Tir, a unique paradigm for host cell attachment and signalling by a microbial pathogen
Upon translocation into host cells, Tir inserts into the apical membrane such that the N- and C-terminal portions of Tir are cytosolic, with helical transmembrane domains traversing the host apical membrane. The central (∼ 100-residue) portion, known as the Intimin-Binding Domain (IBD), forms two helices separated by a hairpin loop on the external surface of the host cell (Kenny, 1999; Luo et al., 2000). This loop binds the C-terminus of intimin with high affinity (Kd of ∼ 10 nM for EPEC and less than 1 nM for EHEC) thus facilitating extremely tight attachment of these pathogenic E. coli to the host cell (DeVinney et al., 1999; Ross and Miller, 2007). Based on sequence and antigenic variation within the 280 aa variable C-terminal region, a classification scheme for intimin was developed, becoming an important epidemiological marker alongside the O:H serotypes. The most common intimin types are α, expressed by EPEC belonging to the EPEC1 evolutionary branch; β, expressed by EPEC2 and some non-O157 EHEC (e.g. O26) and γ, which is mainly found in EPEC O55:H7 and EHEC O157:H7 (McGraw et al., 1999). Although intimin types are strain-specific, they are interchangeable between the strains and can interact with any Tir IBD (DeVinney et al., 1999).
In addition to its function as a bacterial receptor, Tir serves as a signalling effector that elegantly initiates the recruitment of host adaptors and actin nucleators from inside the host cytosol, a strategy commonly used by intracellular pathogens to exert pathophysiological effects. Although EPEC and EHEC Tirs are homologous in structure and share ∼ 60% identity (Kenny, 1999) they use divergent mechanisms of signalling that then converge on the activation of common actin nucleators to produce pedestals that are indistinguishable when examined by electron microscopy or fluorescence imaging of actin (Fig. 1).
Figure 1. Actin pedestals as visualized by transmission electron microscopy appear as electron dense protrusions beneath intimately attached bacteria (A). By immunofluorescence microscopy, actin pedestals can be visualized with phalloidin-conjugated fluorophores. EHEC (blue; DAPI stained) induces actin pedestals on infected HeLa cells (B) and HCT8 intestinal cells (C) stained with Alexa-488-conjugated phalloidin (green). White arrowheads point to typical actin pedestals for these cell types. Tir stained with Alexa-568 (red) colocalizes with actin at the tips of actin pedestals, forming yellow foci beneath intimately adhered bacteria.
Download figure to PowerPoint
Both intimin and Tir appear to be adapted for multimerization. The beta barrel region of intimin, which anchors intimin to the bacterial outer membrane, and the IBD of Tir each encode dimerization domains, giving rise to a model in which the two proteins might efficiently form higher-order ‘lattices’ essential for downstream signalling (Luo et al., 2000; Touzé et al., 2003).
Within an infected host cell, Tir interacts with a number of host proteins through its N- and C-terminal cytosolic domains. The N-terminal tail is functionally interchangeable between EHEC and EPEC Tirs and binds various focal adhesion proteins, thereby potentially linking Tir to the cytoskeleton (Goosney et al., 2000; Kenny, 2001). However, such activity is insufficient for actin pedestal formation (Kenny, 2001). Both EPEC and EHEC require the C-terminal tail, the portion of Tir that is most divergent between these two highly related pathogens (DeVinney et al., 1999). Notably, intimin binding enhances post-translocation modification of the C-terminus of Tir inside the host cell. These modifications, when fully characterized, were pivotal in revealing the mechanism of actin pedestal formation (Kenny et al., 1997; Campellone et al., 2004a; Campellone and Leong, 2005).
