The Type VI Secretion System in Escherichia coli and Related Species

Abstract

The type VI secretion system (T6SS) is a multiprotein complex widespread in Proteobacteria and dedicated to the delivery of toxins into both prokaryotic and eukaryotic cells. It thus participates in interbacterial competition as well as pathogenesis. The T6SS is a contractile weapon, related to the injection apparatus of contractile tailed bacteriophages. Basically, it assembles an inner tube wrapped by a sheath-like structure and anchored to the cell envelope via a membrane complex. The energy released by the contraction of the sheath propels the inner tube through the membrane channel and toward the target cell. Although the assembly and the mechanism of action are conserved across species, the repertoire of secreted toxins and the diversity of the regulatory mechanisms and of target cells make the T6SS a highly versatile secretion system. The T6SS is particularly represented in Escherichia coli pathotypes and Salmonella serotypes. In this review we summarize the current knowledge regarding the prevalence, the assembly, the regulation, and the roles of the T6SS in E. coli, Salmonella, and related species.

INTRODUCTION

The adaptation of bacterial species in their ecological niche relies not only on specific regulatory circuits to adapt the metabolism and the growth to the extracellular conditions, but also on the release of molecules – siderophores, exopolysaccharides, and/or protein toxins – in the milieu. To facilitate their transport through the physical barriers that the membranes represent, protein toxins are specifically selected and secreted by dedicated machineries named “secretion systems.” Depending on the nature of the machine itself and on the mechanism of transport of the toxins, these secretory pathways are numbered I to IX. Most of these pathways, including type I (T1SS), type II (T2SS), type III (T3SS), type IV (T4SS or conjugation), type V (autotransporters, intimin/invasin, and two-partner pathways including contact-dependent growth inhibition systems), and curli/fimbriae/chaperone-usher pathways, are assembled and active in Escherichia coli and related species such as Salmonella, Shigella, Enterobacter, and Citrobacter, and therefore details regarding their architecture, assembly, mechanism of transport, as well as the effectors they deliver are described in the corresponding chapters in EcoSalPlus. We will describe here one of the most recently identified secretion pathways, the type VI secretion system (T6SS).

The T6SS is a multiprotein machine, widespread in Gram-negative Proteobacteria, with an overrepresentation in Gammaproteobacteria (1, 2, 3, 4). However, T6SS-like machines have been identified and characterized in other phyla such as in Bacteroidetes (5). The assembly of this secretion system requires 13 different subunits, which are usually encoded within a single genetic locus on the chromosome (2, 6) (Fig. 1A). Basically, the T6SS can be viewed as a syringe-like structure anchored to the cell membrane by a transenvelope complex (3, 4, 7, 8, 9) (Fig. 1B). The T6SS syringe is evolutionarily, structurally, and functionally related to the puncturing tails of contractile tailed bacteriophages (Fig. 1C), although it is not known whether the genes encoding this structure emerged from cooption of bacteriophage genes (9, 10). A large diversity of toxin effectors has been identified in recent years, from effectors promoting actin modification and disabling eukaryotic cells to peptidoglycan hydrolases targeting competing bacteria. Therefore, the T6SS is a versatile weapon targeting cells ranging from bacteria to mammalian hosts. An example of T6SS-mediated interbacterial killing between enteroaggregative E. coli (EAEC) and E. coli K-12 is shown in Fig. 2.

Figure 1.

Genetic organization and general architecture of the T6SS. (A) Schematic representation of the T6SS core genes. Genes are specified by a letter corresponding to the Tss nomenclature (“A” corresponding to “TssA”) or by their vernacular, usual names (Hcp, VgrG, PAAR, and ClpV). The color code is shared with panels B and C. (B) Architecture of the T6SS. The membrane complex, composed of the TssJ lipoprotein (orange) and the TssM (blue) and TssL (red) inner membrane proteins, is indicated (OM, outer membrane; PG, cell wall; IM, inner membrane). The different regions of the tail (spike, tube, sheath, and baseplate) are shown. (C) Architecture of a contractile tailed bacteriophage. Components that are shared with the T6SS (spike, tube, sheath, and baseplate) are depicted with the same color code (LTF, long tail fibers).

Figure 2.

Interbacterial competition between E. coli cells. Time-lapse fluorescence microscopy recordings of green fluorescent protein-labeled EAEC T6SS⁺ cells (green) in the presence of mCherry-labeled T6SS⁻ prey bacterial cells (red) in T6SS-3 inducing conditions (one image every 7.5 min). Prey cells that are killed and not present in the next frame are indicated by white arrows. Scale bar is 5 μm.

MECHANISM OF ACTION OF THE T6SS

In recent years, several aspects of the mechanism of action of the T6SS have been described. Genetic, biochemical, and structural characterization of the different T6SS subunits or protein complexes has defined the overall architecture of this secretion apparatus. Time-lapse fluorescence microscopy recordings have provided a dynamic view on how the system is assembled and works. A model for the mechanism of action of the T6SS has been proposed based on these data as well as on the knowledge on similar contractile structures such as tail bacteriophages or pyocins (3, 4, 8, 9) (Fig. 3). T6SS biogenesis starts with the assembly of the membrane complex (MC) and the tail assembly platform – or baseplate complex (BC) (Fig. 3A). The syringe, composed of the inner tube, capped by the needle spike, and wrapped by a contractile structure, the sheath, then polymerizes to form a several hundreds of nanometers-long tubular structure (Fig. 3B and C). During T6SS assembly, toxin effectors are loaded into the tube or associate with the spike trimer. Contraction of the sheath propels the inner tube/spike, allowing perforation of the target cell membrane and delivery of the effectors (Fig. 3D). The contracted sheath is then disassembled and recycled by a dedicated ATPase named ClpV (Fig. 3E and F).

Figure 3.

Mechanism of action of the T6SS. The biogenesis of the T6SS starts with the assembly of the TssJLM membrane complex (MC) and recruitment of the baseplate complex (BC) (A), which serves as a platform for polymerization of the tail tube/sheath structure (B, C). During elongation of the tail structure, effectors (red balls) can be loaded inside the inner tube lumen or attached to the VgrG spike. Following contact with a prey cell, the sheath contracts and propels the inner tube/spike toward the target, allowing penetration and delivery of the effectors (D). Once contracted, the ClpV AAA+ ATPase is recruited to the apparatus for recycling sheath subunits (E, F). The MC (and BC ?) might be reused for a new round of assembly.

