Vol. 15, No. 14, pp. 1725-1752, July 15, 2001

REVIEW
Protein secretion and the pathogenesis of bacterial infections

Department of Microbiology & Immunology, UCLA School of Medicine, Los Angeles, California 90095, USA

	Instruments of bacterial warfare

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

The skin, the oral cavity, and the gastrointestinal tract of humans are colonized with bacteria. Microbial entry into the blood or blood-circulated tissues is hindered by anatomical barriers. The barriers consist of epithelia as well as membranes that are fortified by layers of collagen and other connective tissues. Following the breakdown of a barrier, the dwindling of an immune system, or an attack by particularly virulent bacteria, microbes gain entry into deeper tissues and multiply within newly conquered space. Human disease is the result of such bacterial multiplication. Microbial entry into circulated tissue is accompanied by an immune response. Immune cells recognize bacterial products (lipids, carbohydrates, peptidoglycan, or protein decorations) and respond by attracting macrophages, polymorph-nuclear leukocytes, or other immune cells in an effort to kill the invading pathogen (Medzhitov and Janeway 1999; Aderem and Ulevitch 2000). Many bacterial pathogens have evolved to enter and multiply within blood-circulated tissues (Finlay and Falkow 1997). The underlying pathogenic strategies are remarkably diverse and often result in unique disease symptoms. Nevertheless, all mechanisms of bacterial manipulation of the host organisms can be viewed in three principal categories: microbial adhesion, secretion of toxins into the extracellular milieu, and injection of virulence factors into host cells. There are three rules of thumb that bacterial pathogens must consider if they want to mount a successful infection.

If you want to invade a hoststick to it

Vibrio cholerae adhere to human intestinal tissues and secrete a toxin that, once engulfed by epithelial cells, causes a fulminant diarrhea. Mutants lacking the Vibrio surface adhesin (TCP pili) cannot stick to the intestines and fail to cause disease (Herrington et al. 1988). Microbial adhesion to host tissues can be mediated by individual proteins or by sophisticated organelles such as pili. Pili form fibrous structures that emanate from the bacterial surface and display an adhesive property at the tip. Pili are often distributed over the bacterial surface (pertrichious pili, e.g., type I pili) or they are located at a single site (polar pili, e.g., type IV pili). Some pili are retractable (type IV pili) and provide for adhesion and bacterial movement. Other pili are involved in conjugation, that is, the exchange of genetic information between cells, and provide for adhesion, retraction, and the transport of DNA. Flagella, another surface organelle of bacteria, are more flexible and longer than pili. Flagella do not display adhesive properties but function as a rotating propeller that allow bacteria to swim during infection.

If you don't want to get killedbring your weapons and fight

Invasin is an adhesive outer membrane protein of Yersinia pseudotuberculosis, a microbe that colonizes lymphoid tissues after invading the intestines of humans (Isberg and Falkow 1985). Expression of invasin in Escherichia coli K-12 is sufficient for a nonpathogenic microbe to invade epithelial cells (Isberg et al. 1987). Nevertheless, invasin alone can not confer the ability to cause disease. What does it take to generate a potent pathogen such as Y. pseudotuberculosis? Protein secretion machines are the instruments of microbial warfare! Many bacteria secrete toxins into the extracellular milieu during infection. For example, Staphylococcus aureus secretes several exotoxins that damage host cell membranes (Dinges et al. 2000). Although these compounds diffuse within extracellular fluids, their concentration is highest in the immediate vicinity of staphylococci, generating "mine fields" that keep immune cells at bay. Other toxins act at a distance from the invading pathogen much like a bullet or an "intelligent" missile. Tetanus toxin circulates in blood and adheres to neuronal receptors (Schiavo et al. 2000). Following endocytosis, the engulfed toxin exerts its pathogenic property inside the target cell and cleaves proteins to prevent the fusion of synaptic vesicles with the plasma membrane. The infected host dies from respiratory paralysis while the underlying infection itself, typically contamination of a minor wound with Clostridium tetani, is located elsewhere and plays only an indirect role in the outcome of the disease. Bacteria have evolved a bounty of mechanisms for toxin secretion: the Sec pathway, autotransporter, type I-IV secretion, and toxin release systems (e.g., tetanus toxin). The type III and type IV secretion systems of Gram-negative pathogens can be viewed as bacterial arms for close combat with immune cells. A protein conduit is formed between the pathogen and the host cell that provides for the rapid and massive deposition of proteins (type III and type IV) or for the transport of DNA (type IV).

If you want to invade another hostget out if you can

The killing of a human host allows Clostridium tetani to multiply as these anaerobic microbes can only grow in tissues that are not circulated with oxygenated blood. But the killing doesn't provide microbes with a mechanism for transmission to a new host. Several enteric pathogens, for example, Salmonella, Shigella, or V. cholerae, have elegantly solved the problem by spreading via fecal-oral contamination. During V. cholera infections enough cholera toxin is secreted to cause a fulminant diarrhea with large amounts of infectious microbes while carefully avoiding the rapid killing of the host. Again, it is protein secretion that matters as too much of a toxin is not always a good thing and can limit the success of a pathogen. This article describes the principles of protein secretion by pathogenic microbes.

	Gram-positive and Gram-negative microbes

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Bacteria can be classified with the staining procedure of Christian Gram. Gram-positive bacteria retain the dye crystal violet, which is removed with ethanol from the envelope of Gram-negative bacteria (Popescu and Doyle 1996). The different staining properties are caused by differences in envelope structure. Gram-positive microbes elaborate a single plasma membrane (also called inner membrane [IM]; Fig. 1A) followed by a thick cell wall layer (CW) that functions as a surface organelle for the display of carbohydrates and proteins (Fig. 1; Ghuysen and Hackenbeck 1994). A double membrane surrounds Gram-negative bacteria, enclosing the periplasmic space and peptidoglycan layer between two lipid bilayers (Fig. 1; Inouye 1979). The outer membrane (OM) of Gram-negative bacteria functions as a surface organelle for the display of proteins and carbohydrates (Inouye 1979). Assembly of some bacterial envelope structures, for example, outer membrane or cell wall, is essential for the growth and the replication of microbes (Ghuysen and Hackenbeck 1994). In contrast, the secretion of proteins into the extracellular milieu appears to be essential for bacterial multiplication in infected host tissues, but it is not required for microbial growth under laboratory conditions (Pugsley 1993a).

