Sec-Dependent Secretion across the Inner Membrane

The GSP of gram-negative bacteria is a group of secretory pathways unified by the common requirement of the Sec system for translocating proteins across the IM (for a review, see reference 18). Proteins translocated via the Sec system carry an amino-terminal signal sequence that is cleaved by a signal peptidase when the protein is released into the periplasmic space. The genome of H. pylori strain 26695 has 1,590 open reading frames (62), of which 517 have a putative signal sequence indicating that these proteins could be secreted via the GSP (2). This figure may be an overestimate since it is based on sequence similarities to a peptide motif derived from the characteristics of known signal peptides. Nevertheless, it indicates that a substantial proportion of proteins synthesized by H. pylori are destined to cross at least the IM. In comparison, Haemophilus influenzae has a similar number of putative genes (1,680), but only has 330 predicted signal peptides (8).

Homologs of essential Sec-related genes besides secA and a set of signal peptidases are found in H. pylori. At the same time, some factors important for the sec system in Escherichia coli are not found in H. pylori. One of these is the SecB chaperone. The presence of a signal sequence on a protein slows down the folding of the whole protein, allowing chaperones like SecB to bind and prevent premature folding that would interfere with export. SecB binds to a group of proteins targeted for secretion and also guides them to exit sites in the membrane. A factor reported to direct the interaction of SecB with this set of proteins is the trigger factor, and a homolog of this factor is found in the H. pylori genome. Apparent orthologs to secB are also missing in some of the other bacterial genomes recently sequenced, despite the presence of other components of the Sec system (61). Taken together, this makes it likely that a protein with a "SecB-like" function is present in H. pylori even though no open reading frame has been annotated as secB.

An alternative route for targeting to the membrane is via the signal recognition particle (SRP). It recognizes the signal sequence and, together with cell division FtsY protein, targets mainly hydrophobic proteins with multiple membrane-spanning regions to the cell membrane for export. The SRP is composed of the Ffh protein and a 4.5S-RNA species. Both ffh and ftsY homologs are present in H. pylori, suggesting the presence of this pathway. A second "missing" gene in H. pylori is secE. E. coli SecE, SecY, and SecG make up the membrane-located protein translocase unit together with the auxiliary SecD and SecF. In E. coli only SecE and SecY are essential translocase components; thus, it is surprising to find homologs in H. pylori to all E. coli translocase components except SecE.

Secretion across the Outer Membrane

Missing Pathways

The second terminal branch of the GSP is the chaperone/usher pathway, responsible for export and assembly of (adhesive) virulence-associated structures such as P and type 1 pili of uropathogenic E. coli (60). For export through this pathway only two accessory proteins are needed for transfer across the OM: a chaperone guiding the protein through the periplasmic space, assisting in folding and preventing the protein from premature interactions, and an usher protein for translocation through the OM. No component of this pathway has been found in H. pylori and, in addition, no surface organelles besides the flagella have been described.

The third terminal branch of the GSP is type II secretion, also called the main terminal branch of the GSP. The type II secretion pathway is closely related to the biogenesis of type IV pili and is considerably more complex than the autotransporter pathway, requiring 12 to 16 accessory proteins (for a review, see reference 46). No example of this export mechanism has been described for H. pylori, nor have homologous genes to those encoding type II proteins been identified in the annotated genomes of H. pylori strains 26695 or J99. It is not very surprising not find type II secretion candidate genes absent from the H. pylori genome, since, in fact, only a few of the bacterial genomes sequenced have components for type II secretion in their annotated sequences.

The TAT Secretion System

A new Sec-independent system for translocation of proteins across the IM has recently been identified, and orthologs of the essential genes of this system have been found in the H. pylori genome (48). Like the Sec system, it requires an amino-terminal signal peptide but with specific different features. Most noticeable is the requirement of two consecutive arginine residues preceding the hydrophobic core region of the signal peptide. This feature served to give the new system the name "twin arginine translocation," or Tat-system. The system was first found in chloroplasts of maize where it constitutes a ΔpH-driven transport system for protein transport across the thylakoid membrane. The Hfc106 protein is necessary for this pathway in maize, and homologs to hfc106 are found in a wide variety of bacterial genomes (for a review, see reference 13). In E. coli a set of genes homologous to hfc106 has been found; two of them, tatA and tatB, are organized as part of an operon, and a third named tatE is in an unlinked position. The gene tatB is essential for the pathway to function, because a knockout of this gene completely blocks Tat-dependent secretion. Neither tatA nor tatE is essential, although efficiency is reduced in knockouts of either gene. In a double tatA/tatE knockout the pathway is inactive. Interestingly, a tatA knockout in E. coli can be functionally complemented by the H. pylori tatA homolog (48). However, E. coli tatB cannot be complemented by H. pylori tatB, which is in agreement with the fact that the tatB genes of the two bacteria are much more divergent than tatA. Nevertheless, it is unusual that the most essential gene of the pathway is the least conserved between the two bacteria.