EPEC pedestal formation pathways
Host modification of the C-terminus of EPEC Tir is critical for its activity. This domain contains multiple tyrosine, threonine and serine residues that are available for potential host phosphorylation. Serine/threonine phosphorylation, which changes the apparent molecular weight of Tir on SDS-PAGE where it appears as a 90 kDa protein (similar to the MW of Hp90), has been suggested to play a role in Tir insertion into the apical membrane, but otherwise does not contribute to actin pedestal formation. Tyrosine phosphorylation of EPEC Tir, on the other had, is absolutely required for efficient actin polymerization (Kenny et al., 1997). Multiple protein kinases, including members of the Abl and Src kinase families, have been localized to EPEC pedestal tips and thus implicated in pedestal formation; broad inhibition of both these classes of kinases by non-specific pharmacological inhibitors can inhibit EPEC pedestals (Swimm et al., 2004; Hayward et al., 2009). However, EPEC can still induce pedestal formation on a variety of cultured cells lacking individual kinases from the Abl or Src families, suggesting that EPEC Tir is capable of utilizing multiple, redundant, tyrosine kinases for Tir phosphorylation in vitro (Swimm et al., 2004).
Of the Tir C-terminal tyrosine residues, phosphorylation of residue Y474 (in EPEC1 strain E2348/69) is the most critical for focal actin assembly and triggers the major pathway by which EPEC1 strains initiate pedestal formation in vitro (Fig. 2, pathway 1) (Kenny et al., 1997; DeVinney et al., 2001). In clinical isolates, this critical tyrosine residue may be in other close positions due to small insertions or deletions. This pedestal formation pathway utilizes the host adaptor Nck, which contains an SH2 domain capable of binding a 12-residue region of EPEC Tir that includes the critical phosphorylated Y474 (Gruenheid et al., 2001; Campellone et al., 2002). Nck, found at the tips of EPEC-induced actin pedestals, also contains three SH3 domains that may directly or indirectly recruit the host actin nucleation promoting factor N-WASP (Gruenheid et al., 2001; Buday et al., 2002). N-WASP, in turn, recruits and activates the actin nucleating Arp2/3 complex (Snapper and Rosen, 1999; Miki and Takenawa, 2003).
Figure 2. EPEC1 and EHEC use different mechanisms of Tir activated signalling cascades that converge on common host actin nucleation factors, N-WASP and Arp2/3 to form actin pedestals. The major EPEC1 actin pedestal pathway requires phosphorylation of the critical Tir residue Y474, which recruits Nck (pathway 1). Nck can activate N-WASP directly or through WIP. The major EHEC actin pedestal pathway requires Tir residues NPY458 and the EHEC effector EspFU/TccP (pathway 2). EspFu/TccP is linked to Tir via host adaptors IRTKS and IRSp53 and activates N-WASP. Minor EPEC actin assembly pathways utilize Tir residues Y474 or Y454 to initiate Nck independent activation of N-WASP (pathways a and b). A minor actin assembly pathway for EHEC utilizes EspFu, but does not require N-WASP for Arp2/3 activation (pathway c).
Download figure to PowerPoint
Due to similarities in the use of Nck and N-WASP by vaccinia virus for ‘actin comet tail’ formation, a similar mechanism of actin assembly has been proposed for EPEC (Campellone and Leong, 2005). Vaccinia produces A36R, a Nck-binding protein, that, like EPEC Tir, is tyrosine phosphorylated by the host. An eight-residue sequence in the A36R Nck-binding region is nearly homologous to the Y474 region of EPEC Tir (Campellone and Leong, 2005). Recruitment of Nck by A36R leads to indirect activation of N-WASP through the intermediary adaptor, WASP Interacting Protein (WIP) (Moreau et al., 2000). WIP contains N-terminal proline-rich domains that are capable of binding Nck's SH3 domains; its C-terminal WASP-binding domain (WBD) activates N-WASP by binding to the N-WASP N-terminal WH1 domain (Moreau et al., 2000). Ectopic overexpression of the WBD decreases vaccinia actin comet tail formation, presumably by a dominant-negative effect, and various mutations in the WH1 domain of N-WASP can abrogate its recruitment and actin comet tail formation by vaccinia (Moreau et al., 2000).
A role for WIP in pathogen-mediated actin assembly appears to depend on the experimental system. Garber and colleagues showed that both EPEC and vaccinia are still able to induce actin assembly in a cell line lacking WIP (Garber et al., 2012). N-WASP also contains a proline-rich domain that is capable of binding Nck directly, and thereby leading to its activation (Antón et al., 1998). As is the case for the kinases(s) responsible for EPEC Tir phosphorylation, seemingly disparate observations concerning the mechanism of N-WASP activation by Nck and the role of WIP may indicate a degree of redundancy in the N-WASP activation pathway.