We will describe in the following sections the genetic organization, the prevalence of T6SS gene clusters, the regulatory mechanisms underlying their expression, and the structure, assembly, and roles of this secretion machine, emphasizing the current knowledge on the T6SS in E. coli, Salmonella, and related species.

GENETIC ORGANIZATION AND PREVALENCE OF T6SS GENE CLUSTERS IN E. COLI, SALMONELLA, AND RELATED SPECIES

Type VI secretion system genes are distributed in Gram-negative Proteobacteria with an overrepresentation in Gammaproteobacteria (1). Therefore, T6SS genes are found in most E. coli and Salmonella species, with the exception of the E. coli B and K-12 laboratory strains. The genes encoding components and toxins of the type VI secretion system are usually clustered and grouped into genetic islands (1, 6). The GC content of these regions is generally different from that of the core genome, suggesting that they have been acquired by horizontal gene transfer (1, 2, 6). These gene clusters encode the 13 core components of the T6SS, i.e., all the subunits required to assemble a functional apparatus. Additional genes present in these clusters encode toxins and antitoxins, adaptor proteins that bind both machine components and toxins, as well as auxiliary proteins required for the assembly of the apparatus or the recruitment and proper delivery of the toxins (3, 4, 9). According to the gene organization and to homologies/similarities, the E. coli T6SS gene clusters categorize in three distinct phylogenetic groups: T6SS-1 to T6SS-3 (Fig. 4 and 5). This observation suggests that these clusters were present in common ancestors or that genetic exchanges occurred between all these strains that may share the same environment. However, the former hypothesis is likely because (i) each of the phylogenetic groups is found in both intestinal (AIEC, EAEC, EHEC, EPEC…) and nonintestinal (UPEC, APEC, MNEC…) strains and (ii) these groups are not found in bacteria that share similar environments such as Salmonella or Enterobacter species (see below). Among these three phylogenetic groups, the T6SS-1 (Fig. 5A) and T6SS-2 (Fig. 5B) gene clusters are the most commonly found in E. coli chromosomes. For example, the prevalence of T6SS-1, T6SS-2, and T6SS-3 in APEC genomes is 14.6, 2.4, and 0.8%, respectively (11, 12). Interestingly, 85% of the T6SS⁺ APEC strains belong to the virulent phylogenetic groups (11, 12). It is worthwhile to note that, in general, the T6SS-2 gene cluster is overrepresented in pathogenic strains with high virulence traits. This is particularly clear for enteroaggregative E. coli (EAEC), in which T6SS-2⁺ strains (e.g., 042) cause diarrhea, whereas T6SS-2⁻ strains (e.g., 17-2, 34b) fail to elicit diarrhea in human volunteers (13). However, this prevalence information should be taken with care, because this does not necessarily mean that T6SSs are directly involved in pathogenesis, but they may prepare the ground for virulence factors by clearing the niche of potential bacterial competitors. Indeed, T6SS-1 and T6SS-2 gene clusters are also found in nonpathogenic strains of E. coli such as E. coli W (14; Fig. 5B); however, in this strain, the T6SS-1 cluster is inactivated by insertion of a mobile element (14), and it is not known whether it is functional.

Figure 4.

Phylogenetic tree of selected T6SS gene clusters. T6SS gene clusters catagorize in 5 phylogenetic groups (A to E) (1, 2). The distribution of the E. coli-associated T6SSs (T6SS-1 to 3, red) and Salmonella-associated SPI T6SSs (green) is shown, as well as that of E. cloacae and C. rodentium (blue) and the model T6SSs from P. aeruginosa, V. cholerae, Edwardsiella tarda, and Francisella tularensis (black). The figure has been prepared with phylogeny.fr using the sequences of the TssF core component homologues (similar results were obtained with TssB homologues) (115).

Figure 5.

Organization of T6SS-1 to -3 gene clusters. Genes encoding the T6SS-1 (A), T6SS-2 (B), and T6SS-3 (C) in the indicated E. coli strains are shown schematically. Homologous genes are colored similarly (see box below). When predictable, putative phospholipase effector/immunity pairs (Tle1/Tli1, Tle3/Tli3, or Tle4/Tli4) or rhs genes are indicated. Open reading frames with unknown function are shown in white. Genes into brackets are not present or not identical in all the strains listed. Genes were identified using the SecReT6 database (116).

With the exception of Shigella sonnei and Salmonella enterica serotype Gallinarum, which carry T6SS-2-like gene clusters (Fig. 5B), the T6SS genetic organization in Salmonella, Citrobacter, or Enterobacter species is distinct from the E. coli T6SS-1 to T6SS-3 loci. In S. enterica serotypes, T6SSs belong to five phylogenetically distinct families, named as the Salmonella pathogenicity island (SPI) they are encoded on. T6SSs are found in SPI-6 in S. enterica serotype Typhimurium, SPI-19 (similar to T6SS-2) in S. enterica serotypes Dublin, Gallinarum, or Enteridis, SPI-20 and SPI-21 in S. enterica serotype Arizonae, and SPI-22 in Salmonella bongori (15, 16) (Fig. 4 and 6). However, SPI-6 T6SS remnant genes are found in S. enterica serotypes Enteridis and Gallinarum, suggesting these genes had been lost during evolution. These T6SS loci are characterized by the presence of noncore genes inserted in between the core elements (17). In S. enterica serotype Typhimurium LT2, the organization of the SPI-6 core genes is different from the E. coli species but rather shares synteny with distantly strains such as Burkholderia mallei and Ralstonia eutropha (17). Therefore, Salmonella T6SSs derive from non-E. coli clusters and have evolved from the original cluster(s) by the acquisition of noncore modules. These modules might have been transferred between strains as a hcp-tae4-tai4 module encoding an Hcp protein, and an amidase/immunity pair is found in both S. enterica serotype Typhimurium and Enterobacter cloacae, although the core genes differ (18). The addition of these distinct modules during evolution may confer specialized functions to these T6SSs.

