Autotransporter Proteins

IAN R. HENDERSON¹ AND JAMES P. NATARO^2*

[SECTION EDITOR: ALISON O’BRIEN]

Posted November 1, 2005

Bacterial Pathogenesis and Genomics Unit, Division of Immunity and Infection, University of Birmingham, Birmingham B15 2TT, United Kingdom,¹ and Center for Vaccine Development, Department of Pediatrics, University of Maryland School of Medicine, 685 W. Baltimore St., Baltimore, MD 21201²

*Corresponding author. Phone: (410) 706-8442, Fax: (410) 706-6205, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

AUTOTRANSPORTER SECRETION

Translocation of proteins outside gram-negative bacteria necessitates passage through the inner membrane (an energy-dependent process), the periplasm, and the outer membrane. This formidable series of obstacles can be overcome by sophisticated biological processes. In fact, the magnitude of this challenge is illustrated by the fact that only a few solutions to the problem of the gram-negative barrier have been identified, yet with typical resourcefulness, vast numbers of gram-negative species have exploited these few mechanisms via horizontal genetic exchange. Currently, five major gram-negative secretion systems (numbered I to V) are recognized (42). This classification is based on similarities among the components of the systems and the mechanisms of translocation across the outer membrane. The fundamental processes of each system are described elsewhere in this volume.

Perhaps the simplest secretion mechanisms are those we include under the rubric of type V secretion. This category comprises proteins secreted via the autotransporter system (type Va), the two-partner secretion pathway (type Vb), and the recently described AT-2 system (type Vc) (42). The unifying characteristic of the type V secretion family is that all the secreted proteins are translocated across the outer membrane via a transmembrane pore formed by a β-barrel structure (43, 99). In addition, from homology comparisons and the similarities in the gross anatomy of the autotransporter and two-partner secretion pathways, it appears that these two subfamilies are related (37). The Yersinia adhesin YadA has recently been presented as the prototype of a novel class of bacterial adhesin belonging to the AT-2 subfamily which, due to their outer membrane topology, could be considered as surface-attached oligomeric autotransporters; this pathway has thus been designated type Vc (45, 73, 89). Figure 1 provides a schematic overview of the type V secretion pathways.

Fig. 1Schematic overview of the type V secretion systems. The secretion pathway of the autotransporters proteins (type Va) is depicted on the bottom left of the diagram, the two-partner system (type Vb) is in the center of the diagram, and the Oca family (type Vc) is on the right of the figure. The four functional domains of the proteins are shown: the signal sequence, the passenger domain, the linker region, and the β-domain. The autotransporter polyproteins are synthesized and generally exported through the cytoplasmic membrane via the Sec machinery. Effector proteins with an unusual extended signal sequence, which purportedly mediates Srp-dependent export, are found in all three categories of type V secretion. Once through the inner membrane, the signal sequence is cleaved and the β-domain inserts into the outer membrane in a biophysically favored β-barrel structure that forms a pore in the outer membrane. After formation of the β-barrel, the passenger domain inserts into the pore and is translocated to the bacterial cell surface where it may or may not undergo further processing. Figure courtesy of Henderson et al. (42).

Pohlner et al. (85) were the first to propose a model for the autotransporter secretion pathway based on the relationship between the gene encoding the gonococcal immunoglobulin A1 (IgA1) protease and its extracellular product. Since this initial description many more proteins that follow this pathway of secretion have been identified among the gram-negative bacteria (41, 42). In all cases the primary structure of the autotransporter protein is reminiscent of the gonococcal IgA1 protease, comprising four modular domains (47). Proteins secreted by this pathway possess (i) an N-terminal signal sequence for inner membrane translocation, (ii) a functional passenger domain that can be surface exposed or released into the extracellular milieu and represents the mature species, (iii) a linker region necessary for translocation of the passenger domain through the outer membrane, (iv) a C-terminal region involved in the formation of a transmembrane pore. This chapter will focus on the function of the Escherichia coli and Salmonella autotransporters for which a considerable amount of literature is available.

The precise events required for autotransporter translocation through the outer membrane are the subject of intense scrutiny. The original model for outer membrane translocation comprised a threading of the N-terminal domain through a monomeric outer membrane channel formed by the β-domain. Several investigators reported that large heterologous passenger species were unable to undergo outer membrane translocation. Recently, however, Veiga et al. (101, 102) reported that a translocation-incompetent passenger species was able to impede the translocation of competent species when coexpressed in the same cell. This report was accompanied by high-resolution electron photomicrographs demonstrating the formation of hexameric ring structures by the purified β-domain. The investigators concluded that the outer membrane channel formed by the β-domain was actually a ring of (perhaps six) β-barrels; the proposed model implies that translocation of the passenger species would occur through the pore formed by the oligomeric ring.

The original model, that autotransporter outer membrane translocation takes place via a monomeric β-barrel, was revived on publication of the crystal structure of the Neisseria NalP β-domain (77). The protein was derived by refolding of a purified β-domain lacking the NalP passenger; biophysical studies suggested that the protein was folded into its native state. The crystal structure revealed a monomeric β-barrel with the C-terminal α-helix of the NalP passenger locked inside, in near perfect agreement with the initial model of autotransporter secretion as presented by Pohlner et al. (85). However, several aspects of this crystal structure are not fully compatible with experimental observations. Prominently, the dimensions of the β-barrel pore were calculated to be smaller than 2 nm, a figure predicted by electrophysiological studies and by calculation of the sizes of proteins and folded motifs that are able to pass. Thus, a pervasive sense in the field is that essential elements of autotransporter secretion are as yet unrevealed. Two models have been offered to reconcile these disparate observations. Tommassen has suggested that the translocation of passenger species requires the Omp85 complex, which has been shown to mediate membrane insertion of other outer membrane proteins (105). It is possible that Omp85 provides an alternative pore instead of or in addition to that of the β-domain. An alternative model entails a concurrent folding and insertion of the β-domain, predicating a transiently larger pore during the insertion and folding process. Reconciliation of these disparate models requires further experimental data.

The function of the C-terminal end of the passenger has also received some attention. Observations by Dutta et al. (23), Velarde et al. (103), and Oliver et al. (75, 76) suggest that proper folding of the C-terminal end of the passenger is necessary for outer membrane translocation of the full passenger species. Moreover, several groups have shown that some passenger mutations that are incompetent for outer membrane translocation can be rescued by native species (23, 74, 76). These observations imply a special contribution for the folded structure of the passenger C-terminal region, perhaps serving as a chaperone for the N terminus, a scaffolding for N-terminal folding, and/or a function in recognition of the outer membrane pore in the process of translocation.

FUNCTIONS OF AUTOTRANSPORTER PROTEINS

The first autotransporter to be described for either Salmonella or E. coli was the Tsh serine protease of an avian pathogenic strain of E. coli (86). Additional autotransporters were soon recognized, but genomic efforts have spurred a vast expansion in the number of known or predicted autotransporter species. Indeed, examination of the finished and unfinished E. coli and Salmonella genomes has revealed a plethora of autotransporters that can be separated into three distinct categories on the basis of their functional passenger domains: (i) the serine protease autotransporters of the Enterobacteriaceae (SPATEs), (ii) adhesins homologous to the AIDA-I family, and (iii) lipases belonging to the GDSL family of serine proteases (see Table 1).

