Type II Secretion in Escherichia coli

MARCELLA PATRICK, MIRANDA D. GRAY, MARIA SANDKVIST, AND TANYA L. JOHNSON^*

[SECTION EDITOR: TRACY PALMER]

Posted 19 October, 2010

Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, MI 48109–0620

^*Corresponding author. Mailing address: Department of Microbiology and Immunology, 3726 Medical Science Building II, University of Michigan Medical School, 1150 West Medical Center Drive, Ann Arbor, MI 48109–0620. E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

INTRODUCTION

Type II Secretion Systems

The double membrane of the gram-negative cell envelope adds a level of complexity for secreted proteins as they must cross two physical barriers to reach the exterior of the cell. Gram-negative bacteria have therefore evolved several proteinaceous systems to secrete proteins into their surroundings. Some of these pathways transport proteins across the bacterial cell envelope in a single step, including the signal sequence-independent type I, III, and IV secretion pathways. On the other hand, the signal peptide-dependent secretion systems (type II secretion [T2S] and autotransporter pathways) involve a two-step process. In these latter pathways, proteins containing an N-terminal signal peptide are first translocated across the inner membrane via the Sec or Tat systems. After removal of their signal peptides and release into the periplasm, the mature proteins cross the outer membrane (for reviews, see references 35, 58, and 97). In the case of the type II secretion system (T2SS), proteins fold into their final or near-final tertiary structure before outer membrane translocation (97).

The T2SS supports the secretion of many virulence factors, including toxins and proteases, and therefore secretion via this pathway is regarded as a major virulence mechanism. Bacteria that utilize a T2SS include human, animal, and plant pathogens (98). In humans, these bacteria are often the cause of serious diseases such as cholera, other severe diarrheal diseases, and pneumonia. In addition to its role in pathogenesis, the T2S pathway has been implicated in transporting proteins that have a role in symbiosis or other associations between bacteria and eukaryotic hosts.

Depending on the bacterial species, 12 to 15 genes have been identified as essential for T2S, and the genes and gene products have been designated by the letters A to O and S (Fig. 1) for most species. The organization of the T2S gene cluster is relatively well conserved, with most of the genes appearing in a single operon. A large amount of research studying model organisms such as Klebsiella oxytoca, Pseudomonas aeruginosa, Dickeya sp. (formerly Erwinia chrysanthemi), and Vibrio cholerae has given us a reasonably good picture of the individual components that make up the T2S machinery (35, 58, 97). The precise mechanism of the secretion process is not as well understood, however. One thing that has become apparent is that the T2S apparatus shares an evolutionary history with type IV pili (35, 83). While this chapter will focus principally on the T2S transport process, it should be mentioned that much of what has been learned about the T2SS complex can also be applied to type IV pilus biogenesis.

Fig. 1Type II secretion system (T2SS) gene alignments. The genes comprising the T2SSs from individual species are shown schematically. Homologous components are colored similarly; arrowheads show orientation. Dashed lines indicate a deleted region in the operon and do not represent linkage relationships.

Escherichia coli and T2S

The T2SS is widespread among gram-negative bacteria such as V. cholerae, K. oxytoca, P. aeruginosa, Legionella pneumophila, and E. coli. E. coli is an extremely diverse group of bacteria that contribute to normal intestinal microbiota, but also include intestinal pathogens of global significance such as enteropathogenic E. coli (EPEC, associated with infant diarrhea), enterohemorrhagic E. coli (EHEC, an important food-borne pathogen), enterotoxigenic E. coli (ETEC, associated with traveler's diarrhea), enteroinvasive E. coli (EIEC, associated with bacillary dysentery), enteroaggregative E. coli (EAEC, associated with persistent diarrhea in children), and DAEC (diffusely adhering E. coli, also associated with diarrhea in children). Pathogenic E. coli are also important causes of extraintestinal infections. Uropathogenic E. coli (UPEC) is the major cause of community-acquired urinary tract infections, while the meningitis-associated E. coli (MNEC) is the most common cause of gram-negative meningitis in newborns. Most E. coli appear to possess one or two complete T2S operons encoded chromosomally; however, there are exceptions to this rule. In EHEC, for instance, the T2S genes are present on the large virulence plasmid, pO157 (104).

The T2SS is expressed by EPEC strain E2348/69 at 37°C and is required for virulence (128). In EHEC strain O157:H7, the T2SS has been shown to promote the bacterium's adherence and intestinal colonization. Mutants not expressing the T2SS were more than fivefold less able to colonize infant rabbits (51). In UPEC, T2SSs have been demonstrated to be important in spreading between sites in the urinary tract and in avoiding efflux in the mouse bladder (66). Finally, in ETEC, T2S proteins are responsible for secretion of one of the major virulence factors leading to diarrhea, heat-labile enterotoxin (LT). Expression of the etxAB (eltAB) operon, which encodes LT, is repressed by the nucleoid-structuring protein (H-NS) when cells are grown at low temperatures (115), i.e., temperatures outside the human body. The expression of LT may also be regulated by osmolarity (67, 116) and pH (67). The T2S genes of ETEC have been shown to constitute one transcriptional unit, with a σ⁷⁰ promoter identified at the start, and are also transcriptionally controlled by H-NS in response to low temperatures (128). Regulation in ETEC therefore seems to occur with both expression of the secretion apparatus and LT, thus ensuring the coordinated production and secretion of LT at the appropriate time and place.

