Correspondence: Ross E.Dalbey, Department of Chemistry, The Ohio State University, 100 West 18th Avenue, Columbus, OH 43210, USA. Tel.: +1 614 292 2384; fax: +1 614 292 1532; e-mail: firstname.lastname@example.org
Gram-negative bacteria assemble many proteins into the inner and outer membranes and export a large number of proteins to the periplasm or to the extracellular medium. During the billions of years bacteria have been around, they have evolved a number of different pathways with sophisticated machines to accurately and efficiently move proteins from one location to another. In this review, we first introduce specific proteins that are representative substrates of the protein transport pathways and describe their function. Then, their specific routes from synthesis to their destinations are described mentioning the signal peptide that may initiate their export and discuss what is known about the folding state of the substrates during transport. The membrane translocation device involved, the energy source required for transport, and whether a chaperone is needed will be discussed.
Gram-negative bacteria contain a myriad of proteins that are destined to the cell surface where they carry out their specific functions. The surface contains two membranes, inner and outer. Between the two membranes, in the periplasm, a meshwork of peptidoglycan is localized that gives the cell its shape and rigidity. Facing the outside of the cell are surface proteins and structures that play various important roles. Only the inner membrane is directly accessible for a newly synthesized protein, and its assembly into the membrane is defined as the insertion process. Proteins of the periplasm and outer membrane are first exported through the inner membrane and then inserted into the outer membrane.
In addition to transporting proteins to the cell surface, many proteins are secreted from the cell into the extracellular medium. Remarkably, in Gram-negative bacteria, there are over six secretion pathways that are used to accomplish this process. Secretion of protein components is often essential for pathogenicity of the bacteria and other functions such as biofilm formation, modulation of the eukaryote host, and nutrient acquisition. The proteins can be exported directly from the cytoplasm out of the cell by a 1-step process or by a 2-step process where the protein is first exported to the periplasm and then moved across the outer membrane.
This review will describe the different pathways that are used to export proteins to the different compartments of the bacterial cell or the extracellular medium. It will be written from the point of view of a bacterial protein that follows one of the various export routes and uses different translocation machineries to reach its destination. For a more detailed review on secretion-related surface structures, please see the contributions in this issue by Thanassi et al. (2012).
Sorting proteins into the inner membrane
The inner membrane of the Gram-negative proteobacterium Escherichia coli contains more than 1000 different integral proteins. The main pathway for insertion of these proteins into the inner (cytoplasmic) membrane of E. coli is the Sec pathway (for review, see Xie & Dalbey, 2008). The Sec translocase catalyzes the translocation of polar loops across the membrane and inserts apolar segments into the membrane. A subset of proteins is inserted into the membrane by the YidC insertase (see Kiefer & Kuhn, 2007, for review). These latter Sec-independent proteins possess only short hydrophilic domains that are translocated to the periplasmic side of the membrane. However, YidC is involved in the membrane insertion of both the Sec-dependent and Sec-independent membrane proteins.
It is worth noting that the Sec pathway also plays a prominent role in protein export and secretion in bacteria (Fig. 1). The Sec machinery is the main machinery that promotes the export of proteins to the periplasm. Also, most lipoproteins and β-barrel proteins localized to the outer membrane are initially translocated across the inner membrane by the Sec system. Similarly, the surface pili are assembled on the cell surface from subunits that are first translocated across the inner membrane by the Sec apparatus. Furthermore, the proteins that are secreted out of the cell by a 2-step mechanism enter the periplasm by the Sec pathway. Finally, the components that make up the secretion apparatus of the Type 1-6 secretion system are membrane inserted or exported by the Sec system.
Signal peptidase (Lep), lactose permease (LacY)
Signal peptidase (Lep) and lactose permease are two intensely studied inner membrane proteins. Lep contains two transmembrane segments and a large carboxyl-terminal domain exposed to the periplasmic space. Lep is a type 1 signal peptidase that is responsible for processing exported proteins by cleaving their signal sequences at the periplasmic surface of the inner membrane (Paetzel et al., 2002). Lep is one of the few membrane proteins that are targeted to the inner membrane by signal recognition particle (SRP) and subsequently requires the motor ATPase SecA for translocation of a hydrophilic domain across the SecYEG channel (Wolfe et al., 1985; de Gier et al., 1996). Lac permease is a proton-driven lactose transporter residing in the inner membrane and is a typical multi-spanning protein with 12 transmembrane (TM) segments (Ito & Akiyama, 1991; Nagamori et al., 2004). Lac permease requires SecYEG, but not SecA for membrane insertion.
First, SRP is required for targeting Lep, Lac permease, and most integral membrane proteins to the inner membrane (Ulbrandt et al., 1997; Wickstrom et al., 2011a). The bacterial SRP is comprised of the polypeptide Ffh and the 4.5S RNA (Luirink & Sinning, 2004). Ffh is essential for membrane insertion of Lep and Lac permease (Macfarlane & Muller, 1995; de Gier et al., 1996). In the targeting step, SRP interacts with a hydrophobic segment of the membrane protein as it is being synthesized by the ribosome. Ffh contains a groove that binds the hydrophobic segments. The SRP/nascent protein chain/ribosome complex is delivered to the membrane-associated SRP receptor FtsY (Luirink & Sinning, 2004). This targeting event requires GTP to be bound to both the SRP and SRP receptor at the membrane. Following GTP hydrolysis, SRP and SRP receptor dissociate, allowing them to be used again for protein targeting.
After targeting, the nascent chain of Lep is transferred to SecA and the SecYEG translocase. SecA is required for translocation of the large periplasmic region of Lep as is observed for many proteins translocating a hydrophilic loop or tail region across the membrane (Kuhn, 1988; Deitermann et al., 2005). SecA uses its two-helix finger (that may move up and down as ATP is hydrolyzed) to thread portions of the protein chain, 20–30 residues at a time, into the membrane-embedded SecYEG translocon (Economou, 2008; Zimmer et al., 2008). Whereas the large hydrophilic domain of Lep is translocated by the Sec translocase, the hydrophobic segments are inserted by the Sec machinery into the bilayer. For Lep, each transmembrane segment inserts individually at the interface of Sec/YidC (Houben et al., 2004). This explains why the membrane assembly of Lep uses both the Sec translocon and YidC. How does the Sec machine carry out these challenging tasks? A possible answer to this question came from the determination of the structure of an archaeal Sec complex (van den Berg et al., 2004). The structure revealed that SecY constitutes a channel with an hourglass shape, whereas the transmembrane SecE and SecG proteins are at the periphery of the channel. In the center of the channel, there is a pore ring comprised of six hydrophobic amino acids. A helical plug on the extracytoplasmic side of the membrane helps to close the pore. It was proposed that first the plug moves out of the center of the channel and thereby allows the hydrophilic region of a protein to move through the pore ring. The hydrophobic segments of a membrane protein substrate bind to the lateral gate region formed mainly by TM2 and TM7 and then exit the channel via YidC to partition into the lipid bilayer (Beck et al., 2001). When YidC is depleted from the cell, Lep is believed to remain within the Sec translocon for a longer time period because it is inefficiently inserted into the lipid bilayer without YidC. This explains why overexpression of Lep blocks the export of Sec-dependent exported proteins under YidC depletion conditions because the translocon is jammed (Samuelson et al., 2001).
