A sophisticated multi-step secretion mechanism: how the type 3 secretion system is regulated


For correspondence. E-mail kolbe@mpiib-berlin.mpg.de; Tel. (+49) 30 28460 332; Fax (+49) 30 28460 301.


Many Gram-negative pathogens utilize type 3 secretion systems (T3SSs) for a successful infection. The T3SS is a large macromolecular complex which spans both bacterial membranes and delivers effector proteins into the host cell. The infection requires spatiotemporal control of diverse sets of secreted effectors and various mechanisms have evolved to regulate T3SS in response to external stimuli. This review will describe mechanisms that may control type 3 secretion, revealing a multi-step regulatory strategy. We then propose an updated model of T3SS that illustrates different stages of secretion and integrates the most recent structural and functional data.


Bacterial infections are a substantial portion of serious human diseases worldwide. For successful infection, bacterial pathogens have evolved several mechanisms, including various secretion systems, through which they manipulate host cells and escape the immune response (Cossart and Sansonetti, 2004).

Many Gram-negative bacteria employ type 3 secretion systems (T3SSs) to colonize host cells and escape immune response or establish symbiosis (Dai et al., 2008). The T3SS is a macromolecular complex for transporting bacterial effector proteins into the host cell (Galan and Wolf-Watz, 2006). In bacterial pathogens, upon secretion through the T3SS, the effector proteins are thought to be translocated inside the host cell by a pore (translocon) formed in the host membrane (Blocker et al., 1999). Once the effectors are inside the host cell, they modulate cellular functions to the benefit of the pathogen (Galan and Wolf-Watz, 2006).

The T3SSs from different bacteria diverged during evolution, maybe as a consequence of their adaptation to hosts or niches within the host. T3SSs can be classified into seven phylogenetic families, based on the genetic analysis of their components (Troisfontaines and Cornelis, 2005). Most of the T3SSs from animal pathogens belong to one of the three aforementioned families (Cornelis, 2010). The secretion apparatuses from Yersinia spp. (pathogenic species) and Pseudomonas aeruginosa belong to the first family, the archetypes of the second family are represented by Shigella flexneri and Salmonella typhimurium Pathogenicity Island 1 (SPI-1), whereas the third family includes the T3SSs from enteropathogenic (EPEC) and enterohaemorrhagic (EHEC) Escherichia coli and the S. typhimurium Pathogenicity Island 2 (SPI-2) (Cornelis, 2010). The T3SSs share a common evolutionary origin with the bacterial flagellum, as reviewed elsewhere (Troisfontaines and Cornelis, 2005).

This review will focus on the second T3SS family, with reference to some members of the other families to underline likenesses or differences. However, most of the components of the T3SSs share protein sequence, structural and functional similarities across the families and the mechanisms of type 3 secretion seem to be conserved as well.

Since effector delivery is required only at some stages of the infection process, the transport through the T3SS is spatiotemporally regulated (Deane et al., 2010; Marteyn et al., 2010; Yu et al., 2010). Therefore, gaining insights into the mechanisms that regulate T3SS assembly and effector secretion might be an important requisite to fight bacterial infections and develop specific therapies and drugs.

The T3SS is composed of over 20 different proteins (Galan and Wolf-Watz, 2006) which assemble in a syringe-shaped structure embedded in both bacterial membranes. Three-dimensional structural analyses of the T3SS from S. typhimurium Pathogenicity Island 1 (SPI-1) (Schraidt and Marlovits, 2011) and S. flexneri (Hodgkinson et al., 2009) were performed by electron microscopy (EM) of isolated particles. Based on these EM studies, the T3SS can be divided into (i) the extracellular needle with a distal tip, (ii) the ring-shaped basal body (Fig. 1) and (iii) an intracellular compartment called the cytoplasmic ring (C ring). In particular, the cryo-EM and negative-stain transmission electron microscopy (TEM) maps of the basal body of S. typhimurium SPI-1 and S. flexneri are available at a resolution of up to 5–10 Å and 21–25 Å respectively (Hodgkinson et al., 2009; Schraidt and Marlovits, 2011; Kosarewicz et al., 2012). Recently, two detailed models of the T3SS needle from S. typhimurium SPI-1 and S. flexneri have been determined by solid state NMR combined with Rosetta modelling (Loquet et al., 2012) and by remodelling of the crystal structure of the needle protein into cryo-EM map respectively (Fujii et al., 2012). These studies elucidate even finer details of the needle structure.