E. coli K-12 also contains genes encoding a T2SS, but while some of the predicted proteins of the K-12 T2SS share a high degree of similarity with the T2SS proteins of ETEC, others have no significant homology. The E. coli K-12 T2S operon is not expressed under standard laboratory conditions. When expressed on a plasmid in an H-NS mutant background, however, the E. coli K-12 T2SS was shown to secrete a protein encoded by a gene directly downstream of its genomic location, a chitinase (ChiA) (41).

The T2SS in other E. coli may also be under regulatory control, but not necessarily with the same mechanisms as those controlling the ETEC T2S genes. Analysis of other gram-negative bacteria has revealed that regulation of T2S pathways varies greatly from species to species. In P. aeruginosa, for example, one T2S pathway has been found to be regulated by quorum sensing (21), while another T2SS is only expressed under phosphate-limiting conditions (10). A quorum-sensing system has also been identified for Erwinia carotovora, a plant pathogen that secretes cell wall-degrading enzymes via a T2SS. When a strain with a mutation in a quorum-sensing gene was examined, both the production of T2S-dependent exoenzymes and the ability of the mutant strain to propagate and cause disease were reduced (60). In addition, expression of the T2S genes was induced in early stationary phase, suggesting that these genes may also be under quorum-sensing control (75). In V. cholerae, the T2S genes appear to be regulated by RpoE, an alternative sigma factor that responds to extracytoplasmic stress (26).

In contrast to regulated expression of the T2S genes in many species, expression of the T2S genes in Aeromonas hydrophila may be constitutive. Reverse transcriptase-PCR analysis of mRNA production in A. hydrophila has shown that at least one gene in the T2S operon is constitutively expressed (110). The expression of the A. hydrophila T2S-dependent serine protease AspA, however, does appear to be regulated by quorum sensing (111). It is possible that constitutive expression of the secretion apparatus is necessary for species that secrete substrates in several different environments.

SECRETED PROTEINS

A wide variety of proteins are secreted via the T2SS. Many of these proteins are associated with degradation of macromolecules or destruction of various tissues. T2S substrates include chitinases, cellulases, pectin lyases, toxins, proteases, lipases, phospholipases, and nucleases, to name a few (Table 1). Secretome projects are only now beginning to uncover the number of proteins that depend on T2SSs for their transport into the extracellular environment (25). While only a few substrates are known in E. coli, the proteins discovered so far have proven to be critical for the pathogenicity of these bacteria.

Heat-Labile Enterotoxins

Perhaps the best-characterized substrates of the T2SS in E. coli are the heat-labile enterotoxins (LTs) of ETEC. LT proteins are highly homologous to the cholera toxin of V. cholerae, particularly in structure (Fig. 2) (108). Like cholera toxin, LT is a member of the AB₅ family of enterotoxins, where the mature toxin is constructed of a symmetric ring of five identical B subunits joined together with a single A subunit. Although the cholera toxin genes of V. cholerae are phage encoded (121), the etxAB operon is most often located on a plasmid (47) in ETEC. LT genes, however, can also be found within lysogenic phage (113) or on the chromosome (56, 86), suggesting that the genes encoding LTs may have originated from ancestral cholera toxin genes (124).

Fig. 2Comparison of heat-labile enterotoxin (LT-IIb) and cholera toxin structures. The structures of LT-IIb (left, PDB ID code 1TII) and cholera toxin (right, PDB ID code 1XTC), portrayed such that their A subunits (brown) are in a similar orientation. The individual B subunits are in red, magenta, green, blue, and cyan.

After synthesis in the cytoplasm, the individual A and B polypeptides are transported as precursor proteins across the inner membrane and into the periplasm by the Sec apparatus. Once in the periplasm, the N-terminal signal sequences are removed from the polypeptides and the subunits assemble into a mature AB₅ holotoxin. The folding and assembly of the subunits is catalyzed by DsbA, a periplasmic isomerase. Loss of DsbA by gene inactivation greatly reduces the amount of LT holotoxin formed (129).

Once fully assembled, the LT holotoxin is transported across the outer membrane via the T2SS. Inactivation of the T2S pathway by mutagenesis of the outer membrane protein GspD completely abolished LT secretion across the outer membrane and resulted in an accumulation of the toxin in the periplasmic space. Release of LT from the periplasm could then be restored when the gspD mutant was complemented with a plasmid copy of the wild-type gspD gene (114).

In contrast to cholera toxin, which is freely secreted away from the cell; most ETEC strains do not release LT into the extracellular environment, but rather the toxin largely remains associated with the bacterial cell surface (Fig. 3) (53, 69, 123). Following secretion through the T2SS, LT binds to lipopolysaccharide (LPS). Extracellular LT is only detected in association with outer membrane vesicles released from the cell. Interestingly, the ability of AB₅ toxins to bind to the bacterial cell surface, rather than be released into the extracellular environment in a soluble form, is species specific rather than toxin specific. Both LT and cholera toxin can bind to E. coli LPS and remain associated with the cell. These interactions have been shown to occur via the inner core sugar 3-deoxy-D-manno-octulosonic acid of LPS. Neither LT nor cholera toxin is able to bind to V. cholerae LPS, however, because the sugar of V. cholerae LPS is phosphorylated (52). It has been proposed that the reduced toxicity of ETEC compared with V. cholerae could be explained, in part, by this difference in locations of the AB₅ toxins (109). Because V. cholerae appears to secrete its toxin away from the cell rather than tightly binding toxin to the surface of the bacteria, more cholera toxin may come into contact with host receptors.

