Cytoskeleton-modulating effectors of enteropathogenic and enterohemorrhagic Escherichia coli: a case for EspB as an intrinsically less-ordered effector


D. Hamada, Division of Structural Biology (G-COE), Department of Biochemistry and Molecular Biology, Graduate School of Medicine, Kobe University, 7-5-1 Kusunoki-cho, Chuo-ku, Kobe 650-0017, Japan
Fax: +81 78 382 5816
Tel: +81 78 382 5817


Enterohemorrhagic and enteropathogenic Escherichia coli produce various effector proteins that are directly injected into the host-cell cytosol through the type III secretion system. E. coli secreted protein (Esp)B is one such effector protein, and affects host-cell morphology by reorganizing actin networks. Unlike most globular proteins that have well-ordered, rigid structures, the structures of type III secretion system effectors from pathogenic Gram-negative bacteria, including EspB, are often less well-ordered. This minireview focuses on the functional relationship between the structural properties of these proteins and their roles in type III secretion system-associated pathogenesis.


enterohemorrhagic Escherichia coli


enteropathogenic Escherichia coli


E. coli secreted protein


type III secretion system


translocated intimin receptor


Gram-negative pathogenic bacteria maintain a type III secretion system (T3SS) that functions in secreting virulence factors directly into the cytosolic space of host cells [1]. Among such virulence factors, several effector proteins influence the morphology of actin filaments that maintain host-cell morphology and cell–cell contacts.

In the case of enteropathogenic or enterohemorrhagic Escherichia coli (EPEC or EHEC, respectively), effectors involved in actin reorganization include E. coli secreted protein (Esp)B [2,3], EspFU [4–6] and EspL2 [7,8]. By interacting with host proteins involved in the regulation of actin morphology, these factors control morphological changes in filaments, thereby allowing the formation of actin-based pedestals that underlie bacterial attachment sites on the host-cell membrane.

The work of our group focuses on the role of EspB in host-cell actin reorganization [9], in particular, how the conformational properties of EspB contribute to the multifunctionality of this protein [10]. EspB binds to various host proteins including α-catenin [2], α1-antitrypsin [11] and myosin [12]. To create a pore on host-cell membranes, EspB and EspD bacterial proteins form a complex that binds to an EspA needle on the bacterial membrane [13]. Assembly of this complex results in a conduit that links bacterial and host-cell membranes.

In this minireview, we describe the multifunctional roles that EspB plays in pedestal formation and in mediating morphological changes in infected host cells. We compare the structural properties of EspB and other T3SS effectors in various pathogenic Gram-negative bacteria and describe how the ‘intrinsically less-ordered’ nature of these effectors contributes to pathogenesis.

EspB as an effector of actin filament reorganization

The EspB (or EarB) gene product was first identified as an important factor in EPEC attachment [14] and later characterized as a T3SS-secreted protein required for signal transduction [15]. During its function in attachment, EspB associates with EspD [13] at the tip of a hollow EspA filament formed on the bacterial cell surface [13], resulting in the formation of a pore on a host cell (Fig. 1) that serves as a conduit between bacterial and host-cell membranes. Pore formation allows the secretion of T3SS virulence factors into host cell.

Figure 1.

 Schematic representation of roles of EspB. EspB secreted into host-cell cytosol binds to the C-terminal region of the α-catenin, destabilizing E-cadherin/β-catenin/α-catenin complexes at adherence junctions that mediate cell–cell contacts and cytoskeletal morphology. Binding of EspB to the C-terminal region of α-catenin promotes the dissociation of N-terminal interactions of α-catenin with β-catenin. Thu, during this process, EspB does not merely compete with β-catenin for α-catenin binding, but in fact induces a conformational change in the N-terminal region of α-catenin by binding to the C-terminal region. EspB-bound α-catenin shows enhanced affinity with actin filaments and also promotes bundling of actin filaments that accumulate at pedestals formed underneath the attachment site of bacterial cell.

In addition, EspB functions as a signal transducer or effector. It is secreted via a T3SS needle into the host-cell cytosol, where it participates in the rearrangement of actin molecules that promote morphological changes in host cells and pedestal formation [16,17]. Although EspB has the potential to form pore structures together with EspD, EspA is required to translocate this protein into the host cytosol [18]. As an effector, EspB binds to host proteins, including α-catenin [2] and myosin [12], that regulate cytoskeletal morphology by controlling actin network formation. After binding, EspB redirects the activity of these regulator proteins to generate actin-based cytoskeletal pedestals that are the basis for EHEC and EPEC attachment sites (Fig. 1). EspB is therefore required both as a pore-forming protein and a signal effector during EHEC and EPEC pathogenesis.

When bound together, EHEC EspB promotes the action of α-catenin in bundling actin filaments, in opposition to the action of actin-related protein 2/3 in promoting actin filament branching [9]. This activity is consistent with EspB/α-catenin colocalization at pedestals, as well as the role of EspB in reorganizing actin filaments and host proteins associated with cell morphology. In binding to α-catenin, EspB also promotes the dissociation of α-catenin from the E-cadherin/β-catenin/α-catenin complex at cell–cell adherence junctions [9], which probably leads to the destabilization of cell contacts [19] and facilitates bacterial penetration through intestinal epithelium.