View larger version (16K): [in this window] [in a new window]   Figure 1.   The Sec pathway transports proteins across the bacteria plasma membranes. (A) The SecA and SRP pathway of Gram-positive bacteria translocates proteins across the plasma membrane. The cell wall envelope (CW) of some organisms is thought to be impermeable for polypeptides, however factors for protein transport across the cell wall have hitherto not been identified. Proteins are translocated across the plasma membrane of Gram-negative bacteria in a post-translational (B) or cotranslational manner (C). (IM) Inner membrane. (B) Signal peptide (black box) bearing precursors are recognized by SecA and SecYEG. Upon substrate binding and ATP hydrolysis, SecA pushes the precursor through the SecYEG translocon in a manner that also requires SecDF and YajC. After the signal peptide is removed by signal peptidase, the translocated protein is released into the periplasm. (C) Signal peptides of nascent polypeptides are recognized by the signal recognition particle, SRP, which is composed of Ffh and Ffs. This interaction stalls ribosomal translation, which is restored once the SRP complex interacts with the SRP receptor, FtsY, and the ribosomes is docked on the translocon. The membrane protein YidC is also thought to be involved in the cotranslational secretion of membrane proteins. (OM) Outer membrane. Drawings adapted from Duong et al. (1997).

	Secretion across the plasma membrane

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Protein secretion into the extracellular milieu by Gram-positive pathogens requires transport of polypeptides across the plasma membrane and the cell wall envelope (Fig. 1A; Smith et al. 1978, 1980, 1981; Schneewind et al. 1992). This process has been studied very little. Genome sequencing of a variety of Gram-positives showed that many of the secretion genes, which were initially identified in E. coli (Schatz and Beckwith 1990), are also present in these organisms. Transport of proteins across the plasma membrane of Gram-negative organisms leads to secretion into the periplasm but not to secretion into the extracellular milieu envelope (Fig. 1B,C; Pugsley 1993a). The thin peptidoglycan layer of Gram-negative organisms is not thought to function as a permeability barrier for folded proteins. Gram-negative organisms have evolved dedicated secretion systems that transport polypeptides beyond the outer membrane.

The Sec pathway has evolved to translocate polypeptides across the plasma membrane. Proteins destined for transport by this pathway are synthesized as signal peptide bearing precursors (Lingappa and Blobel 1980; Silhavy et al. 1983; Gennity et al. 1990). Signal peptides are generally positioned at the N terminus and consist of one or more positively charged amino acids followed by a stretch of 10-20 hydrophobic amino acids (Silhavy et al. 1983; Gennity et al. 1990). Once the precursors have been translocated across the membrane, the signal peptides are removed by signal peptidase (Chang et al. 1982; Dalbey and Wickner 1985). Some polypeptides insert into the plasma membrane using a noncleavable signal peptide, also referred to as a signal/anchor sequence (type I membrane protein; Davis and Model 1985; Davis et al. 1985; Singer 1990). Other polypeptides insert into the plasma membrane by using both a cleavable signal peptide and a second hydrophobic sequence that is located downstream (stop transfer or membrane anchor sequence-type II membrane protein; Lingappa et al. 1978). The Sec pathway was first characterized by searching for E. coli suppressor mutants that restore secretion of an outer membrane protein with a defective signal (Emr et al. 1981). Another search used LacZ fusions to the C terminus of secreted proteins, resulting in hybrids that jam the Sec pathway as LacZ folds rapidly into a nonsecretable conformation (Silhavy et al. 1976, 1977). Temperature sensitive mutations in sec genes allowed the internalization of a larger fraction of LacZ hybrids, thereby conferring a Lac-up phenotype (Oliver and Beckwith 1981). These genetic approaches have been complemented by biochemical studies measuring the Sec-mediated translocation of precursor proteins into membrane vesicles (Muller and Blobel 1984; Brundage et al. 1990). The membrane-embedded translocation machinery is composed of SecYEG and YajC, which together with the membrane proteins SecDF and cytoplasmic ATPase SecA, are sufficient to promote the translocation of precursor proteins into the lumen of membrane vesicles in vitro (Fig. 1B; Duong and Wickner 1997).

The translocation pore of the E. coli Sec pathway is composed of three membrane proteins, SecYEG, which can accept substrates in two ways (Duong et al. 1997; Pohlschroder et al. 1997). Signal peptide-bearing precursor proteins are maintained in a secretion-competent state by binding to chaperones, for example, SecB (Randall 1992). Bacillus subtilis and other Gram-positive bacteria make do without SecB, and other chaperones presumably function as substitute (Kunst et al. 1997). Translocation substrate is transferred to SecA, an ATPase that undergoes conformational rearrangements upon interacting with the secretion machinery, a process that is thought to push precursor proteins through the translocation pore (Economou and Wickner 1994). Once polypeptides have been translocated, signal peptidase removes the signal peptide from the precursor and mature protein is released into the periplasmic space (Dalbey and Wickner 1985). A second mechanism of protein secretion involves the cotranslational translocation of membrane proteins (Ulbrandt et al. 1997). The signal recognition particle (SRP), Ffh and 4.5 S RNA (ffs) in E. coli, binds to signal peptide-bearing nascent polypeptides, an interaction that is thought to stall ribosomal translation (Poritz et al. 1990). Once the SRP complex is bound to its receptor (FtsY in E. coli) and the ribosome has docked on the translocation pore, translation resumes and presumably provides the force to translocate polypeptides across the membrane (Miller et al. 1994). Three membrane proteins, SecD, SecF, and YajC, associate with the SecYEG pore and appear to regulate SecA-dependent translocation activity (Fig. 1C; Duong and Wickner 1997). The membrane protein YidC appears to be required for the insertion of polytopic membrane proteins into the plasma membrane (Samuelson et al. 2000); an association of YidC with the SecYEG translocase has not yet been revealed.