Table 1Functional classification and distribution of the E. coli and Salmonella autotransporters

SPATEs

Members of the SPATE family are proteins from E. coli and Shigella spp., which, like the Neisseria and Haemophilus influenzae IgA1 proteases and Hap, possess a consensus serine protease motif (43). Tsh was the first of the SPATEs to be described. Since the description of Tsh, numerous of investigators have described SPATE proteins in the different pathotypes of E. coli and in Shigella. SPATE proteins have several features in common: (i) all possess an unusual extended signal sequence; (ii) the serine protease active site is reminiscent of a chymotrypsin clan protease; (iii) unlike the IgA1 proteases, none of the SPATE family has been shown to cleave IgA1; (iv) the serine protease motif of SPATE proteins does not have a role in cleavage of the passenger domain from the β-domain; (v) the points of cleavage of passenger domains from the β-domains are identical; (vi) each SPATE member is among the predominant secreted proteins of their respective pathogens; (vii) SPATEs are highly immunogenic proteins. While the full contribution of these proteins to pathogenesis remains elusive, and no universal contribution has been suggested, specific phenotypes have been reported for various members of the SPATE family (see below).

Evolution of the SPATEs.

To date no SPATE has yet been characterized in a nonpathogenic organism. However, E. coli is a highly clonal species represented by four major phylogenetic groups (A, B1, B2, and D), based on techniques such as multilocus sequence typing or multilocus enzyme electrophoresis. Most extraintestinal pathogenic E. coli are derived from the B2 phylogenetic group, with group D contributing the second largest group of extraintestinal clones. In contrast, diarrheagenic pathotypes occur almost exclusively in the A, B1, D, and ungrouped phylogenetic clusters. A recent examination of the E. coli reference (ECOR) collection revealed that SPATE proteins were clustered in the B2, D, and a subgroup of the A phylogenetic branches (Fig. 2) (80a). Many of the strains comprising the ECOR collection have been isolated from healthy people or animals. Furthermore, the absence of the SPATEs from the majority of the B1 strains, and the remainder of the A subgroup, appears unusual despite the fact that at least one SPATE, and in many cases several, have been identified in all the diarrheagenic and extraintestinal pathovars of E. coli. In addition, genome sequencing has revealed the presence of a SPATE protein, termed Boa (for bongorii autotransporter), in Salmonella bongorii (80a). To date this remains the only species of Salmonella to possess a SPATE protein.

Fig. 2Distribution of SPATEs among the ECOR collection. Phylogenetic tree of the ECOR isolates showing the distribution of vat- and SPATE-encoding genes where each locus is represented by a shaded box as indicated in the figure. The number of the ECOR isolate is given in boldface, and each of the major phylogenetic branches is indicated. The complete complement of virulence-associated loci is preferentially associated with the B2 phylogenetic cluster. Figure adapted from Parham et al. (80a).

Despite their high levels of homology, the SPATE proteases demonstrate distinct substrate specificities (22). Previously, Henderson et al. (43) suggested that the SPATE family of autotransporters could be divided into two groups: one demonstrating cytopathic activity and exemplified by Pet of enteroaggregative E. coli (EAEC), the other exhibiting preference for extracellular targets, as illustrated by Pic of EAEC and Shigella. Recently, Dutta et al. (22) explored the phylogenetic relationships of a subset of SPATE proteins using split decomposition analyses of the complete passenger domains. Whereas this investigation was unable to establish a correlation between the phylogenetic groupings and the biological activity, it did reveal evidence of significant homologous recombination among family members. In addition, highly similar proteins such as Pet and Sat (53% identity) were not found to share peptide specificities despite their common ability to cause cytopathic effects on cultured cells and to cleave spectrin, a component of the mammalian cell cytoskeleton.

Since this initial study several additional members of the SPATE subfamily have been described for E. coli, including EspI, EatA, EpeA, EaaA, EaaC, and Boa. Split decomposition phylogenetic analyses incorporating these new members of the SPATE subfamily of autotransporters revealed that the original bifurcating phylogenetic pattern proposed by Henderson et al. (43) continues to be valid for the passenger domains (Fig. 3) (Henderson et al., unpublished data). However, the same evolutionary pattern does not hold for the β-domains; EpeA and EspP have virtually identical β-domains, yet the corresponding passenger domains reside on separate branches of the bifurcating phylogram (Fig. 3). These analyses reinforce the evidence for significant homologous recombination among the passenger and β-domains. These analyses have also suggested horizontal gene transfer of Boa from E. coli to S. bongorii.

Fig. 3ClustalX phylograms of amino acid sequence alignments of full-length SPATE passenger domains (A) or the β-domains (B). Trees were further tested for reliability using bootstrap analysis, yielding results of 96.9% (A) and 99.4% (B). In panel A the known substrates or effect for each SPATE is indicated.

Tsh and Hbp.

The use of cosmid screening and subcloning to analyze adherence mechanisms of an avian pathogenic E. coli (APEC) that causes disseminated infection in birds led to the identification of Tsh (for temperature-sensitive hemagglutination) (86). The predicted amino acid sequence of the secreted Tsh protein was shown to be homologous to four serologically distinct H. influenzae iga genes and with Neisseria gonorrhoeae IgA1 protease, indicating that this protein belonged to the autotransporter family (86). Further studies demonstrated that E. coli K-12 expressing a recombinant version of the tsh gene produced a 106-kDa extracellular protein and a 33-kDa outer membrane protein, a phenomenon consistent with the IgA1 protease autotransporter pathway.

Subsequently, a novel heme-binding protein, termed the hemoglobulin-binding protease (Hbp), was characterized in an E. coli strain isolated from a human wound infection (79). Hbp is 99.9% identical with Tsh, with only two amino acid differences in the passenger domain. In both cases the tsh and hbp genes are located on a ColV-type plasmid, near the colicin V genes (21). For the purposes of this chapter Tsh and Hbp are considered to be the same protein.

The region of Tsh that is homologous to the serine-type IgA1 proteases is limited mainly to the 900 N-terminal amino acid residues of Tsh (86), and like the IgA1 proteases Tsh possesses a 7-amino-acid serine protease motif that includes the active site serine (S²⁵⁹). Site-directed mutagenesis of the serine protease motif abolished protease activity but not hemagglutinin activity or, in contrast to the IgA1 proteases, the extracellular secretion of the passenger domain (86). Furthermore, correlation of Tsh expression and hemagglutination activity appears to be a very complex phenomenon, influenced by strain and environmental conditions; for both APEC and recombinant E. coli K-12 strains expressing the tsh gene, only the whole bacterial cells and not the cell-free supernatant exhibited hemagglutination (96). The contribution of Tsh to virulence is unknown. Tsh does not cleave human or chicken IgA (96). Otto et al. (79) have reported that that Hbp interacts with and degrades human hemoglobin, then subsequently binds the released heme. Dutta et al. (22) demonstrated that Tsh possessed the ability to cleave mucin and human coagulation factor V (Fig. 4). Dozois et al. (21) found that experimental inoculation of chicken with the wild-type E. coli strain χ7122 and an isogenic tsh mutant demonstrated that Tsh may contribute to the development of lesions within the air sacs of birds but is not required for subsequent generalized infection manifesting as perihepatitis, pericarditis, and septicemia. More recent results suggest the synergy of abscess formation by E. coli and Bacteroides fragilis can in part be explained by the capacity of B. fragilis to intercept Hbp and iron from heme to overcome the iron limitations imposed by the host (78).