Figure 6.

Organization of T6SS gene clusters in Salmonella, Enterobacter, and Citrobacter. Genes encoding the T6SS in the indicated strains are shown schematically. Homologous genes are colored similarly (see box in Fig. 5). When predictable, rhs genes are indicated. The rhs^main and rhs^orphan open reading frames shown to undergo rearrangements (94) are indicated in the S. enterica serotype Typhimurium SPI-6 gene cluster, as well as the Tae4/Tai4 effector/immunity pairs in S. enterica serotype Typhimurium SPI-6 and E. cloacae. Open reading frames with unknown function are shown in white. Genes were identified using the SecReT6 database (116). Note that the transcription of the C. rodentium tssM gene, interrupted by an early stop codon, is rescued by frameshifting (114).

A remarkable difference between the T6SSs in Escherichia and Salmonella strains is the nature of the toxin effectors. T6SS-1-like clusters generally encode effectors belonging to phospholipases. T6SS-2 clusters have recombinant hot spot (Rhs) elements bearing putative activities, whereas S. enterica SPI-6 and E. cloacae T6SS gene clusters encode amidases and Rhs-linked antibacterial activities (Fig. 5B and 6).

In addition to the main T6SS gene clusters, additional islands encoding Hcp, VgrG, PAAR, and putative toxins could be found disseminated on genomes. Because Hcp and VgrG have been shown to be carriers for the transport of the effectors, the existence of Hcp/VgrG islands suggests that they correspond to additional modules that adapt to the core machine for the delivery of specific toxins. In several instances, these small islands are inserted within the core gene cluster. As described above, this is particularly visible in the case of the S. enterica sp. Typhimurium SPI-6 gene cluster, in which additional islands are inserted within core genes (17, 19) (Fig. 6). It is also very clear when comparing T6SS gene clusters from distinct E. coli species, e.g., a vgrG-tle-tli-paar fragment is found in the T6SS-1-like T6SS operons in EAEC 042, AIEC LF82, and UPEC UT189, but with differences in the vgrG gene and in the effector-immunity pair (Tle effectors of families 1, 3, and 4, respectively) (Fig. 5A).

Most of the strains contain several copies of T6SS gene clusters; this multiplicity likely corresponds to various lifestyles and thus might be reflected by the regulatory mechanisms and the target cells (20). However, little information is available regarding whether these clusters are differently regulated, or have distinct functions or similar functions in different conditions. In EAEC 17-2, two T6SSs are encoded within the pheU pathogenicity island (21). Both T6SSs have antibacterial activities (22) but are expressed in different conditions: while the T6SS-1 cluster is under the control of the Fur repressor and hence induced during iron starvation (23), the T6SS-2 cluster is under the control of AggR, the aggregation master regulator, and is expressed in host cells or in synthetic media mimicking the macrophage environment (21). The regulatory mechanisms and function of the E. coli and Salmonella T6SSs are described in more detail in the corresponding paragraphs below.

BIOGENESIS AND ARCHITECTURE OF THE T6SS

The T6SS proteins assemble two modules with different evolution histories. As described above, the cytoplasmic syringe-like structure derives from the bacteriophage contractile tail or coevolved with it from a common ancestor. The assembly of both structures follows a similar sketch: the Hcp tail tube protein polymerizes to form the inner tube and is tipped by the VgrG membrane-penetrating needle. A sheath-like structure, constituted of the TssBC proteins, polymerizes in an extended, metastable conformation around this inner tube. This two-layered tubular edifice, usually hundreds of nanometers long, is assembled on a platform called the baseplate complex (10). The BC is tethered to the cell envelope via contacts with the second module, the membrane complex (24, 25). This MC is composed of three subunits distributed into the inner and outer membranes. Two of these subunits share sequence homologies with two components of the type IVb secretion system found in Legionella pneumophila, Coxiella burnetii, or ColIb-P9 plasmids (1, 6).

The best characterized T6SS in E. coli strains and related species is the EAEC T6SS-1 (Sci-1) machinery. The biogenesis of the T6SS starts with the formation of the membrane complex. Most cells assemble 1 to 3 MCs that remain static (26). Fluorescence microscopy experiments showed that its localization is not spatially restricted to the cell pole or to the septum, but rather that it is randomly distributed in the cell envelope (26). The MC is composed of three proteins (Fig. 7A and B): the outer membrane TssJ lipoprotein and the inner membrane TssL and TssM proteins (27, 28, 29). The EAEC TssJ lipoprotein is tethered to the outer membrane via acyl chains but faces the periplasm (28). Its tridimensional structure has been solved (see Fig. 7C; 30). TssJ has a transthyretin fold, i.e., a β-sandwich of two β-sheets. Two loops, notably those connecting β-strands 1 and 2, mediate contact with the TssM C-terminal domain (26, 30, 31). TssM is a 130-kDa large protein composed of three transmembrane segments in its N-terminal third, whereas two-thirds of the protein extends through the periplasm, from the inner membrane to the outer membrane-anchored TssJ lipoprotein (29, 30). The crystal structure of the C-terminal portion of the periplasmic region of TssM in complex with TssJ has recently been solved (Fig. 7C). It is composed of two domains. The structure confirmed that contacts between the two partners are established by interactions of the TssM C-terminal β-domain with loops L1-2 and L3-4 of TssJ (26). Interestingly, this β-domain is followed by an α-helix (colored purple in Fig. 7C) that inserts into the outer membrane, the insertion being facilitated by the TssJ lipoprotein (26). The 300-amino-acid loop located between TM2 and TM3 shares an NTPase fold, although the presence of Walker A and B motifs is not conserved among TssM homologues. TssM contacts TssL, which is composed of a ∼200-amino-acid cytoplasmic domain anchored to the inner membrane by a unique C-terminal transmembrane segment, categorizing TssL as a C-tail protein (32). The proper insertion of this C-tail protein requires the YidC protein and the contribution of the DnaK cytoplasmic general chaperone (32). The structure of the EAEC TssL cytoplasmic domain has been reported (33). It is composed of two bundles of three α-helices, with a general shape resembling a fish hook (Fig. 7D; 33). In several instances, the TssL C-terminus is fused to an additional domain of the OmpA/Pal/MotB family that mediates contact with the peptidoglycan (34). In the T6SS-1 of EAEC and other E. coli pathotypes, the TssL protein interacts directly with an additional component, TagL, which carries this motif. In vivo and in vitro studies have shown that this domain anchors the T6SS to the cell wall, and that mutations preventing TagL interaction with the peptidoglycan abolish T6SS function (27). The structure of the EAEC TssJLM complex has been recently solved at 11.6-Å resolution by negative stain electron microscopy (Fig. 7E; 26). The TssJLM complex has an overall rocket-shape structure with 5-fold symmetry. It is composed of 10 copies of each protein, and its base is composed of the TssL and TssM cytoplasmic and intramembrane domains. The TssM periplasmic domains form 10 arches starting from the base and converging to the tip of the structure in two layers of 5 pillars. The assembly of the MC starts from the outer membrane and progresses inward: the TssJ lipoprotein recruits TssM, and then TssL is added to the TssJM complex (26). The T6SS tail assembles on the MC that also serves as channel for the passage of the Hcp/VgrG needle during sheath contraction. The interior of the membrane complex has a size sufficient to accommodate the VgrG protein. The tip complex closes the structure at the outer membrane and it has been proposed that large conformational changes occur – notably a reorganization of the pillars – to allow the passage of the needle upon sheath contraction (26). Indeed, conformational changes in the periplasmic portion of TssM have been reported in Agrobacterium tumefaciens (35).