Although the major pathway for EPEC pedestal formation requires Nck, Nck-independent mechanisms of actin pedestal formation have also been uncovered, reflecting redundant pedestal formation pathways (Fig. 2, pathways a and b). EPEC infection of Nck-deficient murine embryonic fibroblasts still generate roughly one-fifth the number of the actin pedestals generated upon infection of wild-type fibroblasts. Most of this activity, also requires Y474, reaffirming its role as a key residue for triggering actin assembly by EPEC Tir. However, even in the absence of Y474, about 2–5% of bound EPEC can generate actin pedestals, a rate more than 10-fold higher than for a Tir-deficient EPEC strain. An additional tyrosine residue at position 454 of EPEC Tir is critical for this minor Nck-independent actin assembly activity (Campellone and Leong, 2005). Initially considered of marginal importance, this pathway was later shown to be highly related to the major pathway of actin pedestal formation by EHEC.
EHEC pedestal formation pathways
A critical first indication that EHEC and EPEC utilized distinct pedestal formation pathways came from studies showing that tyrosine phosphorylated proteins were absent from EHEC actin pedestals (DeVinney et al., 2001). In fact, EHEC Tir lacks the equivalent of EPEC Tir Y474, and EHEC does not require Nck for actin pedestal formation (Kenny, 1999; DeVinney et al., 2001). Consistent with distinct pathways of actin assembly, EHEC tir cannot complement EPEC tir mutants for actin pedestal formation unless modified to encode the critical Y474-containing region of EPEC Tir (Kenny, 2001; Campellone et al., 2002). Co-infection of cultured tir and eae (encoding intimin) deletion mutants of EPEC and EHEC revealed that the latter requires an additional EHEC-specific effector for pedestal formation (DeVinney et al., 2001; Kenny, 2001). These observations pointed to different mechanisms of actin pedestal formation for EPEC and EHEC (Fig. 2).
Experiments using chimeric Tir proteins suggested that the key to differences in EPEC and EHEC Tir lie in the C-terminus. While the N-terminus and IBD of EHEC and EPEC Tir are highly similar and functionally interchangeable, their cytosolic C-terminal domains are only ∼ 41% identical (DeVinney et al., 1999). Further studies of this region revealed that an NPY sequence at residues 456–458 of EHEC Tir is critical for actin pedestal formation (Brady et al., 2007). EHEC Tir residue Y458 corresponds to EPEC Tir Y454, the residue previously shown to be essential for triggering one of the minor Nck-independent actin assembly pathways in EPEC pedestal formation, described above.
Instead of utilizing tyrosine phosphorylation and Nck recruitment, the EHEC Tir NPY-mediated pathway acts through a different category of host adaptors (Fig. 1, pathway 2). The NPY458 sequence of EHEC Tir, and likely the equivalent sequence of EPEC Tir as well, binds to IRTKS and IRSp53, members of the I-Bar subfamily of membrane deforming and remodelling proteins. These proteins link to Tir via their N-terminal IMD domains, which also form ‘zeppelin'-shaped homodimers associated with convex membrane protrusions (Vingadassalom et al., 2009; Weiss et al., 2009; de Groot et al., 2011; Zhao et al., 2011). The C-terminal regions of IRTKS and IRSp53 contain additional motifs, such as an SH3 domain, that link their membrane deforming activity to signalling proteins and actin regulators (Vingadassalom et al., 2009; Weiss et al., 2009).
A paradox in this Tir NPY-mediated pedestal pathway is that it promotes pedestal formation with low efficiency for EPEC, yet with high efficiency for EHEC (Campellone and Leong, 2005; Campellone et al., 2006). This riddle was solved by the identification of an additional non-LEE-encoded bacterial effector, termed EspFU (a homologue of the effector EspF encoded on prophage U) or TccP (Tir cytoskeleton coupling protein) (Garmendia et al., 2004; Campellone et al., 2004b). Found in EHEC O157:H7, EspFU/TccP also greatly enhances NPY458 mediated pedestal formation if expressed in EPEC1 (Campellone et al., 2006).