Interestingly, the first and second hydrophobic segments of Lep are membrane inserted using two different mechanisms. The first segment (NoutCin orientation) initiates translocation of the N-terminus of the protein even under conditions where the function of the Sec machinery is impaired (Lee et al., 1992). In contrast, the second hydrophobic segment requires the Sec machinery to initiate insertion and promote translocation of the large periplasmic C-terminal domain of Lep (Dalbey et al., 1987). Crosslinking studies show that both hydrophobic segments of Lep can be crosslinked to YidC, indicating that Lep inserts into the membrane at a site containing SecYEG and YidC (Houben et al., 2000; Samuelson et al., 2000), and to SecY (Fig. 2) (Houben et al., 2002). Hydrophobic segments 1 and 2 of Lep are oriented in an Nout Cin and Nin Cout fashion, respectively, consistent with the positive inside rule (Heijne, 1989; Laws & Dalbey, 1989; San Millan et al., 1989).
SecYEG, but not SecA, is required for membrane insertion of Lac permease (Ito & Akiyama, 1991) where all 12 TM segments need to be inserted into the membrane bilayer. Currently, it is not exactly known how all 12 segments leave the translocon. However, from studies with the 6-spanning aquaporin and the eukaryotic Sec translocon, it is thought that each TM segment of the multispanning membrane protein leaves the SecY channel one by one (Sadlish et al., 2005).
Interestingly, the Sec-dependent Lac permease requires YidC for proper folding (Nagamori et al., 2004). If YidC is depleted in the cell, the inserted Lac permease is folded improperly; the binding of two monoclonal antibodies that recognize folded epitopes of Lac permease is inhibited. This supports the idea that YidC may play a direct role in the helix–helix packing of the membrane proteins, and its task is to promote folding of the already translocated proteins (Beck et al., 2001).
YidC insertase pathway
Bacteriophage Pf3 coat protein and Foc subunit of the ATP synthase
Aside from working in the Sec pathway, YidC can function alone as a true membrane insertase in a separate pathway (Samuelson et al., 2000; Serek et al., 2004). This Sec-independent YidC-only pathway is evolutionarily conserved and is also operational in mitochondria and chloroplasts for the insertion of proteins into the inner membrane and thylakoid membrane, respectively (Wang & Dalbey, 2011). It is still not clear how many Sec-independent substrates are inserted by the YidC insertase, but the membrane levels of many proteins are reduced by YidC depletion (Price et al., 2010; Wang et al., 2010; Wickstrom et al., 2011b).
The single-spanning Pf3 coat protein is one of the most studied proteins of the YidC-only pathway. Pf3 is the major coat protein of the filamentous phage Pf3 that infects Pseudomonas aeruginosa. To assemble progeny phage, the coat proteins are first inserted into the inner membrane, a process that can be studied in E. coli (Rohrer & Kuhn, 1990). The Pf3 coat protein is thought to bind to the membrane surface by the interaction of the positively charged residues at the carboxyl-terminus of the protein to the negatively charged phospholipid head group of the membrane lipids. At the membrane surface, the Pf3 coat protein interacts with YidC which then catalyzes the integration of the hydrophobic stretch into the membrane and translocation of the flanking hydrophilic region (Chen et al., 2002). Binding of Pf3 coat protein to YidC induces a conformational change within YidC that alters its periplasmic domains (Winterfeld et al., 2009; Imhof et al., 2011). YidC is sufficient for the membrane insertion of the Pf3 coat protein as it can insert into YidC proteoliposomes, but not into protein-free lipid vesicles (Serek et al., 2004). Translocation of the tail is promoted by the proton motive force (pmf) (Rohrer & Kuhn, 1990). The acidic residues in the tail are essential for the N-tail translocation step (Kiefer et al., 1997). The negatively charged residues are also a major determinant for the orientation of the protein in the membrane. When the charged residues within the C-tail and N-tail were interchanged, the membrane orientation of the inserted Pf3 coat protein was completely inverted (Kiefer & Kuhn, 1999).
Subunit c of the F1Fo ATP synthase is the best-studied endogenous membrane protein that goes by the YidC-only pathway. This ATP synthase consists of a proton-conducting channel formed by the subunits a, b, and c in the bacterial inner membrane and a trimer of both subunits α and β in the cytoplasm. They are connected by the γ, δ, ε proteins and also functionally link proton transport with ADP phosphorylation. Interestingly, subunit c of the F1Fo ATP synthase inserts by the YidC-only pathway (Yi et al., 2003, 2004; van Bloois et al., 2004). For this protein, the two short hydrophilic tails of about seven and three residues at the N-terminus and at the C-terminus, respectively, are translocated into the periplasmic space. YidC proteoliposomes are sufficient to insert subunit c where it can form the Foc oligomer (Van Der Laan et al., 2004). After membrane insertion, subunit c proteins oligomerize into ring structures of defined numbers, which differs among various species (Stock et al., 1999; Pogoryelov et al., 2005).
Recently, a cryo-EM study at low resolution proposed that YidC functions as a homodimer to insert proteins. YidC was found to sit at the exit channel of the ribosome near the predicted protein L23 (Kohler et al., 2009). This study speculated that YidC functions as a channel and that a pore structure is formed in between TM2 and TM3 of one subunit and TM2 and TM3 of a second subunit. Interestingly, of the six transmembrane segments of YidC, TM2, and TM3 were previously shown to play important roles for YidC function. Point mutations that inactivate YidC have been found within TM2 and TM3 (Jiang et al., 2003; Yuan et al., 2007). Clearly, a high-resolution structure of YidC is needed to understand how YidC works at a molecular level.
Transport of proteins into the periplasmic space
Translocation across the inner membrane to the periplasmic space in E. coli requires the Sec translocase or the twin arginine translocase (Tat) pathway. While substrates are translocated through the Sec channel in an unfolded state (Fig. 2) (for review, see Driessen & Nouwen, 2008), the Tat machinery can export proteins across the membrane in a folded state (for review, see Lee et al., 2006). Many of the Tat substrates in Gram-negative bacteria have metal cofactors that are introduced by the biosynthetic machinery localized in the cytoplasm, prior to export.
The maltose-binding protein (MBP) and the disulfide bond protein (DsbA)
MBP is a periplasmic binding protein for maltodextrans and interacts with the maltose transporter (MalFGK2) to deliver maltose sugars. MBP is exported by the SecA/SecYEG machinery across the inner membrane, and it typifies Sec-dependent exported proteins that also require the chaperone SecB for targeting to the SecYEG translocase. MBP is made as a precursor with an amino-terminal signal peptide that is essential for export. The export of MBP requires the molecular chaperone SecB that has antifolding activity and prevents preMBP from folding into a stable protease-resistant conformation in the cytoplasm (Randall & Hardy, 2002; Zhou & Xu, 2005). SecB is a tetramer, a dimer of dimers (Muren et al., 1999), and each monomer of SecB has a simple α- and β-fold (Xu et al., 2000). The SecB tetramer has two clefts, ~70 Å in size, that are thought to bind the unfolded preprotein (Xu et al., 2000). The exported protein/SecB complex interacts with SecA via the SecB 8-stranded antiparallel β-sheet contacting two regions of SecA (Zhou & Xu, 2003; Suo et al., 2011). This leads to the transfer of the exported protein to SecA, which then catalyzes stepwise translocation through the SecYEG channel (Fig. 2).
Unlike MBP and most periplasmic proteins, the exported protein DsbA uses the SRP pathway for targeting to the Sec machinery (Schierle et al., 2003; Arts et al., 2007; Francetic et al., 2007). DsbA is a component of the periplasmic protein oxidation system that is devoted to introduce disulfide bonds to ensure the correct folding of many periplasmic proteins (Mamathambika & Bardwell, 2008). DsbA is synthesized with a signal sequence that shows an increased hydrophobicity compared to the signal peptides recognized by SecA (Shimohata et al., 2005). The targeting of pre-DsbA by SRP leads to co-translational export (Schierle et al., 2003). In contrast, many substrates go by the SecB/SecA targeting pathway and are inserted post-translationally (Randall & Hardy, 1995).