Figure 7.

Architecture and structure of the T6SS membrane complex. (A) The tssJ, tssL, and tssM genes that encode the components of the membrane complex. (B) Schematic representation of the TssJ, -L, and -M proteins: TssJ is an outer membrane (OM)-tethered lipoprotein, whereas TssL and TssM are inner membrane (IM)-embedded proteins. In T6SS-1, the membrane complex comprises an additional protein, TagL, which binds to the peptidoglycan (PG) layer (not depicted here) (27). (C) Crystal structure of the complex between the soluble fragment of TssJ (orange) and the two C-terminal domains of the TssM periplasmic segment (light and dark blue) including the C-terminal helix that inserts into the outer membrane (in purple) from EAEC T6SS-1 (Protein Data Bank [PDB]: 4Y7O) (Reprinted from reference 26 with permission). (D) Crystal structure of the cytoplasmic domain of TssL from EAEC T6SS-1 (PDB: 3U66) (33). (E) Negative stain electron microcopy structure reconstruction of the EAEC TssJLM complex (lower panel, EMDB: 2927) (adapted from reference 26 with permission) (scale bar is 50 nm). The position of the outer (OM) and inner (IM) membranes are predicted based on the presence of detergent micelle and the putative location of the transmembrane segments of TssM, respectively. In the upper panel is shown a top view of the TssJLM complex in which crystal structures of the TssJ-M complex (panel C) are docked, highlighting the presence of two concentric layers closing the channel at the outer membrane.

Once assembled, the MC serves as a docking station for the BC, i.e., the assembly platform of the T6SS tail (Fig. 8A and B). The T6SS BC is composed of the TssE, TssF, TssG, TssK, and VgrG subunits, and assembles independently of the MC (25). In bacteriophages, the minimal baseplate is composed of six wedges (comprising the gp6, gp53, and gp25 proteins in bacteriophage T4) that assemble around the spike complex (36). TssE is the T6SS counterpart of the bacteriophage gp25 wedge protein, and the structure of the EAEC TssE can be modeled based on the structures of gp25 homologues (Fig. 8C; 1, 6, 37). This subunit has been suggested to be connected to the sheath. Although we still lack structural information on TssF and TssG, these proteins share limited homologies with the phage T4 gp6 and gp53 proteins, respectively (25). In agreement with phage baseplate structure, a complex comprising TssE, TssF, TssG, and VgrG could be purified from EAEC (25). In addition, it has been shown that TssF, TssG, and TssK form a stable complex in Serratia marcescens (38). Once assembled, the BC is recruited to the MC via multiple contacts including TssK-TssL, TssK-TssM, and TssG-TssM interactions (24, 25). The hub of the bacteriophage – and probably of the T6SS – baseplate is the spike complex. The structure of the T6SS spike protein, VgrG, from uropathogenic E. coli CFT073 has been reported (Fig. 8D; 39). This membrane-penetrating needle is a trimer with a base that connects to the inner tube, followed by a region composed of repeats that form a highly stable three-stranded β-helix or β-prism. The structure of the β-prism of the E. coli O157 VgrG protein is also known (Fig. 8D; 40). In most cases, an adaptor protein called PAAR interacts at the tip of VgrG and sharpens its extremity (41). The VgrG trimer sits on the Hcp inner tube. Hcp proteins assemble hexameric rings with an internal lumen of ∼40 Å and an external diameter of ∼110 Å. The structures of several Hcp proteins from diverse bacteria have been reported in the literature, including that of EAEC (Fig. 8E; 42). The Hcp rings stack on each other on a head-to-tail organization, and this assembly is strictly controlled in vivo by T6SS core components including baseplate subunits and TssA (25, 43). The assembly of the inner tube is coordinated with that of the sheath but the rigid tube serves as template for sheath polymerization (43). The sheath, composed of the TssB and TssC proteins, is the contractile structure that assembles in an extended, metastable conformation that stores the energy required to propulse the inner tube (44). By using time-lapse fluorescence microscopy, it has been shown that the assembly of the 600-nm-long T6SS tube/sheath is completed in ∼20 s and could be maintained in the extended conformation for several hundreds of seconds (22, 44). However, how the extended sheath is stably maintained requires further investigation. The atomic structure of the T6SS sheath in its contracted form has been solved by cryoelectron microscopy (Fig. 8F; 45, 46, 47). The sheath is a helical structure composed of 6-TssB/TssC heterodimer strands. Contacts between the heterodimers involve the formation of a 4-stranded β-sheet handshake domain comprising two β-strands from TssC, one from the next TssC on the same strand, and one from TssB from the neighboring strand. This assembly mechanism therefore connects heterodimers of the same strand, as well as with those of the next strand (47). The contraction of the sheath is a fast event that occurs in less than 5 ms (44). Although the propulsion of the inner tube or the delivery of effectors has not been imaged yet, the contraction of the sheath coincides with the lysis of the prey cell (Fig. 9; 22). Once contracted, an N-terminal helix of the TssC subunits protrudes from the structure and recruits the ClpV ATPase that will recycle the sheath subunits (44, 48, 49, 50).