Fig. 4Substrate specificity of the SPATE proteins. (A) Cleavage of protein substrates by SPATE proteins. I, Spectrin. One microgram of spectrin purified from sheep red blood cells was incubated with 1 μg of each SPATE protein overnight. The reaction products were separated on SDS–6% PAGE gels. II, Pepsin. Three micrograms of purified pepsin was incubated with 1 μg of each SPATE for 1 h at 37°C. The reaction products were separated on SDS–12% PAGE gels. III, Human coagulation factor V. A total of 2.5 μg of purified coagulation factor V was combined with 2 μg of each SPATE protein in a 40-μl total volume and incubated overnight. (B) Cleavage of bovine submaxillary mucus. Ten micrograms of each SPATE was incubated overnight at 37°C on medium containing 1.5% agarose and 1% bovine submaxillary mucus. Mucin was stained with 0.1% amido black. (C) Effects of SPATEs on HEp-2 cell monolayers are shown by oil immersion light microscopy of Giemsa-stained HEp-2 cells after treatment with SPATE proteins at 500 nM for 5 h. Rounding of cells is shown with Pet and Sat (arrows). Figure adapted from Dutta et al. (22).

Vat.

Recently, it was reported that certain avian pathogenic strains of E. coli induced a phenotype similar to that associated with the vacuolating cytotoxic activity of VacA, the major Helicobacter pylori autotransporter protein (91). In an effort to identify the factor responsible for this vacuolating phenotype, Parreira and Gyles (82) looked for SPATEs in a strain of E. coli (Ec222) isolated from a septicemic chicken. This approach revealed the presence of a gene encoding a SPATE protein with high homology to Tsh (75% amino acid identity). By using chromosome walking techniques, these researchers demonstrated that Vat was encoded on a pathogenicity island inserted adjacent to the thrW tRNA gene. The presence of vat at this position has been demonstrated for the neonatal meningitis strain E. coli RS218 (Henderson et al., unpublished data) and the uropathogenic E. coli strains CFT073 and 536 (94, 106). Henderson et al. (unpublished data) have found that Vat may be a common virulence factor of extraintestinal pathogenic E. coli. It is worth noting as well that Heimer et al. (36) recently characterized Vat in E. coli CFT073, though in their study they designated the protein Tsh. For the purposes of this chapter, and to distinguish it from the Tsh/Hbp protein described above, the protein will be referred to as Vat.

Culture of wild-type and vat null mutants of E. coli Ec222 in the presence of chicken embryo fibroblasts demonstrated that vat was necessary and sufficient for the vacuolating phenotype of this septicemic strain of E. coli. Further investigations using purified Vat demonstrated that this protein also possesses the ability to induce vacuolation in chicken embryo fibroblasts as assessed by staining with neutral red (82). Vat remains the only member of this branch of the SPATEs (see Fig. 3) to demonstrate cytopathic activity.

The importance of Vat in the infectious process was demonstrated by using two routes of infection in the chicken model: inhalation and wound inoculation. In both models the wild-type strain was significantly more virulent that the isogenic vat mutant as assessed by mortality, lesion formation, and recovery of the bacteria from the infected chickens (82). Heimer et al. (36) demonstrated by reverse transcriptase PCR that vat was expressed in mice during experimental urinary tract infection, but the contribution of Vat to infection in this model was not reported.

Pic and PicU.

In 1998, a SPATE termed Pic was described as a secreted 116-kDa protein recognized by sera from children who were infected during an outbreak of EAEC in Mexico (71; C. Eslava, J. M. Villaseca, M. R. Morales, A. Navarro, and A. Cravioto, Abstr. 93rd Gen. Meet. Am. Soc. Microbiol. 1993, abstract number B-105, p. 44). Henderson et al. (38) found that Pic was identical with a protein previously termed ShMu (shigella mucinase), which is encoded on the Shigella she pathogenicity island (87). The protein was renamed Pic (protease involved in intestinal colonization) due to its broader distribution. Like Vat, Pic is encoded on the chromosome. However, the sequences flanking pic are different in EAEC and Shigella flexneri. In S. flexneri the she pathogenicity island is inserted at the pheV tRNA between the genes yqgA and yghK; the site also encodes another SPATE, called SigA (see below). In EAEC the pic pathogenicity island is inserted at pheU and lacks the sigA gene (80). These data suggest that the pic gene can undergo horizontal gene transfer independent of the pathogenicity island.

Purified Pic catalyzes gelatin degradation, and degradation can be abolished by disruption of the predicted proteolytic active site. Moreover, functional analysis of purified Pic implicated this factor in mucinase activity, serum resistance, and hemagglutination; these activities were not observed for the mutant form possessing a defective proteolytic site (38). Further studies demonstrated that Pic could also cleave pepsin, human coagulation factor V, and spectrin in a manner similar to the cytopathic toxins EspP and Pet (see below) (22). However, no cytopathic effects, similar to those induced by EspP and Pet, have been observed even after prolonged exposure of mammalian cells to the purified Pic protein. Harrington et al. have shown that a pic mutant in EAEC strain 042 is outcompeted by wild-type 042 by a factor of over 3 logs in the streptomycin-treated mouse model, though the precise mechanism of this effect is not known (S. M. Harrington, I. R. Henderson, and J. P. Nataro, unpublished data).

The pic gene has a unique characteristic among the autotransporter proteins: embedded within the pic (she) gene on the opposite strand is encoded the ShET1 enterotoxin, which comprises five 7-kDa B subunits and one 20-kDa A subunit (29). Analysis of ShET1 expression in a simulated intestinal fermentor using lacZ fusions to set1B demonstrated that expression of ShET1 was significantly increased under conditions similar to those of the human intestine (4).

Recently PicU, a homologue of Pic (96% identity at the amino acid level), was characterized in uropathogenic E. coli. Parham et al. (81) demonstrated that Pic and PicU possessed similar functional activities in in vitro assays, including the ability to degrade mucin, spectrin, pepsin, and coagulation factor V. Subsequent investigations by Heimer et al. (36) revealed that picU was expressed during experimental infection in the mouse model of urinary tract infection. However, cochallenge of wild-type organisms and the picU mutant in the mouse model suggested that PicU-encoding strains had no competitive advantage in mixed infection. Nevertheless, because PicU is a secreted protease, it is possible that protein secreted by the wild-type strains may complement the picU deficiency of the mutant strains. Indeed, challenge of mice separately with wild-type and picU-mutant strains revealed that the wild-type parent colonized the bladder to a greater extent and displayed higher levels of neutrophil infiltration into the lumen, epithelium, and submucosa; however, while these data showed a trend they were not statistically significant (P = 0.2). Thus, the role of PicU in uropathogenic E. coli infection remains enigmatic.

