Type II Secretion and Pathogenesis

CLASSIFICATION OF SECRETION PATHWAYS

While the outer membrane provides a natural barrier and acts together with multidrug pumps to protect the gram-negative cell from harmful agents such as detergents, disinfectants, dyes, certain antibiotics, and toxins, it also poses a challenge with respect to the uptake of nutrients and the excretion of by-products (105). For this purpose, the outer membrane has been equipped with different pathways to selectively transport molecules both into and out of the cell (104, 106). Although it has been accepted that the outer membrane barrier allows for passage of small molecules, until very recently it was thought to prevent the extracellular release of larger macromolecules such as proteins. Results from studies accumulated within the last 10 to 15 years have challenged this view and clearly demonstrated that the outer membrane is not inert with regard to transport of such molecules. In fact, these studies have suggested that every gram-negative organism may have the capacity to selectively secrete proteins across the two membranes to the extracellular environment. What is perhaps surprising is that, although extracellular secretion is a relatively rare event, several highly specialized pathways have evolved for this purpose (see below). At the moment it is not understood why certain proteins are secreted via one pathway and not another. It is possible that the choice of secretion pathway for individual proteins is determined in part by the function they perform at the extracellular site where they are delivered, as was recently suggested by Lory (87).

Six pathways for extracellular secretion have been identified to date. These include the signal sequence independent pathway (type I), the main terminal branch of the general secretion pathway (type II), the contact-dependent pathway (type III), the type IV pathway, theBordetella pertussis filamentous hemagglutinin secretion pathway (two-partner secretion [TPS] pathway) and the autotransporter pathway (for reviews, see references 7, 16, 23, 26, 51, 63, 64, 131, and 133). Some of these pathways transport proteins across the bacterial cell envelope in one single step. For example, translocation via the type I or III pathway occurs across the inner and outer membrane simultaneously, bypassing the periplasmic compartment. Although the number of accessory gene products required for successful secretion varies among these pathways from 3 for type I to at least 20 for type III, both of these pathways support the secretion of unfolded proteins and neither utilizes the sec system. In contrast, the type II, TPS, and autotransporter pathways involve two steps in which proteins containing an N-terminal signal peptide are first translocated across the cytoplasmic membrane via the secmachinery. Then, following removal of their signal peptides and release into the periplasm, the mature proteins cross the outer membrane in a separate step. As the name implies, secretion through the autotransporter pathway does not require any accessory factors. The respective C-terminal domains of these secreted proteins are thought to insert into the outer membrane, each forming a pore that allows transport of the N-terminal portion to the cell surface. The mature portion of the protein is then released from this pore structure by proteolytic cleavage (51, 120). In contrast, outer membrane translocation through the type II pathway requires the function of at least 12 gene products and occurs by a mechanism that is still obscure. This pathway can shuttle a large number of seemingly unrelated proteins in their folded states to the extracellular environment (see below). Translocation through the outer membrane via the TPS pathway may only depend on the pore-forming protein FhaC (63). Finally, the type IV pathway is primarily involved in the mobilization of DNA either between bacteria or from bacteria to plant cells; however, recently the type IV system has been found to also transport proteins, such as B. pertussis toxin andHelicobacter pylori CagA antigen (16, 112). While the DNA is thought to be transferred as a DNA-protein complex in one single step across the cell envelope, pertussis toxin appears to be transported to the periplasmic compartment in asec-dependent manner and then translocated across the outer membrane in a separate step (21).

TYPE II SECRETION PATHWAY

The type II pathway was first discovered in Klebsiella oxytoca, where it was found to be required for secretion of the starch-hydrolyzing lipoprotein, pullulanase (32). Since then, this pathway has been discovered in a number of bacterial species, including several extracellular pathogens. These include human pathogens such as Vibrio cholerae and Pseudomonas aeruginosa, fish pathogens such as Aeromonas hydrophila, and plant pathogens, including Erwinia chrysanthemi, Erwinia carotovora, and Xantomonas campestris (Fig. 1). In these species several of the type II secretion genes have been inactivated and shown to be essential for the translocation of multiple proteins (Table 1). These include several different plant cell wall-degrading enzymes of the Erwiniaspecies and X. campestris and exotoxin A, elastase, phospholipase C, alkaline phosphatase, and lipases LipA and LipC ofP. aeruginosa (86, 95, 159). In V. cholerae this pathway supports the secretion of cholera toxin (CT), hemagglutinin-protease (HAP), and chitinase (25, 28, 137). In addition, while extracellular neuraminidase and lipase activities can be detected with wild-type V. cholerae grown on appropriate indicator media, these activities are abolished in strains carrying mutations in the eps type II secretion genes (137; M. Bagdasarian, unpublished data).

• Open in new tab
• Download powerpoint

Fig. 1.

Alignment of genes encoding type II secretion pathways. Individual species with the names of their secretion genes and a schematic representation of the genes A to O and S shown as boxes with arrowheads to indicate their orientation. The designations of thexcp genes of P. aeruginosa and P. putida differ from those of other species. Spaces are introduced to indicate that genes are not adjacent. Lines connecting genes indicate that they are linked but interspersed by opposing promoters or other genes. The names of the O genes are vcpD(pilD) in V. cholerae, tapD in A. hydrophila, xcpA (pilD) in P. aeruginosa, andpilD in L. pneumophila. The gene alignment was done using previously published sequences (1, 4, 6, 15, 29, 30, 32-34, 40, 44, 49, 59, 61, 66, 67, 82, 89, 129, 137, 140, 144). Sequences of open reading frames from S. putrefaciens, C. crescentus, Y. pestis, and G. sulfurreducens were obtained from The Institute for Genomic Research and Sanger Centre via the NCBI microbial genome database.