Figure 8.

Architecture and structure of the T6SS tail complex. (A) The tssA, tssB, tssC, tssE, tssF, tssG, tssK, hcp, vgrG, and paar genes that encode the components of the tail complex (blue, sheath subunits; black, inner tube subunit; green, spike subunits; pink, baseplate subunits). (B) Schematic representation of the T6SS tail complex (same color code as panel A). (C) Structural model of EAEC T6SS-1 TssE based on the bacteriophage gp25 crystal structure (PDB: 4HRZ). (D) Composite structure made with the crystal structures (from bottom to top) of the UPEC CTF073 VgrG1 protein (PDB: 2P57) (39), the E. coli O157 EDL933 VgrG β-helical prism (PDB: 3WIT) (40) and the E. coli O6 PAAR protein (PDB: 4JIW) (41). (E) Crystal structure of the EAEC T6SS-1 Hcp hexamer (left, top view; right, side view) (PDB: 4HKH) (42). (F) Cryoelectron micrograph of a contracted T6SS sheath from V. cholerae (left panel, scale bar is 100 nm) and atomic-resolution cryoelectron structure of the TssB-C complex (PDB: 3J9G) (adapted from reference 47 with permission).

Figure 9.

T6SS sheath contraction coincides with target cell lysis. Time-lapse fluorescence microscopy recordings of EAEC producing fluorescently labeled sheath subunits (TssB-sfGFP) in the presence of mCherry-labeled T6SS^- E. coli K-12 prey cells (one image every 7.5 min). The time lapse highlights the assembly and the contraction (white arrow) of the T6SS sheath, followed by the lysis of the target cell. Scale bar is 1 μm. Adapted from reference 22 with permission.

FUNCTION AND EFFECTORS

The type VI secretion system garnered interest because of its ability to target both eukaryotic and prokaryotic cells, therefore delivering a broad diversity of toxins including nucleases, phospholipases, peptidoglycan hydrolases, NAD(P)⁺ glycohydrolases, or cytoskeleton-remodeling enzymes (51, 52, 53, 54, 55). These toxins are independent polypeptides confined into the Hcp tail tube lumen, bound to the tip of the VgrG needle directly or via adaptor proteins, or covalently linked as additional C-terminal domains to VgrGs (41, 54, 56, 57, 58, 59, 60, 61, 62).

The type VI secretion system recently emerged as one of the key players during bacterial pathogenesis. However, beside the fact that T6SS expression is usually coregulated with virulence factors, the role of the T6SS in the infection process could be either direct or indirect. In a few instances, including in pathogenic strains of E. coli (see below), the T6SS has been shown to be directly involved in bacterial virulence, such as mediating adhesion to host cells or participating to the survival into macrophages or to systemic proliferation, but the molecular details on how the apparatus – or specific secreted proteins – interferes with the host cells have not been defined. In Vibrio cholerae, it has been shown that the T6SS disables macrophage by interfering with the dynamics of the cell cytoskeleton. This ability depends on the C-terminal domain carried by the VgrG1 spike, which possesses actin cross-linking activity, therefore curbing actin dynamics, preventing cell movement, and inhibiting further phagocytosis of bacterial cells (63, 64, 65). The Aeromonas hydrophila VgrG1 C-terminal extension has been proposed to carry actin-targeting ADP-ribosyltransferase activity (66). Additional T6SS effectors with antihost activities include phospholipases in Pseudomonas aeruginosa, toxins that induce membrane fusion (such as the Burkholderia thailandensis and Burkholderia pseudomallei VgrG5 C-terminal extensions) or interfere with microtubule dynamics (67, 68, 69, 70).

The direct role of the T6SS for virulence toward mammalian models of infection has been challenged by the discovery that the vast majority of T6SSs characterized so far are involved in bacterial growth inhibition. The T6SS is used to deliver antibacterial effectors with peptidoglycan hydrolase (amidase [Tae], glycoside hydrolase [Tge]) or phospholipase (Tle) activities into the periplasm of the target prey cell (71, 72, 73, 74). These amidases, muramidases, and phospholipases belong to various families that hydrolyze bonds within the peptidic stems (Tae1-4 families) or glucosidic chains (Tge1-3 families) of the peptidoglycan or ester bonds of phospholipids (Tle1-5 families), respectively (52, 53, 54, 72, 74). Toxins with nuclease (Tde) and NAD(P)⁺ glycohydrolase activities have been reported and therefore should reach the cytoplasm for their action (55, 75, 76). How these toxins are transported across the inner membrane is not clearly defined, but it has been proposed that the translational elongation factor Tu (EF-Tu) contributes to the translocation of the NAD(P)⁺ glycohydrolase effector in P. aeruginosa (55). Producing cells are protected from their own effectors or the effectors of their siblings by the coproduction of specific protein inhibitors – or immunity proteins – that usually bind and inhibit the activity of the cognate toxins (52, 53, 54). T6SS⁺ bacteria, including closely related species, carry different and unique sets of antibacterial effectors, supporting a leading role for T6SS effectors in inter- and intrabacterial competition and in reshaping bacterial communities (52).

Functions and Effectors in E. coli, Salmonella, and Related Species

Phenotypes associated with T6SS in E. coli and Salmonella

Most of the E. coli and Salmonella T6SSs studied so far have been shown to participate in adherence to biotic and abiotic surfaces, in bacterial competition, or in virulence toward various models of infection (Table 1). Taken together, the available information on E. coli T6SSs point to a role of T6SS-1 and T6SS-3 for antibacterial activity and of T6SS-2 for pathogenesis.

Table 1.