Both SRP- and SecB-targeted substrates converge at the Sec translocase at the membrane surface (Valent et al., 1998). The targeted preprotein enters the Sec translocation channel where the signal peptide binds to the lateral gate. This induces a conformational change thereby opening the channel and moving the plug helix from the pore center to a peripheral location 30 Å away, near SecE. Interestingly, a signal peptide can activate the Sec channel in trans and promote translocation of mature protein (Gouridis et al., 2009). These studies suggest that the signal peptide functions as an allosteric activator of the translocase. Once the channel is activated and in the open state, the mature region of the polypeptide chain can be moved through the pore ring to the other side of the membrane (Cannon et al., 2005).
Trimethylamine N-oxide reductase (TorA) and the septal ring protein (SufI)
Many of the identified proteins that are exported in a folded state by the twin arginine translocation (Tat) pathway contain cofactors. Tat substrates are characterized by’a twin arginine signal peptide with the consensus sequence ZRRXφφ (where Z represents a hydrophilic residue, X can be any residue, and φ is a hydrophobic residue) in the amino-terminal region followed by a hydrophobic stretch (Natale et al., 2008). This is different from the signal sequences recognized by the Sec components that were described above.
TorA is a periplasmic enzyme that contains a molybdopterin guanine dinucleotide cofactor and is involved in anaerobic respiration. Export of the folded TorA with the co-factor bound requires the TatABC complex (Fig. 2) consisting of TatA, TatB, and TatC (Bogsch et al., 1998; Sargent et al., 1998; Weiner et al., 1998). Knockouts of the genes encoding any one of these Tat components result in the accumulation of enzymatically active TorA in the cytoplasm of the cell, supporting the idea that the cofactors are introduced and folded prior to export. Strikingly, the molybdenum redox factor must first be added to TorA in order for TorA to be exported (Jack et al., 2004). To ensure that the Tat substrate proteins are assembled properly prior to export, a proof-reading mechanism prevents export until proper cofactor assembly has occurred (Jack et al., 2004). For example, the TorD chaperone is involved in the assembly of the molybdenum cofactor onto TorA and performs a proof-reading function, where the TorD chaperone has been shown to bind to the TorA Tat signal peptide. Binding of TorD to the signal peptide may function to prevent premature targeting to the Tat translocase (Palmer et al., 2005). In the periplasm, TorA forms a complex with the c-type cytochrome TorC, a single-spanning membrane protein with a large periplasmic domain containing five hemes (Gon et al., 2001).
The Tat-dependent SufI is a component of the septal ring structure of constricting bacterial cells. SufI has no metal ion or any other bound cofactor and therefore has been employed as a model substrate for cell-free systems (Tarry et al., 2009). Using the in vitro system, it was shown that the pmf is required for the translocation of SufI, while ATP is not required (Yahr & Wickner, 2000). Translocation of preSufI across the membrane does not depend on a pH gradient but rather depends on two electrical potential requiring steps (Bageshwar & Musser, 2007). Note that Tat substrates have also been identified that are anchored to the inner membrane by N-terminal lipid anchors or C-terminal transmembrane anchors (Hatzixanthis et al., 2003; Storf et al., 2010).
In E. coli, the TatABC machinery transports folded Tat substrates. TatBC is involved in recognition of the Tat signal peptide (Alami et al., 2003) and most likely TatA, which forms channel complexes of variable diameters, mediates the translocation event (Gohlke et al., 2005). One attractive model is that the signal sequence of the substrate first binds to the TatBC complex and then associates with TatA homooligomer in a pmf-dependent manner. The substrate is then transported across the membrane by the TatA oligomeric pore within the TatABC complex.
Sorting of proteins to the outer membrane
Two pathways are known for sorting proteins to the outer membrane after they are translocated across the inner membrane via the SecYEG or TatABC machineries. Lipoproteins go by the Lol pathway [for review, see Tokuda & Matsuyama (2004)] while β-barrel proteins go by the β barrel assembly machinery (BAM) pathway (for review, see Knowles et al., 2009).
Braun's lipoprotein Lpp
One of the most studied substrates of the Lol system is the Braun's lipoprotein (Lpp). Lpp is the major lipoprotein in the outer membrane and connects the peptidoglycan to the outer membrane. It is synthesized with a signal peptide (Inouye et al., 1977) that is cleaved off the preprotein by signal peptidase II (lipoprotein signal peptidase) (Fig. 3). All lipoproteins contain the lipobox with the common motif Leu-Ala/Ser-Gly/Ala-Cys spanning the −3 to +1 positions (Hayashi & Wu, 1990). An invariant cysteine residue at the +1 position is modified with diacylglycerol after the processing step, and the N-terminus is further modified by a fatty acid (Sankaran & Wu, 1994).
The Lol pathway is essential for the transport of lipoproteins from the inner to outer membrane. Five proteins, namely LolABCDE, are all essential for viability of E. coli. Tokuda and coworkers elegantly elucidated the entire Lol pathway by using an in vitro cell-free system and by studying the release of the Braun's lipoprotein in spheroplasts from the inner membrane outer surface into the medium. Initially, they identified a periplasmic protein (LolA) that was essential for the release step. LolA acts as a chaperone for the transfer of the protein from the inner to outer membrane (Matsuyama et al., 1995). Lpp is transferred from the LolA–Lpp complex to LolB, the outer membrane receptor, prior to outer membrane insertion (Fig. 3) (Matsuyama et al., 1997; Okuda & Tokuda, 2009). The fatty acid chains of Lpp are protected by a hydrophobic β barrel and an α-helical lid that both, LolA and LolB, possess (Takeda et al., 2003).
LolCDE uses ATP hydrolysis to release Lpp or other translocated lipoproteins from the outer face of the inner membrane (Fig. 3) (Yakushi et al., 2000). LolD, a peripheral inner membrane subunit facing the cytoplasm, contains the Walker and ABC transporter motifs. This ATP hydrolyzing component forms a complex with LolC and LolE, each predicted to have four transmembrane segments. The LolCDE transporter has a stoichiometry of 1 : 2 : 1 with respect to C, D, and E subunits. The membrane-embedded LolC and LolE components bind the lipoproteins, destined for the outer membrane. To avoid transport to the outer membrane, inner membrane lipoproteins contain an inner membrane retention signal (typically a negatively charged residue at the +2 and +3 position) that prevents them from binding to the LolABC transporter (Gennity & Inouye, 1991).
The maltoporin LamB
The last few years have been an exciting time for scientists studying the transport of β-barrel proteins to the outer membrane with the discovery of the BAM complex. The BAM machinery is able to recruit unfolded β barrel proteins in the periplasmic space for outer membrane assembly.
A workhorse of this pathway is the LamB protein substrate. LamB transports external maltose and maltodextrins across the outer membrane into the periplasm. LamB forms trimers in the outer membrane with each monomer consisting of an 18-stranded β-barrel. It is synthesized as a precursor, translocated across the inner membrane by SecA and SecYEG and cleaved from its signal sequence by signal peptidase I (Lep) (Driessen & Nouwen, 2008). In the periplasmic space, LamB and other outer membrane proteins (OMPs) require chaperones to guide them to the outer membrane. The periplasmic chaperones SurA, Skp, and DegP have been implicated in the transport of OMPs to the outer membrane. SurA seems to be the most important because its depletion results in a reduction of LamB and other β-barrel proteins in the outer membrane (Fig. 3). Silhavy and coworkers have proposed that in E. coli there are two chaperone pathways; one involving SurA and the other involving Skp and DegP (Sklar et al., 2007). Whereas SurA is mainly involved in the folding of the OMPs, Skp/DegP functions in preventing aggregation by binding unfolded OMPs. To do so, the jellyfish-like structure of Skp may hold the proteins between its tentacles (Walton et al., 2009).