Although the first type II secretion genes were identified for K. oxytoca, A. hydrophila, and P. aeruginosa and sequenced 10 years ago (31, 32, 36, 67), the first documented secretion mutant was identified 15 years before this (57). In that study, a chemically induced V. cholerae mutant strain that did not secrete CT was isolated. V. cholerae has been, and continues to be, the subject of intense research, as it remains a significant health threat in several parts of the world, where it causes a severe and life-threatening diarrheal disease. Its primary virulence factor, CT, is in large part responsible for the symptoms of cholera and is probably the most intensely studied protein of all those known to be secreted via the type II secretion pathway. CT is a multimeric protein complex and is nearly identical to theEscherichia coli heat-labile enterotoxin (LT) (146). The crystal structures of LT and CT confirm that both toxins are composed of a pentameric ring of identical B subunits of 11.6 kDa each and one A subunit of 28 kDa (145, 163). Both toxins bind via their B subunit pentamer to the GM₁ganglioside on intestinal epithelial cells (98). After entering the epithelial cells, the A subunit ADP-ribosylates the stimulatory G protein of the adenylate-cyclase complex. The resulting elevation of cyclic AMP leads to a rapid and profuse efflux of water and ions from the cells, with diarrhea as a consequence (146).

The folding, assembly, and secretion pathways of CT and the almost identical LT have been analyzed in great detail, and CT serves as a prototype for other proteins secreted via the type II pathway. The individual toxin subunits are first produced as precursor proteins with typical N-terminal signal peptides. After translocation through the cytoplasmic membrane via the sec pathway, the signal peptides are removed and the mature subunits are released into the periplasm, where they assemble noncovalently into an AB₅holotoxin complex in a process that is assisted by the disulfide isomerase DsbA (TcpG) (54, 55, 116, 162). Only when assembled does the toxin complex traverse the outer membrane in a second step that requires VcpD and 12 additional gene products that collectively comprise the extracellular protein secretion apparatus (Eps) (Fig. 1) (41, 54, 92, 113, 137, 138). The B subunit pentamer carries the information for outer membrane translocation, and the A subunit is secreted only by virtue of its association with the B₅ complex (56). Based on these early studies and the finding that the periplasmic toxin complex is very similar, if not identical, to the secreted form of the toxin, it was proposed that proteins are secreted in their fully or nearly fully folded forms via the type II pathway. The suggestion that folding is a prerequisite for outer membrane translocation has been confirmed with a number of other proteins. However, recent studies suggest that, at least in some cases, there still may be some structural differences between periplasmic secretion competent proteins and their extracellular forms (18). The best examples of this are the elastase ofP. aeruginosa and HAP of V. cholerae, which are both secreted as proforms (13, 47, 96), and it is only when these proteins have reached the extracellular environment that their propeptides are removed.

SEQUENCE AND MUTATIONAL ANALYSIS OF THE TYPE II SECRETION GENES

Depending on the species, between 12 and 15 genes have now been identified as essential for type II secretion, and the homologous genes and gene products have been designated, for most species, by the letters A through O and S (125) (Fig. 1). The level of homology between the genes of different species varies from 25 to 40% identity for the C, L, and M genes to 60 to 80% identity for the E, F, and G genes. The genetic organization of the type II secretion gene cluster is relatively well conserved (Fig. 1). Most of the genes appear to be organized in a single operon, and several of the genes are overlapping. If variations are detected, they are usually found at the 5′ and 3′ ends of the gene clusters (Fig. 1). Minor differences between the type II secretion gene clusters of relatively closely related species have also been observed. For instance, the C and D genes ofP. aeruginosa (designated xcpP andxcpQ, respectively) are transcribed in the opposite direction from the E to N genes (xcpR to xcpZ), as well as being transcribed from a different promoter (Fig. 1). This is in contrast to the genetic organization of most other species, including Pseudomonas putida (30), where the C to N genes are transcribed in alphabetical order, most likely from the same promoter (Fig. 1). Comparison of the secretion genes of two closely related Erwinia species also shows that one of the genes, outN, is missing in E. chrysanthemi, while it is present and required for secretion in E. carotovora(83). The O gene, which in some species is the last gene of the type II secretion operon, is homologous to the prepilin peptidase gene required for processing of type IV prepilin subunits. In species where the O gene is not linked to the rest of the type II secretion genes, it is found associated with a subset of genes required for type IV pilus biogenesis. In these species the prepilin peptidase exhibits a dual function, i.e., it is required for processing and methylation of the type IV prepilin subunits as well as the prepilin-like components encoded by the type II secretion genes G, H, I, J, and K (12, 111, 149). In V. cholerae, for instance, VcpD (PilD) is required for CT secretion via the Eps apparatus and has been shown to process the precursor form of EpsI as well as the prepilin forms of the type IV pilin subunits PilA and mannose-sensitive hemagglutinin (MSHA) A (MshA) (41, 92). DNA composition analysis of the secretion genes of V. cholerae suggests that the eps genes are relatively old and that the vcpD gene has been acquired recently by horizontal gene transfer (50). It is possible that theeps cluster originally contained an epsO gene at the distal end but that when the vcpD gene was acquired, theepsO gene became redundant and was subsequently lost.