SepA.

Benjelloun-Touimi et al. (5) described a Tsh homologue that is a major extracellular protein of S. flexneri (designated SepA for Shigella extracellular protein). SepA, like all members of the SPATEs, possesses a serine protease active site, but a natural substrate has not been found for SepA. Nevertheless, investigation of the proteolytic activity of this protein using a wide range of synthetic peptides conjugated to p-nitroanilide found that SepA hydrolyzed several of these substrates and the activity was inhibited by the serine protease inhibitor phenylmethylsulfonyl fluoride (PMSF) (6). Several hydrolyzed oligopeptides have been described as specific substrates for cathepsin G, a serine protease produced by polymorphonuclear leukocytes, which has been proposed to play a role in inflammation. However, unlike capthesin G, SepA degraded neither fibronectin nor angiotensin I and had no effect on aggregation of human platelets.

The presence of sepA on the virulence plasmid of S. flexneri serotypes 2a and 5, as well as the recognition of SepA by the sera of monkeys infected with Shigella, suggest that SepA might be involved in Shigella pathogenicity (5). However, SepA is absent from S. flexneri serotype 6. Characterization of a sepA mutant indicated that SepA was not required for entry into cultured cells or for intracellular dissemination. Nevertheless, a sepA mutant demonstrated a reduced ability to induce both tissue inflammation and mucosal atrophy in the rabbit ligated ileal loop model. Such results indicate that SepA may have a role in tissue invasion, although this hypothesis remains to be proven (5). However, the presence of the other SPATE autotransporters Pic (see above) and SigA (see below) may provide an element of redundancy and complementation such that the true contribution of SepA to infection was not apparent in these studies.

EatA.

In the classic enterotoxigenic E. coli (ETEC) paradigm, the pathogen adheres to the small intestinal mucosa by virtue of specific colonization factors (CFAs) and subsequently elaborates heat-labile, heat-stable, or both enterotoxins in the small intestine, causing fluid secretion without significant mucosal inflammation. For many years these factors were considered to be solely responsible for ETEC-mediated disease. However, recent investigations have suggested that the pathogenesis of ETEC infection is more complex than originally proposed. Thus, toxin-negative mutants can cause residual diarrhea in volunteers, and several studies have demonstrated an increase in fecal leukocytes, elevated fecal cytokines, and lactoferrin (68). The implication of these studies is that additional virulence factors are required for ETEC disease, and several additional virulence factors, including the autotransporters TibA (see below) and EatA, have recently been described (25, 57, 83).

EatA is a member of the SPATE family of autotransporters and is encoded on the large virulence plasmid of the classic ETEC strain H10407. Mature EatA is a 110-kDa extracellular protein with 70% amino acid identity to the S. flexneri SPATE SepA (see above). EatA demonstrates the same specificity as SepA for p-nitroanilide-conjugated oligopeptides, cleaving peptides that have been identified as substrates for cathepsin G. Site-directed mutagenesis of the residues within the predicted serine protease catalytic triad abolished the ability of EatA to cleave the oligopeptides. Furthermore, these mutations had no effect on the secretion of EatA (83).

Comparison of wild-type ETEC H10407 and its isogenic eatA mutant in a rabbit ileal loop model of infection indicated that there were no differences in the quantity of fluid accumulated 16 h postinfection. In contrast, fluid accumulation was not as pronounced in the eatA mutant 7 h postinfection when compared with wild type, despite the fact that similar numbers of bacteria could be recovered from the loops (83). These data suggest that the differences were not due to bacterial numbers or an altered survival rate. After 7 h postinfection the eatA mutant did not demonstrate the focal areas of mucosal destruction and leukocyte infiltration observed for the wild-type organism (83). These data suggest that while EatA is not absolutely required for infection it may act to accelerate ETEC virulence.

Pet.

In 1998, a second SPATE member was described in EAEC. This 108-kDa protein was recognized by the sera from the same population of Mexican children that recognized Pic (73; Eslava et al., Abstr. 93rd Gen. Meet. Am. Soc. Microbiol. 1993). In contrast to Pic, Pet (plasmid-encoded toxin), encoded by the EAEC virulence plasmid pAA, does not possess the ability to degrade mucin (27). Other investigations revealed that Pet had the ability to degrade casein, gelatine, factor V, and pepsin and that this activity depended on the serine protease active site (22).

As mentioned previously a feature of EAEC infection is mucosal damage, and indeed several children who succumbed during the Mexican outbreak demonstrated necrotic lesions on their intestines. This mucosal damage was assumed to be the result of one or more elaborated cytotoxins. Navarro-García et al. (72) found that purified Pet induces temperature-, time-, and dose-dependent cytopathic effects on HEp-2 and HT29/C1 cells. Further investigations revealed that Pet acts as a cytoskeleton-altering toxin, inducing contraction of the cytoskeleton, dissolution of actin microfilaments, and release of the cellular focal contacts in cell monolayers, followed by complete cell rounding and release of the cell from the substratum. In addition, these cytopathic activities depend on the serine protease activity of Pet, as they are inhibited by protease inhibitor PMSF and are not induced by Pet S260I, a serine protease active site mutant (72). However, despite the cytopathic effects observed for Pet, it was unclear whether these effects were due to general proteolytic activity acting outside the cell and mediated by the Pet serine protease active site, or whether this was a specific targeted effect.

Subsequent investigations revealed that Pet was internalized by HEp-2 cells and that this specific uptake was required for the induction of cytopathic effects (Fig. 5). Uptake of Pet into mammalian cells appears to occur via a retrograde transport mechanism, because treatment of cells with brefeldin A inhibits the cytopathic and cytoskeletal effects observed when cells are treated with native Pet. The use of Western blotting and confocal microscopy demonstrated that the serine protease activity was not required for uptake because both native Pet and Pet S260I could be observed inside the epithelial cells (69). Furthermore, using a scanning-linker mutagenesis approach Dutta et al. (23) were able to demonstrate that the Pet passenger domain could be divided into three sections: (i) the protease domain comprising about the N-terminal 400 amino acids, (ii) a cell-binding and internalization domain consisting of approximately amino acids 400 to 800, and (iii) the linker region encompassing approximately the C-terminal 100 amino acids. Taken together, these data demonstrate that the proteolytic site is necessary for toxicity and that the target for Pet is an intracellular moiety (23, 69).

Fig. 5Activity of the enteroaggregative E. coli Pet toxin. Internalization of Pet (A) and a serine protease motif mutant (Pet S260I) (B) into HEp-2 cells. HEp-2 cells were treated with either Pet or Pet S260I for 1 h. Actin cytoskeleton is labeled with green and Pet or Pet S260I is labeled with red. Note the perinuclear localization of Pet or Pet S260I inside the cells. Effect of Pet (C) or Pet S260I (D) on fodrin redistribution in HEp-2 cells. HEp-2 cells were treated with either Pet or Pet S260I for 3 h. Actin cytoskeleton is labeled with blue, Pet or Pet S260I with red, and fodrin with green. Note that Pet but not Pet S260I cause cytoskeletal damage and fodrin redistribution (arrows) and it is possible to detect a delayed interaction between Pet S260I and fodrin due to inability to cleave it (yellow dots). Scanning electron photomicrographs of in vitro-cultured human colonic tissues infected with the EAEC strain 042 (E) and the pet mutant strain JIF1 (F). For the tissue shown in panel E the surface of the colon is markedly abnormal, as manifested by increased crypt apertures (white arrowhead), prominent mucosal crevices (white arrow), goblet cell pitting (black arrowhead in a circle), and rounding of epithelial cells (black arrow in a circle). Figure courtesy of Henderson et al. (42).