A big breakthrough in this field was the identification of the Omp85 class of proteins – the key component of the BAM machine – that catalyzes the β-barrel insertion and folding event (for review, see Tommassen, 2010; Ruiz et al., 2006; Walther et al., 2009). Omp85 is essential for viability of Neisseria meningitidis and is required for folding and outer membrane localization of several porin proteins (Voulhoux et al., 2003). Along with a β-barrel domain, Omp85 proteins contain a periplasmic region with several polypeptide transport-associated (POTRA) domains that bind β-barrel proteins destined to the outer membrane by recognizing a motif that is found within the outer membrane porin proteins at their C-terminus (Robert et al., 2006; Kim et al., 2007; Tommassen, 2007). Omp85 is evolutionarily conserved. Homologues are found in mitochondria and chloroplasts where they also catalyze insertion and translocation of β-barrels (Paschen et al., 2003; Voulhoux et al., 2003; Wiedemann et al., 2003; Schleiff & Soll, 2005).
In E. coli, Silhavy and coworkers discovered an Omp85 homologue called BamA (formerly called YaeT) that is essential for outer membrane insertion of LamB and other OMPs (Fig. 3) and is part of a complex made up of four additional lipoproteins: BamB, C, D, and E (formerly called YfgL, NlpB, YfiO, and SmpA) (Wu et al., 2005). The BAM complex and the SurA chaperone are the minimal components needed for outer membrane insertion (Hagan et al., 2010). Recently, Hagan et al. (2010) showed that urea-diluted OmpT (in the presence of SurA) can insert into BamABCDE liposomes.
To provide insight into how the BAM complex may function, the structures of various Bam components have been solved (Kim et al., 2007, 2011; Kim & Paetzel, 2011). This includes the structure of a BamA construct that contains the first four POTRA domains that are involved in substrate binding of outer membrane proteins such as LamB (Kim et al., 2007). Each of the POTRA domains contains a three-stranded β sheet with two α helices on top. Based on the structure of the POTRA domains, Kahne and coworkers proposed that substrates bind to the POTRA domains of BamA by a lateral alignment of the β-strands. This mechanism would allow the POTRA domains to mediate the folding and assembly of the β barrel into the outer membrane.
There are several bacterial proteins that are exported across the outer membrane by secretion systems and produce surface structures (Fig. 4). These secretion systems produce the type I pili, curli, and type IV pili. Regarding the nomenclature of the pili and the secretion systems (next section), the type designation is misleading and was started independently for classifying pili and secretion systems. The pili play a very important role for adherence onto a surface and to the host cells and are therefore critical for pathogenic bacteria to infect their host. The subunit proteins of all pili types are first exported into the periplasm via the Sec translocase and then assembled into the growing pilus.
In addition, flagella are bacterial surface structures that are required for the movement of the bacteria. Flagellin, the major filament protein, is exported to the tip of the filament by a type 3 secretion system (T3SS) described in the secretion section.
Type I and P pili
Escherichia coli type I pili, and the P pili of pyelonephritic E. coli, are fibers that consist of a pilus rod of ~2 μM in length that is formed mainly by the major pilin and a short thin component of 7–8 nm that is joined on the distal tip of the fibrillum. The pili subunits such as FimA are made in a precursor form with N-terminal signal peptides and are exported across the inner membrane by the Sec translocase (Fig. 4). The signal peptide is then proteolytically removed by signal peptidase 1.
The cell surface assembly of the pilin subunits to produce type 1 and P pili occurs by the conserved chaperone-usher (CU) pathway (for review, see Kline et al., 2010). The dedicated chaperone (FimC for type I pili and PapD for P pili) interacts with the pilin subunit in the periplasmic space (Fig. 4). The chaperones function to stabilize the pilin proteins so that they do not polymerize prematurely in the periplasm. The subunit proteins have a noncanonical immunoglobulin fold lacking the C-terminal β-strand, whereas the chaperones have two immunoglobulin-like domains that interact with each subunit providing the missing β-strand in a process termed donor strand complementation (Choudhury et al., 1999; Sauer et al., 1999).
The outer membrane usher is critical for the transfer of the pilin subunit from the chaperone onto the growing pili polymer (Fig. 4). The usher consists of two 24-stranded β-barrels forming a twin pore machinery (Remaut et al., 2008). While only one of the pores is used as the translocation channel, both are used for the recruitment of the chaperone–subunit complexes. Pilin subunits assemble onto the pili by a concerted β-strand displacement mechanism similar to what has been described for the binding of the chaperone with the pilin subunit (Remaut et al., 2006). In a zip-in-zip-out mechanism, the’complementing β-strand is provided by the amino-terminal extension sequence (Nte) of the incoming subunit. Evidence for such a mechanism comes from the structure of the pilin subunit (missing the amino-terminal extension) that was crystallized together with the Nte domain derived from a second pilin subunit (Verger et al., 2007). Polymerization of the pilin subunits proceeds by the missing strand complementation of the next pilus subunit in the usher (Remaut et al., 2006). In the P pili system, the polymerization is terminated by binding a PapH subunit that lacks a binding groove for Nte and therefore is unable to bind another pilin subunit (Verger et al., 2006).
Sepsis and the proinflammatory response are promoted by Enterobacteriacaea that express curli pili. These pili are highly flexible and stable filaments, 4–7 nm in size, and promote cell aggregation. The major pilin subunit CsgA is synthesized with a signal sequence of 20 amino acid residues (Fig. 4). After its Sec-dependent membrane translocation and processing by signal peptidase 1, it associates with CsgB that has been exported into the periplasm and is then secreted across the outer membrane by a pore-forming lipoprotein, CsgG (Fig. 4) (Robinson et al., 2006). CsgB and CsgA are assembled on the surface of the outer membrane presumably by a nucleation–precipitation mechanism (Fig. 4) (Hammar et al., 1996). By this mechanism, the CsgA subunits polymerize on the cell surface into a fibrillar structure (Bian & Normark, 1997). Filament assembly also requires the periplasmically exposed CsgE and CsgF proteins that interact with CsgG. The curli pili assembly is an attractive example for an extracellularly controlled assembly mechanism.
Structural analysis of the pili and computer modeling suggest that the curli pili are composed of cross-β intertwined Csg A subunits each containing 5-stacked β-strands (White et al., 2001; Barnhart & Chapman, 2006).
Type IV pili
Type IV pili of Gram-negative bacteria are surface filaments that can be as long as 4 μM and are 5–8 nm in diameter. Type IV pili are involved in virulence, phage transduction, DNA uptake during transformation and biofilm formation. The pilus filament, formed by the pili subunits assembling into a helical structure, provides the flexibility and strength that is needed for adhesion, twitching, and gliding motility.
PilA is the main subunit that forms the primary structural component of the pili. Other minor pilins FimU, PilV, PilW, PilX, and PilE are also incorporated into the assembling type IV pilus structure (Giltner et al., 2010). The pilin subunits are made as precursors with signal peptides that are processed by a type IV peptidase on the amino-terminal side of the hydrophobic domain. PilA is targeted to the membrane by the SRP pathway and then translocated across the membrane by the Sec translocase. The type IV signal peptidase proteolytically removes the signal peptide and cleaves at the amino-terminal side of the signal peptide hydrophobic region at the cytoplasmic surface removing the basic region (Paetzel et al., 2002). This is in contrast to type 1 and 2 signal peptidases that cleave at the periplasmic surface on the carboxyl-terminal region of the hydrophobic region. Another unusual feature of the type IV signal peptidase is that it methylates the N-terminal amine of the mature region of the protein.