The secretion pathways in K. oxytoca and theErwinia species have been reconstituted in laboratory strains of E. coli (83, 123). In this background, Possot and colleagues have shown that all pulgenes except for pulB, pulH, and pulN, are required for secretion (123). A similar result was obtained with the E. chrysanthemi out secretion genes expressed in E. coli (83). In this case, allout genes, with the exception of outH, were found to be required for secretion of Erwinia pectate lyase. In contrast, an E. chrysanthemi outH mutant was defective for extracellular secretion (83). The discrepancy in requirement for the H gene in Out-dependent secretion is not fully understood; however, it could be due to a copy-number effect (chromosomal versus plasmid expression) or differences between the species. It is worth mentioning that even though laboratory strains ofE. coli are not thought to transport proteins to the extracellular environment under normal growth conditions, they do contain what appears to be a complete set of type II secretion genes, some of which may have compensated for the missing gene(s) in those studies (Fig. 1). However, the level of expression of the E. coli gsp secretion genes is extremely low (124). The lack for requirement of pulN in pullulanase secretion is consistent with the finding that several of the type II secretion gene clusters do not contain an N gene (Fig. 1). The requirement for the A gene has also been observed only for A. hydrophila, and the requirement for the B gene has been observed only forErwinia species and A. hydrophila (60, 66, 83). Likewise, the need for an S protein in type II secretion has been demonstrated only for Erwinia and K. oxytoca (45, 143).

It is interesting to note that in P. aeruginosa a second D gene with overlapping function was identified by Martinez and colleagues (94), who noticed that in a strain carrying a mutation in the type II secretion gene D (xcpQ), a low level of extracellular secretion could still be detected. This secretion was contributed by another D homologue, xqhA (94). Subsequently, when this gene was inactivated in the xcpQmutant background, extracellular secretion was completely abolished. Furthermore, when the complete genome sequence of P. aeruginosa was released, it was noted that the xcpgenes have several homologues, including a complete second set of secretion genes (148; A. Filloux, personal communication).

The function and interactions of the individual components of the type II secretion apparatus have been reviewed recently and will not be discussed in great detail here (37, 131, 133). However, a summary of the current knowledge will be presented. Protein D is a member of the secretin family that includes proteins required for type IV pilus biogenesis, filamentous phage extrusion, and type III secretion (42). These proteins are present in the outer membrane, where they are thought to form gated secretion pores (8, 72, 83-85, 90, 107, 108, 132). The secretins exhibit ion-conducting properties and an oligomeric structure that can be visualized by electron microscopy (8, 14, 85, 90, 107, 108). Proper outer membrane insertion of protein D is assisted by protein S in K. oxytoca and E. chrysanthemi(27, 45, 143), but the requirement for the S protein in other species has not been confirmed. The presence of sequences homologous to the S gene in the E. coli O157:H7 pO157 plasmid and the Yersinia pestis chromosome may suggest that the S gene is restricted to Enterobacteriaceae (Fig. 1). Although specifically required for outer membrane translocation, the remaining components of the type II secretion apparatus appear to be associated with the cytoplasmic membrane, where several of them have been found to interact (5, 11, 58, 70, 88, 123, 126-128, 134-136, 141, 151). However, recent studies suggest that there may also be interactions between some of the cytoplasmic membrane components and the secretin (10, 24, 75, 121). In addition, outer membrane translocation requires the proton motive force of the cytoplasmic membrane (78, 122). Taken together, these findings suggest that the components of the secretion apparatus form a multiprotein complex that spans both membranes.

Several of the type II secretion genes, including the D, E, F, and O genes, are homologous to genes required for assembly and export of type IV pili. These are surface appendages that play a major role in colonization and virulence of a number of bacterial pathogens, including P. aeruginosa, Neisseria gonnorhoeae, and V. cholerae (20, 46, 71, 74, 110). Moreover, proteins G, H, I, J, and K are similar to the type IV prepilin subunits with respect to their size and sequence similarity within the N-terminal domains. Because of these similarities, it was suggested that they might form a pilus-like structure (109), and a recent paper by Sauvonnet and colleagues demonstrated that the components of the type II secretion apparatus can form a pilus-like structure when their genes are overexpressed (139). However, when the genes were expressed from a chromosomal location or from a low-copy-number plasmid, no pilus structure was observed, even though secretion of pullulanase could be detected. In support of this finding, a pilus structure encoded by the xcp type II secretion genes of P. aeruginosa has been identified (D. Nunn, personal communication). These results suggest that a subset of the type II secretion genes has the capability to form a pilus-like structure; however, it is not known whether the pilus structure per se directly supports secretion. Different models for pilus-mediated secretion have been proposed. The cytoplasmic membrane-anchored type II secretion pilus may act as a piston, and through extension and retraction it may push the secreted proteins through the gated pore (37, 133, 143). Support for this model comes from recent studies showing that the type IV pilin subunits assemble into filaments prior to outer membrane translocation and that the type IV pili can forcefully retract (99, 157). In this model, protein E, which has been found to exhibit autophosphorylation activity and is present on the cytoplasmic side of the cytoplasmic membrane, may regulate this process via interaction with proteins L, M, and C and with the secretin in the outer membrane (5, 121, 123, 127, 133, 134, 136). Alternatively, the E protein may supply energy for the outer membrane translocation process by promoting polymerization of the pilin-like proteins G, H, I, J, and K (133). Because the type II secretion pilus formed long filaments that bundled into higher-ordered structures when the pul genes were overexpressed, an alternative role of the pilus in secretion would be to channel secreted proteins past surface structures such as capsules or past aggregated bacteria present in biofilm (Fig. 2) (133, 139).