Figure 1.

Structural organization of the T3SS needle complex. Schematic representation of the main substructures of the T3SS needle complex.

The T3SS needle has a variable length (45–150 nm) depending on the bacterial species. Since the channel of the needle has a diameter of 1–3 nm (Fig. 1), it allows the passage of unfolded or partially folded proteins (Cornelis, 2010) (Fig. 1). At the proximal end, the needle is connected to the basal body by an inner rod, which is thought to be located in the periplasmic region and is anchored to a socket-like structure (Fig. 1) (Galan and Wolf-Watz, 2006). On the extracellular side, the needles of most T3SSs end with a tip, which is proposed to interact with the translocon. The translocon is composed of two hydrophobic proteins called translocators, which enable the transit of effector proteins into the host cell (Blocker et al., 2003). Overall, the T3SS is suggested to form a continuous channel connecting the bacterial cytoplasm to the host cell cytoplasm. Inside bacteria, the entrance of the channel is defined by a cup-like structure, located at the centre of the cytoplasmic face of the base (Galan and Wolf-Watz, 2006) and is surrounded by the proximal ring of the T3SS basal body (Fig. 1). Both the cup and the proximal ring can interact with C ring components, thus influencing protein transport through T3SS (Botteaux et al., 2010; Barison et al., 2012).

The C ring is composed of cytoplasmic protein complexes which recruit the needle components and the effectors for secretion (Lara-Tejero et al., 2011). The C ring interacts with an ATPase (Jouihri et al., 2003; Morita-Ishihara et al., 2006), suggested to belong to the AAA+ ATPase protein family (ATPases associated with diverse cellular activities) (Galan and Wolf-Watz, 2006) or the F1-ATPase family (Pozidis et al., 2003; Muller et al., 2006). Before secretion, most of the effectors are localized in the cytoplasm in complex with their corresponding chaperones, which also play an important role in targeting the effector proteins to the T3SS (Stebbins and Galan, 2001; Lilic et al., 2006; Lunelli et al., 2009; Lokareddy et al., 2010). The ATPase is required for the dissociation of chaperone–effector complexes and subsequent unfolding of the effectors (Akeda and Galan, 2005). Moreover, it is needed for T3SS function, most likely by providing energy for the secretion process (Galan and Wolf-Watz, 2006). Proton motive force is also required for the secretion process, but the mechanism is not clearly understood (Wilharm et al., 2004; Minamino and Namba, 2008; Paul et al., 2008).

Assembly of the T3SS basal body and needle

The assembly of the basal body except the cup employs a sec-dependent pathway and is initiated by a group of membrane proteins called the export apparatus (Wagner et al., 2010a). Following the basal body assembly, the inner rod and the needle assemblies are carried out by the T3SS itself (Galan and Wolf-Watz, 2006).

The C ring components Spa33, MxiK and MxiN form complexes together with the ATPase Spa47 in S. flexneri (Jouihri et al., 2003; Morita-Ishihara et al., 2006; Johnson and Blocker, 2008). The homologues of C ring components, YscQ, YscK, YscL and the ATPase YscN in Yersinia pestis are suggested to form complexes as well (Jackson and Plano, 2000). Likewise, the homologues in S. typhimurium SPI-1, SpaO, OrgA and OrgB, respectively, assemble in a complex called the sorting platform (SP) (Fig. 2A). The SP has been demonstrated to recruit needle components, translocators and effectors in a sequential manner to define the order of protein secretion (Lara-Tejero et al., 2011). Since this protein complex is conserved across species, it will be referred to as the SP in the following sections. The SP is also required for needle assembly in S. flexneri since strains missing any of its components lack the T3SS needle (Jouihri et al., 2003; Morita-Ishihara et al., 2006).

Figure 2.