Phenotypes and effectors associated with T6SS in E. coli, Salmonella, and related species

T6SS	Pathotype/serotype	Strain	Activity	Effectora	References
T6SS-1	EAEC	17-2	Biofilm	–	28
			Antibacterial	Tle1	77
	APEC	TW-XM	Biofilm	–	78
			Antibacterial	Tle4	78
T6SS-2	APEC	DE719	Attenuated virulence in ducks Reduced intracellular in chicken macrophages	–	12
		SEPT362	Attenuated virulence in chicks	–	81
		TW-XM	Penetration of the blood-brain barrier during cerebral infection	–	78
	MNEC	K1	Invasion of human brain microvascular endothelial cells	–	84
T6SS-3	EAEC	17-2	Antibacterial	–	22
SPI-6	S. Typhimurium	LT2	Affect replication in macrophage and systemic dissemination in mice and in chicks	–	17, 80, 86, 87, 90
			Antibacterial	Tae4	19, 93
			Antibacterial	Rhsorphan	94
	S. Typhi	Ty2	Systemic infection in mice	–	85
	E. cloacae	ATCC13047	Antibacterial (putative)	Tae4	18, 73
SPI-19	S. Gallinarum	287/91	Colonization of the gastrointestinal tract and systemic dissemination in chicks	–	89
CTS1	C. rodentium	IC68	Antibacterial	–	79

^a –, unknown.

T6SS-dependent bellicose behaviors toward neighboring bacteria have been evidenced for EAEC 17-2 (T6SS-1 and T6SS-3; 22, 77), APEC TW-XM (T6SS-1; 78), and Citrobacter rodentium (79). S. enterica serotype Typhimurium LT2 has also been recently reported to have antagonistic activities against E. coli and Salmonella species in a SPI-6 T6SS-dependent manner (19). However, in S. enterica serotype Typhimurium, the T6SS is upregulated in the late stages of infection, once the bacterium is internalized in eukaryotic phagocytic cells, whereas no upregulation could be observed in mixed bacterial cultures (17, 19, 80). These observations suggest that the antibacterial activity is activated inside the host. Interestingly, the T6SS is under a regulatory network that also controls virulence factors including proliferation genes. Because the S. enterica serotype Typhimurium population inside phagocytic cells is usually clonal, Brunet et al. proposed that the T6SS may serve to eliminate disabled S. enterica serotype Typhimurium cells of the progeny, i.e., the cells unable to produce the T6SS immunity and proliferation genes (19).

A role of the T6SS for biofilm formation has been reported for EAEC 17-2 (T6SS-1; 28) and APEC SEPT362 (T6SS-2; 81). In APEC, the defect in biofilm is accompanied by decreased adherence to epithelial cells (81). However, it is likely that these phenotypes are caused by impacts on fimbriae gene regulation or by perturbations of the biofilm structure due to the antibacterial activity. Indeed, deletions of T6SS-2 genes in APEC SEPT362 affect the expression of type 1 fimbriae and flagella, two extracellular structures required for adhesion and biofilm formation (81, 82).

T6SS-dependent pathogenesis toward hosts has been tested for a number of strains (Table 1). From the available data, and in agreement with the overrepresentation of T6SS-2 in virulent strains, no defect in virulence has been noted for T6SS-1 or T6SS-3 mutants, whereas mutations in T6SS-2 gene clusters usually impact colonization, survival, or invasion. The EAEC 17-2 T6SS-1 and T6SS-3 are not required for virulence toward Caenorhabditis elegans and intestinal survival within BALB/c mice respectively (21, 28, 77) and mutants in the UPEC CFT073 T6SS-1 gene cluster do not present colonization defects in CBA/J mice bladders and kidneys (83). In contrast to T6SS-1 and T6SS-3, the APEC DE719 and SEPT362 T6SS-2 display attenuated virulence and decreased systemic dissemination in chicks or ducks, and reduced intracellular survival in chicken macrophage cells (12, 81). In APEC strain TW-XM, T6SS-2 is necessary for cerebral infection and penetration of the blood-brain barrier (11). Similarly, the MNEC K1 T6SS-2 is required for internalization in human brain microvascular endothelial cells (84). In MNEC K1, T6SS-2 carries two Hcp proteins. Analyses of the phenotypes of mutations in these two genes showed that Hcp1 is required for efficient binding to brain endothelial cells, whereas Hcp2 induces stress fiber formation, cytoskeleton rearrangements, cytokine and chemokine release, and cell apoptosis via activation of the caspase 8 pathway (84). Therefore, the function of this apparatus is necessary at two different stages of the infection process, probably by the Hcp1- and Hcp2-specific transport of distinct effectors.

In S. enterica serotype Typhi, the SPI-6 genes are required for systemic infection in a humanized mouse model (85), whereas conflicting data have been reported for the S. enterica serotype Typhimurium SPI-6 T6SS regarding replication, survival, and proliferation in macrophages or in mice (17, 80, 86, 87). However, the most important effect on replication within phagocytic cells is observed with the tai4 mutation, a gene that encodes the immunity to the Tae4 amidase, suggesting that this defect is indirect and due to self-intoxication caused by the loss of the immunity (17, 19). The S. enterica serotype Gallinarum SPI-19 is required for survival and growth within chicken macrophages and for efficient colonization of the chick gastrointestinal tract and internal organs (88, 89). It is interesting to note that the phenotypes associated with the deletion of the SPI-6 T6SS gene cluster in S. enterica serotype Typhimurium can be rescued by the expression of the S. enterica serotype Gallinarum SPI-19 T6SS, suggesting that both T6SSs perform similar functions despite their phylogenetic differences (90).

E. coli and Salmonella T6SS toxins

In E. coli and Salmonella strains, few effectors have been characterized in detail, but their presence and putative activities can be easily predicted. They are organized in tandem with genes encoding small proteins that likely correspond to cognate immunity proteins. Furthermore, they usually co-occur with hcp, vgrG, or paar genes.

