Conserved features of type III secretion

Authors

  • A. P. Tampakaki,

    1. Institute of Molecular Biology and Biotechnology (IMBB), PO Box 1527, GR-71110 Heraklion, Crete, Greece.
    2. Technological and Educational Institute of Crete (TEI), GR-7110 Heraklion, Crete, Greece.
    3. University of Crete, Department of Biology, PO Box 2208, GR-71409 Heraklion, Crete, Greece.
    Search for more papers by this author
    • A.P.T. and V.E.F. have equally contributed to this work.

  • V. E. Fadouloglou,

    1. Institute of Molecular Biology and Biotechnology (IMBB), PO Box 1527, GR-71110 Heraklion, Crete, Greece.
    2. University of Crete, Department of Biology, PO Box 2208, GR-71409 Heraklion, Crete, Greece.
    Search for more papers by this author
    • A.P.T. and V.E.F. have equally contributed to this work.

  • A. D. Gazi,

    1. Institute of Molecular Biology and Biotechnology (IMBB), PO Box 1527, GR-71110 Heraklion, Crete, Greece.
    2. University of Crete, Department of Biology, PO Box 2208, GR-71409 Heraklion, Crete, Greece.
    Search for more papers by this author
  • N. J. Panopoulos,

    1. Institute of Molecular Biology and Biotechnology (IMBB), PO Box 1527, GR-71110 Heraklion, Crete, Greece.
    2. University of Crete, Department of Biology, PO Box 2208, GR-71409 Heraklion, Crete, Greece.
    Search for more papers by this author
  • M. Kokkinidis

    Corresponding author
    1. Institute of Molecular Biology and Biotechnology (IMBB), PO Box 1527, GR-71110 Heraklion, Crete, Greece.
    2. University of Crete, Department of Biology, PO Box 2208, GR-71409 Heraklion, Crete, Greece.
    Search for more papers by this author

E-mail kokkinid@imbb.forth.gr; Tel. (+30) 812810394455; Fax (+30) 812810394351.

Summary

Type III secretion systems (TTSSs) are essential mediators of the interaction of many Gram-negative bacteria with human, animal or plant hosts. Extensive sequence and functional similarities exist between components of TTSS from bacteria as diverse as animal and plant pathogens. Recent crystal structure determinations of TTSS proteins reveal extensive structural homologies and novel structural motifs and provide a basis on which protein interaction networks start to be drawn within the TTSSs, that are consistent with and help rationalize genetic and biochemical data. Such studies, along with electron microscopy, also established common architectural design and function among the TTSSs of plant and mammalian pathogens, as well as between the TTSS injectisome and the flagellum. Recent comparative genomic analysis, bioinformatic genome mining and genome-wide functional screening have revealed an unsuspected number of newly discovered effectors, especially in plant pathogens and uncovered a wider distribution of TTSS in pathogenic, symbiotic and commensal bacteria. Functional proteomics and analysis further reveals common themes in TTSS effector functions across phylogenetic host and pathogen boundaries. Based on advances in TTSS biology, new diagnostics, crop protection and drug development applications, as well as new cell biology research tools are beginning to emerge.

Introduction

Beauty, intrigue and cures in plague, wilt and blight

The elucidation of the mechanisms underlying bacterial pathogenesis is a major focus of microbiological research, both for practical applications (new diagnostics and antibiotics) and to gain insight into the complex intermolecular interactions underlying the interplay between bacterial and host cells. Genetic and biochemical analyses of bacteria and the analysis of microbial genomes have shown that pathogens are distinguished from their non-pathogenic relatives by the presence of pathogenicity genes, often organized in clusters, the so-called pathogenicity islands. As a result, despite the large variety of symptoms and diseases in humans, animals and plants induced by invasive pathogenic bacteria, even distantly related pathogens harbour closely related virulence genes. This point has become particularly apparent for a set of approximately 20–25 genes that together encode one of the most recently unravelled pathogenic mechanisms which is termed ‘type III secretion’. By this mechanism extracellularly located bacteria that are in a close contact with a eukaryotic cell deliver proteins into the host cell cytosol. While the type III secretion apparatus is conserved in pathogens across the plant/animal phyllogenetic divide, the secreted proteins differ considerably and the diseases range from bubonic plague and septicimia in humans to localized lesions, systemic wilting and blights in plants.