The assembly of type IV pilus on the periplasmic side of the inner membrane requires 12 or more proteins with components located in the inner membrane, the periplasm, and the outer membrane. Here, we will discuss the type IV pili system of P. aeruginosa because it has been extensively studied. This system has two ATPases: PilB that drives polymerization of the pilus and PilT that drives depolymerization of the pilus (Burrows, 2005; Jakovljevic et al., 2008).
PilC, the inner membrane transporter together with the assembly PilB ATPase, extrudes the pilin subunits that face the periplasm (that are anchored to the membrane by the hydrophobic segment) from the membrane allowing them to assemble onto the growing pilus filament (Fig. 4) (Ayers et al., 2009). Accordingly, the filament is pushed through the periplasm, the peptidoglycan, and across the outer membrane via the outer membrane secretin PilQ. PilQ has a dodecameric structure and is similar to the secretins found in the type 2, 3, and 4 secretion systems (Collins et al., 2001).
DNA uptake and phage transduction involve the retraction of the pili, promoted by the disassembly of the pilus filament. It is proposed that the PilT ATPase rapidly depolymerizes the pilus with up to 1500 subunits per second released (Burrows, 2005). Therefore, the same type IV pilus assembly machinery acts in two directions for the assembly and disassembly processes.
Secretion out of the cell
In Gram-negative bacteria, secretion out of the cell can occur in two fundamentally different ways. One-step secretion involves export from the cytoplasm into the extracellular medium or into another cell using a continuous tunnel spanning the inner membrane, the periplasm, and the outer membrane. Type 1, 3, 4, and 6 secretion systems use a 1-step secretion mechanism that does not involve periplasmic intermediates. Substrates of the Type 3 secretion pathway are transported via a flagellum-like injectisome from the cytoplasm of the bacterial cell across the inner and outer membrane to the cytoplasm of a eukaryotic host cell. The injectisome is composed of a needle (located in the extracellular space) that is attached to the basal body residing in the inner membrane, periplasm, and outer membrane of the bacterial cell. In Type 4 secretion, a large trans-envelope device is formed at the bacterial surface that can secrete effector proteins and DNA into the extracellular medium and into another cell. Type 6 secretion was recently discovered in pathogenicity islands of Burkholderia.
2-step secretion involves first export to the periplasmic space followed by transport across the outer membrane. Type 2 and 5 secretion systems use a 2-step mechanism. The Sec translocase is typically used first to export these proteins to the periplasm, although in some cases, the Tat machinery exports these proteins. The periplasmic intermediate is then exported across the outer membrane in the second step. The Type 2 and 5 secretion systems known today secrete extracellular enzymes and proteins including many that are involved in pathogenesis (exotoxins).
Hemolysin HlyA and the heme-binding protein HasA
Proteins that are secreted by the Type 1 secretion system (T1SS) include many adhesins, proteases, and toxins that are delivered into a host cell. The secretion of the E. coli hemolysin is an excellent model system of the T1SS.
Hemolysin HlyA is secreted into the medium where it’may lyse red blood cells. Hemolysin contains an uncleaved C-terminal signal that is essential for export. The C-terminal signal peptide allows hemolysin to bind to the ATPase HlyB. Substrate binding induces the formation of a continuous tunnel (comprised of HlyB/HlyD/TolC) spanning the inner and outer membrane (Fig. 5) (Thanabalu et al., 1998). The C-terminal signal of HlyA is sufficient for secretion and can direct, in some cases, the secretion of other proteins (Kenny et al., 1991; Nakano et al., 1992).
T1SS is comprised of three components (HlyB, HlyD, and TolC) (Fig. 4). HlyB is an ABC transporter in the inner membrane that is a homodimer, each monomer with six predicted transmembrane segments. ATP hydrolysis is critical for transport; mutations of the Walker motif within HlyB inhibit ATP hydrolysis and secretion (Koronakis et al., 1993). The HlyB ATPase forms a complex with the adaptor protein HlyD, independent of the TolC protein (Thanabalu et al., 1998). HlyD is a trimeric inner membrane protein with a large periplasmic domain that contacts the outer membrane component TolC in the presence of substrate HlyA.
The T1SS exports unfolded protein molecules as demonstrated with the heme-binding HasA protein of Serratia marcescens, another well-studied T1SS. HasA functions as an iron scavenger and delivers iron to the outer membrane bacterial receptors for uptake. Interaction of HasA in the cytoplasm with the SecB chaperone has been demonstrated by coimmunoprecipitation. In accordance, secretion of HasA is strongly inhibited in a secB deletion strain (Delepelaire & Wandersman, 1998). When the SecB protein binds to HasA, it keeps it in an unfolded or loosely folded state and slows its folding by three orders of magnitude (Wolff et al., 2003). Recently, Bakkes et al. (2010) showed with a HlyA-MBP fusion protein that the rate of folding determines whether a substrate is secreted (Bakkes et al., 2010).
X-ray crystallography studies of TolC revealed an amazing structure (Koronakis et al., 2000). TolC is a homotrimer that forms a 12-stranded barrel in the outer membrane and a periplasmic tunnel that is 10 nm long. When substrate is bound to HlyB, the TolC channel entrance is open to allow the substrate to pass through the TolC tunnel that spans the entire envelope. The TolC entrance region that contacts HlyB is believed to open’and close by an iris-like mechanism where the TolC α-helices are moved from a twisted closed state to an open untwisted state (Koronakis et al., 2004). How the opening and closing is regulated is not exactly clear at present.
Pullulanase and cholera toxin
Lipase, elastase, cellulase, alkaline phosphatase, and exotoxin are among the substrates of the Type 2 secretion systems (see Filloux, 2004). This includes the cholera toxin secreted by Vibrio cholerae that consists of a pentameric ring of CtxB and binds to the CtxA protein that harbors the enzymatic activity responsible for adenylate cyclase activation in the target cell (O'Neal et al., 2005). The best-studied T2SS is from Klebsiella oxytoca, which has been investigated over the years by the Pugsley laboratory. This type II system secretes pullulanase, a surface-anchored lipoprotein that enables K. oxytoca to grow on branched maltodextrin polymers.
Pullulanase is first exported to the periplasm by the Sec system, and then, the T2SS is used for its secretion across the outer membrane (Fig. 6). The T2SS also known as the ‘general secretion pathway’ (GSP) is the major secretion pathway in Gram-negative bacteria. The Type 2 system is homologous to the ancient type IV pilus assembly system described in the surface structure section (Hobbs & Mattick, 1993; Nunn, 1999). For secretion of pullulanase across the outer membrane, up to 16 different proteins are required. Surprisingly, only two reside in the outer membrane (Filloux, 2004). Numerous T2SS systems in different bacterial species have been characterized, and the protein components of the system have been individually named. To unify the nomenclature between species, the common term Gsp is used to describe the common protein subunits. Like the type IV pili system, the T2SS system has an ATPase, a type IV signal peptidase, and a secretin. Most of the proteins span the inner membrane and have domains exposed to the periplasmic space.