The genome of the Gram-positive pathogen S. aureus contains secAYEG and yajC similar to E. coli and B. subtilis (Blattner et al. 1997; Kunst et al. 1997). A secB gene could not be found, however, a second set of secretion genes, secA-2 and secY-2, was identified. As observed for B. subtilis, the S. aureus genome encodes for a secDF fusion gene but not for single secD and secF genes (Bolhuis et al. 1998). Two signal peptidase genes (spsA and spsB) are present in the S. aureus chromosome (Cregg et al. 1996), whereas ffh and ftsY are present in single copy only. It is not clear whether the presence of two sets of secretion genes results in the assembly of two secretion pathways in S. aureus. The cell wall envelope of some Gram-positive bacteria acts as a diffusion barrier for the movement of proteins. Few studies have addressed whether additional factors are required to facilitate secretion of proteins across the bacterial cell wall. B. subtilis PrsA is a peptidyl-prolyl isomerase and tethered to the bacterial plasma membrane via thioether diacylglyceride modification (Jacobs et al. 1993; Kontinen and Sarvas 1993). PrsA is thought to act on translocated proteins by catalyzing their folding and release into the extracellular medium (Vitikainen et al. 2001).

	Bacterial surface organelles

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Surface proteins of Gram-negative organisms are inserted into the outer membrane. The outer membrane is an asymmetric bilayer with an inner phospholipid leaflet and an outer leaflet that is largely composed of lipopolysaccharide (LPS; Raetz 1987). Five or six acyl side chains of LPS create the hydrophobic environment and are attached to the phosphorylated saccharide KDO (Imoto et al. 1983; Qureshi et al. 1983; Takayama et al. 1983). The phosphoryl groups of neighboring LPS molecules are complexed with magnesium ions, an ionic interaction that is important for the integrity of the membrane envelope. Attached to KDO and protruding on the bacterial surface are long polysaccharide chains, whose composition is highly variable among bacterial species (Rietschel 1984) and even subject to regulated variation within organisms (Ernst et al. 1999). Proteins destined for the outer membrane are translocated across the inner membrane via the Sec pathway. A large number of folding factors appear to act on outer membrane proteins (Missiakas and Raina 1997; Danese and Silhavy 1998), which assume -barrel structures that expose the alternating hydrophobic residues of -sheets at the interface with membrane lipids (Cowan et al. 1992). The outer membrane is tethered to the peptidoglycan layer via murein (Braun) lipoprotein, a short helical polypeptide that is covalently linked to the peptidoglycan at the C-terminal end (Braun and Hantke 1974). The N terminus of lipoprotein is inserted into the inner leaflet of the outer membrane with its diacylglyceride decoration (Hantke and Braun 1973). Electron microscopic studies suggest the existence of adhesion zones between inner and outer membranes (Bayer 1979; Lopez and Webster 1985). The physiological relevance of these adhesions has been under debate for a long time. It is not yet clear whether the adhesions truly represent interacting lipid bilayers and whether adhesions play a role in the assembly of the outer membrane and its proteins. Pulse-labeling experiments revealed the appearance and disappearance of soluble intermediates in pathways that leads to the insertion of outer membrane proteins (Stader and Silhavy 1988; Brissette and Russel 1990). These observations certainly do not support a model whereby outer membrane proteins are transported via lipid adhesions.

Several outer membrane proteins have been characterized as receptors for host proteins or tissues (Fig. 2A; Isberg and Falkow 1985; Isberg et al. 1987). One well-studied example is Yersinia pseudotuberculosis invasin. The N-terminal domain of invasin is inserted into the outer membrane (Leong et al. 1991), whereas the C-terminal domain protrudes 180 Å on the bacterial surface using a string of IgG-like domains as a folding module (Hamburger et al. 1999). Positioned at the C-terminal end of invasin is the adhesive part of the molecule with a folded structure resembling that of C-type lectins (Kelly et al. 1999; Luo et al. 2000). Invasin binds 1 integrin surface receptors of host cells (Isberg and Leong 1990) and this mechanism is thought to be instrumental during the intestinal uptake of Y. pseudotuberculosis and Y. enterocolitica, leading to bacterial infection of lymphoid tissues within the intestines (Pepe and Miller 1993). Intimins are another group of receptors displayed on the surface of E. coli as well as several other Gram-negative pathogens and provide a function that allows bacteria to inject virulence factors via the type III pathway (see below) into host cells (Kenny and Finlay 1997). Intimin binds to Tir, a secreted bacterial protein that inserts into host cell membranes where it is phosphorylated (Kenny et al. 1997). The mechanism of target cell selection for Tir insertion has not been established. The structures of intimin and the intimin/Tir complex have been solved and consist of three immunoglobulin folds connecting a C-type lectin-like receptor binding domain to the outer membrane (Kelly et al. 1999; Luo et al. 2000).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Display of proteins on the bacterial surface. (A) Gram-negative bacteria anchor surface proteins in the outer membrane (OM). After translocation and signal peptide cleavage of precursor proteins, outer membrane proteins (OMPs) are folded in a manner that requires the peptidyl-prolyl isomerases SurA and Fkp. Skp and the presumed cofactor lipopolysaccharide are thought to promote membrane insertion of OMPs. (CW) Cell wall; (IM) inner membrane. (B) Gram-positive bacteria anchor surface proteins to the cell wall envelope. Signal peptide bearing surface proteins are initiated into the Sec pathway. The C-terminal sorting signal (hatched box), which is composed of an LPXTG motif, a hydrophobic domain, and charged tail, retains translocated polypeptides within the secretory pathway. Sortase, a membrane anchored transpeptidase, cleaves the sorting signal between the threonine and the glycine of the LPXTG motif and links the C-terminal carboxyl group to the amino group of peptidoglycan cross-bridges within the cell wall envelope. Drawings adapted from Duong et al. (1997) and Navarre and Schneewind (1999).