The genes coding for what are now recognized as structural type III secretion system (TTSS) components were first described as a contiguous cluster, designated ‘hrp’ (hypersensitive reaction and pathogenicity) in a plant pathogen, shortly after the initial cloning of the avr genes (now grouped into the TTSS effector class) from a related pathogen. A dominant theme in plant–microbe interactions was then, and still remains today, the elucidation of the molecular basis of Flor's ‘gene-for-gene’ hypothesis for race-cultivar specificity and the underlying concept of ‘basic compatibility’ (Elingboe, 1976), while no similar concepts regarding animal host–bacterium interactions had been advanced. Important insights into fundamental questions of bacterial pathobiology came with the recognition, in subsequent years, of the TTSS as a complex multiprotein channel dedicated to moving the effectors from the pathogen to the host.

Various aspects of TTSS have been extensively reviewed.1 This microreview will highlight the importance of the remarkable functional, sequence and structural homologies among the TTSS of bacterial pathogens in understanding TTSS function.

An overview of the system

TTSSs are found in an increasing number of Gram-negative bacteria pathogens of plants, vertebrate and invertebrate animals. S. typhimurium, Yersinia spp. and E. coli have two TTSSs, playing their role at different stages of infection. The presence of TTSSs in commensals (Photorhabdus luminscens, Sodalis glossinidius, Sitophilus zeamais), as well as in plant symbiotic rhizobia (Rh.) and in some ‘non-pathogenic’ prokaryotes (Pseudomonas fluorescens) may open unsuspected pages in the life history book of these and other microorganisms.1

The TTSS of pathogens is viewed as an inter-kingdom protein transfer device that: (i) is capable of injecting structurally diverse protein substrates across the two bacterial and the eukaryotic cell membrane directly into the eukaryotic cytosol; (ii) is triggered when the pathogen comes in close contact with host cells; and (iii) also functions in vitro under appropriate conditions that mimic key parameters of the host environment.

TTSSs share a common morphology

Low-resolution electron microscopy studies have revealed the morphology of the TTSS apparatus of animal pathogens (Salmonella, Shigella, EPEC), while for plant pathogens only the extracellular part of the apparatus has been visualized (Kubori et al., 1998; Tamano et al., 2000; Sekiya et al., 2001; He and Jin, 2003). These studies showed a basically common structure which is astonishing similar to the flagellar hook–basal body complex. It is composed of two distinct parts (Fig. 1, Table 1): (i) an elongated, hollow extracellular structure, termed ‘needle’ in animal pathogens and ‘Hrp pilus’ in plant pathogens; (ii) a cylindrical base, similar to the flagellar basal body which crosses the two bacterial membranes and ensures the stabilization of the whole structure upon the cell envelope (Fig. 1). An additional structure that protrudes from the inner membrane to the bacterial cytosol and is referred to as bulb, is occasionally observed in secreton preparations of Shigella (Blocker et al., 1999). The bulb is analogous to a flagellar cytoplasmic complex composed of the ATPase and other proteins and part of it could correspond to the flagellar C-ring.

Figure 1.

Schematic representation of the bacterial flagellum (A) and type III secretion systems (TTSSs) in Yersinia (B), Eschericia coli (C) and Pseudomonas syringae (D). The basal body of flagellum consists of proteins organizing the C-, MS-, P- and L-rings. Only conserved proteins in the TTSSs (and their flagellar homologues) have drawn and are marked by similar position and colouring. The P rod of flagellum consists of FlgB, FlgC, FlgF and FliE and has no homologous proteins in TTSSs. The question mark in TTSS indicates that a channel structure in the IM has not been identified yet. The major constituent of the needle is YscF in Yersinia, EspA in E. coli, HrpA in P. syringae. Pore-forming proteins are drawn with the same colour in the eukaryotic cell membrane (EM): LcrV, YopB, YopD in Yersinia, and EspB, D in E. coli. In plant pathogens, putative translocator proteins are HrpZ (P. syringae) and HrpF (X. campestris). Structurally characterized proteins are shown schematically. The homologous proteins FliN (PDB code: 1O6A) and HrcQB (PDB code: 1O9Y) have been crystallized in the dimeric and tetrameric forms respectively, and show extensive structural similarities. Other known 3-D structures are those of the C-terminus of FliG (PDB code: 1LKV), FliC (PDB code: 1UCU) and LcrV (PDB code: 1R6F). The representation of the needle complexes, the EspA and the flagellar filament is based on low-resolution structural studies (Cordes et al., 2003; Daniell et al., 2003; Mimori et al., 1995). OM, outer membrane; PL, peptidoglycan layer; IM, inner membrane of the bacterium; EM, eukaryotic membrane; CW, cell wall of the plant cells. The various components are not drawn to scale.