Model of the secretion process through the T3SS. Binding of Spa40-like protein to the SP and to the ATPase triggers needle assembly (A). Following needle completion, the Spa32-like protein binds to cleaved Spa40-like protein and terminates secretion of needle components (B). Secretion of Spa32-like protein followed by assembly of tip complex mediated by the SP interacting with cleaved Spa40-like protein (C). The tip complex senses the presence of host cell (D) which results in opening of the tip and in translocator secretion via a phosphorylation-dependent recruitment of the SP at the proximal ring (E). Formation of the translocon complex signals MxiC-like protein release and effector secretion. Order of effector secretion is regulated at transcriptional level by IpgC-like protein and MxiE-like protein (F).

Like strains missing C ring components, Spa40 deletion mutants cannot form needles in S. flexneri (Botteaux et al., 2010). Spa40 belongs to a family of four-transmembrane helix proteins with a large C-terminal cytoplasmic domain. Spa40 is conserved in several species and is evolutionary related to FlhB, the flagellar component responsible for switching the specificity of export substrates (Minamino and Macnab, 2000).

Similar to Spa40 (Deane et al., 2008a), Spa40-like proteins such as S. typhimurium SpaS (Zarivach et al., 2008), Yersinia pseudotuberculosis YscU (Lavander et al., 2002; Sorg et al., 2007) and EPEC E. coli EscU (Zarivach et al., 2008), display autoproteolytic activity on the NPTH sequence in their cytoplasmic domain. Instead of dissociating, the resulting proteolytic fragments however remain tightly bound to each other. The presence of a cytosolic domain in Spa40-like proteins suggests that it is localized in the T3SS cup (Fig. 1). Indeed, this hypothesis is supported by comparison of the transmission electron microscopy (TEM) maps obtained from isolated T3SSs of different S. flexneri mutants (Botteaux et al., 2010). Moreover, the cytoplasmic domain of Spa40 interacts with the cytoplasmic domain of the membrane protein MxiA, which is also believed to be localized in the cup. The function of Spa40-like protein autocleavage in T3SS regulation remains unclear. The phenotypes of several non-cleavable mutants have been studied and discordant data have been reported. In S. flexneri, a strain without the Spa40 NPTH motif lacked the needle portion (Botteaux et al., 2010), whereas a strain expressing a non-cleavable Spa40 with a point mutation in the NPTH sequence showed only reduction in translocator secretion (Shen et al., 2012). S. typhimurium strains expressing non-cleavable SpaS mutants revealed a reduction in needle component and translocator secretion (Zarivach et al., 2008). In Yersinia spp., a ΔNPTH strain and a point-mutated non-cleavable YscU strain showed no translocator secretion (Lavander et al., 2002; Sorg et al., 2007). Finally, EPEC point-mutated non-cleavable EscU-expressing strains showed decreased translocator secretion (Zarivach et al., 2008; Thomassin et al., 2011). However, regardless the autoproteolytic activity, Spa40-like proteins seem to play a role in needle assembly and both non-cleavable and cleavable Spa40-like proteins have been demonstrated to interact with SP components in S. flexneri and Yersinia spp. (Riordan and Schneewind, 2008; Botteaux et al., 2010). Thus, we can hypothesize that the interaction between the Spa40-like protein and the SP is important for the needle assembly. However, this has to be confirmed by experimental data. Since the Spa40-like protein autocleavage has been suggested to occur immediately after its translation and folding in the cytoplasm, it is likely that the cleavage is neither the trigger for needle component secretion nor the switch from secretion of needle components to translocator secretion (Zarivach et al., 2008).

The T3SS needle has a defined length which is regulated by the Spa32-like proteins (Fig. 2A and B) (Tamano et al., 2002; Edqvist et al., 2003; Journet et al., 2003). S. flexneri, S. typhimurium and Yersinia strains lacking Spa32-like proteins display longer T3SS needles (Kubori et al., 2000; Tamano et al., 2002; Edqvist et al., 2003). The mechanism of needle length regulation is under debate and various models have been proposed to explain it. Mutational analysis of the Spa32-like protein YscP (Journet et al., 2003) and studies on the needle-length regulation in Yersinia spp. (Wagner et al., 2010b) support the ‘molecular ruler’ model, which has also been used to explain the regulation of flagellar hook length in S. typhimurium (Hirano et al., 1994). According to this model, the partially folded Spa32-like protein is anchored to the base and tip of the needle complex, with its N- and C-terminus respectively (Fig. 2A). When the Spa32-like protein adopts an extended conformation, the needle stops subunit polymerization (Fig. 2B). Complementation of Δspa32 S. flexneri with mutant spa32 and spa32 homologues suggests a ‘molecular tape measure’ mechanism instead (Moriya et al., 2006; Botteaux et al., 2008). During this process, the Spa32-like protein is secreted periodically to sense the length of the needle, although the sensing mechanism is the same as proposed for the ‘molecular ruler’ model. Statistical studies of the needle length in Yersinia spp. mutants comparing the two models support the ‘molecular ruler’ model over ‘molecular tape measure’ model (Wagner et al., 2010b).