E. coli T6SS-1 gene clusters encode putative phospholipases (Table 1, Fig. 5A). They localize upstream the vgrG genes suggesting that, as shown for V. cholerae (91), they use the VgrG needle as a carrier for their transport. Interestingly, close inspection of these phospholipase genes suggests that they belong to different families: while the AIEC LF82 or UPEC CFT073 T6SS-1 clusters carry putative phospholipases of the Tle3 family, those present on the EAEC 042 and APEC TW-XM genomes are closely related to phospholipases of the Tle1 and Tle4 families, respectively (Fig. 5A; 77, 78). Indeed, recent data have shown that the Tle1 protein from EAEC 17-2 is delivered by the T6SS-1 into bacterial target cells where it exerts phospholipase 1 activity (77). The producing strain is protected from the action of its own toxin and that of sister cells by Tli1, an outer membrane lipoprotein that binds Tle1 with nanomolar affinity and inhibits its phospholipase activity (77). In agreement with the genetic organization, Tle1 is transported as a cargo by binding to a C-terminal extension of the VgrG spike protein (77).

E. coli T6SS-2 gene clusters, as well as the S. enterica serotype Gallinarum SPI-19 T6SS-2-like T6SS, usually contain genes encoding Rhs elements (Table 1), but the activity carried by these Rhs proteins cannot be inferred from in silico prediction algorithms. In addition to the main T6SS gene cluster, S. enterica serotype Gallinarum possesses an hcp island that encodes an Hcp protein, an amidase of the Tae3 family and its cognate Tai3 immunity (15, 73). No gene with putative toxin activity is found within E. coli T6SS-3 gene clusters.

The S. enterica Arizonae SPI-21 T6SS carries two pairs of S-type pyocins/immunity, as well as a specialized Hcp protein corresponding to a fusion between a traditional Hcp protein to an effector domain (15).

The S. enterica serotype Typhimurium SPI-6 T6SS encodes Rhs elements as well as an amidase of the Tae4 family (73) (Table 1, Fig. 6). The structure of Tae4 is available, alone or in complex with its cognate Tai4 immunity protein (92, 93). Tae4 is a dl-endopeptidase with a typical NlpC/P60 domain. It hydrolyses the d-Glu/meso-diaminopimelic acid (mDAP) bond of peptidoglycan peptidic stems. A dimer of Tai4 binds to Tae4 with a K_D of 3×10⁻¹⁰ M and inhibits Tae4 activity by inserting the L4 protruding loop of one Tai4 monomer into the Tae4 catalytic pocket (92, 93). Contacts between Tae4 and Tai4 are stabilized by the α-helix of the second Tai4 subunit (92, 93). A similar Tae4/Tai4 pair is encoded within the E. cloacae SPI-6-like T6SS (73). Indeed, cross-immunity between the S. enterica serotype Typhimurium and E. cloacae Tae4/Tai4 pairs has been demonstrated (18). Finally, an original mechanism has been revealed in the case of the S. enterica serotype Typhimurium Rhs (94). The gene encoding the full-length Rhs protein (Rhs^main) is followed by a nontranslated region encoding an orphan C-terminal extension (Rhs^orphan). However, serial passages in broth or within the mouse induces a duplication of the region and a genetic chromosomal rearrangement that results in the production of a chimera Rhs protein constituted of the Rhs^main core and the Rhs^orphan C-terminal extension with antibacterial activity. This elegant mechanism therefore provides a selective advantage to cells of the evolved bacterial lineage because it enables them to maintain the immunity to Rhs^main and to deploy a new toxin that is active against ancestral cells (94).

Finally, two genes present within the Citrobacter freundii and E. cloacae T6SS gene clusters encode proteins with MIX domains, an N-terminal sequence associated with several T6SS toxins (95).

REGULATORY MECHANISMS

T6SS gene clusters are tightly regulated to adapt their expression to the environmental conditions. In agreement with the broad diversity of T6SS targets and activities, T6SS gene clusters are not subjected to a unique regulation, but instead have hijacked most of the transcriptional and posttranscriptional regulatory mechanisms known in bacteria (20, 96): two-component systems, transcriptional activators and repressors, histone-like nucleoid associated proteins, quorum sensing, alternative sigma factors, small regulatory RNAs, etc. In addition, a posttranslational phosphorylation-dependent pathway has been identified and characterized in pseudomonads. This signal transmission involves sensing of specific stimuli and activation of a transenvelope-transducing cascade comprising the TagFQRST proteins and leading to the PpkA-dependent phosphorylation of the forkhead-associated FHA protein (97, 98, 99, 100, 101). The reversibility of the activation is secured by the dephosphorylation of FHA by the PppA phosphatase (97).

Regulation of T6SS Gene Clusters in E. coli, Salmonella, and Related Species

Transcriptional regulation

T6SSs have been studied in detail in Pseudomonas, Agrobacterium, and Vibrio species. Hence, we have a comprehensive picture of the regulatory mechanisms underlying expression of the T6SS gene clusters in these bacteria. By contrast, with the exception of the enteroaggregative E. coli and S. enterica serotype Typhimurium T6SSs, only very little is known on the regulation of T6SSs in other E. coli and Salmonella serotypes.

Enteroaggregative E. coli

The EAEC 17-2 strain genome encodes two complete sets of T6SS genes (families T6SS-1 and T6SS-3). The T6SS-1 family cluster (also called sci-1) is under the control of the Fur repressor (23, Fig. 10). Fur – for Ferric uptake regulator – is the main regulator that couples iron homeostasis to gene expression (102). In presence of iron, Fur binds to two Fur boxes present in tandem in the promoter sequence of this cluster, overlapping with the transcriptional −10 (Fur⁻¹⁰) and −35 (Fur⁻³⁵), preventing RNA polymerase recruitment and therefore turning OFF the T6SS-1 genes (23). This very simple mechanism is complexified by an epigenetic circuit depending on the action of the DNA adenine methyltransferase, Dam, which couples T6SS expression to the replication state. The Fur⁻¹⁰ box contains a GATC motif that is recognized and methylated by Dam. When Fur is bound to the Fur⁻¹⁰ box, the site is not methylated. However, under low iron conditions, Fur is dislodged and the T6SS-1 genes are turned ON. If cells replicate, the Fur⁻¹⁰ box is methylated after the first replication, preventing Fur from binding back, therefore turning the T6SS genes under a constitutive ON state. Fur therefore controls the passage between the OFF and ON states whereas Dam is a sensor of replication and controls the passage between the reversible and constitutive ON states (23) (Fig. 10). In agreement with these data, the EAEC T6SS-1 is activated in minimal media or in iron depletion conditions (24). However, this mechanism is unlikely to be conserved between the T6SS-1 clusters shared by pathogenic E. coli because no Fur box could be readily identified in their promoter regions.