Fig. 3Schematic model of LT secretion. LT toxin subunits are translocated across the inner membrane via the Sec translocon and fold into an AB₅ structure (shown in ribbon structure) in the periplasm of the ETEC cell. (1) After holotoxin formation, the assembled LT is targeted to the T2SS by an unknown mechanism. (2) Once engaged with the T2SS, LT is transported across the outer membrane. (3) Upon exposure to the extracellular space, LT binds to LPS on the bacterial cell surface (4), and as outer membrane vesicles are budding off the cell surface, LT is released. (5) B subunits bind to gangliosides on the host cell and facilitate the entry of the toxin by an unknown mechanism, initiating a series of cellular events leading to diarrhea.

During a mutational screen designed to find ETEC genes involved in pathogenesis, a second operon was discovered and shown to be required for maximal LT secretion (38). Fleckenstein et al. (38) demonstrated that deletion of leoA, a gene within this second operon that encodes a protein homologous to bacterial and eukaryotic GTPases, greatly reduced LT secretion. Exactly how this operon contributes to LT release from the cell is unclear, but it appears to play some role in the formation and protein content of outer membrane vesicles produced by ETEC.

The host cell surface ganglioside GM1 receptor recognized by the LT is dictated by the B subunits of the holotoxin (43). Once the B subunits bind to this membrane receptor, they facilitate entry of the toxin into the host cell (109). Following endocytosis, LT is transported to the endoplasmic reticulum where the A subunit is reduced and the resulting catalytic A1 subunit retrotranslocates to the cytosol (73, 117). Here, A1 ribosylates the stimulatory G proteins leading to activation of adenylate cyclase and elevated production of cAMP. Elevation of cAMP results in activation of cAMP-dependent protein kinase A and phosphorylation of chloride channels; stimulating Cl⁻ secretion and reducing the absorption of Na⁺ followed by massive efflux of water (109). Indeed, LT directly contributes to the severity of diarrhea seen in the host (12). Results from piglet infection studies also suggest that LT can lead to enhanced colonization of the intestinal epithelium by ETEC strains by affecting bacterial attachment to the epithelium, as well as causing a general increase in the concentration of bacteria in the small intestine (132).

ChiA

The only known T2S substrate in E. coli K-12 is the chitinase, ChiA (41). Chitinases degrade chitin, a polymer of N-acetylglucosamine (GlcNAc) that is a major component of the cell walls of fungi, crustaceans, and other organisms. Chitin is one of the most abundant biopolymers in nature and is used by many microorganisms as a source of carbon and nitrogen. Use of chitin as a nutrient, however, requires degradation of the GlcNAc polymer into monomers by the action of chitinases and chitobioses. In other gram-negative bacteria such as V. cholerae and L. pneumophila, chitinase secretion has been well documented (25, 74, 125), and the enzyme has been shown to be a substrate of the T2SS (22, 25, 85). Indeed, when the T2SS of E. coli K-12 was kept under H-NS suppression, ChiA was not secreted, and the vast majority of the protein accumulated in the periplasm (41).

As reported for other bacterial chitinases (22, 122), E. coli ChiA was able to degrade ethylene-glycol chitin, a lysozyme substrate (40). Bacteria expressing chitinases with broad specificity could be important for additional aspects of competition and survival, such as promoting pathogenicity or other interactions with host organisms. Along these lines, there has been a renewed interest in examining the effect of chitin, and presumably the action of chitinases and other chitin-binding proteins, on the pathogenicity of Vibrio and Legionella spp. other than their role in carbon acquisition. In particular, chitin has been shown to induce proteins involved in colonization (25, 93), induce natural competence (80), and promote the survival of the bacteria during exposure to stomach acid (81) and low temperatures (6).

Interestingly, the remnants of a second T2SS are present in the E. coli K-12 strain, MG1655, at a second chromosomal location, suggesting that K-12 strains may have originally possessed another T2S pathway, which has since been lost. One gene from the pathway that still remains in K-12 strains is pppA, encoding a prepilin peptidase (42). Francetic et al. (42) cloned the pppA gene from E. coli K-12 and demonstrated that it is functional.

StcE

StcE (secreted protease of C1-esterase inhibitor) is a metalloprotease encoded by the large virulence plasmid pO157 of EHEC O157:H7 (70). Immediately downstream of stcE is the T2SS operon etpC-O. In view of this genetic linkage and the presence of a candidate cleavable N-terminal signal peptide in the StcE protein sequence, Lathem et al. (70) investigated whether StcE was, in fact, secreted by the etp T2SS. Upon disruption of etpD, the gene encoding the putative T2SS outer membrane secretin StcE was no longer detectable in culture supernatants by immunoblot analysis. Complementation in trans with etpD restored the ability of the etpD mutant strain to secrete StcE, thus confirming that the defect in T2S was responsible for the absence of extracellular StcE (70).

Expression of the stcE gene is upregulated by the global regulator Ler (LEE-encoded regulator). By inducing Ler expression in O157:H7, the level of extracellular StcE was increased nearly 12-fold over uninduced conditions (70). Ler also plays an important role in the expression of other secretion systems in E. coli. The LEE pathogenicity island is common to enteropathogens that cause disease through attaching and effacing lesions, and it encodes many proteins that are required for virulence, including the proteins required for type III secretion. Upon Ler binding, expression of many genes in the LEE pathogenicity island is permitted (33).