Discovered before the Tir NPY458 residues and IRTKS/IRSp53, EspFU/TccP provided structural insights that helped elucidate both the identity of its linker to EHEC Tir and the mechanism of N-WASP activation. EspFU/TccP consists of an N-terminal secretion signal, followed by (in nearly all isolates) between three and eight repeating elements (Garmendia et al., 2005; 2006). Each functional repeating unit contains two structurally and functionally distinct domains, an N-terminal helical section that binds and activates N-WASP, and a C-terminal proline-rich domain with multiple canonical SH3 binding PxxP motifs that are recognized by the SH3 domain in the C-terminus of IRTKS and IRSp53 (Garmendia et al., 2006; Cheng et al., 2008; Aitio et al., 2010).
EspFU/TccP acts through structural mimicry and, remarkably, has evolved to outcompete some endogenous host ligands, making it a most efficient hijacker of mammalian signalling pathways. SH3 domains are an abundant and common motif for affecting cellular interactions. The proline-rich domains of EspFU/TccP recognize the SH3 domain of IRTKS with high affinity (Aitio et al., 2010), potentially allowing it to outcompete Eps8, a host actin modulator that appears to be a major binding partner of IRSp53 (Funato et al., 2004; Disanza et al., 2006). Like EspFU/TccP, Eps8 and has been shown to bind the SH3 domains of IRTKS and IRSp53 via multiple PxxP motifs (Aitio et al., 2010). Although the IRTKS/IRSp53-binding sequences of Eps8 and EspFU/TccP are nearly identical, the latter binds to the IRTKS SH3 domain with higher affinity due to the presence of a tryptophan, rather than an alanine, between tandem PxxP motifs. Thus, EspFU/TccP's tryptophan ‘switch’ may allow EHEC to usurp control of IRTKS/IRSp53 from natural host ligands (Aitio et al., 2012).
Similarly, the N-WASP-activating helical domain of EspFU/TccP acts by high-affinity mimetic displacement. N-WASP's C-terminal VCA domain, which functions to activate Arp2/3, is normally maintained in an autoinhibited state, due to an intramolecular interaction between an amphipathic helix in this domain and a hydrophobic groove located the GBD domain in the central portion of N-WASP (Snapper and Rosen, 1999; Miki and Takenawa, 2003; Cheng et al., 2008). Endogenous N-WASP activators such as Cdc42 disrupt this autoinhibition by binding to the GBD and indirectly disrupting this interaction by altering the GBD structure (Snapper and Rosen, 1999; Miki and Takenawa, 2003). In contrast, EspFu/TccP's helical domain is structurally similar to the helix in the N-WASP VCA domain and can directly displace it from the GBD, thereby activating N-WASP (Cheng et al., 2008).
A notable finding is that more than 99% of EspFU/TccP alleles possess at least three repeats of the functional (i.e. helical- and proline-rich-containing) domains, suggesting that multiple repeats confer a selective advantage to EHEC (Garmendia et al., 2005). Multiple EspFU/TccP repeats synergistically enhance activation of Arp2/3 in vitro (Campellone et al., 2008; Sallee et al., 2008). In addition, we speculated above that the dimerization domains of intimin and Tir might promote a higher-order structure conducive to efficient actin signalling. In fact, the presence of dimerization activity in IRTKS/IRSp53 indicates that three of the four components in this signalling cascade, i.e. intimin, Tir, IRTKS/IRSp53 each have the capacity to multimerize and the fourth, TccP/EspFU, encodes multiple repeat elements. A compelling hypothesis is that EHEC utilizes a series of both bacterial and mammalian components to promote the formation of a large multimeric complex at the host cell membrane, one that efficiently generates filamentous actin and enhances EHEC's ability to attach to colonic epithelium.