Figure 10.

Regulation of the EAEC T6SS-1 gene cluster. (A) Schematic representation of the promoter organization of the EAEC T6SS-1 gene cluster. The location of the −10 and −35 transcriptional elements (blue), of the Fur-binding sequences (red) and of one of the GATC sites (green) are shown. (B) Regulatory mechanism of the EAEC T6SS-1 gene cluster (23). In iron-replete conditions, a Fur dimer (red balls) represses the expression of the T6SS-1 gene cluster by binding to the Fur⁻¹⁰ box, which overlaps with the −10 element (OFF). When iron is limiting, the −10 element is available for the RNA polymerase allowing expression of the T6SS-1 genes (ON). Upon replication, the GATC site is methylated (CH₃) and by preventing Fur binding allows Fur-independent, constitutive expression of the T6SS-1 gene cluster.

The EAEC T6SS-3 gene cluster (also called sci-2 or aai) is activated in synthetic media mimicking the macrophage environment such as Eagle’s medium (21, 22). DNA microarrays and quantitative reverse transcription-PCR have demonstrated that this activation depends on an AraC-like transcriptional regulator called AggR (21, 103). Although no consensus binding site has been defined for AggR, this activator regulates other EAEC virulence factors such as the Aaf fimbriae, the dispersin, and the dispersin transporter (103).

No data are available yet for the T6SS-2 gene cluster found in the EAEC 042 strain. Although it remains to be experimentally tested, it has been proposed that regulation of the T6SS-2 EC042_0229 gene involves the synergistic action of the cyclic AMP receptor protein (CRP) and the nucleoid-associated protein Fis by an original mechanism requiring Fis-dependent compensation of the non-optimal spacing between the CRP and RNA polymerase binding sites (104).

Salmonella enterica

In S. enterica serotype Typhimurium LT2, the SPI-6 T6SS gene cluster is controlled by the SsrA/B two-component system, one of the major regulatory pathways of Salmonella virulence (80). The SsrB response regulator binds to and positively regulates most SPI-2 promoters including those controlling expression of the T3SS genes (105). By contrast, SsrA/B exerts a negative control on the SPI-6 T6SS gene cluster (17, 80), probably by direct SsrB binding on distinct promoter regions (106). The expression of T6SS genes encoded within the SPI-6 pathogenicity island are not detected under laboratory in vitro conditions (80, 107); however, promoter reporter and transcriptional profiling studies showed that the expression of these genes is activated in the late stages of macrophage and epithelial cells infection (17, 108). In addition to SsrB, the SPI-6 T6SS genes are silenced by the histone-like nucleoid-structuring H-NS protein (19, 109, 110). H-NS binds to A/T-rich motifs and polymerizes to spread and silence the genes by preventing access to the RNA polymerase or activators. H-NS is thus a xenogenic silencer that usually represses horizontally acquired genes and islands (111). Because T6SS genes are clustered in these islands, H-NS is probably involved in the regulation of many T6SS gene clusters in pathogenic E. coli, but this needs to be addressed.

In S. enterica serotype Typhi Ty2, several regulators have been identified such as RcsB, PmrA, and Hfq, but their contribution for the activation of the T6SS genes is relatively weak (112). In E. cloacae, several genes of the T6SS cluster are under the control of a LuxR/acylhomoserine lactone-dependent quorum-sensing mechanism and therefore respond to the population density (113).

In addition to the EAEC T6SS-3 cluster, shown to be induced in media mimicking the macrophage environment, most of the E. coli and Salmonella T6SS genes are induced in in vivo conditions. This has been reported for the S. enterica serotype Typhimurium SPI-6 T6SS genes, which are upregulated during macrophage infection (80, 108) and for the S. enterica serotype Gallinarum SPI-19 T6SS genes, which are upregulated after internalization by murine or avian macrophages (88). These data suggest that these T6SS gene clusters might play a direct role during infection or that host-mediated activation of the anti-bacterial activity might help to clear the niche and to enable efficient colonization.

Transcriptional frameshifting and posttranslational activation

In C. rodentium, the tssM gene is subjected to transcriptional frameshifting. The tssM gene is interrupted by a premature stop codon, but a poly(A) slippery tract located upstream of the stop codon induces incorporation of additional adenosines in the RNA and hence the synthesis of two TssM length variants, both required for the activity of the Type VI secretion apparatus (114). Yet, whether the frameshifting efficiency is influenced by environmental cues or by regulatory factors is unknown.

Finally, it is interesting to note that genes encoding the phosphorylation-dependent posttranslational pathway found in Pseudomonas and Agrobacterium species are present in none of the E. coli, Salmonella, Citrobacter, or Enterobacter strain genomes sequenced so far, except for those encoding FHA proteins that are found associated with E. coli T6SS-2 and Salmonella SPI-19 clusters (Fig. 5B and 6). However, the contribution of FHA for the assembly or the activation of these systems has not been reported in these strains.

CLOSING REMARKS AND FUTURE DIRECTIONS

This review summarizes the current knowledge on the T6SSs present in E. coli and related species. It is clear that the recent years have provided a detailed view on the architecture and mechanism of assembly of this apparatus. However, the regulatory mechanisms underlying the expression of these gene clusters, the effectors delivered by these machineries, and the function of these T6SSs during host infection remain enigmatic for most enterobacterial pathogens. Further studies will provide a better understanding of the T6SS contribution in the ecological niche of these strains or for pathogenesis. Similarly, although a number of antibacterial effectors with amidase, peptidoglycan hydrolase, phospholipase, and DNase activities have been identified or predicted, it remains to determine whether phospholipases and DNases might be targeted into eukaryotic host cells and to identify antieukaryotic-specific effectors. These toxins would therefore be interesting targets for the development of drugs that will interfere with these toxins, not only for human health, but also in the cases in which the bacterial pathogen targets poultry or cattle.