Surface proteins of many different Gram-positive bacteria are tethered to the cell wall envelope by a mechanism requiring a C-terminal sorting signal with an LPXTG motif (Fig. 2B; Schneewind et al. 1992, 1993). After initiation of surface proteins into the Sec pathway, the sorting signal is recognized and cleaved by sortase, a membrane-associated transpeptidase (Mazmanian et al. 1999; Ton-That et al. 1999). The carboxyl group of threonine within the LPXTG motif is amide linked to peptidoglycan cross-bridges, thereby tethering the C-terminal end of surface proteins to the cell wall envelope (Schneewind et al. 1995; Ton-That et al. 1997; Navarre et al. 1998). A second targeting mechanism for surface proteins involves the binding of polypeptides to designated envelope structures, cell wall teichoic acids, or lipoteichoic acids (Sanchez-Puelles et al. 1990; Baba and Schneewind 1996, 1998; Jonquieres et al. 1999; Varea et al. 2000). Teichoic acids are often composed of polyribitolphosphate or polyglycerophosphate backbone structures (Fischer 1997). Species-specific esterified decorations of these molecules appear to be required for the anchoring of some surface proteins to the cell wall envelope (Holtje and Tomasz 1975). Bacterial crystalline layers are surface organelles that are assembled from secreted polypeptides (Sleytr et al. 1993). After secretion, the polypeptides bind to carbohydrate receptors on the bacterial surface and aggregate into regularly shaped arrays, which form the new surface layer (Ries et al. 1997; Egelseer et al. 1998). For example, surface layer proteins of B. anthracis adhere to pyruvylated cell wall polysaccharide (Mesnage et al. 2000). The Gram-negative organism Aeromonas hydrophila employs a type II secretion pathway (see below) to assemble an array of paracrystalline protein on the bacterial surface (Thomas and Trust 1995a,b).

	Secretion pathways of Gram-negative bacteria

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Research over the past two decades has identified a large number of secreted proteins that play important roles during the pathogenesis of infections caused by Gram-negative bacteria. The pathways whereby these proteins are transported have been separated into several classes: (1) type I pili (Soto and Hultgren 1999; Sauer et al. 2000); (2) auto-transporters; (3) type I secretion (Koronakis and Hughes 1996; Binet et al. 1997); (4) type II secretion (general secretory pathway [GSP]; Russel 1998) and type IV pili (Nunn 1999); (5) type III secretion (Hueck 1998; Galan and Collmer 1999) and flagella (Macnab 1992); and (6) type IV secretion (Christie and Vogel 2000) and DNA conjugation (Pansegrau and Lanka 1996). The classification is based on the molecular nature of the transport machineries and their catalyzed reactions. Many excellent reviews that provide a more detailed and in-depth treatise of the various subjects have been published recently for each of the six classes of secretion pathways (Macnab 1992; Koronakis and Hughes 1996; Pansegrau and Lanka 1996; Binet et al. 1997; Hueck 1998; Russel 1998; Galan and Collmer 1999; Soto and Hultgren 1999; Sauer et al. 2000; Christie 2001; Plano et al. 2001; Sandkvist 2001). This review will emphasize recent findings about secretion machines and the various mechanisms whereby proteins can be transported.

	Assembly of type I pili

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Type I pili (Duguid et al. 1955; Brinton 1959) allow E. coli and other Gram-negative bacteria to attach to host cells during the initial stages of infection (for detailed review, see Soto and Hultgren 1999; Sauer et al. 2000). Pili are essential for bacterial colonization of the urinary tract as the invading microbes are confronted with an additional defense barrier, the flow of urine. Employing adhesion and retraction properties of pili, bacteria advance against the flow of urine and colonize the lower (cystitis) or upper (pyelonephritis) urinary tract of humans. E. coli strains associated with pyelonephritis display Pap pili, structures that are composed of three parts: (1) the tip adhesin; (2) the fibrillum; and (3) the pilus rod (Lindberg et al. 1987; Kuehn et al. 1992). The base of the pilus rod is embedded in the outer membrane of Gram-negative bacteria. Pap (type I) pili are assembled from six pilins, all of which are encoded by a single large operon (Soto and Hultgren 1999). To identify the position of pilins within the pilus, wild-type E. coli and mutants lacking single pilin genes were analyzed by electron microscopy and immuno-gold labeling (Norgren et al. 1984; Baga et al. 1987; Lindberg et al. 1987; Kuehn et al. 1992; Jacob-Dubuisson et al. 1993). PapG pilin (G in Fig. 3) functions as an adhesive molecule and is located at the pilus tip (Fig. 3; Lindberg et al. 1987; Lund et al. 1987). The N-terminal domain of PapG binds digalactoside, a glycolipid that is located on the surface of host cells (Kallenius et al. 1980; Leffler and Svanborg-Eden 1981; Bock et al. 1985). The C-terminal domain of PapG is incorporated into the fibrillum (Hultgren et al. 1989) via the adapter protein PapF (Lindberg et al. 1987; Kuehn et al. 1992; Jacob-Dubuisson et al. 1993). The fibrillum, which consists mostly of PapE subunits, is attached to the pilus rod via a second adapter protein, PapK (Jacob-Dubuisson et al. 1993). PapA is the major pilin subunit and assembles to form the pilus rod (Normark et al. 1983; Norgren et al. 1984). PapH terminates the polymerization of PapA and is presumably positioned at the pilus base within the plane of the outer membrane bilayer (Baga et al. 1987). PapJ is required to prevent the release of growing pili into the extracellular milieu (Tennent et al. 1990). Deletion of papJ, encoding a periplasmic pilin-like subunit, results in the release of pili into the culture medium, a phenotype that is similar to that of papH mutants (Tennent et al. 1990). Plasmid-encoded papJ complements the phenotype of papJ but not that of papH mutants, suggesting that PapJ and PapH fulfill distinct functions during pilus assembly (Tennent et al. 1990). PapJ encompasses a Walker box domain (ATP-binding site), however, its significance is not clear as the periplasm is not believed to contain nucleotides.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Assembly pathway of type I pili. Six different pilins assemble into a structure that is composed of a tip protein PapG (G), a fibrillum consisting of PapF, PapE, and PapK, as well as the pilus rod consisting of PapA. After Sec-mediated translocation across the plasma membrane, pilin associate with the periplasmic chaperone PapD (D). Pilin-PapD complexes dissociated at the assembly site, that is, the outer membrane usher PapC (C), in a series of donor strand exchange reactions. Drawing adapted from Sauer et al. (2000).