Table 1.  Approximate sizes of TTSS components.
Structural componentsLength (nm)Width (nm)Inner (outer) diameter (nm)
  1. Sal, Salmonella; Shi, Shigella; EPEC, enteropathogenic Escherichia coliI; Pst, P. syringae pv. tomato; Yer, Yersinia.

  2. Sal, Kubori et al. (1998); Shi1, Blocker et al. (1999); Shi2, Tamano et al. (2000); Shi3, Blocker et al. (2001); EPEC, Sekiya et al. (2001); Pst, Jin et al. (2003); Yer, Cornelis (2002).

Transmembrane part (basal body)10Shi1/32Shi2/31EPEC21Shi12–3Shi3
 Lower rings20Sal/11Shi240Sal/26Shi2/18EPEC 
 Upper rings18Sal/9Shi220Sal/15Shi2/17EPEC 
Extracellular part (needle/pilus)80Sal/45Shi2/60–80Yer/32–688EPEC/several µmPst13Sal/11Shi1/8Shi2/12EPEC2–3Shi3/2Yer(6–8Pst/6–7Yer)
Cytoplasmic part (bulb)27Shi144Shi1 

The TTSS extracellular appendages have different characteristics (Fig. 1). The pilus is a flexible, thin and long structure (Table 1) while the needle is stiff and shorter. The EPEC needle is extended by an additional large structure, the EspA filament (Daniell et al., 2001).

Both, the Hrp pilus and the needle are physically linked to the basal body and are assumed to play analogous functions. It is believed that via these structures the attachment to the host cell membrane is established and that they function as conduits for the translocation/secretion of effectors. Although the first role remains to be proven, direct evidence exists for the latter: in situ immunogold labelling experiments show that secretion occurs only at the sites of Hrp pili which secrete proteins only from their tips (Jin et al., 2001).

The extracellular appendages have a common architectural pattern

In analogy with the flagellum filament, which is built by many copies of the flagellin protein, the TTSS extracellular appendages are assembled through the stepwise polymerization of a major component, i.e. HrpA in P. syringae and E. amylovora, HrpY in R. solanacearum, MxiH in Shigella, PrgI in Salmonella, YscF in Yersinia, PscF in P. aeruginosa and EscF and EspA in EPEC (Aizawa, 2001; Koebnik, 2001; He and Jin, 2003). The monomers are small, hydrophilic proteins with low sequence similarities. They exhibit patterns of heptad repeats at their C- and/or N-termini which generally indicate associating α-helices. This is consistent with the crystal structure of flagellin and the electron cryomicroscopy structure of the flagellar filament (Samatey et al., 2001; Yonekura et al., 2003).

The mechanisms which control the length of these appendages are not well understood. For the Shigella needle and the flagellar hook a common mechanism is assumed through which the length depends on the production level of the individual subunits (Muramoto et al., 1999; Tamano et al., 2000). Journet et al. (2003) showed that the length of the Yersinia needle depends on the size of the YscP protein. The authors proposed that YscP controls the length of the needle acting as a ‘molecular ruler’ during the stepwise assembly of the injectisome and that the two ends of YscP act as anchors connected to the basal body and the growing tip of the needle. Through this arrangement YscP can signal the export apparatus to stop exporting YscF when the mature needle length is reached.

X-ray fibre diffraction and electron microscopy show that the flagellar hook and filament, the EspA filament and the Shigella needle are helical assemblies with their major subunits arranged as an extended cylindrical structure with a central channel diameter of about 20–30 Å (Mimori et al., 1995; Cordes et al., 2003; Daniell et al., 2003). Even though MxiH and EspA are significantly smaller than flagellin, their assembly has helical symmetry parameters that are very similar to the flagellum filamentous structure (MxiH: ≈ 5.6 subunits/turn, 24 Å the pitch of helix, 4 Å axial rise; EspA: ≈ 5.6 subunits/turn, 26 Å the pitch of helix, 4.6 Å axial rise; flagellin: ≈ 5.5 subunits/turn, 26 Å the pitch of helix, 4.7 Å axial rise).