• Open in new tab
• Download powerpoint

Fig. 2.

Regulation of secretion. Three steps in the colonization of a general surface, such as the surface of plant leaves, the mucosal epithelium, or a solid surface in an aquatic milieu. Secretion of virulence factors occurs only when the bacteria have reached a critical concentration or retrieved the appropriate signal from the surface.

IDENTIFICATION OF ADDITIONAL SPECIES WITH TYPE II SECRETION GENES

Several of the species that encode the type II secretion pathway can be isolated from multiple ecological niches (Table 1). V. cholerae can not only colonize the human gastrointestinal tract but also live in the aquatic environment either as a free-swimming organism, as an organism associated with plankton, or as a member of biofilm attached to chitinous or abiotic surfaces. The opportunistic human pathogen P. aeruginosa, which frequently causes infections in cystic fibrosis patients, cancer patients, and burn victims, can survive and grow in soil as well as the aquatic milieu and can infect both animals and plants. Another species that can be isolated from soil, water, and plants is Burkholderia cepacia (Table 1). This species is also associated with infections in cystic fibrosis patients, where it can cause necrotizing pneumonia.B. cepacia has recently been found to contain a type II secretion pathway that is required for protease and lipase secretion, and sequences homologous to epsF-epsK have been deposited in the GenBank database (accession number AF127982 ). Sequences homologous to the epsC-epsN genes have also been identified in the related Burkholderia pseudomallei, which is responsible for melioidosis, a life-threatening, systemic bacterial infection. In this species the type II secretion genes are required for secretion of protease, lipase, and phospholipase C (33). It is interesting to note that the order of the secretion genes of B. cepacia and B. pseudomallei differs from that of the secretion genes of other species. Instead of being upstream of the D gene, the C gene is inserted between the F and G genes (Fig. 1). Furthermore, the orientation of this gene is opposite that of the rest of the genes in the cluster.

Recently, the first intracellular pathogen to express a type II secretion pathway was identified. Two different laboratories have identified type II secretion genes in Legionella pneumophila(4, 44, 81), the causative agent of a severe form of pneumonia called Legionnaires' disease (154). As many as eight different proteins may depend on these genes for extracellular secretion. Some of these proteins are listed in Table 1. L. pneumophila is a facultative intracellular pathogen; it can be found outside the human host in the freshwater environment either as a free-living organism in biofilms or as an intracellular pathogen of protozoa (154). The type II secretion genes identified to date in L. pneumophila include the lspF-lspKgenes and the prepilin peptidase gene pilD (Fig. 1) (4, 44, 81). Sequence analysis of DNA 1 kb downstream oflspK suggests that L. pneumophila may lack the L, M, and N genes, unless they are present elsewhere in the genome (44). Early release of sequence information from theL. pneumophila sequencing project at the Columbia Genome Center provides evidence that sequences homologous to the epsC, epsD, and epsE genes are present in this species (N. P. Cianciotto, personal communication). By taking advantage of this information, mutations in the putative lspD andlspE genes have been introduced and found to prevent extracellular secretion of protease (N. P. Cianciotto, personal communication).

As mentioned above, type II secretion genes have also been identified in E. coli K-12 (9, 147); however, they do not appear to be expressed under normal laboratory growth conditions (40). In a study aimed at identifying conditions that increase the level of E. coli gsp transcription, it was found that expression from the gspC promoter was increased two- to threefold in a global regulatory mutant lacking the nucleoid-structuring protein H-NS (124). Although the expression was improved in this mutant background, it was still extremely low and no secreted protein could be detected.

Francetic and colleagues suggested that ChiA is a chitinase, which may be secreted via the type II secretion pathway in E. coliK-12 (12). The chiA gene is located downstream of the type II secretion genes and is separated from them by only two open reading frames (ORFs). The expression of the chiA gene is, like the secretion genes, regulated by H-NS (38). In addition, ChiA exhibits a low-level homology to the V. cholerae endochitinase, which is secreted via the Eps pathway (25, 137). However, even in the hns mutant background ChiA accumulated in the periplasmic compartment, and it was only when the E. coli gsp genes were overexpressed from a plasmid in the hns mutant that chitinase secretion could be observed (39).

In contrast to E. coli K-12, the food-borne pathogenE. coli 0157:H7 does not carry type II secretion genes on the chromosome (117; V. Burland, personal communication). From an analysis of the two E. coli genomes using a genome browsing tool (www.genome.wisc.edu ), it is evident that the region containing the type II secretion genes as well as the chiA gene inE. coli K-12 is replaced by novel DNA in E. coliO157. Instead, the large 92-kb plasmid pO157, which is found in most enterohemorrhagic E. coli (EHEC) O157:H7 strains, carries a set of type II secretion genes, etpC-etpM, etpO, andetpS (Fig. 1) (15, 140). This is the first described case in which the type II secretion genes are present on a plasmid. It is not known if these genes encode a functional secretion apparatus or what the substrate for this putative apparatus is. It is possible that the etp genes may be required for secretion of a protein encoded by one of the more than 20 novel ORFs on the plasmid or a chromosome-encoded protein. When the distribution ofetp genes among other E. coli strains was determined by hybridization and PCR, etp-like sequences were found in all EHEC strains of serogroup O157 but were less likely to be identified in other serogroups (140). No positive signal was obtained with enterotoxigenic, enteropathogenic, enteroinvasive, or enteroaggregative E. coli. On the other hand, when probes of the E. coli K-12 gspGH genes were used in hybridization studies, homologous sequences were restricted to enteroinvasive and enterotoxigenic E. coli(147). No positive signal was obtained for theSalmonella or Shigella species tested.