Assembly of Pap pili occurs in two stages (Fig. 3). Pilins are first translocated by the Sec pathway into the periplasmic space (Baga et al. 1984, 1987; Norgren et al. 1984; Lindberg et al. 1986; Tennent et al. 1990). Each pilin is engaged by the PapD chaperone (D), thereby preventing premature aggregation of subunits within the periplasm (Kuehn et al. 1991, 1993). During the second stage of pilus assembly, pilin-chaperone complexes associate with PapC (C), the usher protein in the outer membrane (Dodson et al. 1993). PapC forms a pore structure with an inner diameter of 2-3 nm that is large enough to accommodate folded pilin subunits (Thanassi et al. 1998). Pilin-chaperone complexes are thought to occupy the pore structure and newly formed pili emanate on the bacterial surface, assembled by successive addition of pilins at the outer membrane base (Saulino et al. 2000).

How is an ordered assembly of pilins achieved? A model of kinetic partitioning has emerged that assumes that assembly sites (PapC) display altered affinity for subunits when engaged with different pilins. PapG, the fibrillar tip component, complexed with the PapD chaperone displays the highest affinity for the empty PapC usher (Dodson et al. 1993). The chaperone of PapC:PapG-PapD complexes could be dissociated by incoming PapF-PapD complexes. The chaperone of PapC:PapG/PapF-PapD could in turn be dissociated by PapE-PapD and so forth, until PapA-PapD dissociates PapC:PapG/F/E_n/K/A_n-PapD to form the PapG/F/E/K/A_n+1-PapD product.

Structural analysis of pilins and pilin-chaperone complexes has provided a more detailed understanding of pilus assembly. PapD is composed of two globular domains, each with an IgG-like fold (Holmgren and Branden 1989). Pilin subunits assume a similar IgG-like fold, however the seventh -sheet of the IgG-barrel is absent, resulting in the exposure of a hydrophobic groove on the molecular surface (Choudhury et al. 1999; Sauer et al. 1999). Exposure of this hydrophobic surface likely causes pilin aggregation (Lindberg et al. 1989; Kuehn et al. 1991). PapD stabilizes bound pilins by providing the missing -sheet in a mechanism that has been referred to as "donor strand complementation" or "domain swapping". The N terminus of pilin subunits consist of alternating hydrophobic amino acids much like the complementing strand of the PapD chaperone and has been implicated in subunit interactions. The PapD chaperone of PapG-PapD, complexed with the outer membrane usher PapC, could be dissociated by the N terminus of PapF, thereby completing the IgG-like fold (Eisenberg 1999; Sauer et al. 1999). Pilus assembly can thus be viewed as a series of domain swapping reactions in which the seventh -sheet of the pilin -barrel is provided by either the periplasmic PapD or the N-terminal extension of incoming pilins. The specificity of pilin selection into the growing pilus presumably is determined by the interaction of the hydrophobic groove of the pilin subunit complexed with the usher and the structure of the N terminus in subsequent pilin subunits.

How is pilus assembly terminated? PapA is the most abundant pilin both in the bacterial periplasm and in the assembled pilus. PapH is present only in small amounts (Normark et al. 1983; Baga et al. 1987). Experimental increase in the cellular concentration of PapH shortens the length of pili dramatically. In contrast, a reduced concentration of PapH results in elongated pili and in loss of pili into the extracellular milieu (Baga et al. 1987). It is not yet known whether PapH, the terminator of assembly, assumes a complete IgG-like fold or an altered hydrophobic surface that prevents other pilin subunits from binding (Barnhart et al. 2000). As assembled subunits exit the periplasmic space, pilins are sequestered from the periplasm and cannot be removed from the structure by additional domain swapping reactions. What is the energy source that drives pilus assembly? It appears that assembly or disassembly may be fueled solely by gradients of concentration. During assembly, the concentration of pilin-chaperone complexes is high and that of outer membrane assembly sites low, favoring the dissociation of PapD and pilins. During disassembly, which has not yet been studied extensively, the concentration of "empty" chaperone may be high, promoting the removal of pilins from the pilus base. Alternatively, PapJ may discharge pili out of the usher into the extracellular medium when concentrations of other pilins are low.

The inner diameter of the PapC usher pore (2-3 nm) permits assembly and extrusion of folded pilins. In contrast, the 7-nm pilus rod is too large to exit the usher (Thanassi et al. 1998). Assembled rods can be unraveled into linear fibers with a diameter of 2 nm (Abraham et al. 1983; Bullitt and Makowski 1995; Thanassi et al. 1998). Presumably, polymerized PapA subunits undergo a rearrangement, forming a wider assembly that may stabilize the pilus rod. Knowledge of the atomic structure of PapA has been used to build a molecular model for a 7-nm-wide rod (Choudhury et al. 1999; Sauer et al. 1999). The pilus is generated by helical right-handed rotation, with 3.3 PapA subunits per turn, and a central cavity of 1.5-2.5 nm, dimensions that have been observed in native pili (Gong and Makowski 1992; Bullitt and Makowski 1995).