Amino acid sequence conservations in TTSS core components

A subset of 8–10 proteins is highly conserved among all TTSS and the flagellum. These proteins are located in the inner bacterial membrane or in the cytoplasm and are loosely associated with the membrane. They form a conserved core of the basal body of TTS apparatuses. It has been proposed that this core could be involved in the recognition of a universal secretion signal. This hypothesis could explain the promiscuous character of the apparatus demonstrated in a number of studies (Anderson et al., 1999; Rossier et al., 1999; Subtil et al., 2001), i.e. the Yersinia YopE effector can be secreted by the TTSS of the plant pathogens Erwinia, Xanthomonas and Pseudomonas, the Yersinia TTSS can secrete the Pseudomonas AvrB and AvrPto proteins and the TTSS of Xanthomonas secretes the Ralstonia PopA and Pseudomonas AvrB proteins.

The conserved proteins have been grouped into families of homologues (Hueck, 1998) and many studies concern their location and function (Table 2). The YscR/S/T families are equivalent to the flagellar proteins FliP/Q/R, which have been proposed to form a universal structure for membrane channelling (Aizawa, 2001).

Table 2.  Equivalent (or putatively equivalent) proteins from the TTSSs of Yersinia, P. syringae and flagellum.
Animal pathogensYersiniaPlant pathogens P. syringaeFlagellumProtein characteristics, function and location
YscRHrcRFliPIntegral inner membrane proteins with periplasmic extensions, participate in the rod formation
YscSHrcSFliQ 
YscTHrcTFliR 
LcrD (YscV)HrcV (HrpI)FlhALcrD and FlhA: regulation
HrcV: secretion
YscUHrcUFlhBIntegral inner membrane proteins with periplasmic extensions, participate in the rod formation
YscJHrcJFliFFliF: forms the MS-ring
YscJ and HrcJ: lipoproteins, putative connectors of the secretion apparatus across the periplasm
YscQHcQBaFliN/FliYFliN/Y: peripheral cytoplasmic protein, part of the switch complex that is connected with the MS-ring
YscNHrcNFliISoluble components, ATPases
YscCHrcC  Outer membrane proteins, may be considered the functional equivalents of flagellum P- and O-rings
YscOHrpOFliJFliJ: protein of the export apparatus, cytoplasmic, proposed a chaperone-like activity
Proteins have limited sequence similarities
YopNHrpJ Hydrophilic proteins
YopN: located on the bacterial surface, a putative stop-valve (also characterized as effector/translocator).

Members of the LcrD and YscU families participate in the rod formation. The HrpI (E. amylovora) and the lipoproteins YscJ (Yersinia) and MxiJ (Shigella) are involved in the TTSS-dependent secretion and share sequence homologies with the central domain of FliF, which is responsible for the formation of the M-ring (Deng and Huang, 1999). YscU and YscP coordinately regulate the secretion of effector Yops and the YscF needle protein. YscU, similar to its flagellum homologue FlhB, is proteolitically cleaved at the strictly conserved site NPTH located in its cytoplasmic C-terminus. Even though the cleavage does not seem to be required for function, it may ensure a non-toxic protein conformation for the bacterial cell (Edqvist et al., 2003; Lavander et al., 2003).

Members of the YscN family are cytoplasmic proteins possessing an ATP binding motif resembling the β-subunit of F0F1-ATPase. ATP-hydrolysis activity has been demonstrated for some family members and appears to be coupled with the oligomerization state of the protein (Claret et al., 2003; Pozidis et al., 2003; Akeda and Galan, 2004).

YscO from Yersinia is a mobile component of the TTSS required for effector secretion. The putative flagellar counterpart (FliJ) is a cytoplasmic protein for which chaperone activity specific for rod/hook export substrates has been proposed (Payne and Straley, 1998; Minamino et al., 2000). Members of the YscQ family are peripheral cytoplasmic proteins and will be discussed in more detail below.

Structural motifs from conserved TTSS components suggest conserved protein–protein interaction networks

The recent elucidation of the crystal structure of the C-terminal domain of the HrcQB protein (HrcQB-C; PDB code: 1O9Y; Fadouloglou et al., 2004) provided a first insight into the structural characteristics of the highly conserved components of TTSS structural proteins. This work was soon followed by the determination of the structure of the C-terminal domain of the flagellar motor switch protein FliN from Thermotoga maritima (PDB code: 1O6A; Joint Center for Structural Genomics, unpubl.). Both proteins are members of the YscQ family. In P. syringae pathovars there are two proteins (HrcQA and HrcQB) in the place of the complete HrcQ. HrcQA is homologous to the N-terminal and HrcQB to the C-terminal part of the HrcQ. In flagellum, FliN/Y is the major component of the switch complex, which is connected with the MS ring. FliN/Y participates in the formation of a ring-shaped assembly known as the C-ring. Sequence homologies among the YscQ family members are restricted to the C-terminal segment, which comprises approximately 80 residues. Both structures determined correspond to the conserved segment. Attempts to crystallize the complete HrcQB protein have been to date unsuccessful.