Sec-dependent translocation across the IM requires the substrate protein to be unfolded. On the other hand, it has been reported that proteins transported by the Tat pathway in most cases bind redox cofactors, which seem to fold and even oligomerize before crossing the membrane (47, 54). A motif search in the H. pylori genome reveals a number of genes encoding proteins with the twin arginine motif typical of the Tat system in their signal sequences, and as expected, among them are genes coding for proteins using redox cofactors.

Flagellar Export

Motility is mediated by the sheathed flagella in H. pylori (22) and is necessary for colonization in animal models (17). In the context of this chapter the flagellar export apparatus potentially is interesting owing to its similarity to type III export systems.

The prototype flagellum is anchored to the bacterium by a flagellar basal body that spans both inner and outer membranes and is structurally similar to the type III secretion apparatus (34). Flagellar subunits are secreted through a channel within the basal body and polymerize at the distal end of the growing flagellum. Recently, in a different system, it has been demonstrated that the flagellar transport apparatus of Yersinia enterocolitica also functions as a protein export system for a pathogenesis-related phospholipase, placing the flagellar export/assembly system under a new light (66). To date, a similar phenomenon has not been reported for H. pylori. Based on homologies to better defined systems, the genome of H. pylori 26695 contains more than 40 genes suggested to encode either structural components of flagella or proteins involved in its biogenesis and regulation (62). Two structural H. pylori genes have been demonstrated to encode flagellin FlaA and FlaB (35), and a third structural gene encodes FlgE, the hook protein joining the flagella filament to the basal body (38). Most of what is known about the flagellar export apparatus (the flagellar basal body) of H. pylori is based on the presence of genes homologous to others encoding parts of the flagellar export machinery in other bacteria. Knocking out fliI and fliQ, which are proposed to encode components of the transport apparatus, rendered H. pylori aflagellated and nonmotile (26, 45). The organization of genes involved in flagellar synthesis in H. pylori is remarkedly different from that of other bacterial genomes. In H. pylori flagellar genes are scattered over the genome, contrary to their clustered presence in other bacterial genomes.

Altruistic Autolysis

The proteins discussed in this section, urease, heat shock proteins, and superoxide dismutase, are found only in the cytoplasm of most bacterial species. In H. pylori these proteins are found also on the bacterial surface and in the growth media (19, 23, 56). Since in other species they are not exported proteins and also lack signal sequences for export by the GSP, the mechanisms through which these proteins are excreted are intriguing, especially in the case of urease. This enzyme hydrolyzes urea to ammonia and carbon dioxide. The ammonia formed can neutralize hydrochloric acid of the gastric juice, creating a more suitable microenvironment around the bacteria. It can be used also for metabolic purposes by the bacterium and has toxic effects on epithelial cells. Urease activity is necessary for pathogenesis in animal models (15, 16) and is the basis for the highly reliable urea-breath test for detecting H. pylori infection.

Urease and the heat shock protein HspB are located in the cytoplasm as well as on the bacterial surface both in vitro and in vivo (14). The urease located on the bacterial surface is reported to be catalytically active since inactivation of extracellular urease with a membrane-impermeable urease inhibitor makes H. pylori susceptible to acid (33). The enzyme is a 550-kDa multimer consisting of six ureA and six ureB subunits. For activity it needs nickel ions and a complex of accessory proteins to import the ions, to incorporate them into the apoenzyme, and to produce the active holoenzyme. An obvious obstacle for export of an active urease is its large size, unless it can be exported as subunits.

Altruistic autolysis, the "unselfish" autolysis of some bacteria for the benefit of the whole population, has been reported for some pathogenic bacteria. Autolysis of Streptococcus pneumoniae appears to enhance virulence (4) and is genetically regulated (40). Autolysis of Neisseria gonorrhoeae is suggested to be a way for the bacteria to release DNA that can be taken up by other nonlysed bacteria, i.e., transformation, leading to increased antigenic variation of pili subunits (27, 53).

Autolysis has been reported also for H. pylori. It has been suggested that enzymatically active urease and HspB are exported in this way from lysed bacteria and adsorbed to the surface of other living bacteria (43).