	Autotransporter

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Pathogenic Neisseria, N. gonorrhea, and N. meninigitis, infect humans and survive in for prolonged periods of time their tissues. Within their host Neisseria must escape the immune response, in particular secretory IgA antibodies in urinary and oral mucosa fluids that facilitate complement-mediated killing and opsonization of microbes. Neisseria secrete IgA protease, an enzyme that cleaves antibodies on mucosal surfaces (Halter et al. 1984). IgA protease is synthesized as a pre-proenzyme (1528 residues; Pohlner et al. 1987). An N-terminal signal peptide initiates the precursor into the Sec pathway (Fig. 4). After cleavage of the signal peptide by signal peptidase, the proenzyme resides in the bacterial periplasm and is presumed to be only partially folded. The C-terminal -domain of IgA protease (residues 1254-1528) assumes a -barrel structure that inserts into the outer membrane and functions as an autotransporter for the N-terminal domain. Once the N-terminal protease domain is exposed on the bacterial surface, it cleaves the proenzyme at the junction between the N-terminal and the C-terminal domain (residues P1117-A1118). The cleaved N-terminal domain of the proenzyme is released from the bacterial surface and acts as a diffusible virulence factor, which matures into the 106-kD IgA protease and the small stable -protein. Fusion of signal peptide bearing polypeptides to the N terminus of the -domain results in autotransport of the hybrid polypeptide across the outer membrane (Klauser et al. 1990). Expression of IgA protease or -domain fusions in E. coli also result in autotransport, suggesting that this pathway requires no specific machinery factors other than the Sec pathway (Klauser et al. 1992). Autotransporters with a C-terminal -domain similar to that of Neisseria IgA protease have been found in several other Gram-negative organisms.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Autotransporters. Neisseria IgA protease is synthesized as a pre-proenzyme. After Sec-mediated translocation across the cytoplasmic membrane, the C-terminal domain of the proenzyme inserts into outer membrane and translocates the N-terminal domain through its lumen. The N-terminal domain acts as a protease that cleaves its peptide tether with the C-terminal domain (arrowhead) and is thereafter released into the extracellular medium. Drawing adapted from Pohlner et al. (1987).

	Type I secretion

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Bacteria secrete pore-forming toxins and degradative enzymes as a mechanism of countering host defenses. Several of these proteins are transported across the bacterial envelope using ATP-binding cassette (ABC) containing transporters (type I secretion). E. coli HlyA is a well-studied example for type I secretion and is discussed here (Fig. 5). Pathogenic E. coli secrete hemolysin (HlyA) into the extracellular milieu (for detailed review, see Koronakis and Hughes 1996; Binet et al. 1997). HlyA is a lipid-modified polypeptide with a domain that is composed of 11-17 nine-amino-acid repeats (LxGGxGND; Felmlee and Welch 1988). The repeat domains bind calcium and are thought to interact with host cells, triggering HlyA insertion into the plasma membrane and leakage of the cytoplasmic contents of target cells (Ludwig et al. 1988; Boehm et al. 1990a,b; Ostolaza et al. 1995). Similar repeat domains have been identified in secreted proteins of other Gram-negative bacteria and these polypeptides are collectively referred to as the family of repeat toxins (RTX; Coote 1992).

View larger version (25K): [in this window] [in a new window]   Figure 5.   Type I secretion of repeat toxins. E. coli HlyA interacts with its cognate ABC transporter (HlyB, ATP-binding cassette) and the trimeric membrane fusion protein HlyD in a proton motive force (PMF) dependent manner. HlyA is translocated simultaneously across the inner and outer membrane in a reaction that requires ATP hydrolysis in the bacterial cytoplasm and interaction of trimeric HlyD with the trimeric outer membrane transporter TolC. Drawing adapted from Stanley et al. (1998) and Zgurskaya and Nikaido (2000).

After synthesis in the bacterial cytoplasm, HlyA is modified by N-acylation of two lysine residues with myristate or palmitate (Stanley et al. 1994, 1998) in a reaction that requires the product of hlyC, the acyl carrier protein (ACP), and ATP (Issartel et al. 1991). HlyC functions as an acyl-transferase to decorate the -amino groups of lysine within HlyA using thioester-linked fatty acids (ACP) as substrate (Hardie et al. 1991; Issartel et al. 1991). Knockout mutations of hlyC and acp abolish lipid modification and the hemolytic property of HlyA but do not affect secretion of the unmodified polypeptide across the double membrane envelope of E. coli (Ludwig et al. 1987). HlyA polypeptide is not cleaved during the secretion process (Felmlee et al. 1985a,b). Fusion of HlyA to the C terminus of cytoplasmic reporter proteins resulted in secretion of the hybid proteins into the extracellular milieu (Fig. 5; Gray et al. 1986; Ludwig et al. 1987; Mackman et al. 1987). Deletion of HlyA sequences from the hybrid polypeptides identified a 20-60 residue C-terminal peptide signal that is necessary and sufficient for secretion of reporter proteins (Gray et al. 1989; Koronakis et al. 1989; Hess et al. 1990). Further mutational analysis of the C-terminal secretion signal revealed several amino acids involved in the signal recognition (Kenny et al. 1992, 1994).

Knockout mutations of hlyB, hlyD, and tolC abolish secretion and cause HlyA to reside in the bacterial cytoplasm (Wagner et al. 1983; Koronakis et al. 1988; Wandersman and Delepelaire 1990). hlyB encodes an inner membrane protein with an ATP-binding cassette (ABC transporter; Delepelaire and Wandersman 1991; Wang et al. 1991; Gentschev and Goebel 1992). TolC is a trimeric outer membrane protein that forms a central, water-filled cavity. The crystal structure of TolC revealed two domains, an N-terminal domain with four -sheets typical of outer membrane proteins and a C-terminal domain with four 10-nm -helices that extend into the periplasm (Koronakis et al. 2000). Assembled trimeric TolC forms a 12-stranded -barrel in the outer membrane with an inner diameter of 3.5 nm (Koronakis et al. 2000). Unlike outer membrane porins, the interior of the -barrel is occluded as the C-terminal -helices of TolC taper to closure and prevent leakage of periplasmic contents across the outer membrane (Koronakis et al. 2000). TolC appears to be involved in several E. coli transport processes that are similar to the type I secretion mechanism, including the efflux of antibiotics, heavy metal ions, detergents, and solvent (multi-drug efflux). HlyD spans the inner and the outer membrane of E. coli and binds directly to both TolC and HlyB (Schulein et al. 1992; Letoffe et al. 1996; Thanabalu et al. 1998).