Finally, just as EPEC encodes multiple actin assembly pathways, EHEC triggers yet another alternative pathway for pedestal formation in a cell culture model (Fig. 2, pathway c). Low levels of actin assembly can be observed when embryonic fibroblasts from N-WASP knockout mice are infected with an intimin-expressing E. coli strain that delivers EHEC Tir and EspFU/TccP. This N-WASP-independent pathway depends on multiple repeats of EspFU/TccP, again suggesting a functional role for multiple repeats (Vingadassalom et al., 2010) but the critical elements of this pathway are yet unidentified.
TccP2/EspFM-induced actin polymerization in EPEC2 and non-O157 EHEC
Typical EHEC O157:H7 also possesses a pseudogene of EspFU/TccP encoded on prophage CP-933 M/Sp4, and thus called TccP2/EspFM, which lacks an intact type III secretion signal (Ogura et al., 2007). Beta-glucuronidase-positive and sorbitol-fermenting EHEC O157 strains harbour an intact tccP2/espFM gene as well as an espFU/tccP gene (Ogura et al., 2007). Although they carry divergent N-terminal secretion signals, EspFU/TccP and TccP2/EspFM share the same repeat structure, and EspFU/TccP and TccP2/EspFM are functionally interchangeable in EHEC O157 strains (Ooka et al., 2007).
Unexpectedly, EspFU/TccP was also found in EPEC strains of the serotype O119:H6, which express Tir that contains both NPY458 and Y474 equivalents (Ooka et al., 2007). Infections of cultured cells have shown that EPEC O119:H6 can simultaneously utilize the Tir:Nck and Tir:IRTKS/IRSp53:EspFU/TccP pathways (Fig. 3) (Whale et al., 2007). Later, this infection strategy was found to be common among EPEC strains belonging to the EPEC2 evolutionally lineage and non-O157:H7 strains, as they encode both a functional TccP2/EspFM and a Tir protein that can be tyrosine phosphorylated (Whale et al., 2006; Ogura et al., 2007). These results show the existence of yet another level of redundancy and suggest that Tir-induced actin polymerization is under strong selective pressure.
Figure 3. EPEC2 and non-O157 EHEC can use the major pathways of both typical EPEC1 and EHEC O157 to generate actin pedestals. Tir in these strains contain both a Y474 equivalent, which initiates actin assembly through Nck (pathway 1), and an NPY458 equivalent, which initiates actin assembly through IRTKS/IRSp53 and an EspFU/TccP homologue, TccP2/EspFM (pathway 2).
Download figure to PowerPoint
Map and EspH impact on Tir-induced actin polymerization
Map and EspH have both been implicated in filopodia formation and modulation of Tir-induced actin polymerization in undifferentiated epithelial cells. While Map is a Cdc42 guanine nucleotide exchange factor (GEF) (Huang et al., 2009), EspH is a RhoGEF inhibitor that acts by competitively binding to the tandem Dbl-homology and pleckstrin-homology (DH-PH) domains of Dbl-family mammalian RhoGEFs (Dong et al., 2010). Ectopic expression of EspH leads to cell detachment, while infections with strains either missing or overexpressing EspH result in short or elongated pedestals respectively. The cytotoxicity of EspH could be offset by expression of the bacterial RhoGEFs EspM2 and EspT (Wong et al., 2012a). EPEC infection leads to accumulation of WIP at the tip of the Tir-induced actin pedestals (Garber et al., 2012; Wong et al., 2012b) in a process involving EspH, which itself taggers actin polymerization in a Tir-dependent mechanism (Wong et al., 2012b).
Activation of Cdc42 by Map leads to transient filopodia formation during early stages of infection; interaction of Map with NHERF1 via its PDZ-biding motif is needed for filopodia stabilization (Berger et al., 2009). Recovery from the filopodial signals requires phosphorylation of Tir tyrosine Y474 and activation of the actin polymerization pathway as either infection of Nck-proficient cells with EPEC expressing TirY474 or infection of Nck-deficient cells with wild-type EPEC results in persistence of filopodia (Berger et al., 2009). The relevance of these pathways in EPEC-induced microvilli effacement was recently reported following infection of polarized Caco-2 cells (Dean et al., 2013). Taken together, these data show the existence of an elaborated cross-talk between Tir and other effectors that subvert actin dynamics.