The C-terminal domain of the HrcQB protein forms an elongated, gently curved homotetramer (Fig. 2A). The four monomers assemble into two tightly bound homodimers that are packed together to form a dimer of dimers. The secondary structure of each monomer consists of five strands and one short helix. The C-terminal domain of FliN crystallizes as a dimer, which shares extensive structural similarities with the HrcQB-C dimer (Fig. 1). Multiple sequence alignments among the YscQ family members reveal well-conserved residues and regions of high similarity (Fig. 2B). The conservation pattern extensively coincides with the secondary structure elements found in HrcQB-C and the C-terminal domain of FliN implying fold similarities among the aligned sequences of TTSS proteins. Interestingly, the distribution of conserved residues on the molecular surface of HrcQB-C is markedly asymmetric with the majority of them being concentrated on the concave side (Figs 2C and 2D). The concave side of the molecule could thus interact with other conserved elements of TTSS. The dimer–dimer interface in the HrcQB-C structure comprises segments of high homology and is highly suggestive of conserved interactions via a recurring structural motif for protein–protein interfaces in the supramolecular structure of the TTSS apparatus.

Figure 2.

A. Schematic representation of the HrcQB-C structure. Each chain is individually coloured and labelled.
B. Multiple sequence alignment of HrcQB-C with the C-termini of homologues from animal and plant pathogens and the flagellum identified by a Blast search with an identity of 26% or higher. The numbering scheme corresponds to the sequence of the full length HrcQB while the secondary structure elements shown on top are those of the HrcQB-C structure. Strictly conserved residues are highlighted, similar residues are boxed.
C and D. Two different views of the molecular surface of HrcQB-C coloured according to the degree of amino acid sequence conservation. Multiple sequence alignment with identity 47% or better was used. The colouring code is from blue for non-conserved to orange for highly conserved residues.

Potential interactions for HrcQB (or HrcQB-C) and other TTSS components have been investigated via genetic and biochemical experiments (Fadouloglou et al., 2004). These have demonstrated that in the P. syringae secretion apparatus HrcQB (or HrcQB-C) is cytoplasmic or loosely associated with a membrane protein and strongly binds to HrcQA, a TTSS protein for which a transmembrane domain is predicted on the basis of sequence data. These results are fully consistent with earlier findings from the flagellum counterparts of the HrcQB/HrcQA pair, the FliN/FliM proteins, the association of which results in the formation of a multiprotein, ring-shaped assembly known as the C-ring, in which most likely the C-terminal parts of both proteins participate. Based on the analogies between the HrcQB/HrcQA and the FliN/FliM proteins, it has been proposed that the C-terminal domains of the former may associate in the P. syringae Hrp secretion apparatus to form a cytoplasmic assembly analogous to the flagellar C-ring. The structural similarity between the C-terminal domains of FliN and HrcQB supports this hypothesis. Sequence similarities between HrcQA and HrcQB which exhibit a characteristic pattern (Fadouloglou et al., 2004), suggested that both proteins could associate in the formation of a C-ring-like assembly, via a structural motif similar to the one occurring in the dimer–dimer interface of the HrcQB-C structure. There is a broad analogy in the pattern of the interactions described above between conserved (HrcQB, FliN) members of the YscQ family and the less well-conserved (HrcQA, FliM) components of the P. syringae TTSS or flagellum systems, with observations made in other TTSSs. For example, the conserved YscQ family proteins Spa33 (C-terminal domain) from Shigella flexneri and YscQ from Yersinia have been found to interact with the non-conserved components MxiK, MxiN and YscK, YscL respectively (Jouihri et al., 2003). Despite low sequence similarities it has been proposed that MxiN is analogous to YscL and MxiK analogous to YscK. Furthermore, similar interactions, which are conserved between virulence-associated and flagellar TTSSs (Jackson and Plano, 2000), have been observed for YscN family members. Thus protein interaction networks start to emerge, which are consistent with the weak sequence similarities observed between components of various systems.