To date, several substrates have been identified for StcE. C1-esterase inhibitor is a member of the serine protease inhibitor family in humans, while the others, glycoprotein 340 and mucin 7 and CD43 and CD45 are salivary and neutrophil glycoproteins, respectively (45, 70, 112). StcE can bind these substrates and change the glycoprotein's activity in both cleavage-dependent and non-cleavage-dependent ways. Additionally, purified StcE has been shown to significantly reduce the viscosity of human saliva (45). Because the known StcE substrates are glycosylated proteins with defensive roles in the body, and because the expression of StcE's mucinase activity has been shown to be coregulated with proteins of the LEE pathogenicity island elements, it has been hypothesized that StcE contributes to the pathogenesis of E. coli O157:H7 by degrading the protective layer of mucins and glycoproteins on host immune and intestinal cells (45).

Other enteropathogens also secrete proteins with mucinase activity via the T2SS. For instance, the HA/protease of V. cholerae has been shown to degrade fibronectin, ovomucin, and lactoferrin and is highly homologous to the P. aeruginosa elastase protease (37, 49). Both proteases depend on the T2SS for exit from the cell (11, 85). Using the T2SS to export proteins involved in degradation of the mucosal layer appears to be a common theme seen in bacteria that require access to epithelial cells for successful colonization and/or invasion.

YodA (ZinT)

Besides the protease StcE, the genome of O157:H7 contains at least one other T2S-dependent protein. This second T2S substrate, YodA (ZinT), has been shown to be essential for adherence of EHEC to cultured HeLa cell monolayers and to aid in the colonization of the infant rabbit intestines (51). The authors demonstrated YodA's (ZinT’s) dependence on the T2SS by replacing etpC, the first gene in the pO157-encoded etp operon, with an antibiotic cassette. The etpC mutant showed almost no detectable YodA (ZinT) in the supernatant, whereas this protein was readily identified in the supernatant from the wild-type strain (51). These results strongly suggest that YodA (ZinT) is a substrate of the EHEC T2SS.

YodA (ZinT) was originally identified in E. coli K-12 as a cadmium-induced protein that is translocated from the cytoplasm to the periplasm following exposure to cadmium stress (90). Besides transcriptional regulation by cadmium, YodA (ZinT) has been demonstrated to bind a variety of metal ions (including cadmium, mercury, nickel, and zinc) via a lipocalin-like domain (24, 61). Lipocalins are a family of small, usually secreted, proteins that are often involved in transport (39). Because the growth of the E. coli K-12 yodA (zinT) mutant was found to be impaired in zinc-limited conditions, and because zinc was one of the metals that bound to purified YodA (ZinT), it has been proposed that YodA (ZinT) may have a function in zinc homeostasis (61).

In E. coli K-12, YodA (ZinT) is localized first to the cytoplasm and then to the periplasm, but its secretion to the extracellular environment has not been observed, probably because the genes encoding the E. coli K-12 T2SS are usually transcriptionally silenced (41). In EHEC, not all YodA (ZinT) is extracellularly secreted, and it is not yet clear whether, like LT, the protein becomes associated with the outer membrane and/or becomes incorporated into vesicles. Also like LT, another protein has been implicated in the release of YodA (ZinT) in addition to the EHEC T2SS. Mutation of the protein AdfO (adherence factor encoded on CP-933O phage) reduced secretion of YodA (ZinT) and three other proteins in EHEC O157:H7 by at least 2-fold (51). As with the LT and LeoA relationship, it is unclear exactly what role AdfO is playing in YodA (ZinT) secretion.

DraD

In addition to the ability of pathogenic E. coli to secrete mucins and other degradative enzymes via the T2SS to adhere and invade mucosal epithelial layers, other proteins termed adhesins and invasins are also used in promoting infection. These proteins are often located on extended surface appendages known as fimbriae. Although each family of adhesin proteins is generally associated with a specific human disease, the Dr family of fimbriae from E. coli is a notable exception, as its members are associated with both diarrheal (DAEC) and urinary tract infections (UPEC) (71). Dr fimbriae are so named because of their ability to bind the Dr blood group erythrocytes through the decay-accelerating factor (CD55) (79).

Dr fimbriae are polymeric structures built from repeating monomers of the DraE fimbrial subunit and may have a minor fimbrial subunit, DraD, serving as the tip adhesin (7). Fimbrial subunits like DraE cross the inner membrane in an unfolded form via the Sec system. After interacting with a periplasmic chaperone, the subunits must be targeted to the outer membrane for the assembly into the fimbriae fiber, followed by transport of the fiber across the outer membrane to the cell surface via the chaperon/usher pathway. Assembly of Dr fimbriae does not require the tip adhesion DraD. In fact, examination of isolated Dr fimbriae suggests that most Dr fimbriae are made up of only DraE subunits, and DraD is surface located in structures independent from the Dr fimbriae (130).

When the chaperon/usher pathway was deleted in UPEC, DraD, unlike DraE, was still exported to the cell surface (131); suggesting that this adhesin uses a different system to translocate the outer membrane. With the exception of DraD, all fimbriae subunits have a highly conserved N-terminal extension that participates in subunit-subunit interactions. In contrast, DraD is synthesized as a protein precursor with an N-terminal cleavable signal sequence. In accordance with this observation, DraD, but not the other proteins encoded by the Dr fimbriae operon, has recently been shown to be secreted via the T2SS in UPEC. Inactivation of the T2SS by mutagenesis of the gspD gene blocked DraD transport to the cell surface (131).