The energy transducing component of the Type 2 transport system is GspE (the secretion ATPase) which functions as an oligomer (Possot & Pugsley, 1994; Camberg & Sandkvist, 2005). As suggested from work on the GspE protein of Xanthomonas campestris (XpsE), oligomerization of the secretion ATPase might occur only upon ATP binding. To energize export of the T2SS substrate, GspE associates with the inner membrane GspL protein, together with the integral inner membrane proteins GspC, GspF, and GspM (Sandkvist et al., 1995; Possot et al., 2000), and promotes the assembly of the pseudopilins into a pseudopilus in the periplasm. The assembly of the pseudopilus is required for the secretion of pullulanase (Fig. 6). The different pseudopilin subunits (GspG, H, I, and K) are synthesized with N-terminal signal peptide and are exported by the Sec system. The signal peptide of these subunits is processed by GspO, a type IV signal peptidase that methylates the amino-terminal residue of the mature proteins as well (Nunn & Lory, 1993). Type 2 secretion across the outer membrane is most likely facilitated by the assembly of the pseudopilus structure in the periplasm that acts as a piston to push secreted proteins through the outer membrane secretin (Sauvonnet et al., 2000). The secretin forms a dodecameric channel with an inner cavity of 76 Å (Fig. 6) (Nouwen et al., 1999). Strikingly, a pilus comprised of GspG can be formed on the’cell surface when the complete K. oxytoca T2SS is overexpressed in E. coli (Sauvonnet et al., 2000). Similarly, a pilus is formed when XcpT of X. campestris is overproduced in P. aeruginosa (Durand et al., 2003).
In the T2SS, the secreted proteins are folded in the periplasm prior to secretion. For example, pullulanase obtains its disulfide bond prior to secretion (Pugsley, 1992), and the B pentamer of the cholera toxin is already formed in the periplasm before its secretion (Hirst & Holmgren, 1987). Although it is not known how the substrates are recognized for protein export, a linear secretion signal can be excluded. Binding of the secreted protein to secretin appears to be complex and the best guess is that there is a structural motif within the folded secreted protein that is presented to the secretin (Sauvonnet et al., 1995; Shevchik et al., 1997; Palomaki et al., 2002; Reichow et al., 2010).
Piston-mediated pushing of substrates across the outer membrane secretin is supported by the recent structural data (Reichow et al., 2010). Using the cryo-EM structure of secretin and X-ray structures of other components (Korotkov & Hol, 2008), a model was generated of how the cholera toxin is pushed through the secretin. Initially, in a priming stage, the pseudopilus tip interacts with the substrate. Then, as the pseudopilus grows by the assembly of pseudopilin subunits, it pushes the exoprotein – cholera toxin – into the secretin complex. This leads to the interaction with the constriction point within the secretin channel and triggers gate opening and secretion.
Yops and flagellar proteins
The T3SS of the Yersinia proteins causing plague in humans is related to the assembly pathway of bacterial flagellum. T3SS is found in pathogenic and symbiotic bacteria that infect both animal and plant cells.
The T3SS substrates, also called effector proteins, are exported across three membranes: the inner and outer membrane of bacteria and the plasma membrane of the eukaryotic target cell. The T3SS substrate proteins are synthesized with N-terminal uncleaved signals and require export-specific chaperones to prevent tight folding (He et al., 2004). The substrates are exported out of the cell in a 1-step process. The secretion system involves more than 25 proteins that are highly conserved among the pathogenic bacterial species and several show sequence similarities to flagellar assembly genes. The secretion system is usually activated by a prior contact of the bacteria with the host cell (Pettersson et al., 1996).
Yops (for Yersinia outer membrane proteins) are arguably the best-studied T3SS-exported proteins. Indeed, the T3SS was discovered in Yersinia pestis where Yops were shown to be exported from the cytoplasm of the Gram-negative bacteria directly into the cytoplasm of the eukaryotic host rather than into the extracellular medium (Rosqvist et al., 1994; Sory & Cornelis, 1994; Cornelis, 2006). Secretion of Yops requires amino-terminal signals that function as binding sites for the cytoplasmic chaperone SycD (Buttner et al., 2008). The substrate is then believed to move through the export apparatus energized by the cytoplasmic YscN ATPase (Fig. 7) (He et al., 2004). The secretion device used in T3SS is an injectisome that functions as a molecular syringe to inject Yops (or other T3SS substrates) into the eukaryotic host cell (Fig. 7) (Kubori et al., 1998). The injectisome has a basal structure characterized with ring structures in the inner membrane (C ring), in the periplasm, and in the outer membrane. A needle-like extension protrudes from the outer ring.
The T3SS substrates that are transported from the cytoplasm move through the basal body and directly through the needle of the injectisome. The double ring in the inner membrane is formed by the YscJ protein (Yip et al., 2005) where Ysc stands for Yop secretion protein component. In addition, YscR,S,T,U,V and LcrD reside in the inner membrane and join the C ring together with the soluble protein YscQ and the YscN ATPase. In the outer membrane, the secretin YscC forms a double ring structure of the injectisome that forms the channel in the outer membrane in which T3SS substrates are exported across (Fig. 7) (Koster et al., 1997). Recently, ring models of the secretin were constructed and placed within the injectisome (Spreter et al., 2009).
The first secreted proteins (YscC, F, O, P and YscX; not to be confused with the T3SS Yops) make up the external part of the injectisome. For example, YscF is the major protein forming the needle. After the needle reaches a specific length, the export system secretes the T3SS YopB and YopD proteins to localize them to the host membrane. They are required for translocation of the T3SS effector Yops into the host cell and most likely the secreted proteins move through the central channel of’the needle, in a manner similar to that proposed for the flagellins (Fig. 7). The effector Yops are translocated into the eukaryotic cell via a channel formed by YopB and YopD in the plasma membrane of the target cell (Tardy et al., 1999).
The flagellum uses a T3SS for export of the hook-filament junction protein, the filament-capping protein, and flagellin, the major subunit of the filament in flagellum. These secreted proteins become incorporated into the filament (Fig. 7) (see Macnab, 2004, for review). Like the Yop-secreted proteins, the recognition signal of flagellar secreted proteins is located in the amino-terminal region of the T3SS substrates. These flagellum secreted proteins also require substrate-specific chaperones (Fraser et al., 1999; Auvray et al., 2001).
As seen with the T3SS injectisome, the flagellum has a basal body of similar architecture. The components comprising the flagellar export complex, which are essential for the formation of the flagella, are: FliH, I, J, O, P, Q, R and FlhA, B. FlhAB and FliOPQR are inner membrane proteins localized within the central pore of the ring basal body, and FliHIJ are peripheral cytoplasmic proteins (Fig. 7). This export complex first secretes proteins needed to form the external parts of the basal body. Once formed, then the device can secrete type 3-secreted proteins such as flagellin. The FliI ATPase (homologous to the YscN ATPase) drives the export of the individual flagellar components (Fan & Macnab, 1996). ATP hydrolysis is controlled by FliH that interacts with the FliI hexamer. FlhA and FlhB interact with each other and might function as the transmembrane export pore because they are known to interact with the substrate (Minamino & Macnab, 2000).
The energy sources required for flagellar assembly and Type 3 secretion are the pmf and ATP hydrolysis (Galan, 2008). The pmf is believed to drive the movement of the protein substrates through the central channel (Minamino & Namba, 2008; Paul et al., 2008). ATP hydrolysis by the Flil cytosolic ATPase is important for unfolding and presenting the substrate to the secretion machinery (Galan, 2008). However, FliI is not essential and flagellar assembly can occur in a pmf-dependent manner in the absence of the ATPase FliI and its regulatory protein FliH, although inefficiently (Minamino & Namba, 2008).
Ti-plasmid and the RP4-plasmid
Type 4 secretion is observed in both Gram-negative and Gram-positive bacteria. The evolutionary origin of the system is the conjugational system. T4SS is employed for the transport of virulent proteins or DNA into eukaryotic cells as well as for the conjugative transfer of plasmids from one bacterium to another (Fronzes et al., 2009a). The T4SS systems can be understood as a protein secretion system, where the secreted protein is covalently linked to the DNA and the DNA is then transported by a piggy-back mechanism.