Blattner and colleagues at the University of Wisconsin are in the process of sequencing the genome of another pathogenic E. coli strain, the highly virulent uropathogenic E. colistrain CFT073, which is responsible for acute pyelonephritis. In contrast to E. coli O157:H7, this pathogen has kept the type II secretion genes on the chromosome, and these genes are essentially identical to the corresponding E. coli K-12 genes (www.genome.wisc.edu ).

The recently released genome sequence of the economically important plant pathogen Xylella fastidiosa, which affects production of citrus fruits and grapes, revealed the presence of a type II secretion gene cluster homologous to that of the related X. campestris (144). Genes encoding putative substrates for this secretion pathway were also present. These include cellulase and polygalacturonase, which are known virulence factors secreted via the type II pathway in Erwinia and X. campestris(Table 1).

Analysis of DNA sequences deposited in the National Center for Biotechnology Information (NCBI) microbial genome database of completely and incompletely sequenced genomes reveals that additional species harbor DNA homologous to the type II secretion gene cluster. The genomes of Shewanella putrefaciens, Caulobacter crescentus, and Y. pestis, sequenced by the Institute for Genomic Research and Sanger Centre, contain ORFs that display homology to the eps genes of V. cholerae. The genetic organization of the secretion genes in S. putrefaciens is identical to that of V. cholerae andA. hydrophila (Fig. 1). In fact, when comparing the type II secretion genes from different species identified to date, theeps genes exhibit the overall highest level of homology to the secretion genes in A. hydrophila and S. putrefaciens. S. putrefaciens is a pathogen rarely found in humans; only a few cases of bacteremia have been reported. It is an inhabitant of marine ecosystems and is often associated with spoiling fish (43). C. crescentus is also an environmental species that is not associated with human disease but has been studied extensively for its asymmetric cell division and cell differentiation (65, 142). Of all species having the O gene as the last gene of the secretion gene cluster, the O gene inC. crescentus is transcribed in the opposite orientation compared with the rest of the secretion genes (Fig. 1). No ORF homologous to epsN gene was detected in this species when using the standard setting for the BLAST search of the NCBI microbial genome database. However, there is a stretch of approximately 870 nucleotides (nt) between the putative M and O genes that would be large enough to accommodate an N gene, which is approximately 750 nt in length (137). Y. pestis, the causative agent of the black death, or plague, which killed nearly half of the population of Europe in the 14th century also contains ORFs homologous to theepsC-epsL and vcpD genes (Fig. 1). Immediately downstream of these ORFs, there is also an S gene. Using the standard setting for the BLAST analysis, no ORFs homologous to theepsM and epsN genes were identified. However, there is a stretch of DNA of approximately 600 nt between the L and O genes in Y. pestis. This spacing would allow for an M gene but most likely not for an N gene. The epsM gene of V. cholerae is 497 nt (137). Consistent with this is the finding that all type II pathways identified to date have an M gene, while only about half of the species have an N gene (Fig. 1). Y. pestis has been extensively studied for its ability to secrete virulence factors via the contact dependent type III secretion pathway; however, the presence of another extracellular secretion pathway in this species has not been previously noted. What the type II pathways in Y. pestis, S. putrifaciens, or C. crescentusmight be transporting is not known. One can only speculate that they may transport a metalloprotease, lipase, phospholipase, or chitinase, as do the type II secretion pathways of other species (Table 1). Consistent with this is the identification of ORFs in Y. pestis and S. putrefaciens whose translated products exhibit homology to V. cholerae endochitinase, while sequences present in the S. putrefaciens genome are homologous to V. cholerae HAP and the lipase LipA. Finally,C. crescentus contains DNA homologous to V. cholerae lipA.

The anaerobic Geobacter sulfurreducens (17, 22), isolated from aquatic sediments, is also likely to contain a type II secretion apparatus. Genes homologous to theepsC-epsI genes are present in one sequence contig, while sequences homologous to the epsA and epsJ genes are present on others. The genomes of many other species contain sequences homologous to the D, E, and F genes of the type II secretion gene cluster. However, since these sequences are also homologous to genes required for type IV pilus biogenesis, it is not possible at the moment to determine whether these sequences encode components of the type II secretion apparatus.