The ‘molecular ruler’ and ‘molecular tape measure’ models, however, leave important questions unanswered. Specifically, in the ‘molecular ruler’ model the T3SS channel would be too narrow to accommodate needle components and the ruler protein simultaneously, whereas the ‘molecular tape measure’ model lacks a clear understanding of how the needle completion is signalled to the cytoplasm. A third model proposed by Marlovits et al. is based on the observation that in S. typhimurium SPI-1 the inner rod and the needle are assembled simultaneously. According to this model, the relative concentrations of the cytoplasmic needle and inner rod components and the rate of assembly of the inner rod determine the length of the needle (Marlovits et al., 2006). However, the role of Spa32-like proteins in needle length regulation is not clearly explained by this model. Spa32 has been demonstrated to bind to Spa33 and cleaved Spa40, suggesting a role of these interactions in needle length regulation in S. flexneri (Morita-Ishihara et al., 2006; Botteaux et al., 2008; 2010).

Taken together, we propose a model of the needle assembly process. Spa40-like protein binds to the SP and the ATPase. In this configuration, the SP can load the rod and the needle components, triggering the needle assembly (Fig. 2A). To finish the needle assembly (Fig. 2B), Spa32-like protein, activated by needle completion, could mediate the termination of needle components secretion through interaction with Spa40-like protein.

Assembly of the tip complex and activation of secretion

After needle assembly, the tip complex is positioned at the distal end of the needle (Fig. 2C and D) (Espina et al., 2006; Sani et al., 2007; Veenendaal et al., 2007). Cleavage of Spa40-like protein is required for the efficient delivery of tip proteins in Yersinia spp. (Sorg et al., 2007) and S. flexneri (Shen et al., 2012).

In S. flexneri, the translocator IpaD is located at the tip of the needle (Olive et al., 2007; Veenendaal et al., 2007). Similarly, the translocators LcrV and SipD have been shown to form the tip complex in Yersinia spp. and S. typhimurium SPI-1, respectively (Mueller et al., 2005; Lara-Tejero and Galan, 2009), suggesting its conservation across different species. The presence of a tip in S. typhimurium SPI-2 and EPEC however, has not been shown yet. Protein binding and structural studies have confirmed the interaction between SipD and the needle subunit PrgI in S. typhimurium (Lunelli et al., 2011; Rathinavelan et al., 2011) (Fig. 2D). Similarly, it has been reported that IpaD interacts directly with the needle subunit MxiH in S. flexneri (Zhang et al., 2007). Five IpaD or LcrV molecules may form the tip in S. flexneri and Yersinia spp. respectively (Mueller et al., 2005; Epler et al., 2012). In S. flexneri, the translocators IpaD and IpaB colocalize at the tip of the needle in the absence of host cell (Olive et al., 2007; Veenendaal et al., 2007). However, in S. typhimurium the translocator SipB is displayed on the tip only upon contact with the host cell (Lara-Tejero and Galan, 2009). The binding of IpaB to the tip complex depends on the presence of IpaD in S. flexneri. It has been hypothesized that these translocators could assemble in a 4:1 complex (IpaD:IpaB) in S. flexneri (Veenendaal et al., 2007).