Conserved structural features of type III effectors

Comparisons among TTSS effectors from plant and animal pathogens have identified common structural features (Collmer et al., 2002; Fouts et al., 2002; Zwiesler-Vollick et al., 2002; Greenberg and Vinatzer, 2003). TTSS effectors appear to have a modular structure: the amino-terminal region (residues 1–50) carries targeting information (secretion and translocation signal); the region spanning residues 50–150 is the chaperone binding domain in chaperone-dependent effectors, targeting the proteins to the appropriate TTSS; the remainder of the protein directs effector activity. Although TTSS effectors have no classical secretion signals, a minimum of 15 N-terminal residues are necessary for their secretion. The lack of sequence similarity among effectors in this region has prompted investigations to determine the nature of the secretion signal. Systematic mutagenesis of N-terminal codons for several effectors from plant and animal pathogens led to the proposal that signals controlling secretion may reside in the mRNA, rather than in the protein sequence (Anderson and Schneewind, 1999; Mudgett et al., 2000). However, this issue remains controversial. Recent reports identify other shared features of the N-terminal regions in many effectors from plant and animal pathogens. These include five N-terminal hydrophilic residues, absence of acidic residues in the 12 N-terminal positions and recurrence of certain characteristics, such as amphipathicity and richness in Ser and Gln, in the first 50 residues (Lloyd et al., 2001; 2002; Guttman et al., 2002; Petnicki-Ocwieja et al., 2002; Russman et al., 2002).

Whatever the secretion signal is it appears to function universally, i.e. TTSS secretion substrates of one type III machine can be recognized by the TTSS apparatus of another pathogen (Rosqvist et al., 1995; Anderson et al., 1999; Subtil et al., 2001). Taken together, these findings suggest that the effectors might be targeted to the secretion apparatus by a similar mechanism. It is significant, that in the absence of type III chaperons, the effectors can be also secreted through the flagellar system (Lee and Galan, 2004).

Sequence alignments have shown that some effectors from animal pathogens exchibit α-helical coiled-coil propensity, suggesting possible interactions with other proteins via association of α-helices (Delahay and Frankel, 2002), a hypothesis supported by the Yersinia complexes SycN–YscB, TyeA–YopN and SycD–YopD (Day and Plano, 1998; Iriarte et al., 1998; Francis et al., 2000). However, none of the P. syringae effectors recently reported to be secreted by the TTSS is predicted to have a coiled-coil structure.

The crystal structures of animal pathogen effectors such as SptP from Salmonella (Stebbins and Galan, 2000), YopH (Stuckey et al., 1994; Evdokimov et al., 2001a), YopE (Evdokimov et al., 2002a), YopM (Evdokimov et al., 2001b) and LcrV (Derewenda et al., 2004) from Yersinia, ExoS from P. aeruginosa (Wurtele et al. 2001), TiR from EPEC (Luo et al., 2000), AvrPphB (Zhu et al., 2004) and AvrB (Lee et al., 2004) from P. syringae have been determined. SptP is a modular protein composed of two functional domains. Its carboxyterminal part displays sequence similarity to YopH as well as to other eukaryotic tyrosine phosphatases. The crystal structure of SptP revealed that the C-terminal part adopts a canonical tyrosine phosphatase fold. The N-terminal domain possesses GTPase activating protein (GAP) activity and is similar to two other TTSS effectors: the Yersinia YopE and the P. aeruginosa ExoS. The GAP domain of SptP does not resemble the structure of eukaryotic GAPs. Unlike the catalytically equivalent arginine fingers of the eukaryotic GAPs, which are invariably contained within flexible loops, the critical arginine in SptP is part of an α-helix. The structure of SptP is strikingly similar to the GAP domains from P. aeruginosa ExoS and Y. pestis YopE, despite the fact that the three amino acid sequences are not highly conserved.

Common functional features of TTSS effectors

Despite the distinct pathologies and host specificities of the pathogens, many effectors either perform similar functions in the host cells or target similar host cell proteins. Many effectors of plant and animal pathogens have common functions/targets, suggesting that the pathogens have evolved similar functions to hijack the host defence mechanisms. A number of effectors influence host cell signalling by specifically altering host proteins, phosphorylation patterns, suppressing host defences and facilitating effector translocation.

One group of TTSS effectors (AvrRxv/YopJ family) present in animal pathogens (Y. pseudotuberculosis, S. enterica), plant pathogens (X. campestris, P. syringae, E. amylovora, R. solanacearum) as well as Rhizobium spp. (Buttner and Bonas, 2003) interferes with the host signalling pathways and possesses cysteine protease activity. Interestingly, all family members have homologies to the adenovirus protease, based on secondary structure predictions and conservation of catalytic residues. However, protease activity has not been shown for all the effectors in this group and it is not yet known whether all family members target the same host cellular pathways.