REGULATION OF SECRETION

Quorum sensing is the mechanism whereby a number of bacterial species control gene expression in response to cell density. One form of quorum sensing involves the production and detection of acylated homoserine lactones. Activation (or repression) of genes occurs when these diffusable autoinducers accumulate and reach a critical concentration as the cell density increases. This means of cell-to-cell communication to coordinate gene expression occurs apparently both within and between different species and is thought to contribute to successful colonization and infection. The first species identified that depends on quorum sensing for regulated gene expression wasV. fischeri, which requires quorum sensing for bioluminescence (35, 103). Subsequently, the production of many extracellular virulence factors has been found to be regulated in a growth-phase-dependent manner as a result of quorum sensing or autoinduction. In P. aeruginosa, quorum sensing involves two different systems, lasRI and rhlRI (114, 115). These systems overlap and in some instances regulate the same genes. In addition, rhlRI-regulated genes require thelasRI system for maximal expression (118). The genes for several of the substrates for the type II secretion pathway Xcp, including exotoxin A and elastase, are regulated by these systems, as are the xcp secretion genes themselves (19). This dual level of regulation should not only prevent premature secretion of proteins before the cells have reached a critical mass but also conserve energy. If there is no substrate to secrete, there is no reason to produce and assemble the secretion apparatus. LasRI homologues have also been identified for B. cepacia(80). In a B. cepacia cepR mutant, no extracellular protease can be detected, but it is not yet known if CepR regulates protease production and/or biogenesis of the type II secretion apparatus (80). Another recent report suggests that the expression of lipase and phospholipase C is also under quorum-sensing control in B. cepacia (156). In addition to intraspecies regulation, there is also apparent cross talk and interspecies signaling between P. aeruginosa andB. cepacia (97). Cell-free medium taken from stationary-phase cultures of P. aeruginosa increases the production of lipase and protease when added to cultures of B. cepacia (97). A quorum-sensing system has also been identified for E. carotovora, another species with a type II secretion pathway. The virulence of this species is dependent on production and secretion of a number of plant cell wall-degrading enzymes (Table 1). When a strain with a mutation in the E. carotovora epxI gene, which is homologous to the V. fischeri luxl gene, was analyzed, the production of these exoenzymes and the ability of the mutant strain to propagate and cause disease were reduced (69, 119). Finally, expression of theErwinia out secretion genes themselves is greatly increased in early stationary phase, suggesting that these genes may also be under quorum-sensing control (82).

In contrast to expression of the type II secretion genes ofPseudomonas and Erwinia, expression of theexe and eps secretion genes of A. hydrophila and V. cholerae may be constitutive (150; M. Sandkvist and V. J. DiRita, unpublished studies of V. cholerae). Addition of purified A. hydrophilaautoinducer N-(butanoyl)-l-homoserine lactone to cultures of A. hydrophila expressing anexeL::lacZ fusion had no effect onlacZ expression, and reverse transcriptase PCR analysis ofexeD mRNA production in wild-type A. hydrophilashowed that the exeD gene is constitutively expressed (150). While AhyRI does not regulate the exesecretion genes in A. hydrophila, the expression of the secreted serine protease AspA is growth phase regulated by this quorum sensing system (150). In V. cholerae, growth-phase-dependent expression of the type II secretion genes was also not observed since the expression of anepsE::lacZ fusion remained unchanged during exponential and stationary-phase growth (M. Sandkvist and V. J. DiRita, unpublished data). In addition, when wild-typeV. cholerae was cultured and samples were collected during growth to analyze for EpsE protein production by immunoblotting, the same level of EpsE protein per cell mass was produced during all growth phases tested (M. Sandkvist and V. J. DiRita, unpublished data). It is possible that constitutive expression of the secretion apparatus may be necessary for species that secrete substrates in several different environments. For instance, the Eps apparatus must be active not only in the human intestine, where it secretes CT, but also in the aquatic environment, where it is thought to secrete chitinase (Table 1) (25). Although the expression of the eps genes appears to be constitutive, it is still possible that certain growth conditions may affect the activity of the secretion apparatus. Interestingly, maltose has been found to inhibit secretion of CT in vitro (73). When maltose was present, only 22% of total amount of CT was secreted following 6 h of growth. The rest of CT accumulated inside the cells. In contrast, without maltose, 95% of CT was extracellular. Following overnight growth (17 h), the effect of maltose was even greater, with only 5% of the total amount of toxin detected in the growth media. The mechanism of this inhibition is not understood but it appears to be at least partially growth phase dependent, because no inhibition of CT secretion was observed following 3 h of growth. Interestingly, addition of maltose to wild-type V. cholerae cultures also had an adverse effect on secretion of HAP and biogenesis of the type IV pilus MSHA (73). HAP, like CT, is secreted via the Eps pathway (137). MSHA, on the other hand, requires another set of genes for its assembly and outer membrane translocation (46, 91, 93). However, secretion of CT, HAP, and MSHA requires the function of one common component (41, 92). The prepilin peptidase VcpD (PilD) processes the MSHA prepilin subunit as well as the pilin-like proteins EpsG, -H, -I, -J, and -K, which are required for CT and HAP secretion (92). Thus, one explanation could be that maltose interferes with the expression of the vcpD gene or the activity of the VcpD protein, thereby inhibiting secretion.

While the eps genes may be constitutively expressed and extracellular secretion is likely to occur in environments as diverse as the human intestine and the aquatic milieu, expression of the substrates for this pathway is tightly controlled. The regulation of expression of the CT genes ctxAB is complex and differs during in vitro growth and infection (76). In vitro, expression of ctxAB is greatly increased at 30°C in Luria broth (pH 6.5) with low osmolarity and requires the transcriptional activators ToxR, ToxT, and TcpP (48, 53, 101, 102). Interestingly, during colonization of the infant mouse intestine, TcpP is not required for toxin expression. Instead, ctxABexpression requires the toxin-coregulated pilus. In an elegant study, Lee and colleagues showed that the ctxAB and tcpAgenes are sequentially expressed during infection and that expression of ctxAB requires prior tcpA expression (76). This led the authors to suggest that colonization is a prerequisite for toxin production. With the imposition of such a regulatory control, toxin production and secretion will only occur whenV. cholerae has arrived at the intestinal epithelium (76).