STRUCTURAL AND FUNCTIONAL CHARACTERIZATION OF THE T2S APPARATUS

The characterization of the E. coli T2SS has been quite limited; however, their similarities with more well-characterized T2SS in other organisms are relatively high and thus much of the research done on other bacterial T2SS can translate into those of E. coli. Therefore, in this section we will review what is generally known about the workings of the T2S complex based on studies performed with other bacteria, including V. cholerae, K. oxytoca, and P. aeruginosa.

The T2S machinery is composed of at least 12 gene products (T2S:C-N, T2S:O, and any combination of T2S:A, B, and S) (Fig. 1), which probably assemble into a large protein complex spanning the entire cell envelope (Fig. 4). Once transported across the cytoplasmic membrane and folded in the periplasm, proteins to be secreted into the extracellular environment are believed to be pushed through a pore in the outer membrane termed the secretin (T2S:D). This is thought to be facilitated by the polymerization of a pseudopilus (consisting of T2S:G-K) using the energy provided by a cytoplasmic ATPase (T2S:E), which associates with the inner membrane protein complex (T2S:C, L, M, and F) (35, 98). The following sections will break down the individual components of the T2SS and review the current knowledge of how the components interact to form a functional apparatus.

Fig. 4Model of T2SS with LT. The T2SS comprises an ATPase (T2S:E), an inner membrane complex (T2S:C, F, L, and M), pseudopilins (T2S:G, H, I, J, and K), and an outer membrane secretin (T2S:D), which assemble into a complex spanning the entire cell envelope. Fully assembled LT (shown in ribbon structure) engages the T2S apparatus in the periplasm to allow for its extracellular translocation.

The ATPase

T2S is an energy-dependent process. Only one protein encoded by the T2S operon has been characterized to have ATPase activity and that is the cytoplasmic protein, T2S:E. T2S:E is related to a larger family of secretion ATPases, which are known to be involved in many activities including protein secretion, type IV pilus biogenesis, DNA uptake, and archeal flagellar assembly (87). The V. cholerae ATPase was one of the earliest T2S proteins to be crystallized, and from that work it was hypothesized to form a hexameric ring structure based on homologies to other characterized RecA-like ATPases (95). Later functional work supported this hypothesis by demonstrating that a fraction of T2S:E corresponding to the size of a hexamer is capable of hydrolyzing ATP (19). The ATPase activity of T2S:E was stimulated even further by the association of its N-terminal domain with the N-terminal cytoplasmic domain of the inner membrane protein, T2S:L (18); an interaction which is necessary to link T2S:E to the greater T2S complex (3, 89, 91, 99). T2S:E and the cytoplasmic domain of T2S:L themselves are monomeric when purified, but they cocrystalize as heterodimers (3). Although T2S:E and cytoplasmic domain of T2S:L copurify in a 1:1 ratio, the final stoichiometry of the T2S:EL complex in the functional T2SS remains to be determined.

The Inner Membrane Complex

Although proteins secreted through the T2SS engage the apparatus in the periplasm, the majority of the T2S components are found in the inner membrane. T2S:L is a bitopic membrane protein with a large cytoplasmic domain, a short membrane-spanning helix, and a smaller periplasmic domain, and it is believed to dimerize in the membrane (100). Aside from linking the cytoplasmic T2S:E to the membrane, it has been shown to interact with the other inner membrane proteins, T2S:C, M, F, and J (29, 72, 92, 94, 101, 118). The C-terminal periplasmic domain of T2S:L was recently crystallized as a dimer (1). When comparing the periplasmic T2S:L structure, it was found that its closest structural homolog is the periplasmic domain of another T2S inner membrane protein, T2S:M (1). Interestingly, periplasmic domain of T2S:M also forms a dimer (4, 59); however, the periplasmic domains of T2S:L and T2S:M are not homologous at the primary sequence level and their dimer arrangements are not the same (1).

T2S:L and T2S:M are known to form a stable complex that protects them from degradation (89, 92, 100), and it has been demonstrated that residues 84 to 99 in the periplasmic domain of T2S:M are necessary to interact with T2S:L residues 216 to 296 (59, 101). Unfortunately, structural information is not available for these regions in T2S:M and T2S:L; thus, it is still unknown how these proteins form a complex and the relative stoichiometry of said complex. When the periplasmic domain of T2S:M dimerizes, a hydrophobic cleft with a hydrophilic rim is formed, which may bind a yet undefined ligand or peptide. Because the cleft residues in T2S:M's are relatively well conserved, the cleft is likely important for secretion (4).