During secretion, HlyA forms a complex with HlyB and HlyD (Fig. 5). Assembly of trimeric HlyD with HlyB occurs in the presence and absence of secretion substrate (Thanabalu et al. 1998). HlyD trimerization is a requirement for its interaction with TolC (Thanabalu et al. 1998). In the initial phases of transport, HlyA polypeptide is bound to the ABC transporter and the membrane-fusion complex leading to a transient complex with the outer membrane TolC trimer. Several different transport steps can be envisaged: (1) formation of a complex between HlyA and HlyB/HlyD; (2) binding of ATP by HlyA/HlyB_n/HlyD₃, presumably the first committed step of transport; (3) Conformational change of HlyD resulting in the formation of a HlyA/HlyB_n/HlyD₃/TolC₃ complex; (4) HlyB-mediated movement of HlyA across the plasma membrane and ATP hydrolysis; (5) movement of HlyA across the outer membrane. Catalysis of steps 1-4 could occur by alterations in the folding state of HlyB and could be triggered by the binding, hydrolysis, and release of ATP (Koronakis et al. 1993). Further, the proton motive force across the plasma membrane of E. coli plays a critical role in secretion as reagents that disrupt the proton gradient also abolish the initial stages of HlyA transport (Koronakis et al. 1991).

Similar ABC transporters, membrane fusion proteins, and outer membrane pore proteins have been identified in the secretion pathways for RTX toxins and other proteins of many Gram-negative bacteria. It seems that Gram-negative pathogens use this system to transport virulence factors across the bacterial envelope (Fleischmann et al. 1995; Stover et al. 2000). Gram-positive organisms also employ ABC transporters (Kunst et al. 1997). As Gram-positive bacteria lack the outer membrane permeability barrier, the HlyD and TolC components of the Gram-negative apparatus are not required for secretion in Gram-positive organisms. In addition to secretion of proteins in bacteria, ABC transporter are also found in eukaryotes from yeast to humans and are responsible for the efflux and influx of a large number of ions and small molecules (Holland and Blight 1999; Zgurskaya and Nikaido 2000).

	Type II secretion, type IV pili, and filamentous phage

Top Instruments of bacterial... Gram-positive and Gram-negative... Secretion across the plasma... Bacterial surface organelles Secretion pathways of Gram-... Assembly of type I... Autotransporter Type I secretion Type II secretion, type... Type III machines Type IV secretion Conclusions References

Type II secretion

Secretion of type II substrates occurs in two stages (Fig. 6). First, the Sec machinery translocates signal peptide bearing type II substrates across the plasma membrane. Folded secretion substrates are then transported by the type II machinery across the outer membrane. Some well-known bacterial toxins are secreted in this manner. V. cholerae secrete cholera toxin, which is composed of CtxA and CtxB subunits (Lonnroth and Holmgren 1973). Mature CtxB assembles into a pentameric ring structure within the bacterial periplasm (Hirst and Holmgren 1987) and associates with the C-terminal domain of mature CtxA to form cholera toxin, CtxA:CtxB₅ (Cuatrecasas et al. 1973; Lonnroth and Holmgren 1973; Finkelstein et al. 1974; Heyningen 1974). CtxA forms an intramolecular disulfide bond and is proteolytically processed (Cuatrecasas et al. 1973; Lonnroth and Holmgren 1973; Heyningen 1974). Secreted CtxB₅ binds to G_m1 ganglioside , a sphingoglycolipid, on the surface of human intestinal cells (Cuatrecasas 1973; Holmgren et al. 1973; King and Van Heyningen 1973), causing toxin uptake and reduction of CtxA by cytosolic thioredoxin. The reduced CtxA fragment is released from the membrane-bound receptor complex (Sattler and Wiegandt 1975; Tomasi et al. 1979) to activate host cell adenylate cyclase (Cassel and Selinger 1977; Cassel and Pfeuffer 1978; Gill and Meren 1978), resulting in massive diarrhea (Greenough et al. 1970; Schafer et al. 1970; Sharp and Hynie 1971; Field et al. 1972). E. coli enterotoxin (LT), Shigella dysenteriae shiga toxin, and E. coli shiga-like toxins are composed of a similar structural assembly (AB toxins), albeit that the mode of action for shiga toxins is to prevent ribosomal translation of host cells (Schmitt et al. 1999). Type II secretion substrates of other Gram-negative organisms include proteases, alkaline phosphatase, pectate lyase, and elastase, as well as the lipoprotein pullulanase (for further details, see Russel 1998).

View larger version (21K): [in this window] [in a new window]   Figure 6.   Type II secretion across the outer membrane. Klebsiella oxytoca PulA precursor protein is transported by the Sec pathway across the plasma membrane. After substrate recognition by the type II machinery (GSP, general secretory pathway), the fully folded and enzymatically active PulA (A) is translocated across the outer membrane. The outer membrane secretin, GspD, acts as a translocator for PulA and requires binding of the GspS chaperone for insertion into the outer membrane. The type II secretion machinery is assembled from one cytoplasmic component (GspE, an ATPase), five inner membrane proteins (GspC, GspF, GspM, GspL, and GspO), and five pseudopilins (GspG, GspH, GspI, GspJ, and GspK). Pseudopilins harbor a pilin signal peptide (see Fig. 7) and are cleaved by prepilin peptidase (GspO) during Sec-mediated translocation across the plasma membrane. The type II secretion machinery appears to be multifunctional, allowing secretion of proteins across the outer membrane and assembly or retraction of type IV pili. Drawing adapted from Russel (1998) and Sandkvist (2001).