Importantly, EspB binds to the C-terminal vinculin homology domain of α-catenin, whereas formation of α-catenin/β-catenin/E-cadherin complexes at adherence junctions requires the N-terminal vinculin homology domain of α-catenin [9]. Based on these interactions, it was hypothesized that conformational changes in α-catenin mediated by EspB, rather than EspB-blocking interactions with β-catenin, lead to the dissociation of α-catenin from adherence junction complexes (Fig. 1).

EspB also interacts with the actin-binding domain of several myosin proteins, including myosin-1a, -1c, -2, -5, -6 and -10 [12]. By inhibiting interactions between myosins and actins, EspB can prevent both the initiation of phagocytosis and the production of microvillus effacing [12]. It has been reported that deletion of the central domain (amino acids 159–218) of EspB creates a mutant protein that cannot bind to myosin-1c; nevertheless, an EPEC mutant strain carrying this EspB deletion translocated virulence factors to host cells and induced actin reorganization [12]. These results are consistent with experiments that map the α-catenin-binding domain of the EHEC EspB to the N-terminus (amino acids 1–108) [2].

Structural properties of EspB and other T3SS effectors

3D structures of numerous proteins associated with EHEC or EPEC T3SS have been solved [20–28] using X-ray crystallography, solution NMR [29–31] and cryo-EM [32]. However, because of the tendency of this protein to assume a less-ordered conformation, the structural properties of EspB are currently unknown. EspB consists of a substantial amount of α-helical structure, but lacks rigid structures commonly found in globular proteins [10], and therefore is classified as a ‘natively partially folded protein.’

Recently, various functional proteins have been found that maintain almost completely disordered structures even under native conditions. These proteins are called ‘natively unfolded’ or ‘intrinsically disordered’ proteins [33,34]. EspB is basically unfolded but maintains some secondary structures. The structure is therefore more similar to the partially folded or ‘molten globule’ states of globular proteins that accumulate during folding kinetics [35,36]. As shown in Fig. 2A, a far-UV CD spectrum shows that EspB contains α-helical structures but with less-ordered tertiary folds according to the less-dispersed signals in a 15N–1H HSQC spectrum (Fig. 2B) [10]. For this reason, EspB protein is classified as ‘natively partially folded’, rather than ‘natively unfolded’ or ‘intrinsically disordered’ [10]. A similar structural property has been observed with PopD which is a homolog of EspB expressed by Pseudomonas aeruginosa [37]. It should be noted here that, according to the original definition by Ohgushi & Wada [35], the ‘molten globule’ is a partially folded intermediate state with a significant amount of secondary structure, similar to the tightly packed native state, but lacks tertiary contacts. The molten globule state has been considered to be a relatively stable intermediate state that is accumulated during kinetic or equilibrium refolding or unfolding of a globular protein, in contrast to the well-ordered native structures with rigid secondary and tertiary structures under near native conditions or fully unfolded structure without ordered conformations. Use of the term ‘molten globule’ therefore sounds as if it is an intermediate state accumulated during the folding reaction into the tightly packed ordered structures with well-ordered secondary and tertiary structures. However, EspB assumes a ‘partially folded’ structure under native conditions and does not form a tightly packed native structure by itself. To clarify that the structure of EspB is the native state rather than the folding intermediate, we do not use terms such as ‘natively molten globule’ or intrinsically molten globule’, particularly for EspB.

Figure 2.

 Structural properties of EspB. Far-UV CD and 1H-15N HSQC spectra of EspB obtained at 20 °C, pH 7.0. EspB assumes a significant amount of α-helical structure according to CD (A), but less-dispersed signals are observed in HSQC spectra (B), suggesting a lack of rigid conformation. These data indicate that EspB assumes a ‘natively partially folded’ conformation, similar to the ‘molten globule’ state. Spectra-based figures are reproduced from Hmada et al. [10] with permission from the publisher.

Various algorithms can predict disorder regions of proteins from their amino acid sequences and the Predictor of Naturally Disordered Regions (PONDR®; algorithm is one of them [38–40]. This algorithm suggests that EspB contains substantial amounts of disordered and some ordered regions (Fig. 3). However, it should be noted that the predicted ordered regions in this calculation do not neccessarily mean that the regions are folded into well-ordered rigid structures as usually observed for globular proteins and, in this case, they assume less-ordred partially folded structures [10]. Interestingly, the putative α-catenin- (1–108 in EHEC EspB) [2] and myosin-binding regions (159–218 in EPEC EspB) [12] of EspB overlap with regions predicted to assume ordered conformations (Fig. 3). These data suggest that the α-helical structures found experimentally in EspB coincide with ordered regions, and that an ability to assume an α-helical structure may be involved in the recognition of α-catenin or myosin host target proteins.

Figure 3.