In an attempt to define the intracellular target for Pet, Villaseca et al. (104) coincubated purified Pet with erythrocyte membranes. Analysis of the treated mammalian membranes by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) revealed that two high-molecular-weight proteins were cleaved in the presence of Pet. Western blotting identified these proteins as the intracellular proteins α- and β-spectrin. These spectrin proteins possess repetitive stretches of amino acids and form an integral part of the focal adhesion complexes. Indeed, further work showed Pet is capable of cleaving purified spectrin, and fodrin (nonerythroid spectrin) from HEp-2 cells and that these effects depended on the proteolytic activity of Pet because the processing effect was inhibited by PMSF and was not observed for the Pet S260I mutant (104). To corroborate this work Sui et al. (98) demonstrated spectrin condensation and identical cytopathic effects when the Pet functional domain was expressed intracellularly. Moreover, subsequent experiments demonstrated that Pet was able to cleave recombinant fodrin, which contained amino acid repeats 8 to14 fused to glutathione S-transferase (109 kDa); this effect correlated with dose-dependent redistribution of fodrin in HEp-2 cells. Pet is able to cleave fodrin in repeat 11, between M₁₁₉₈ and V₁₁₉₉, inside of the calmodulin-binding domain of fodrin. Such effects and fodrin redistribution in HEp-2 cells were prevented by using the Pet S260I mutant or PMSF (14). Fodrin cleavage as the primary mode of action could account for the cellular changes induced by Pet toxin, because fodrin is one of the main protein components of the focal adhesion complex; thus, degradation of fodrin leads to dissolution of the focal adhesion complex, followed by the eventual release of the cell from the substratum.

Although a likely mechanism of action and a target for Pet have been elucidated, the full contribution of Pet to infection remains to be characterized. However, initial investigations using in vitro organ culture and Ussing chamber experiments indicate that Pet may contribute to the pathological outcomes of EAEC infection. In vitro organ culture experiments with human intestinal mucosa and EAEC strain 042 demonstrated that this strain causes dilation of crypt openings, extrusion of colonic enterocytes, development of intercrypt crevices, and loss of apical mucus from goblet cells (39). In similar experiments an E. coli 042 strain carrying a null mutation in the pet gene was unaffected in its ability to adhere to the colonic explants but was unable to elicit the characteristic mucosal changes observed with the wild-type parent strain (Fig. 4). The effects were restored by trans-complementation of the pet gene (38, 39). Furthermore, Ussing chamber experiments with purified Pet demonstrated similar pathological effects in rat jejunum characterized by increased mucous release, cell exfoliation, and crypt abscesses (71). In addition, purified Pet raised the short-circuit current and decreased the electrical resistance of rat jejunum mounted in an Ussing chamber, an effect that is indicative of mucosal damage and enterotoxicity (71). The ability of Pet to cleave spectrin intercellularly is consistent with the observations derived from the in vitro organ culture experiments: cleavage of fodrin as described above could account for the extrusion of the colonic enterocytes, which would, in turn, lead to crypt dilation. Several membrane channels have been shown to be linked to the fodrin network (19), and therefore cleavage of fodrin could explain the enterotoxicity observed for Pet.

Sat.

Half of all uropathogenic E. coli isolates possess none, or only one, of the known uropathogenic virulence factors and it is reasonable to postulate other, as yet uncharacterized, bacterial factors important in the pathogenesis of urinary tract infections (95). In an attempt to identify additional virulence factors, Guyer et al. (32) examined the proteins released into the extracellular milieu by the uropathogen E. coli CFT073, a strain cultured from blood and urine of a patient with acute pyelonephritis. N-terminal amino acid sequencing of several proteins revealed a 107-kDa species with high homology to the N-terminal region of the Pet passenger domain. A pet probe was used to screen a cosmid library of E. coli CFT073 and identify the gene encoding the 107-kDa species. DNA sequencing revealed that this gene encoded a protein termed Sat, which possessed all the hallmarks of a typical SPATE autotransporter protein; Sat is one of three SPATEs (PicU, Vat, and Sat) present in this pyelonephritic strain of E. coli. Like Pet, Sat exhibited serine protease activity against casein, factor V, and purified spectrin and this activity could be inhibited with the serine protease inhibitor PMSF (22, 32). No proteolytic activity was observed against IgA, hemoglobin, pepsin, or mucin (22, 32).

To determine whether Sat induced cytopathic effects, Guyer et al. (32) incubated purified Sat with HEp-2, Vero, and HK-2 cells. Like Pet, Sat induced morphological changes characterized by cell elongation, release of focal adhesion contacts, and detachment from the monolayer (32). Furthermore, incubation of wild-type E. coli CFT073 revealed cytopathic effects on Vero cells and human adult bladder (CRL-1749) and kidney (CRL-1573) cell lines. These effects were not observed for the isogenic sat mutant (33). The cytopathic effects, release of cells from the substratum, or the vacuolation of cells did not induce cell lysis.

The contribution of Sat to infection was monitored in the murine model of ascending urinary tract infection. Wild-type E. coli CFT073 and its isogenic sat mutant behaved in a similar fashion in terms of colonization of the urine, bladder, and kidney tissue after a 4-day incubation period (32). Nevertheless, mice infected with wild-type E. coli CFT073 produced a strong antibody response to Sat, which indicated that the protease was expressed during infection and might play a role other than colonization of the urinary tract (32). Indeed, further investigations revealed that mice infected with wild-type E. coli CFT073 showed dissolution of the glomerular membranes and vacuolation of proximal tubule cells and that no such effects were observed in kidney sections of mice infected with a sat null mutant. These data suggest that Sat contributes to the pathogenesis of urinary tract infection by inducing damage of kidney cells during upper urinary tract infections (33).

SigA.

As mentioned previously, while characterizing the she pathogenicity island in S. flexneri, Rajakumar et al. (3, 87) found a second IgA1 protease-like homologue called sigA, lying 3.6 kb downstream and in an inverted orientation with respect to pic. Investigation of SigA expression in vitro revealed that it was preferentially expressed at 37°C. Based on the presence of the serine protease motif and similarity to the other SPATEs, SigA was investigated for proteolytic activity. Purified SigA and culture supernatant fractions from wild-type S. flexneri demonstrated degradation of casein. Culture supernatant fractions from a sigA null mutant did not demonstrate proteolytic activity against casein, indicating that, of the three SPATEs expressed by S. flexneri (SigA, Pic, and SepA), only SigA had the ability to degrade casein (2).