A third inner membrane protein, T2S:F, has also been shown to make up part of this inner membrane complex. T2S:F spans the membrane three times with its N-terminal domain in the cytoplasm and its C-terminal domain in the periplasm (2). Only the larger, first cytoplasmic domain has been crystallized and it was found to form dimers. The first 55 residues of T2S:F were too flexible to crystallize, and are therefore thought to potentially interact with other T2S components (2). Using two-yeast hybrids and coimmunoprecipitations, Py et al. (92) demonstrated the formation of an T2S:ELF complex in E. chrysanthemi, but this was conditional on T2S:E and L forming a complex first. Similarly, Arts et al. (9) demonstrated that the T2S:F homolog in P. aeruginosa is unstable unless in the presence of the T2S:E and L proteins. The authors determined the second cytoplasmic domain to be important for this stabilization, which is in contrast to the E. chrysanthemi studies; however, it is feasible that both cytoplasmic domains play a role in interaction with other T2S components (9, 92). The structure of the first cytoplasmic domain of T2S:F was recently solved, and a careful sequence comparison between the first and second cytoplasmic domains shows many similarities between the two that may translate into structural similarities as well (2). Interestingly, T2S:E binds in a pocket within the T2S:L cytoplasmic domain, as demonstrated by cocrystallizing the two proteins. T2S:E fills only a portion of this binding cleft, making it tempting to speculate that T2S:F may also bind within this same pocket in T2S:L (3). Future work is necessary to understand the precise interactions between the various domains of these proteins and other components of the inner membrane complex.

Connecting the Inner and Outer Membranes

At least one other T2S protein is associated with the inner membrane complex, T2S:C. T2S:C is a bitopic inner membrane protein with a periplasmic domain that is divided into two subdomains: the HR domain and either a PDZ or coiled-coil domain. T2S:C has been shown to interact and stabilize the inner membrane T2S:LM complex (44, 72, 118). The interaction of T2S:C's N-terminal 46 residues with the inner membrane proteins T2S:L and M protect them from degradation (72), and similarly, the presence of the T2S:LM complex seems to be necessary to protect T2S:C from degradation and may play a further role in stabilizing T2S:C in its proper conformation to support secretion (78). T2S:C homologs are known to be degraded in the absence of outer membrane protein T2S:D in K. oxytoca and P. aeruginosa (14, 88), and it was recently demonstrated that the periplasmic subdomains of T2S:C and D could be copurified, indicating a direct interaction between these two T2S components (65). Korotkov et al. (65) further determined that the HR domain is primarily responsible for interaction with T2S:D. As T2S:C spans the periplasm, it is a suitable candidate for substrate recognition and/or gating of the T2S:D pore. The second periplasmic subdomain, the PDZ domain, is the most likely candidate to interact with the secreted proteins, another T2S protein, or possibly both (65). Further research into this region of the protein is ongoing; however, it has been shown that the PDZ domain is required for the secretion of most proteins in E. chrysanthemi (15).

The Secretin

T2SSs have only one outer membrane protein, T2S:D, also termed the secretin, which is thought to oligomerize into a pore used in the export of proteins out of the bacterium (16). It is part of a larger secretin superfamily that consists of outer membrane proteins required for T2S, type IV pilus biogenesis, type III secretion, and filamentous phage extrusion. Most secretins require the use of a small lipoprotein termed the pilotin (T2S:S), which binds the C-terminal end of T2S:D and is required for proper positioning of T2S:D in the outer membrane (48, 105). Although many of the secretins that make up the larger family require a pilotin for proper insertion, not all T2SSs have an identified pilotin protein. In fact, only K. oxytoca and E. chrysanthemi have shown their respective T2S:S pilotin to be important for secretin insertion in the outer membrane; however, putative t2s:S genes in other T2S operons (including the E. coli pO157 plasmid-encoded T2SS) have been identified (98, 104). It is possible that the pilotin has not been identified because of limited sequence homology in other T2SSs, or there may be an alternative pathway by which the secretin is inserted in the outer membrane. In support of this latter point, Viarre et al. (119) showed that the secretin for the T2SS in P. aeruginosa, HxcQ, contains a lipid anchor, which is sufficient to target this protein to the outer membrane and allow for correct insertion. Regardless of the mechanism, once properly inserted, secretins are characterized by their ability to form large heat- and detergent-resistant oligomers. They are thought to form rings of 12 to 14 subunits containing a central channel ranging from 50 to 100 Å in diameter that is occluded by a central plug (13, 20, 82). The T2S:D protein has a well conserved C-terminal domain that is protease resistant and thought to be anchored in the outer membrane by amphipathic transmembrane β-strands (13, 46) as well as a more variable N-terminal domain which extends into the periplasm. Recently, Korotkov et al. (63) crystallized the N-terminal domain of the periplasmic domain of T2S:D from ETEC, and showed a region termed the N0:N1 lobe to be easily accessible and thus a good candidate for interacting with other proteins in the periplasm. These interactions may include other T2S components, such as T2S:C and/or T2S:J (29, 65), as well as proteins targeted to, and ultimately secreted by, the T2SS (15, 76, 106).

The Pseudopilins

On the basis of sequence homology between several T2S proteins and type IV pilus biogenesis components, it is believed that the T2S pathway may function similarly to the type IV pilus machinery, which supports the assembly of the pilus, as well as its extension and retraction (83). In particular, the T2S:G-K proteins have been termed the “pseudopilins” because of their sequence similarity to the type IV pilins, with both T2S pseudopilins and the type IV pilins containing a conserved N-terminal leader peptide that is required for transport across the cytoplasmic membrane. Following insertion into the cytoplasmic membrane, the N terminus of the pseudopilins is cleaved and methylated by the prepilin peptidase, which also shows a high degree of homology to the type IV prepilin peptidase (84). The relative abundance of T2S:G protein compared with the other pseudopilins has led to it being termed the major pseudopilin, whereas T2S:H-K proteins have been termed the minor pseudopilins (84).