Several phytopathogenic TTSS effectors appear to suppress host defence responses: the AvrRpm1, AvrB and AvrRpt2 from P. syringae bind to the Arabidopsis RIN4, a potential regulator of basal plant defences (Mackey et al., 2002; 2003). The P. syringae AvrPphF Orf2, VirPphA, AvrPphC and AvrRpt2 interfere with the defence-associated hypersensitive response when it is elicited by another Avr protein (Jackson et al., 1999; Tsiamis et al., 2000). Remarkably, six P. syringae effectors function in a trans-kingdom manner, to inhibit programmed cell death (PCD) elicited by other effectors and the ability of the pro-apoptotic protein Bax to induce PCD in plants and yeast (Abramovitch et al., 2003; Jamir et al., 2004).

Several plant and animal TTSS effectors target small GTP-binding proteins (GTPases) in host cells. These effectors mimic eukaryotic enzymes and are able to regulate signalling pathways that alter cellular processes. The Yersinia YopE, YpkA and YopT directly exert their influence on Rho GTPases, which are the key regulators of eukaryotic actin cytoskeleton (Aepfelbacher and Heesemann, 2001). YopE homologues exist in S. typhimurium (SptP) and P. aeruginosa (ExoS). YpkA shares sequence similarity to eukaryotic serine/threonine kinases and binds to actin and to Rho GTPases. YopT also cleaves Rho GTPases leading to their release from the plasma membrane. It is worth noting that catalytic residues associated with YopT protease activity are conserved in a subset of P. syringae effectors (Collmer et al., 2002; Buttner and Bonas, 2003), though the host targets of these effectors may be different. AvrPphB functions as a cysteine protease specifically cleaving the Arabidopsis PBS1 kinase (Shao et al., 2002).

A subset of TTSS effectors acts as eukaryotic protein tyrosine phosphatases (PTPases). This set includes YopH and SptP from animal pathogens and HopPtoD2 from plant pathogens (Espinosa et al., 2003). The target of YopH is the p130cas and focal adhesion kinase (FAK). Although the host cell targets of HopPtoD2 are not yet known, it has been shown that HopPtoD2 can suppress programmed cell death induced by an activated form of MEKK, a MAPK kinase, raising the intriguing possibility that MAPK proteins could be targets of HopPtoD2.

Global analysis of the recently sequenced P. syringae DC3000 genome has revealed an inventory of candidate effectors with homology to known Avr/Hop proteins and animal pathogen effectors. Among them, HopPtoS1, HopPtoS2 and HopPtoS3 share sequence similarity with ADP-ribosyltransferases (implicated in bacterial pathogenesis in animals by modifying host signal transduction pathways). The similarity to ADP-ribosyltransferases is noteworthy because no such proteins had previously been identified in plant pathogens. It remains to be clarified whether these HopPto proteins have similar targets and activities as P. aeruginosa ExoS (Petnicki-Ocwieja et al., 2002).

Another group of plant and animal TTSS effectors targets the host transcription machinery. The members of the AvrBs3 family, in phytopathogenic bacteria of X. campestris, have nuclear localization signals (NLS), require nuclear localization for their HR-eliciting activity and have typical features of eukaryotic transcription factors (Szurek et al., 2002). The Yersinia effector YopM also contains a functional NLS and is also localized in nuclei of infected host cells suggesting that YopM may bind to the host's transcription machinery (Benabdillah et al., 2004).

A final class of TTSS virulence factors is involved in the formation of extracellular structures on bacterial surface and/or host cell facilitating the translocation of other effector proteins through the host cell plasma membrane (translocators). These proteins possess pore-forming activity suggesting that they may form a translocation pore. This hypothesis is supported by the findings that Yersinia YopB and YopD can be inserted in liposomes, LcrV is capable of creating pores in synthetic lipid bilayers and all interact with each other (Holmstrom et al., 2001). A similar translocation apparatus is also present in Shigella spp. (IpaB, IpaC and IpaD), Salmonella SPI1 (SipB, SipC, SipD) and in P. aeruginosa (PcrV, PopB and PopD) (Buttner and Bonas, 2002). Interestingly, HrpZ from the plant pathogen P. syringae and HrpF from X. campestris bind to the plant plasma membrane and form ion-conducting pores in artificial lipid bilayers (Lee et al., 2001; Buttner et al., 2002). These findings indicate that the mechanism of effector translocation is functionally conserved at the molecular level across the plant–animal pathogen divide.