Klebsiella oxytoca secretes pullulanase, a lipoprotein that forms micelles once it has been secreted into the extracellular milieu (Fig. 6; Wallenfels et al. 1966; Wohner and Wober 1978). Pullulan is a large complex carbohydrate that can not diffuse through bacterial outer membrane pores, whereas the product of pullulanase cleavage, maltotriose, can diffuse (Bender and Wallenfels 1966). When Klebsiella are plated on pullulan as the sole carbon source, the type II secretion of pullulanase is required for bacterial growth (Michaelis et al. 1985; d'Enfert et al. 1987). Cloning of a cluster of 16 Klebsiella genes (pulA-O and pulS) into E. coli can promote the growth of recombinant strains that use pullulan as the sole carbon source (d'Enfert et al. 1987). These and other results suggest that expression of pulC-G, pulI-M, pulO, and pulS is sufficient to promote the type II secretion of pullulanase (PulA) in heterologous organisms (Possot et al. 2000). Once PulA has been translocated by the Sec machinery, its precursor is di-acyl glyceride modified at a cysteine residue and cleaved by type II signal peptidase (Pugsley et al. 1986). An aspartyl residue at position 2 of mature PulA is thought to retain lipid-modified PulA in the outer leaflet of the inner membrane (Poquet et al. 1993b). This mechanism is a prerequisite for type II secretion, as PulA mutants bearing a serine substitution at position 2 are transported by LolAB (Yamaguchi et al. 1988; Matsuyama et al. 1995, 1997) to the inner leaflet of the outer membrane and are not substrate for the secretion machinery (Seydel et al. 1999). Sec translocation and type II secretion of pullulanase can be uncoupled. Expression of PulA without the type II secretion machine results in the accumulation of N-terminally processed PulA in the periplasm (Pugsley et al. 1991). Subsequent expression of the type II secretion machine restores the secretion of PulA into the extracellular milieu (Pugsley et al. 1991; Poquet et al. 1993a).

Type II secretion machines have been referred to as secretons or general secretory pathways and their components have been assigned a unified Gsp nomenclature (Pugsley 1993a). One component, GspD or secretin, is inserted into the outer membrane and oligomerizes into a dodecameric ring structure with an inner diameter of 7.6 nm (Nouwen et al. 1999). Type II machinery-catalyzed transport is thought to move polypeptides through the lumen of GspD. The secretin is not only conserved among type II machines but is also a component of type III machines, the assembly pathway of filamentous phage, and the polymerization of type IV pili (Russel 1994). K. oxytoca GspD requires binding to the outer membrane chaperone PulS (GspS) for proper folding and activity in vivo (Hardie et al. 1996a). Not all GspD homologs require a GspS chaperone for folding (Pugsley et al. 1997). The GspS requirement correlates with the presence of a GspS-binding domain in Klebsiella GspD (Hardie et al. 1996b). Erwinia carotovora and E. chrysanthemi each secrete cellulases via the type II pathway. Expression of cellulase (PelB) in the heterologous organism does not lead to type II secretion (He et al. 1991a; Py et al. 1991). However, replacement of E. carotovora GspD with E. chrysanthemi GspD allowed secretion of E. chrysanthemi PelB by E. crysanthemi, suggesting that GspD may be involved in substrate recognition during type II secretion (Lindeberg et al. 1996). Experiments exchanging GspD from K. oxytoca and E. caratovora corroborated this notion (Possot et al. 2000). Coprecipitation experiments suggest an interaction between PelB and the N-terminal domains of GspD (Shevchik et al. 1997); however, expression of hybrids of PulD and Erwinia OutD in Klebsiella failed to restore secretion of PulA, suggesting that a different region of GspD is responsible for substrate recognition (Guilvout et al. 1999).

Twelve type II machinery proteins (GspBCFGHIJKLMNO) are positioned in the inner membrane (for review, see Russel 1998). One machinery component (GspE) resides in the cytoplasm (Possot et al. 1992; Pugsley and Dupuy 1992; Nunn and Lory 1993; Pugsley 1993b; Reeves et al. 1994; Sandkvist et al. 1995; Thomas et al. 1997). How can cytoplasmic or plasma membrane proteins catalyze secretion when transport of type II substrates across the plasma membrane is accomplished by the Sec pathway? A definitive answer is not yet in sight.

Type IV pili

Several studies have suggested specific functions for several Gsps during the assembly of type IV pili. Type IV pili are expressed by Pseudomonas aeruginosa, Neisseria gonorrhea, and other Gram-negative pathogens (Henrichsen 1975; Fussenegger et al. 1997; Tonjum and Koomey 1997). The pili are thought to retract in a coordinated fashion (Bradley 1980), a phenomenon that has been associated with bacterial gliding motility (Wu and Kaiser 1995). Presumably, pilus adhesion to an immobilized surface and pilus retraction can pull Myxobacteria or Neisseria in one direction. P. aeruginosa type IV pili are involved in adhesion to respiratory epithelia (Alm and Mattick 1997; Hahn 1997) and are composed of the major pilin subunit PilA (Fig. 7; Pasloske et al. 1985). PilA is synthesized as a precursor with a signal anchor sequence that also contains a unique N-terminal prepilin signal sequence (Strom et al. 1991). After translocation of PilA by the Sec machinery, the prepilin signal sequence is cleaved and the liberated amino-group is N-methylated. Both reactions are catalyzed by PilD (GspO) and require S-adenosyl-methionine as a cofactor (Nunn and Lory 1991; Dupuy et al. 1992; Strom et al. 1993). Mature, translocated type IV pilin is thought to polymerize at the plasma membrane (Parge et al. 1995) and the assembled pilus may be extruded through the central cavity of the outer membrane secretin XcpQ (GspD; Parge et al. 1995).

View larger version (25K): [in this window] [in a new window] &nbs