Reports in several organisms have shown that the major pseudopilin component, T2S:G, can assemble into multimers, suggesting that T2S:G may be able to form a pilus-like structure or “pseudopilus” (55). Studies in K. oxytoca and P. aeruginosa have demonstrated that overexpression of the major pseudopilin T2S:G resulted in the presence of bundled surface pseudopili, further illustrating that the T2S pathway is able to assemble a pilus-like structure (31, 62, 120). In addition, the overexpression of T2S:G inhibited secretion through the T2S pathway (31). This result indicates that the pseudopilus may extend through the secretin in the outer membrane when overexpressed and that the formation of the surface pseudopili occludes the pore and prevents secreted proteins from passing through the outer membrane. These findings have led to a model where the pseudopilin components normally polymerize into a shorter periplasmic pseudopilus, which either acts as a piston to push secreted proteins through the outer membrane pore, or as a plug within the secretin that could depolymerize to allow proteins to be secreted.

The exact role of the minor pseudopilins is not known. The minor pseudopilins were found to be required for secretion; however, their overexpression did not inhibit secretion as shown with T2S:G in K. oxytoca and P. aeruginosa (89, 103). T2S:I was the only minor pseudopilin found to be required for the production of surface pseudopili when T2S:G was overexpressed (103). Conversely, deletion of T2S:K increased the length and number of the surface pseudopili observed following overexpression of T2S:G, and overexpression of T2S:K abolished the presence of the surface pseudopili (32, 120). These results suggest that T2S:I may be an initiation factor required for the pseudopilus to form, and that T2S:K may be a termination factor that stops polymerization of the pseudopilus. Deletion of T2S:H and J had no effect on the production of surface pseudopili, indicating that one or both of these proteins may act as a chaperone for the other pseudopilins.

In recent years, the structures for all five pseudopilins have been determined, allowing a better view into what role these proteins may be playing in T2S (62, 64, 126, 127). The structures for the pseudopilins further confirmed the homology with the type IV pilins, with both groups containing an N-terminal α-helix, a variable domain, and a C-terminal antiparallel β-sheet that completes the pilin fold. However, despite this core homology, a great degree of dissimilarity was found within the variable domain between the pseudopilins. Both T2S:H and J have large, elaborate variable domains (126, 127), whereas T2S:I and K have smaller variable regions (64). T2S:K is the largest of the pseudopilins and contains a unique α-domain that is absent in all other pseudopilins. This unique α-domain was found to contain a dinuclear metal binding site and a pair of cysteine residues that form a disulfide bridge (64).

The structure for T2S:G has been resolved for K. oxytoca, P. aeruginosa, V. cholerae, Vibrio vulnificus, and EHEC; however, slight variation occurred among the structures from these species (5, 62, 63). K. oxytoca illustrated a C-terminal β-strand that was observed in a helical conformation in the other species (62). This difference is believed to occur because of a β-strand swap occurring during the K. oxytoca crystallization process. Interestingly, Korotkov et al. (66) found a calcium-binding loop within the C terminus of T2S:G in three organisms: V. cholerae, V. vulnificus, and EHEC. Furthermore, the sequence of the calcium-coordinating side chains was highly conserved among these T2S:G homologs. Disruption of the amino acids coordinating the calcium ion inhibited T2S in V. cholerae, illustrating that calcium is playing an important role during secretion, perhaps acting as a stabilizing factor for T2S:G or within the assembled pseudopilus (63).

The structures for the minor pseudopilins also gave insight into their interactions with each other. T2S:I and J initially crystallized as a heterodimer and were subsequently shown to form a ternary complex with T2S:K (30, 64, 127). It has been suggested that the T2S:IJK complex may be positioned at the tip of the pseudopilus and therefore serves as a cap to the pseudopilus. This hypothesis further supports the finding that T2S:K may act as a termination factor for pseudopilus growth, because the bulky α-domain of T2S:K would prevent the pseudopilus from being able to pass through the secretin pore.

In addition to the structural data, both genetic and biochemical approaches have been utilized to uncover interactions between the pseudopilins. T2S:G was found in heterodimers with T2S:H, I, and J. T2S:H was found to be associated with T2S:G, I, and J, and the absence of T2S:H disrupted the observed T2S:GJ and T2S:IJ interactions. In the absence of T2S:I, T2S:G and H were no longer associated (68). In addition to the tertiary complex of T2S:IJK seen in 3D structures, the existence of a quaternary complex of all four minor pseudopilins has also been shown (30). Furthermore, yeast two-hybrid assays showed that both T2S:H and J can form homodimers and also pointed to a T2S:IJ interaction (29). These findings suggest that the interactions between the pseudopilins are complex and that each individual pseudopilin is playing an important role in the formation of the pseudopilus.

Less is known about the interactions between the pseudopilins and the other components of the T2S complex. Yeast two-hybrid data indicated an interaction between T2S:J and the secretin, T2S:D. This same assay also indicated an interaction between T2S:J and the inner membrane protein T2S:L (29). These data support the hypothesis that the pseudopilins are polymerized in the inner membrane as they build into the pseudopilus, and that T2S:J is part of a complex with T2S:I, H, and K at the pseudopilus tip, which could then interact with the outer membrane.