SigA demonstrated its highest level of homology with Pet, suggesting that it too might act as a cytopathic toxin. Like Pet and Sat, purified SigA caused extensive damage to HEp-2 cells, characterized by release of cellular focal adhesion contacts from the glass substratum, rounding of cells, and detachment (2). Furthermore, the ability of SigA to induce such effects could be inhibited by PMSF, suggesting that the serine protease active site was required for these cytopathic effects. These data suggested that SigA may be a cell-altering toxin with a role in the pathogenesis of Shigella infections. Investigation of wild-type S. flexneri and its isogenic sigA mutant in ligated rabbit ileal loops revealed that the sigA mutants were attenuated, causing a consistent decrease in fluid accumulation (2). Such a decrease would be consistent with an enterotoxic effect similar to that described for Pet.

EspC.

EPEC are a leading cause of infantile diarrhea in developing countries. The pathogenesis of EPEC is largely mediated by a set of secreted proteins, most of which exit the bacterium via a type III secretion mechanism. In 1996, Stein et al. (97) described EspC (for EPEC-secreted protein C), the major exported protein in the supernatants of the prototypical EPEC E2368/69. However, the construction of a deletion in espC by allelic exchange showed that EspC is not necessary for mediating EPEC-induced signal transduction in HeLa epithelial cells and does not play a role in EPEC adherence to or invasion of tissue culture cells (97). Nevertheless, a possible function in EPEC virulence is supported by the fact that EspC is highly immunogenic; human serum collected from a volunteer 28 days after infection with EPEC strongly recognized EspC, while serum collected prior to infection did not (46). Moreover, S-IgA antibodies purified from Mexican women, which inhibit the adherence of EPEC to cells, strongly reacted with a 110-kDa protein, which was assumed to be EspC (60).

The function of EspC in EPEC remains enigmatic; however, it was recently determined that EspC, like Pet, could act as an enterotoxin, as indicated by rises in short-circuit current and potential differences in rat jejunal tissue mounted in Ussing chambers (62). Furthermore, preincubation of EspC with antiserum against the homologous Pet enterotoxin eliminated EspC enterotoxin activity (62). Because EspC and Pet demonstrated similar enterotoxic activity it was of interest to determine whether EspC functioned in a similar fashion to Pet. Despite the fact that EspC cleaved pepsin, factor V, and spectrin in vitro, no cytopathic effects were observed when HEp-2 cells were cultured in the presence of purified EspC (22). To investigate the defect in EspC that prevents it from eliciting cytopathic activity, the interaction of purified EspC with HEp-2 cells was investigated. Thus, Dutta et al. (22) were able to demonstrate that EspC bound to cells as efficiently as Pet but that it was impaired in its ability to enter the cells: Pet can be detected by fluorescence microscopy inside the cells at 15 min, whereas EspC internalization could only be weakly detected after 7 h. Navarro-Garcia et al. (70) were able to demonstrate cytopathic effects on HEp-2 cells using purified EspC; however, these effects were only observed after prolonged incubation (>4 h) and by using a high concentration of EspC (threefold higher than that used to induce Pet cytopathic effects). These EspC-induced effects were characterized by contraction of the cytoskeleton, loss of actin stress fibers, and formation of surface blebs. Like Pet and Sat, these effects depend on the protease activity of EspC as they were inhibited by PMSF and were not observed for the serine protease active site mutant EspC S256I. Furthermore, the cytopathic effects depend on the internalization of EspC as assessed by Western blotting of cellular fractions and fluorescence microscopy with anti-EspC antibodies (70). Surprisingly, EspC appears to mediate these cytopathic effects by a different mechanism than Pet. Whereas Pet cleaves fodrin once at the calmodulin-binding domain, the domain that regulates interaction of fodrin with actin to maintain cytoskeletal integrity, EspC cleaves fodrin multiple times and outside this domain. This suggests that EspC may have a different biological activity and may act to complement the cytoskeletal-altering activity of the proteins secreted via the type III system (TTSS), a system which is absent in EAEC. A synergistic role for EspC and the TTSS is supported by recent investigation of EspC, demonstrating that expression of espC was under the control of regulators of the locus of enterocyte effacement (LEE) TTSS (25). Thus, expression of EspC is unaffected by mutations in the genes encoding the structural components of the LEE apparatus but expression of EspC is coregulated with secretion of LEE effectors (49). This regulation was influenced by temperature, CO₂, and a positive transcriptional regulator PerA (31, 34). In addition, Elliott et al. (24) have found that a LEE-encoded regulator (Ler), which is part of the Per-mediated regulatory cascade, is essential for the expression of multiple LEE-located genes in both EPEC and EHEC as well as EspC.

EspP and PssA.

Brunder et al. (13) described the SPATE protein EspP (for extracellular serine protease plasmid-encoded) encoded on the large virulence plasmid of enterohemorrhagic E. coli (EHEC) O157:H7. Almost simultaneously Djafari et al. (20) described the identical protein PssA (for protease secreted by Shiga-toxin producing E. coli) on the virulence plasmid of STEC O26:H^–. For the purposes of this chapter the term EspP will be used to describe this protein.

Like the other SPATEs, EspP represents the major extracellular proteins of both organisms. Both secreted proteins are highly immunogenic and are detected in sera from infected individuals (13, 20). In vitro assays for EspP function reveals that EspP is a protease capable of cleaving pepsin A and human coagulation factor V (when present in serum but not in purified form), suggesting that degradation of factor V could contribute to the mucosal hemorrhage observed in patients with hemorrhagic colitis (13). Furthermore, EspP also showed serine protease activity in a casein-based assay. Incubation of Vero cells with purified EspP revealed that this protein was capable of inducing cytopathic effects, suggesting a functional importance during infection of the mucosal cell layer by this bacterial pathogen (20).

Other SPATEs.

Several other SPATEs have been identified in a variety of E. coli and Salmonella strains but have only been subjected to limited characterization. In an effort to identify bacterial determinants that possess immunoglobulin-binding activity, Sandt and Hill (92) discovered a family of four cell-surface-localized proteins (EibA and EibC-E) that could bind IgG. Further investigation revealed that these proteins belonged to the Oca class (type Vc) of secreted protein exemplified by YadA (see above). Characterization of the loci encoding these genes in the E. coli strain ECOR-9, which was isolated from the feces of a healthy Swedish child, revealed that eibA and eibC were linked to genes encoding two SPATE autotransporter protein termed EaaA and EaaC, respectively (92). These genes demonstrated 99.4% identity at the nucleotide levels and revealed only eight nonsynonymous differences over the entire length of the protein. Unfortunately, these proteins were not characterized functionally, but because of their association with immunoglobulin-binding proteins and their obvious homology to the IgA1 proteases, it is tempting to speculate that they play a role in degradation of the bound immunoglobulins.

While the existence of Boa (bongorii autotransporter) has been documented for S. bongorii no functional analyses of Boa have been performed. Boa is not present in Salmonella enterica species, indicating that Boa may have been acquired by horizontal gene transfer from E. coli. Attempts to isolate the Boa passenger domain from culture supernatants in a manner similar to purification of other SPATEs has proved unsuccessful, indicating that either Boa is not expressed in S. bongorii under the culture conditions used or that machinery required for Boa expression is absent from S. bongorii.

