roles for membrane-deforming proteins in actin pedestal biogenesis, and uncovered parallels between pedestal constituents and the components that operate during endocytic processes. Finally, the observation that EspFU contributes to colonization in animal models of infection substantiates a long-suggested connection between actin pedestal formation and pathogenesis. As long as pedestal formation continues to support our ability to learn about the regulation of actin assembly and plasma membrane dynamics, the study of Tir- and EspFU-mediated signaling cascades will remain at the leading edge of inquiries into EPEC and EHEC infection. Regulation of actin assembly by Tir and EspFU Acknowledgements I thank John Leong, Taro Ohkawa, and Didier Vingadassalom for comments on this manuscript and Art Donohue-Rolfe, Cindy Lai, John Leong, Loranne Magoun and Saul Tzipori for contributing electron micrographs. I would also like to acknowledge the late David Schauer for his contributions to the C. rodentium field and his scientific input during my graduate training. References 1 Stevens JM, Galyov EE & Stevens MP (2006) Actindependent movement of bacterial pathogens. Nat Rev Microbiol 4, 91–101. 2 Hayward RD, Leong JM, Koronakis V & Campellone KG (2006) Exploiting pathogenic Escherichia coli to model transmembrane receptor signalling. Nat Rev Microbiol 4, 358–370. 3 Tarr PI, Gordon CA & Chandler WL (2005) Shiga-toxin-producing Escherichia coli and haemolytic uraemic syndrome. Lancet 365, 1073–1086. 4 Wick LM, Qi W, Lacher DW & Whittam TS (2005) Evolution of genomic content in the stepwise emergence of Escherichia coli O157:H7. J Bacteriol 187, 1783– 1791. 5 Frankel G & Phillips AD (2008) Attaching effacing Escherichia coli and paradigms of Tir-triggered actin polymerization: getting off the pedestal. Cell Microbiol 10, 549–556. 6 Nougayrede JP, Fernandes PJ & Donnenberg MS (2003) Adhesion of enteropathogenic Escherichia coli to host cells. Cell Microbiol 5, 359–372. 7 Kaper JB, Nataro JP & Mobley HL (2004) Pathogenic Escherichia coli. Nat Rev Microbiol 2, 123–140. 8 Spears KJ, Roe AJ & Gally DL (2006) A comparison of enteropathogenic and enterohaemorrhagic Escherichia coli pathogenesis. FEMS Microbiol Lett 255, 187–202. 9 Campellone KG & Welch MD (2010) A nucleator arms race: cellular control of actin assembly. Nat Rev Mol Cell Biol 11, 237–251. 10 Chesarone MA & Goode BL (2009) Actin nucleation and elongation factors: mechanisms and interplay. Curr Opin Cell Biol 21, 28–37. 11 Stradal TE & Scita G (2006) Protein complexes regulating Arp2 ⁄ 3-mediated actin assembly. Curr Opin Cell Biol 18, 4–10. 12 Lommel S, Benesch S, Rohde M, Wehland J & Rottner K (2004) Enterohaemorrhagic and enteropathogenic Escherichia coli use different mechanisms for actin pedestal formation that converge on N-WASP. Cell Microbiol 6, 243–254. FEBS Journal 277 (2010) 2390–2402 ª 2010 The Author Journal compilation ª 2010 FEBS 2399