The T4SS of Agrobacterium tumefaciens is typical for this secretion system (Alvarez-Martinez & Christie, 2009) and is employed to transfer the tumor-inducing Ti-plasmid into a plant cell. This secretory system requires 12’proteins namely VirB1 to VirB11, and VirD4. The Ti-DNA is covalently linked to the VirD2 protein, the primary substrate of the T4SS. Most Type 4-secreted substrates’require C-terminal uncleaved signal sequences. Chaperones may also be necessary to maintain the protein substrate in a loosely folded or unfolded conformation (Alvarez-Martinez & Christie, 2009). At the cytoplasmic membrane surface, the VirD4 ATPase of A. tumefaciens functions as a receptor to mediate the transfer of a VirD2-Ti DNA strand transfer intermediate to the’T4SS (Fig. 5) (Tzfira et al., 2004). VirD4 recruits the Ti-DNA to the secretion machine by binding to the basic C-terminal tail of the VirD2 protein (Christie, 2004), which is covalently bound to the Ti-DNA (Fig. 5). VirD4 is also able to recruit other effectors by binding the C-terminal signal region of these proteins. The structure of the VirD4 homologue TrwB of E. coli, which is involved in bacterial conjugation, revealed a ring structure similar to that of F1-ATPases and helicases (Gomis-Ruth et al., 2001). Therefore, VirD4 may provide the molecular motor driving transport in the Type 4 system.
The energizing ATPases VirB11 and VirB4 are essential for transport (Fig. 5). The VirB11 ATPase belongs to the family of ATPases that are homologous to the Type 2-secretion ATPases. Recently, the structure of the VirB11 ATPase from Helicobacter pylori (Yeo et al., 2000) revealed that it is hexameric with different conformations of several of the monomers. The authors suggested that dynamic conformational changes within the hexamer might promote the export of substrates (Savvides et al., 2003). In addition, the membrane integrated VirB4 ATPase is probably required for the activity of the secretion channel (Alvarez-Martinez & Christie, 2009).
The substrate Ti-DNA then moves through the secretory system, where it contacts VirB11, VirB6, VirB8, VirB9, and the VirB2 pilin, the main structural component of the pili, in addition to receptor VirD4 (Cascales & Christie, 2004). The core channel components include VirB8, VirB10, and VirB6 in the inner membrane, and VirB9 and VirB7 (and a segment of VirB10) in the outer membrane. Structural work has defined VirB7, VirB9, and VirB10 within the T4SS machinery (see Chandran et al., 2009; Fronzes et al., 2009a, b). Surprisingly, the outer membrane pore is comprised of a two-helix bundle region of VirB10, previously thought to span only the inner membrane. It should be noted that the pilus used in T4SS is not related to the surface pili discussed above; it mediates the specific contact to the target cell. More information on the other components can be obtained by reading (Alvarez-Martinez & Christie, 2009).
Another example of a Type 4 secretion system is the mating pair formation (Mpf) system of E. coli RP4, in which the relaxase protein is exported along with DNA into another bacteria. In conjugation, a linear, single-stranded copy of the DNA is covalently attached at its 5′ end to relaxase in the donor cell. The catalytic tyrosine residue of the relaxase forms a phosphotyrosyl linkage to the DNA which is maintained during the transfer of the DNA by the T4SS into the recipient cell (Pansegrau & Lanka, 1996). The transfer of the DNA is probably coupled to the replication of the second strand in the recipient cell. The relaxase is also involved in the recircularization of the plasmid DNA after the transfer is completed (Draper et al., 2005). The components of the Mpf are basically identical to the T4SS in A. tumefaciens described above.
Because ancestral T4SS are known to secrete only proteins, the transport of DNA may be a late development of the T4SS (Cascales & Christie, 2003). This may also explain why a covalent linkage of the DNA to the relaxase and VirD2, respectively, is required for transport.
The type 5 secretion system includes autotransporters and 2-partner systems. First, the T5SS substrates are exported to the periplasm by the Sec translocase and are then translocated across the outer membrane in a second step. In the case of autotransporters, the secreted proteins are synthesized with an N-terminal passenger domain that is secreted across the outer membrane and a C-terminal translocation domain that forms a β-barrel domain in the outer membrane. Initially, it was assumed that the β-barrel domain promotes export of the passenger across the outer membrane.
The extracellular serine protease EspP
All autotransporters are synthesized in a precursor form with a signal peptidase 1 cleavable signal sequence. Many of these look like typical signal peptides, although a subset of them are characterized with a long amino-terminal extension containing a charged domain with conserved features among autotransporters (Dautin & Bernstein, 2007). In the mature region, a passenger and a translocation domain are present. Passenger domains are comprised of very diverse sequences, but all have structures rich in β-strands that form elongated solenoids (Otto et al., 2005; Junker et al., 2006). The diversity of the sequences correlates with the respective group of the secreted protein that functions as a toxin, protease, or another type of enzymes. The translocation domain at the C-terminus of autotransporters folds into a β-barrel structure, which correlates with its presumed transporter function. A number of autotransporters are cleaved between the passenger and translocation domain during or after translocation across the outer membrane.
EspP protein of Shiga toxin producing E. coli represents an archetypical autotransporter (see Dautin & Bernstein, 2007, for review). EspP is a secreted protease that functions as a virulent factor. Like most autotransporters, EspP undergoes SecA-mediated translocation across the inner membrane by the Sec pathway (Sijbrandi et al., 2003; Peterson et al., 2006; Desvaux et al., 2007). EspP contains an amino-terminal extension to the signal peptide that is essential for secretion, possibly to prevent misfolding of the passenger domain in the periplasm (Szabady et al., 2005). Interestingly, the polypeptide sequence that connects the β-barrel domain and the passenger domain is incorporated into the barrel in the periplasm as determined by analyzing a pretranslocation intermediate (Ieva et al., 2008). The evidence suggests that the EspP β-domain and the embedded passenger peptide segment are integrated into the outer membrane as a single preformed unit. The structures of the translocation domain of many autotransporters reveal an amino-terminal α-helix located within a β-barrel (Oomen et al., 2004; Barnard et al., 2007). In the case of EspP, the α-helix within the β-barrel is cleaved by a novel intra-barrel cleavage mechanism (Dautin et al., 2007). Finally, translocation across the outer membrane requires the folding of a C-terminal 17-kDa passenger sequence on the extracellular side of the membrane suggesting that the energy of folding drives secretion of the passenger domain (Peterson et al., 2010).
Secretion by the autotransporter pathway can clearly tolerate some folding of the passenger domain in the periplasmic space prior to outer membrane translocation (Klauser et al., 1992; Skillman et al., 2005; Jong et al., 2007). The ability to translocate at least a partially folded structure of EspP and other autotransporters raises questions about the self translocation model because the narrow channel of the autotransporter translocation domain can fit only unfolded conformations even if the α-helix within the β-barrel is released.
One attractive model is that the BAM complex is involved in the autotransporter secretion (Meng et al., 2006; Bernstein, 2007). In addition to promoting the insertion and integration of the β-barrel domain into the outer membrane, the BamA channel would facilitate the complete translocation of the autotransporter passenger domain in the C-terminal to N-terminal direction. Recent support for this model involving the BAM machinery comes from work with the E. coli EspP autotransporter using a translocation intermediate. The passenger domain of EspP was shown to interact with BamA (Ieva & Bernstein, 2009).
Filamentous hemagglutinin Fha
The 2-partner system (Tps) is used to transport large virulence factors and contains two proteins: the TpsA protein that is the secreted exoprotein and its cognate TpsB protein that forms the outer membrane channel (see Jacob-Dubuisson et al., 2009, for review). The secreted protein TpsA and outer membrane channel protein TpsB are made with amino-terminal signal peptides and are translocated across the inner membrane by the Sec pathway. Like the autotransporter passenger domains, the secreted proteins are predicted to be rich in β-strands (Kajava & Steven, 2006). The 2-partner pathway is dedicated to secrete large proteins that form long β-helices in their native folded structure (Clantin et al., 2004; Yeo et al., 2007).