Shiga-toxin producing E. coli (STEC) are a prominent cause of food-borne diarrhea in the developed world. Pathogenic STEC strains may possess either one or both shiga toxin variants, Stx1 and Stx2. EHEC comprise a subset of STEC and are characterized by hemorrhagic colitis and in some cases hemolytic-uremic syndrome. Most STEC strains harbor the LEE TTSS and the EspP protease. A portion of STEC induce clinical manifestations similar to EHEC strains but lack the LEE region, and little is known of the virulence factors of these organisms. Members of the SPATE family were identified in two separate studies to characterize the virulence factors of LEE-negative STEC. Thus, Leyton et al. (55) identified EpeA on the large virulence plasmid of E. coli O113:H21 strain EH41 and Schmidt et al. (94) identified EspI on a chromosomal pathogenicity island inserted at the selC tRNA locus in E. coli O91:H^- strain 4797/97. It is worth noting that EspI described by Schmidt et al. (94) is distinct from the type III secreted effector molecule of the same name which was recently described by Mundy et al. (67) in Citrobacter rodentium.

Both EspI and EpeA were assessed for functional activity by in vitro assays. Schmidt et al. (94) demonstrated that EspI was expressed and secreted in a manner consistent with other SPATEs. Furthermore, like EspP, purified EspI could degrade pepsin and this activity could be inhibited by PMSF. Coincubation of human plasma with purified EspI and separately with EspP revealed that both proteins were capable of degrading a 28.5-kDa species of protein. N-terminal amino acid sequencing identified this moiety as apolipoprotein A-I. However, in contrast to EspP, EspI did not elicit cytotoxic effects on Vero cells even after prolonged incubation. Further investigations revealed EspI did not degrade high-density or low-density lipoproteins, IgA1, hemoglobin, haptoglobin, α₂-macroglobulin, thrombin, lactoferrin, transferrin, bovine serum albumin, and collagen type 3 (94). EspI demonstrates a high level of homology with EpeA (58% identity); thus, it was not surprising to find that like EspI, EpeA belongs to the branch of SPATE proteins considered to be noncytotoxic (55) (Fig. 3). Consistent with this, EpeA was unable to induce cytopathic effects on HeLa cells even after prolonged incubation (55). In vitro assays for protein function illustrated that, like Pic, EpeA could act as a mucinase and possessed the ability to degrade gelatine and pepsin. Investigation of EpeA expression revealed that it was maximally expressed at 37°C, and like Pic expression was enhanced during culture in alkaline conditions (pH 8.0) (55). Further investigations are needed to fully assess the contribution of EspI and EpeA to the pathogenesis of LEE-negative STEC.

AIDA-I Autotransporter Family

The largest subfamily of autotransporters is defined by the AidA conserved domain COG3468 and consists of members from a diverse range of animal and plant pathogens including E. coli, S. enterica, Yersinia pestis, Pasteurella multocida, Agrobacterium tumefaciens, Mesorhizobium loti, Neisseria meningitidis, and Brucella melitensis (61). This subfamily, which is composed of more than 55 proteins, possesses some of the best-characterized autotransporter proteins including the S. flexneri mediator of motility IcsA, the major phase-variable E. coli outer membrane protein antigen 43 (Ag43) and the diffuse adhering E. coli (DAEC) adhesin AIDA-I, from which this subfamily derives its name. Despite the fact that this subfamily contains several well-known proteins, most members remain uncharacterized. The passenger domains of these autotransporter proteins display homology with a wide array of proteins secreted via other protein secretion pathways, e.g., the Bordetella pertussis filamentous hemagglutinin secreted via the TPS pathway (see above) (37) and the Vibrio cholerae RtxA protein secreted by a type I pathway (56). Although the characteristics of these proteins have been described elsewhere, it is worth reiterating the salient features of these autotransporter adhesins.

AIDA-I and Its Homologues in E. coli.

The prototypical member of this subfamily of autotransporters, AIDA-I, was first described in E. coli strain 2787 (O12:H27), a DAEC strain isolated from a case of infant diarrhea (8). The passenger domain possessed several characteristics reminiscent of the phase-variable E. coli autotransporter protein Ag43, including that (i) after processing it remains associated with the cell surface through a noncovalent interaction; (ii) it can be released from the cell surface by heating to 60°C; (iii) it displays an aberrant migration pattern on SDS-PAGE under denaturing conditions, migrating with an observed molecular weight higher than the predicted molecular weight (~100 kDa in the case of AIDA-I); and (iv) it possesses a highly repetitive structure consisting of conserved repeats of 19 to 21 amino acids each (7). In contrast to AIDA-I and Ag43, TibA possesses two repetitive regions: one region is similar to AIDA-I but the second shows a high number of proline residues (58). Such proline-rich regions have been implicated in weak nonstoichiometric binding activity and are reminiscent of the B. pertussis pertactin protein (26, 107). Such repetitive structures appear to be a characteristic of proteins associated with adhesive or receptor functions, e.g., filamentous hemagglutinin of B. pertussis (48), matrix-binding proteins of Staphylococcus aureus (35), and mycobacterial heparin-binding hemagglutinin adhesin (84). Indeed, Benz and Schmidt (9) demonstrated that AIDA-I was responsible for the diffuse adhering phenotype of E. coli strain 2787, whereas Henderson et al. (40) showed that Ag43 expression mediates autoaggregation of bacterial cells by permitting cell-cell interactions and subsequent biofilm formation. Furthermore, enzyme-linked immunosorbent assays with purified TibA demonstrated the protein was capable of binding to cultured HCT8 ileocecal cells in a specific and saturable manner and that this binding could be inhibited by anti-TibA polyclonal antibodies (57). Moreover, expression and mutagenesis studies of the tibA gene demonstrated that it was directly responsible for the ability of ETEC to invade cells; chromosomal deletion mutants invaded cells at 15% of the wild-type level, and the full level of invasion could be restored by trans-complementation of the mutant (26).

Subsequent investigations revealed that both AIDA-I and TibA were glycosylated by their respective heptosyltransferases, Aah and TibC (10, 66). For a long time, protein glycosylation was thought to be restricted to the realm of eukaryotes and it was assumed that prokaryotes lacked the necessary enzymes. However, research in recent years has demonstrated that glycosylation is a frequent occurrence in archaea and eubacteria and that these glycosylated proteins often play vital roles in the lifestyles of various organisms (63, 65). Indeed, this glycosylation phenomenon is required for both the adherence and invasion phenotypes associated with AIDA-I and TibA. The occurrence of glycosylation for AIDA-I and TibA suggests that glycosylation may be a much more widespread phenomenon than previously believed, and in particular, may occur frequently among the autotransporter proteins (58).