 Disorder in various T3SS effectors. Predictions of EspB from EHEC (solid line) and EPEC (dotted line), IpaC from Shigella, YopD and YopE from Yersinia, SipC from Salmonella and EspFU and Tir from EHEC. The predictions derived from PONDR® [38–40]. Regions with a PONDR score > 0.5 are predicted to be disordered and those with a score < 0.5 are predicted to be ordered. The PONDR analysis for α1-antitrypsin (a typical natively folded protein with a serpentine fold) and for cyclin-dependent kinase inhibitor p27kip1 (a typical natively unfolded protein) [54] are shown for comparison. Predictions for effector proteins and p27kip1 shown larger regions of predicted disorder relative to natively folded α1-antitrypsin. It should be noted that the predicted ordered regions do not neccessarily assume rigid folded structures usually observed for globular proteins and can form partially folded structures similar to the molten globule state [10].

EspB binds to various proteins including EspA and EspD from bacteria and α1-antitrypsin and α-catenin from host cells. We also found that host proteins other than α-catenin that are involved in the regulation of cytoskeletal morphology can bind to EspB (M. Hamaguchi, I. Yanagihara, K. N. Suzuki & D. Hamada, unpublished data). This indicates that EspB is a typical multifunctional protein. The 3D structures of these EspB target proteins differ significantly (M. Hamaguchi, I. Yanagihara, K. N. Suzuki & D. Hamada, unpublished data). Therefore, it is highly quiestionable how this protein with only 330 amino acid residues manages to recognize these different targets. The structural flexibility of EspB caused by the formation of partially folded structures could be advantageous for its multifunctional properties because its association with various targets of different molecular dimensions and binding surfaces would be facilitated as different conformations can be assumed.

In T3SS proteins from bacteria other than EHEC and EPEC, less-ordered proteins such as IpaC from Shigella flexneri, SipC from Salmonella, PopD from Pseudomonas aeruginosa or YopD from Yersinia pestis, demonstrate functions homologous to EspB (Fig. 3). In complex with IpaB, IpaC forms a pore on host-cell membranes and is also the effector that triggers actin polymerization during the formation of filopodia and lamellipodia [41–43]. SipC is involved in nucleation and bundling of actin filaments via direct binding to actin [44], whereas PopD from Pseudomonas aeruginosa or YopD from Yersinia species also form a pore complex, in this case with PopB [45] or YopB [46], respectively. Similar to EspB [10], some of these other proteins have also been shown to assume disordered or partially folded conformations under native conditions [47,48].

T3SS effector proteins that are not homologous to EspB have also been shown to exhibit ‘natively unfolded’ structures. For example, Yersinia YopE is a cytotoxin that uses GTPase-activating protein activity to target the Rho pathway to induce disruption of actin microfilament structures [49]. The structured region of YopE, which has been resolved using crystallography, has been shown to correspond to a GTPase activator [50]. By contrast, other parts of this protein are disordered entirely in solution, but can assume an ordered structure upon binding to a chaperone [51].

Both EHEC and EPEC encode the translocated intimin receptor (Tir) protein, which localizes to plasma membranes and forms clusters of proteins when bound to the bacterial outer membrane protein, intimin [29,30]. Tir has also been shown to bind the bacterial EspFU/Wiskott–Aldrich syndrome protein complex through either the insulin receptor tyrosine kinase substrate or its homolog, the 53-kDa insulin receptor substrate protein that regulates cytoskeletal organization [4,5]. According to CD spectra, Tir is largely unstructured in solution [52]; the PONDR® algorithm also predicts that large regions of Tir and EspFU have a propensity to form disordered structures (Fig. 3) [52].

This collection of findings suggests that relative structural disorder may be a common feature of T3SS effectors. Less-structured proteins may be favoured for secretion through the narrow T3SS pore, as suggested for flagella T3SS [53], and may also better serve the multiple roles required during pathogenesis.


We have reviewed the role of EspB as an EHEC/EPEC effector and explained how the ‘natively partially folded structure’ of this protein contributes to its multifunctionality. Although a lower proportion of intrinsically disordered proteins is encoded in bacterial genomes relative to eukaryotes [54], structural disorder has also been observed in other T3SS effectors. Like pathogenic viruses [55], these bacterial effectors may have evolved to mimic host protein structural properties in order to regulate the target proteins of host cells. Structural disorder in T3SS effectors may be also an important factor for secretion through T3SS needles.

Various EHEC or EPEC effectors, including EspB, EspFU and EspL2, regulate host-cell actin networks. In the future, clarification of the interplay between these effectors and a detailed analysis of EspB in complex with host targets will provide important insight into these interactions. Via the EspA-mediated T3SS apparatus, EspB is guided to form pore structures in complex with EspD, resulting in a conduit between bacterial and host cell membranes. Structural models depicting this initial stage of infection by bacteria should be allow better understanding of pathogenetic mechanisms of EHEC and EPEC.


This work was supported by Grants-in-Aid for the Global COE program A08 from the MEXT, Japan, and Grant-in-Aid for Young Scientists (B) from Japan Society for the Promotion of Science (to DH).