Stoichiometry of the T2S Complex

Although much is known about the interactions between the various proteins of the T2S apparatus, the relative stoichiometry of the individual components remains elusive. It is probable that the cytoplasmic component, T2S:E exists as a hexamer, while T2S:D is found in an oligomer of 12 or more proteins. Most of the inner membrane components analyzed to date appear to form dimers, although whether they remain so in a functional apparatus has yet to be determined. As the inner membrane proteins are likely interacting with each other (either transiently or in a stable complex), it is anticipated that heterocomplexes form in the membrane. The relative combinations of each protein in those complexes remain in question, however. Likewise, if T2S:C is able to interact with proteins both in the inner membrane and the outer membrane, it seems likely that it would need to conformationally stretch between a 6-fold symmetry in the inner membrane and a larger 12+-fold symmetry in the outer membrane.

Finally, some of the interactions between the T2S proteins are probably transient and it is unknown whether the complex actually exists as a stable entity at any given time. One important question to address is how the various components come together across the entire gram-negative cell envelope to form a functional apparatus. It is known that T2S:C is a key protein in bridging the inner membrane complex to the outer membrane secretin. In addition, Lybarger et al. (78) determined that T2S:D is required for proper localization and focal assembly of the T2S complex in the membrane. These data, in combination with results from the K. oxytoca T2S:D homolog, demonstrate that the secretin can focally localize in the outer membrane in the absence of other T2S components (17) and suggests that T2S:D may be the driving force behind T2S apparatus assembly. Furthermore, T2S:C appears to be able to interact with T2S:D in the absence of any other T2S components and therefore could direct the T2S:L and M proteins (and presumably the rest of the inner membrane proteins) to where the secretin is located (78).

FUTURE DIRECTIONS

Secretion Signal

The interaction between the secretion system and its substrate is an obvious important step in the secretion process. How are proteins that are to be secreted distinguished from periplasmic proteins? Does a specific member of the T2S apparatus recognize substrates? Examination of many different substrates and secretion machineries reveal that there is no simple answer to these questions. A specific secretion signal or motif has not yet been discovered for T2S substrates. Because T2S substrates are likely folded into tertiary or quaternary structures when recognized by the secretion apparatus, it is currently thought that the targeting information probably resides in the final or near-final conformation of the protein. It is only after the folding of the substrate occurs in the periplasm that the secretion system is thought to recognize proteins designated for export into the extracellular environment. A possible result of this is that the secretion signal need not be a linear string of amino acids in the primary sequence of the secreted protein. Alternatively, the sequence may be linear, but only can be recognized by the apparatus when folded into a final conformation.

Some success for identifying the minimum sequence needed for secretion has been made, however. For LT of ETEC, all of the information for secretion is included in the B pentamer portion of the holotoxin. In addition to secretion of B pentamers of cholera toxin and the close homolog LT (50), V. cholerae is capable of secreting the B pentamer of the LT variant, LTIIb, which is less than 10% homologous to CT (23). When the 3D structures of the pentamers are examined, they appear structurally very similar, further substantiating the idea that the secretion signal is not found at the primary sequence level alone (Fig. 2). The 3D structures of several T2S substrates are now available. Even with this information, however, a clear structural motif that is shared between T2S substrates has not been visualized. Clearly, much more work is needed before any conclusions can be made about T2S signals.

Future Challenges

The gene products that assemble into the T2S apparatus function both in pathogenic and nonpathogenic bacteria, where they play a crucial role in the survival and persistence of many gram-negative bacteria. A challenge for the future is to uncover how these products create a large, highly organized, complex capable of secreting a variety of fully folded proteins across the outer membrane of gram-negative bacteria. To do this, numerous interactions between T2S proteins still need to be better understood. Recent structural characterizations along with classical protein-protein interaction experiments are important steps toward completing our picture of how the T2SS assembles and functions in vivo. Further investigation of the stoichiometry and the attempt to define the ordered assembly of the T2S complex should also lead to insights into the mechanism of secretion used by these exported proteins.

While only a small number of T2S-dependent substrates are currently known in E. coli, it is likely that, as has been discovered in other gram-negative bacteria, many more are encoded in the genomes of these organisms. The ability to sequence entire genomes of bacteria inhabiting diverse environments around the globe will also lead to the ability for researchers to explore the role of the T2SS in both environmental, as well as pathogenic, bacteria. Investigation of the T2S secretomes of gram-negative bacteria, including E. coli, will likely increase not only the number of T2S-dependent proteins, but also the types of proteins exported through the system. It is also likely that secreted proteins that do not share any homology to known proteins, and that therefore could be novel substrates secreted by the T2SS, will also be discovered.

Another exciting possibility that can be addressed by genomic sequencing is the existence of variations of the T2SS. One such system, Stt (for second type II), has already been reported in the plant pathogen Dickeya dadantii. Ferrandez and Condemine (34) identified a cluster of genes in this bacterium encoding a second T2SS that contained homologs of all the T2S proteins except T2S:B, H, and O. pnlH, a gene encoding a homolog of the E. carotovora pectinase, was located immediately downstream of the cluster and was therefore considered a candidate for secretion via this second T2SS. In wild-type cells, PnlH was translocated across the inner membrane via the Tat system and apparently targeted to the outer membrane by its uncleaved N-terminal signal peptide. When examined in a stt mutant or in E. coli, PnlH was localized to the outer membrane but could not be detected on the outside of the cell (34). Thus, this alternative T2SS appears to be responsible for translocating PnlH to the extracellular surface of the outer membrane, although it is not responsible for targeting the protein to the outer membrane per se.