Another member of the AIDA-I family, and one of the most studied autotransporter proteins, is IcsA. IcsA was first described by Makino et al. (59) and annotated as VirG. IcsA is expressed on the surface of Shigella strains and is localized to one pole of the bacterium (16, 88). The essential contribution of icsA in the pathogenesis of shigellosis was demonstrated by several human and animal experiments, in which icsA mutants were greatly attenuated (11, 52, 54, 93), and ulcerations and abscesses in the intestinal mucosa were greatly reduced (93). This attenuation was found to be due to the inability of the bacteria to move intra- and intercellularly after invasion. Later, it was demonstrated that IcsA recruits a protein termed N-WASP to the bacterial cell surface, disrupting intramolecular bonds and thereby activating the N-WASP proteins and stimulating actin assembly via the Arp2/3 complex. In addition to N-WASP and the Arp2/3 complex, actin tail assembly requires the VASP, vinculin, profiling, cofilin, and Cdc42 proteins (30). Whereas the exact roles of these proteins in actin tail assembly and motility remains to be determined, it is clear that polar localization of IcsA results in actin accumulation at the pole of the organism. The accumulation of actin in turn leads to intracellular motility of Shigella spp., and formation of an actin tail. In addition, this motility allows intercellular movement, by permitting the cells to breach the membranes separating adjacent cells.

AIDA-I Homologues of Salmonella spp.

In silico analyses of the S. enterica genomes has revealed the presence of six autotransporter proteins: four of which belong to the AIDA-I family, one that belongs to a family of esterases/lipases, and Boa, the SPATE protein mentioned above. Of the four AIDA-I family autotransporters, only ShdA and MisL have been investigated. ShdA was first identified by Kingsley et al. (51) as an attenuated S. enterica mutant in a chick virulence assay. Subsequent investigations demonstrated that ShdA possesses three distinct repeats; three copies of a 102-amino-acid repeat, nine copies of a 63-amino-acid repeat, and four copies of a proline- and glycine-rich 12-amino-acid repeat that, as mentioned previously, is a feature reminiscent of other adhesins (51). Indeed, Kingsley et al. demonstrated recently that an ShdA-glutathione S-transferase fusion bound fibronectin in vitro and that this binding was dose dependent and partially inhibited by antifibronectin antibodies (50). Characterization of a well-defined shdA mutant in the BALB/c murine model of infection revealed that ShdA is essential for S. enterica serovar Typhimurium colonization of the cecum and that inactivation of shdA causes a reduction in bacterial number and the length of time for which bacteria are shed in murine feces (50, 51). Moreover, whole-cell serovar Typhimurium, ShdA, and fibronectin were shown by fluorescence microscopy to colocalize in the murine cecum, strongly suggesting that fibronectin is a receptor for ShdA binding (50).

While MisL demonstrates homology to AIDA-I, ShdA, and several other adhesins and despite its high level of conservation among several S. enterica strains, it appears to be a pseudogene. A serovar Typhimurium misL-lac fusion did not demonstrate any expression under conditions of laboratory growth. Furthermore, a misL mutant exhibited wild-type levels of invasion of epithelial cells and survival within macrophages and was as efficient as the parental strain in its ability to cause a lethal infection in the BALB/c murine model (12). Nevertheless, it remains possible that MisL could be involved in other aspects of infection such as persistence or perhaps host specificity or it may play a role that is unrelated to infection.

Putative Adhesins.

Our recent in silico investigations of the E. coli and Salmonella genomes have revealed the presence of a number of "putative" or "hypothetical" open reading frames with the potential to encode autotransporter proteins (Table 1). Although the predicted proteins demonstrate homology with AIDA-I, a definitive function for these proteins has yet to be elucidated. Nevertheless, some information concerning a few of these proteins may be found in the literature. Indeed, it was suggested previously that the ydeK/U and b1169/1170 loci of E. coli K-12 might represent progenitors of the two-partner secretion systems (108). However, closer examination of the genes indicated that these loci represent out-of-frame autotransporter proteins. When the E. coli genome sequences are compared the b1169/b1170 open reading frame is interrupted at several different places within the locus, but it appears that the two "genes" form an intact locus in S. flexneri 2457T (Henderson et al., unpublished data). In contrast, the ydeK/U locus possesses a run of eight Gs (5'-GGGGGGGG-3') at the point of separation of the two genes. Reduction of this stretch to a run of seven Gs would place this locus in-frame, allowing production of an intact autotransporter. This suggests that the locus can undergo phase variation by slipped-strand mispairing, a process frequently observed in H. influenzae and N. meningitidis (44). While this method of gene regulation has not previously been documented for E. coli, H. influenzae genes remain capable of phase variation when placed in an E. coli background, indicating that E. coli at least have the capacity to utilize this method of phase variation (100).

Dissection of the O-islands from strains of E. coli O157:H7 have revealed that the ypjA gene is located adjacent to the site of insertion of the cryptic prophage CP-933Y. In two strains of E. coli O157:H7, this O-island is missing and the 3' end of ypjA has been deleted, perhaps as a result of prophage excision (53). However, this chromosomal deletion did not affect the ability of the two strains to cause disease, as they were recovered from human infections, suggesting that the product of the ypjA gene does not play a role in the pathogenesis of human E. coli infection. In contrast, a homologue of YcgV has been found in Pseudomonas putida, which is involved in infection of plant seeds but which does not affect the ability of the organism to form biofilms (28). A recent report demonstrated that overexpression of ypjA, ycgV, and yfaL resulted in increased biofilm formation in E. coli K-12, however overexpression of ydeK, yaiT, yejO, or ycgH had no effect on biofilm formation (90).

Esterases (the GDSL Family)

Investigations of the ability of S. enterica to hydrolyze the chromogenic substrate N-acetylphenylalanine β-naphthyl ester led to the identification of the apeR locus that encodes a transcriptional regulator (64). ApeR acts as a negative regulator of transcription of a locus encoding the membrane hydrolase apeE, and this regulatory event is intricately linked to the phoBR phosphate regulatory system (17, 18). The Salmonella-specific apeE gene is located on an island inserted at the tRNA^ArgU locus and encodes a product of a 69.9-kDa protein that is processed to a 67-kDa species by removal of a signal sequence (25 amino acid residues). Amino acid sequence analyses of ApeE revealed that it is a member of an esterases/lipases family in which the active site is represented by the amino acid sequence GDSL and the serine represents the active site residue. This family includes Lip-1 encoding a lipase from Xenorhabdus (Photorhabdus) luminescens, the EstA esterase of Pseudomonas aeruginosa and McaP an adhesin/esterase of Moraxella catarrhalis. Further studies revealed that the ApeE esterase catalyzes the hydrolysis of variety of fatty acid naphthyl esters and of C₆ to C₁₆ fatty acid p-nitrophenyl esters but not peptide bonds (15). Indeed, one method for identification of Salmonella ssp. in the clinical laboratory is hydrolysis of methyl umbelliferyl caprylate, a substrate that fluoresces upon hydrolysis of the ester bond (1, 15). ApeE is uniquely responsible for this diagnostic marker. Neither ApeE nor any similar enzymes have been identified in any of the genome-sequenced strains of E. coli, but E. coli appear to possess several pseudogenes homologous to the C-terminal portion of ApeE but that lack a functional passenger domain (our unpublished observations).

CONCLUSIONS

The autotransporter pathway is emerging as the most common mechanism of protein translocation across the gram-negative outer membrane. Yet due to its relatively recent description and the explosion of autotransporter proteins from genome projects, remarkably little is known about the structure, functions, and mechanism of translocation for these proteins. Future investigations will be needed to resolve these important questions.