How translocator secretion becomes activated is not understood. Non-cleavable mutations in Spa40-like proteins in S. flexneri, Yersinia spp. and EPEC impair secretion, suggesting that these proteins play a role in translocator secretion as well (Sorg et al., 2007; Thomassin et al., 2011; Shen et al., 2012). Absence of the tip proteins leads to constitutive secretion by an unknown mechanism in S. flexneri and S. typhimurium (Menard et al., 1994; Kaniga et al., 1995). In S. flexneri, the IpaD–IpaB complex has been proposed to adopt ‘on’ or ‘off’ states, controlling the passage of proteins through the T3SS (Roehrich et al., 2010). Various stimuli, like host cell contact, changes in calcium concentration, addition of bile salts and Congo red, can trigger secretion (Frank, 2012). Similarly, the assembly and effector secretion through T3SS encoded by S. typhimurium SPI-2 is governed by extracellular pH (Yu et al., 2010). Interestingly, the tip complex, which is suggested to ‘sense’ host cell (Veenendaal et al., 2007), can bind bile salts (Stensrud et al., 2008; Lunelli et al., 2011), suggesting one possible physiological regulator of secretion. The bile salts were shown to both activate and inactivate type 3 secretion in S. flexneri (Pope et al., 1995) and S. typhimurium SPI-1 (Prouty and Gunn, 2000) respectively.

It is important to understand how the presence of the host cell is signalled to the cytoplasmic side of the needle complex leading to the recruitment of SP with bound translocators (Lara-Tejero et al., 2011). Mutations in the needle protein cause changes in conformation and assembly of the tip complex, therefore affecting secretion of translocators (Kenjale et al., 2005; Veenendaal et al., 2007). Additionally, IpaB mutagenesis was shown to perturb tip assembly and effector secretion (Roehrich et al., 2010), suggesting a close inter-dependence between needle protein and tip complex.

Taken together, we hypothesize that the presence of the host cell is sensed via a conformational change in the tip complex, followed by a conformational change in the needle subunits, which subsequently permits secretion (Fig. 2D and E). However, this model has to be confirmed by additional experimental data.

Role of SP-MxiG-like protein interaction in type 3 secretion

Presence of the host cell induces translocator and effector delivery to the T3SS by the SP. Spa33 has been demonstrated to interact with the cytoplasmic domains of MxiG (Morita-Ishihara et al., 2006; Barison et al., 2012) and MxiJ, two main components of the T3SS proximal ring in S. flexneri (Morita-Ishihara et al., 2006). MxiG, a transmembrane protein composed of a periplasmic and a cytoplasmic domain may play an important role in controlling the secretion process. Its cytoplasmic domain includes a forkhead-associated (FHA) phosphothreonine-binding domain, which binds phosphorylated peptides from Spa33 bearing a FHA recognition motif (Barison et al., 2012), but not phosphothreonine amino acid (McDowell et al., 2011). Indeed, it was recently demonstrated that mutations in MxiG that affect phosphoprotein binding reduce secretion of IpaB and impair invasion of HeLa cells (Barison et al., 2012). These findings suggest that the interaction of phosphorylated proteins, possibly SP components like Spa33, with the T3SS proximal ring is required for translocator secretion (Fig. 2E). However, the phosphorylation of SP components, such as Spa33, has to be confirmed by further studies. In addition, McDowell et al. reported that mutations in MxiG did not reduce effector secretion significantly (McDowell et al., 2011). Although the mutations introduced into MxiG were similar, this discrepancy could be due to methodological differences in secretion assays. Further studies are required to test this hypothesis. MxiG homologues in EHEC (Barison et al., 2012) and Chlamydophila pneumoniae (Johnson and Mahony, 2007) are predicted to possess a FHA domain, suggesting that this domain could be conserved across some T3SS families. Although YscD, the putative functional homologue of MxiG in Yersinia spp., contains a FHA domain, the residues involved in phosphothreonine-binding are not conserved (Gamez et al., 2012; Lountos et al., 2012). Interestingly, alanine mutations in this domain increase binding with cytosolic T3SS components and in turn impair secretion, suggesting that FHA domain plays a role in secretion regulation in Yersinia spp. (Gamez et al., 2012).

Based on the available data, we propose that binding of the FHA domain to phosphorylated SP components results in recruitment of the SP and the ATPase to the proximal ring of the T3SS, leading to secretion of the translocators (Fig. 2E).