When V. cholerae is cultured in vitro, HAP production is upregulated in stationary phase (68). This regulation occurs at the transcriptional level by HapR, a 203-amino-acid protein with 71% identity to LuxR of Vibrio harveyi. HapR may directly activate hap expression in V. choleraeor may indirectly regulate HAP production via the stationary-phase sigma factor RpoS. Yildiz and Schoolnik recently showed that inactivation of the rpoS gene in V. cholerae led to reduced levels of HAP production (160). Finally, just as HAP production is increased in late growth phase, the production of extracellular neuraminidase is increased more than 10-fold whenV. cholerae cultures reach stationary phase, and addition of sialic acid to the culture increases the production further (153).

Thus, the expression of most proteins secreted via the type II pathway is growth phase dependent or strictly regulated by environmental signals. In addition, the type II secretion genes themselves are under quorum-sensing control in some species. Whether the regulatory control occurs at the transcriptional level of a particular virulence factor or at the functional level of its secretion apparatus, extracellular secretion of the virulence factor will generally occur only when the bacteria have reached a critical mass and location. This is schematically depicted in Fig. 2, where in the case of V. cholerae, the colonization surface may represent the intestinal epithelium, where CT is secreted or the exoskeleton of crustaceans, where chitinase is released.

SECRETION AND DISEASE

The type II pathway appears to be typically associated with organisms that colonize surfaces, and in most cases these organisms do not invade cells. Several of these species are also known to form biofilms, and mutations in the eps genes interfere with polysaccharide production and biofilm formation in V. cholerae (2). Even so, it is unlikely that the Eps pathway is directly involved in the transport of polysaccharides to the cell surface, because the biosynthesis of polysaccharides appears to be controlled by another pathway in V. cholerae that is homologous to the polysaccharide biosynthesis systems of other species (161). Thus, the defect in biofilm formation in theeps mutants could be due to pleiotropic effects or a weakened outer membrane (137). Alternatively, it is possible that an as-yet-unidentified polysaccharide-processing enzyme required for assembly of polysaccharide on the cell surface could use the Eps system for outer membrane translocation.

The role of the type II pathway in pathogenesis has not been investigated in great detail. However, in species where it secretes known virulence factors, it is assumed to be essential for pathogenesis. In the case of animal and human pathogens, the type II secreted proteins might be required for the establishment of an infection at the mucosal surfaces of the respiratory and gastrointestinal tract. Secreted proteases, lipases, phospholipases, and toxins may aid in the colonization or tissue destruction by defeating host defenses and/or providing nutrients. In V. cholerae, CT and toxin coregulated pilus are the major virulence factors. Both factors have been analyzed in great detail and tested both in animal and human models. The symptoms, including severe diarrhea, vomiting, and headaches, observed with wild-type V. cholerae were considerably attenuated when a toxin-negative mutant was given to human volunteers (52). The number of volunteers with diarrhea and the number of diarrheal stools were reduced. In addition, the mean diarrheal stool volume was decreased from 5 liters for the wild-type strain to 0.2 liters for the toxin mutant. Moreover, ingestion of purified CT caused large volumes of diarrhea in a dose-dependent manner in volunteers (79).

We have hypothesized that the type II secretion pathway Eps is essential for V. cholerae virulence, since it secretes CT. Unfortunately, the eps secretion mutants of V. cholerae have not been tested for their ability to colonize and cause disease in vivo. These mutants grow slowly and are unable to compete with wild-type V. cholerae when cultured together in vitro (Z. Y. Dossani and M. Sandkvist, unpublished data); thus, any results from eps mutant colonization studies in the infant mouse model would be difficult to interpret. However, the role of VcpD, which is required for proper assembly and function of the Eps apparatus, has been tested in virulence and colonization studies. These results indicated that inactivation of the vcpD gene led to a 100-fold increase in the 50% lethal dose (LD₅₀) for the cholera infant mouse model and to an approximately 100-fold reduction in colonization efficiency (92). Since the vcpDmutant is pleiotropic, the reduced virulence of this mutant cannot be contributed to lack of CT secretion only.

Type II secretion mutants of other species have been analyzed for their ability to cause disease. Mutagenesis of the out secretion genes of the plant pathogen E. chrysanthemi produces mutants that are unable to secrete enzymes that digest plant cell walls and are severely attenuated (3, 152). On the other hand, whenP. aeruginosa xcp mutants were tested in a mouse burn infection model, they were found to be as virulent as the parental strain (158). The reason for this result is not clear, although it may suggest that low-level release of virulence factors such as exotoxin A and elastase from the xcp mutants is sufficient to cause disease. Alternatively, the level of toxin and protease gene expression may be increased to compensate for the reduced secretion level in the xcp mutants, or the second type II secretion gene cluster, recently identified in the P. aeruginosa genome, may partly complement the mutations. Similarly, secretion mutants of B. pseudomallei displayed only modest defects in ability to cause disease in the Syrian hamster model (33). The LD₅₀ for these mutants was increased 3- to 13-fold. In the case of L. pneumophila, the type II secretion pathway was found to be required for multiplication in amoebae (44, 81). However, inactivation of thelspG and lspH secretion genes did not have an effect on macrophage cytotoxicity, since like the wild-type strain, thelspGH mutant was able to kill HL-60-derived macrophages (44). On the other hand, a pilD knockout mutant demonstrated severe defects in the ability to grow within and kill macrophage-like U937 cells (81). In addition, the LD₅₀ for the pilD mutant was increased 100-fold compared to that of the wild-type strain administered intratracheally to Guinea pigs (81). These results suggest that the type II secretion pathway in L. pneumophila is not required for growth in macrophages, and that the attenuated phenotype observed with the pilD mutant may be due to lack of biogenesis of a pilus, yet to be identified, since PilD may affect both type II secretion and pilus biogenesis.