Kinetic studies of protein release from H. pylori demonstrate that VacA can be detected at early time-points as a major Helicobacter protein in the culture media, while most H. pylori proteins appear in the media at later times in combination with a lowering of turbidity of the broth media, suggesting cell lysis (50). However, autolysis as a major mechanism for export of urease and HspB has been questioned. Instead, export by specific pathways has been suggested for UreA, UreB, HspA, and HspB on the basis of reported quantitative differences in the relative amount of these proteins found in the growth media compared with their relative abundance in whole H. pylori bacteria (63). The reasoning behind this proposal is that if the proteins were released by lysis, their relative abundance would be the same in intact bacteria as in culture supernatant. Since this does not correspond to the observations, the data support specific export. However, to make this hypothesis more consistent, the relative stability of these proteins must be determined to rule out protein degradation as a contributing factor to the results. Other investigators have suggested that superoxide dismutase is also selectively released from the bacteria (50). In conclusion, there is no clear consensus on how these factors are exported, and the question remains on the mechanisms for export if not by autolysis.

Budding Outer Membrane Vesicles—a Tool for Protein Export

The cell wall of gram-negative bacteria has a feature not seen among gram-positive organisms, namely, the constant shedding during growth of membrane vesicles from the bacterial outer membrane. Outer membrane vesicles (MV) are formed by budding from the OM, thus entrapping periplasmic components in the process. From this perspective, MV can be regarded as small (50- to 250-nm) units of gram-negative cell wall, complete with OM and periplasmic contents (for a review, see reference 5). Even though MV were observed and reported 30 years ago, their existence and importance have become better recognized during the last years. MV of H. pylori bind to target cells followed by uptake and localization to membranous structures. The integrity of MV located intracellularly is maintained for an extended period (20, 55). For a protein to be exported via an MV it must first be located to the periplasm or OM. This would most likely, but not necessarily, involve a Sec-dependent transfer across the IM. The H. pylori VacA toxin is a Sec-dependent autotransporter protein. It is secreted as a soluble oligomer into the growth media but is also found associated with the OM and MV when cultured in vitro and in vivo (20, 31). Purified MV from H. pylori induce vacuole formation in cultured epithelial cells (20, 31). The purified toxin has to be activated by acid to be able to induce vacuoles in eukaryotic cells. This is not the case for the toxin on the bacterial surface or the toxin associated with MV that do not need to be activated by acid treatment. The finding of VacA toxin associated with MV suggests an alternative way for export/delivery of the toxin. Surprisingly, a number of proteins found on the surface H. pylori cells are not associated with its MV. Urease B and HspB could not be detected by Western blot analyses of MV, nor could urease activity be detected in the MV (31). It is unclear whether the proteins are selectively excluded from the MV or simply less strongly associated.

The most intensively studied MV are those of Pseudomonas aeruginosa. They were shown to contain protease, phospholipase C, alkaline phosphatase, and autolysin, all of which are examples of periplasmic proteins (28–30). The membrane composition of MV is similar to the bacteria from which they derive. Some surface features of the MV must, however, be different from those of the bacterial cell wall from which they originate. The curvature of MV is very different from that of the bacterial cell wall, thus making it likely that physical properties and composition of the vesicle membrane and the bacterial OM somehow differ to accommodate the necessary higher curvature. An important difference observed in P. aeruginosa MV is that instead of containing both A- and B-band lipopolysaccharide (LPS) as the bacterial OM, the vesicle membrane only contains B-band LPS. Clustering of B-band LPS into the small domains of vesicles that are being formed has been suggested to allow for the higher curvature of the MV, and also as a driving force in formation of vesicles (30).

Recent work on MV shed from enterotoxigenic E. coli (24) demonstrated that active heat-labile enterotoxin can be shed from bacteria in MV, and that vesicles are enriched in toxin compared to the periplasm of the bacteria from which they derived. Heat-labile toxin was found both inside and on the surface of the vesicles. Vesicule formation in enterotoxigenic E. coli has been proposed to occur at specific sites in the OM with specific protein composition, since the relative abundance of membrane proteins differs between bacteria and MV. Work with MV produced by Shigella flexneri demonstrated the potential of MV for delivery of effector molecules into the host cell cytoplasm. S. flexneri MV containing the membrane-impermeable antibiotic gentamicin were taken up by mammalian cells, and the antibiotic effect was detected in the host cell cytoplasm (29).

There are potential advantages of MV as vehicles for delivery of effector proteins. MV can carry a selection of factors from the bacterial OM. They may protect proteins from degradation and scavenging function by the host to which soluble secreted proteins are subjected. In addition, if the MV display adhesive factors recognizing structures on a target cell, they can be regarded as targeted vehicles for local delivery of high concentrations of virulence factors, thus avoiding the problems of dilution related to secreted proteins. H. pylori appears to make ample use of these strategies.