Conserved features among TTSS chaperones

Efficient secretion and translocation of TTS substrates (i.e. effectors and translocators) frequently requires specific chaperones. Some effectors are escorted by specific chaperones, encoded by a gene located close to the gene for the corresponding effector, while others use broad specificity chaperones, coded by genes within the operons coding for conserved TTSS components (Parsot et al., 2003). For many TTSS effectors chaperones have not been identified. The previously mentioned chaperone binding site (residues 50–150 of the effector) enables the formation of tight complexes between chaperones and secreted proteins (Lee and Galan, 2004).

TTSS chaperones to date characterized (Parsot et al., 2003; Shan et al., 2004) are small (≈ 15 kDa), acidic (pI < 5) proteins with no obvious sequence similarities among them. They may be grouped into categories on the basis of the protein (e.g. effector, translocator) whose secretion/translocation they serve.

Class I chaperones have approximately 130 residues and associate with one (class IA) or several (class IB) effectors. Several structures of the subclass IA chaperones from animal pathogenic bacteria are available (Birtalan and Ghosh, 2001; Luo et al., 2001; Stebbins and Galan, 2001; Birtalan et al., 2002; Evdokimov et al., 2002b). Despite the lack of sequence similarity, all known class IA chaperones exhibit very similar structures (Parsot et al., 2003). In addition, the crystallographically determined chaperone–effector complexes of SicP/SptP from Salmonella (Stebbins and Galan, 2001) and SycE/YopE from Yersinia (Birtalan et al., 2002) have very similar structures. This stereochemical conservation has been interpreted in terms of a three-dimensional secretion signal manifested by the structure of the complex, which is recognized broadly by the TTSS (Birtalan et al., 2002). However, the S. flexneri Spa15 (Class IB) chaperone has the conserved fold of Class IA chaperones but forms a different dimer (Van Eerde et al., 2004). This dictates a different 3-D organization of the chaperone–effector complex in the two situations.

Class II TTSS chaperones bind translocators (Schoehn et al., 2003). They display a pattern of tetratricopeptide-like repeats (TPRs) and are predicted to fold into an all α-helical array. TPRs are imperfect repeats, typically 34-residues long, found in eukaryotic mammalian proteins, such as the mittochondrial Tom70 import receptor, where the cytosolic chaperones Hsp90 and Hsp70 dock (Young et al., 2003). TPRs are also present in regulators of type III secretion (TTS) such as HilA of S. enterica and HrpB of R. solanacearum (Pallen et al. 2003).

Chaperones of the flagellar system constitute a distinct group (class III). They bind to the substrate C-terminal region, which is proposed to mediate interactions between subunits in the flagellum (Parsot et al., 2003). The crystal structure of the Aquifex aeolicus FliS chaperone is an antiparallel four-α-helix bundle, which is unrelated to the structures of other TTS chaperones (Evdokimov et al., 2003).

Prospects

In the last 20 years, research on TTSS has made impressive progress in unravelling well-kept molecular secrets of bacterial pathogenesis. Fuelled by the rapid progress in microbial genomics, the field is blooming in the broader context of infectious diseases, helping to shape a new field dubbed ‘cellular microbiology’.

This review focused on conserved features of TTSS, to highlight basic commonalities in design and function at the ‘threshold of disease’ but, undoubtedly, has not exhausted the subject. In the wake of the flood of new information from structural genomics, proteomics and ‘signalomics’ investigations, many new and important questions emerge while old ones remain unanswered. To name some here should not diminish the importance of others in the reader's mind. Establishing order in TTSS secretion for core, pilus, translocon and effectors requires much additional work. Is the bewildering array of plant pathogen effectors unique to this group and if so why? On a broader level: Is the trans-kingdom protein transfer ability of TTSS restricted to higher eukaryote targets or not (like the Agrobacterial T4SS)? What implications might the presence of TTSS in non-pathogens have? Can the discovery of TTSS in plant symbiotic N-fixing bacteria reorient the research activity on nodulation efficiency and symbiotic nitrogen fixation in new directions, perhaps through analogous strategies to plant disease resistance breeding? Is it possible to develop antibacterial agents that target virulence mechanisms at the level of TTSS or the animal and plant apoptotic cascades?

In summary, data generated by the continuing analysis of TTSS may provide answers to important questions on bacterial pathogenicity and yield important practical applications.

Acknowledgements

This work was partially supported by GSRT grants in the framework of the PENED2001 and the BIOPRO2001 programmes. We thank Dr N.M. Glykos for expert assistance.

Footnotes

Ancillary