MxiC-like protein-dependent switch from translocator to effector secretion

As mentioned above, the SP regulates the order of effector secretion. The ATPase Spa47 interacts with MxiC, a cytoplasmic T3SS protein in S. flexneri. The switch from translocators to early effectors is induced by the secretion and subsequent depletion of MxiC from the cytoplasm (Botteaux et al., 2009; Martinez-Argudo and Blocker, 2010). Cytoplasmic MxiC enhances translocator secretion but inhibits secretion of the effectors. Mutational studies on MxiH suggest that the signal for MxiC secretion is transmitted through the needle. However, the absence of the tip proteins leads to constitutive secretion, independently of MxiC, suggesting that the tip complex acts upstream of MxiC function (Martinez-Argudo and Blocker, 2010). Similarly, S. typhimurium and EPEC/EHEC E. coli strains lacking the MxiC-like protein InvE and SepL, respectively, show reduced secretion of translocators and increased secretion of effectors (Kubori and Galan, 2002; Wang et al., 2008). Unlike MxiC, InvE is not secreted, suggesting a different mechanism for its modulation of translocator and effector secretion (Kubori and Galan, 2002).

SsaL, the MxiC homologue in S. typhimurium SPI-2 T3SS is also required for timely secretion of translocators while suppressing the secretion of effectors (Yu et al., 2010). SsaL forms a complex with SsaM and SpiC which acts as a ‘gatekeeper’ allowing secretion of translocators after acidification of the Salmonella-containing vacuole. Formation of a functional translocon and exposure to neutral pH in cytosol then leads to the dissociation of the ‘gatekeeper’ complex allowing secretion of effectors (Yu et al., 2010). In Yersinia spp. the role of MxiC is filled by the YopN–TyeA complex (Iriarte et al., 1998; Cheng et al., 2001; Ferracci et al., 2005) which exhibits sequence and structural similarities with MxiC (Schubot et al., 2005; Deane et al., 2008b).

In summary, these data indicate that, after successful formation of needle complex, a stimulus triggers the secretion of MxiC-like protein. Release of MxiC-like protein then may cause loading of the sorting platform with effector proteins (Fig. 2F), which are delivered into the host cell in a step-wise manner. Such a stimulus has been proposed to be the formation of a functional translocon complex in the host cell membrane (Martinez-Argudo and Blocker, 2010).

The order of effector secretion is also regulated at the transcriptional level. The effectors can be classified in three families: effectors that are produced independently from T3SS function, effectors whose expression is increased by T3SS activation and effectors expressed only upon T3SS activation (Parsot, 2009). In particular, OspD1, belonging to the first family of effectors, is in complex with its chaperones Spa15 and MxiE. MxiE is the transcriptional activator of the third family of effectors. When present in the cytoplasm, OspD1 acts as an anti-activator of MxiE, resulting in transcriptional repression of the effectors belonging to this family. Upon OspD1 secretion, this repression is relieved inducing expression and secretion of the third family of effectors (Parsot et al., 2005) (Fig. 2F).

Conclusions and future directions

The T3SS is probably the most complex and carefully regulated bacterial secretion system known to date. The necessity for spatiotemporal control of secretion led bacteria to evolve different mechanisms for its regulation.

Although the structure and the function of T3SS has been extensively studied, several important questions remain unanswered, in particular about the regulation of the secretion process. It is unclear how external stimuli transduce a signal into the bacterial cytoplasm, triggering translocator and effector secretion. In particular, the phosphorylation-dependent mechanism, which could regulate translocator secretion does not seem to be conserved in all the T3SS families, as in Yersinia spp. In addition, such a mechanism has to be confirmed in other species, like in S. typhimurium and S. flexneri. From available data, it seems that the C ring components form different hetero-oligomeric complexes rather than one complex. Additional studies are required to understand how the composition and dynamics of such complexes control T3SS depending on the stage of infection. In addition to the needle length regulation process, the assembly of the needle tip complex also needs further investigation.

The T3SS forms an important interface between bacteria and host cells. Given the crucial role of T3SS in life-threatening bacterial infections, focused efforts have to be made in order to gain thorough understanding of this system and prevent such infections.


We would like to thank Anna Brotcke for careful reading of the manuscript. The European Research Council has provided financial support to M. Kolbe under the European Community's Seventh Framework Programme (FP7/2007–2013).