EXPLOITATION OF A SECRETION PATHWAY COMMON TO ALL VIBRIOS FOR TOXIN SECRETION AND HORIZONTAL GENE TRANSFER

Since the type II pathway is widely distributed it was of interest to determine whether all species have the capability to secrete CT. Although species such as P. aeruginosa, K. oxytoca, andE. chrysanthemi contain type II secretion genes homologous to the V. cholerae eps genes, they were unable to secrete the toxin B subunit when a plasmid encoding the B subunit gene was introduced (100). Yet several environmental vibrios other than V. cholerae were found to secrete plasmid-encoded B subunits to the extracellular environment (100). In addition, when the B subunit was expressed in the marineVibrio sp. 60, which is related to Vibrio anguillarum, it was also secreted (77). However, in the extracellular secretion mutant MVT1192, which is unable to secrete protease (62), the B subunit remained in the periplasm (77). When the MVT1192 mutant was analyzed further, we found that the secretion defect could be complemented with the V. cholerae epsE gene (M. Sandkvist and A. Ichige, unpublished data). Likewise, overproduction of an EpsE protein with a mutation in the ATP-binding site in the wild-type Vibrio sp. 60 inhibited protease secretion (Sandkvist and Ichige, unpublished data). This dominant-negative effect of the mutant EpsE protein was also observed for wild-type V. cholerae and is most likely a result of competition between wild-type and mutant EpsE proteins for interaction with EpsL and the rest of the secretion apparatus (134). The findings that the B subunit could be secreted from mostVibrio species and that the V. cholerae epsE gene was able to restore the secretion phenotype in a marineVibrio mutant, suggests that the Eps pathway is common to the Vibrionaceae family. Most of the species in this family are not thought of as human pathogens and are not known to produce enterotoxins. It is therefore likely that the secretion apparatus inVibrio is used primarily for secretion of proteins common to all vibrios, possibly proteases and chitinases. Vibriospecies are common inhabitants of the aquatic environment, and many, if not all, produce and secrete chitinases as well as proteases homologous to V. cholerae HAP. It is believed that the Eps apparatus did not originally evolve for the purpose of secreting CT but rather that CT has evolved in such a way as to be recognized and efficiently secreted by this machinery.

The structural genes for CT are encoded by the filamentous bacteriophage CTXφ, which is related to the E. coli phage f1 (155). The CTXφ genome is 7000 bp in length and chromosomally integrated, but it can also replicate as a plasmid (155). Like the coliphage f1, CTXφ is actively secreted without lysing the host cell. Secretion of CTXφ requires EpsD, the component of the Eps type II secretion apparatus that is likely to form the outer membrane translocation pore (28). This is consistent with the notion that CTXφ does not encode its own D pore for secretion and is in contrast to the homologous E. colif1 phage, which expresses its own outer membrane pore, the EpsD homologue pIV (84, 130). No other Eps protein is thought to be required for the extracellular release of CTXφ, suggesting that the secretory pathways for CT and CTXφ have converged at the site of the EpsD pore (28). This finding is intriguing and suggests that CTXφ has taken advantage of a preexisting secretion pore that is widely distributed within the Vibrionaceaefamily. As such, the EpsD protein has been converted to a virulence-export machinery. In this way, EpsD contributes to the pathogenesis of both the V. cholerae bacterium and the phage by actively secreting the toxin and by releasing the phage to allow for horizontal gene transfer.

CONCLUDING REMARKS

The type II secretion pathway is widely distributed among plant, animal, and human pathogens of the proteobacteria. Representatives are present in all subdivisions of the proteobacteria but are mostly clustered in the gamma subdivision and include species in theAeromonadaceae, Alteromonadaceae, Legionellaceae, Enterobacteriaceae, Pseudomonaceae, Vibrionaceae, andXanthomonadales families. It is possible that this pathway is confined to and is an intrinsic property of the proteobacteria. Alternatively, other bacteria may contain this pathway; however, if they are further removed from and less related to the proteobacteria, their putative secretion genes may not be identified by simple homology searches since the level of homology may be too low. For that reason, identification of this pathway in bacteria belonging to other groups may have to be obtained through functional studies. Although sequencing the secretion genes is important and provides valuable information, it cannot fully explain how proteins cross the outer membrane. Thus, determining the organization of the type II secretion apparatus and identifying the putative secretion signal on the secreted proteins is the focus of intense research at the moment. This information should lead to an increased understanding of the mechanism of secretion through the type II secretion pathway and may generate tools useful for therapeutic interventions. The wide distribution of the type II secretion pathway and its similarity to the type IV pilus biogenesis system suggests that the identification of specific drugs or peptides that can interfere with both pathways may simultaneously block secretion of virulence factors and colonization and could thereby provide a reasonable alternative to antibiotic use.