A prototype Tps system is the Bordetella pertussis system that secretes the filamentous hemagglutinin Fha (TpsA) protein (Jacob-Dubuisson et al., 2004). Fha functions as an adhesin and immunomodulator in the infected eukaryotic host (Inatsuka et al., 2005). Fha and all TpsA members have a conserved 250 amino acid long amino-terminal ‘Tps' domain that is the hallmark of the 2-partner system (Jacob-Dubuisson et al., 2001). This Tps domain is essential for secretion of TpsA proteins in several bacterial systems (Renauld-Mongenie et al., 1996; Grass & St Geme, 2000) and for the binding of TpsA to the TpsB protein in the outer membrane (Hodak et al., 2006). TpsA is believed to be translocated in an extended conformation through the TpsB protein because a globular conformation of TpsA is not conducive to secretion (Guedin et al., 1998).
The cognate TpsB protein in B. pertussis is called FhaC. FhaC possesses an amino-terminal region with two polypeptide transport-associated (POTRA) domains and a carboxyl-terminal β-barrel domain common to the Omp85/TpsB family members (Ertel et al., 2005; Bredemeier et al., 2007; Clantin et al., 2007; Kim et al., 2007). FhaC belongs to the Omp85/Sam50/Toc75 transporter superfamily that inserts proteins into the outer membrane in Gram-negative bacteria and mitochondria and translocates proteins across the chloroplasts outer membrane (Hinnah et al., 1997; Paschen et al., 2003; Voulhoux et al., 2003; Gentle et al., 2004). The structure of FhaC revealed that its pore region is comprised of a 16-strand β-barrel and is occluded by a H1 helix and the loop L6 (Clantin et al., 2007). Conductance experiments are consistent with a pore of about 8 Å in diameter suggesting that the H1 helix and L6 loop are removed from the pore during the secretion step. Given the narrow channel pore, the secreted substrates must be translocated through the pore in an unfolded state.
The favored secretion model for the 2-partner system currently proposes that the Tps domain of TpsA binds to the TpsB POTRA domains and primes the system for translocation. The TpsA portion C-terminal to the Tps domain would initially insert into the channel as a hairpin. Outer membrane translocation of the protein would be driven by the folding of the polypeptide chain on the extracellular region of the cell (Jacob-Dubuisson et al., 2004).
VgrG effector proteins
The recently discovered type 6 secretion system exists in most Proteobacteria that come into close contact with eukaryotic cells. The genetic clusters, which contain as few as 12 to more than 20 genes, are usually found in pathogenicity islands, and the gene expression is upregulated on contact with the host cell (Mougous et al., 2007). The AggR activator, an AraC-like protein, was identified to control the expression of the Type 6 cluster in Burkholderia mallei (Dudley et al., 2006), suggesting an elaborate transcriptional regulation mechanism.
The VgrG-1 protein of V. cholerae is secreted by the type 6 pathway (Fig. 5). It possesses at its C-terminus an RTX actin cross-linking domain, and its activity has been demonstrated to occur in macrophages (Pukatzki et al., 2007). VgrG-1 is synthesized without a cleavable signal sequence and enters the eukaryotic target cell by a T6SS. The C-terminal domains of other VgrG proteins encode a tropomyosin-like and a pertactin-like domain, respectively (Cascales, 2008). The C-terminal effector domain of VgrG-1 can be replaced by a reporter domain and translocated into the target cell (Ma et al., 2009). A most intriguing homology was found between the T6SS VgrG protein and the infection protein device of bacteriophage T4 that forms the tail spike. Electron microscopy of single particles of the N-terminal domain of the secreted outer component VgrG shows a sequence similarity to the central domain of gp5 of T4 phage (Kanamaru et al., 2002; Pukatzki et al., 2007). A trimeric cylinder forming a long needle-like puncturing structure is most likely involved in contacting the host cell. The internal space, 30 Å in diameter, is similar to the T4 phage gp5 structure that allows the passage of the phage DNA. In addition, homology has been found to the T6SS component Hcp that forms a tubular structure in the periplasm to the bacteriophage T4 tail protein gp19 (Leiman et al., 2009). VipA/B proteins also form a tubular structure possibly around the Hcp tube, and architecture similar to the tail core and sheath of phage T4 (Fig. 5). Therefore, it is possible that the Type 6 system has evolved from a bacteriophage and allows the bacteria to use the tip of the T6SS needle to deliver the respective T6SS proteins with enzymatic activity into the host cell.
The inner membrane components IcmF and IcmH of the secretion machine share some similarity with T4SS components forming the inner membrane channel (Das & Chaudhuri, 2003). The energy required to transport effector proteins is probably provided by the ClpV ATPase component that is related to the ClpB protein family involved in protein quality control. ClpV forms oligomeric rings and might operate much like the Type 4 ATPase VirB11 (Schlieker et al., 2005). So far, other effector proteins of the T6SS than VgrG have not yet been investigated in detail.
Proteins use a bouquet of different systems to find their way in or out of the cell. Transport of folded protein substrates across the inner and outer membrane occurs in some of the systems – the Tat machinery and the type 2 secretion machinery (Table 1), respectively. However, most export and secretion systems – the Sec translocase, Type 1 and 3 secretion systems – only transport an unfolded protein chain and also require chaperones to keep exported proteins from folding prematurely in the cytoplasm. A variation of this is observed for substrates inserted into or translocated across the outer membrane by the BAM and the Type 5 system. In these cases, periplasmic chaperones are needed to prevent the folding of the substrates prior to interactions with the outer membrane machineries (Table 1).
Table 1. Export information for different pathway substrates
IM, inner membrane; OM, outer membrane; PP, periplasm; C-SS, C-terminal secretion signal; US, uncleaved signal sequence; SS, signal sequence; RS, reverse signal; RR-SS, twin arginine signal sequence; U, unfolded; F, folded; P, partially folded; pmf, proton motive force.
More variations are observed for the decisive region of the substrate required for transport among the different transport and secretion systems. An amino-terminal signal is required for the substrates of the Sec and Tat system and usually also for the T3SS (Table 1). In contrast, a carboxyl-terminal signal is required for the Type 1 and 4 systems. However, in the Type 2 secretion system, no linear amino- and carboxyl-terminal signal is sensed within the protein for secretion from the periplasm. Rather, it is a folded structure that is recognized by the secretion components, which allows transport across the outer membrane (Table 1).
In general, ATP hydrolysis is required to drive protein translocation across the inner membrane (Table 1). This is observed for the Sec translocase and the Type 1, 2, 3, and 4 systems. However, ATP hydrolysis might be limited to unfolding the substrate as it is entering the translocation machinery. In the case of the lipoproteins, ATP hydrolysis by LolD within the LolCDE complex is required to energize lipoprotein release from the membrane surface. Remarkably, ATP hydrolysis is not required in the flagellar system under certain conditions, but transport still requires the pmf. The pmf (but not ATP hydrolysis) is required for translocation of globular substrates across the inner membrane by the Tat system. Until recently, an important issue in the type 5 secretion system was to unravel which energy source is driving the translocation of proteins across the outer membrane since ATP is absent in the periplasm. A recent study with the EspP autotransporters suggests that the folding of a carboxyl-terminal domain on the extracellular surface promotes the translocation of the passenger domain in the C- to N-terminal direction across the membrane (Table 1). The energetics and the detailed molecular steps that drive these processes in the secretion and export systems still remain to be understood.
This work was supported by a National Science Foundation Grant MCB-1052033 to R.E.D. and by DFG Grant Ku749/5.