|Home | About | Journals | Submit | Contact Us | Français|
The Type 2 Secretion System (T2SS), occurring in many Gram-negative bacteria, is responsible for the transport of a diversity of proteins from the periplasm across the outer membrane into the extracellular space. In Vibrio cholerae, the T2SS secretes several unrelated proteins including the major virulence factor cholera toxin. The T2SS consists of three subassemblies, one of which is the Inner Membrane Complex which contains multiple copies of five proteins, including the bitopic membrane protein EpsL. Here we report the 2.3 Å resolution crystal structure of the periplasmic domain of EpsL (peri-EpsL) from V. parahaemolyticus, which is 56 % identical in sequence to its homolog in V. cholerae. The domain adopts a circular permutation of the “common” ferredoxin fold with two contiguous sub-domains. Remarkably, this permutation has so far only been observed once before: in the periplasmic domain of EpsM (peri-EpsM), another T2SS protein which interacts with EpsL. These two domains are 18 % identical in sequence which may indicate a common evolutionary origin. Both peri-EpsL and peri-EpsM form dimers, but the organization of the subunits in these dimers appears to be entirely different. We have previously shown that the cytoplasmic domain of EpsL is also dimeric and forms a heterotetramer with the first domain of the “secretion ATPase” EpsE. The latter enzyme is most likely hexameric. The possible consequences of the combination of the different symmetries of EpsE and EpsL for the architecture of the T2SS are discussed.
Gram-negative bacteria use sophisticated protein machineries for the extracellular transport of a wide variety of proteins including virulence factors (Cianciotto, 2005). Examples of secreted proteins are cholera toxin, the infective agent of the human pathogen Vibrio cholerae, and heat-labile enterotoxin, a major virulence factor of enterotoxigenic E. coli (ETEC). These hetero-hexameric AB5 toxins are exported in folded form (Hirst and Holmgren, 1987) across the outer membrane by the “Type 2 Secretion System” (T2SS). The T2SS is called the “Extracellular protein secretion” (Eps) system in V. cholerae and related species, and the “General secretory pathway” (Gsp) system in ETEC and other species (Filloux, 2004; Johnson et al., 2006; Pugsley, 1993; Sandkvist, 2001a; Sandkvist, 2001b; Sandkvist et al., 1993; Sandkvist et al., 1997; Tauschek et al., 2002). The T2SS is a complex machinery spanning the inner and outer membranes and consists of 11 or more different proteins, with many of these present in multiple copies. The nomenclature of the T2SS proteins is also quite complex. In this paper, proteins from the Eps system from Vibrio species are referred to as “Eps” followed by a capital letter, while the non-Vibrio T2SS homologs will be called “Gsp” followed by the same capital letter. For instance, the T2SS protein EpsL in the various Vibrio species is called GspL in other species.
The T2SS can be thought to consist of three sub-assemblies: the Outer Membrane Complex, the Pseudopilus, and the Inner Membrane Platform (Filloux, 2004; Johnson et al., 2006; Py et al., 2001). The main component of the Outer Membrane Complex is EpsD, which is thought to open and close during protein secretion. The Pseudopilus consists of five different pseudopilins, EpsG, EpsH, EpsI, EpsJ and EpsK, with EpsG the most abundant and therefore called the “major pseudopilin” of the T2SS. The Inner Membrane Platform consists of five proteins: the two bitopic membrane proteins EpsL and EpsM; the membrane-anchored EpsC; the integral membrane protein EpsF; and the membrane-associated “secretion ATPase” EpsE in the cytosol. The focus of this paper is EpsL.
Several components of the T2SS are related to components of the type 4 pilin biogenesis (T4PB) system (Filloux, 2004). T4PB systems are responsible for a diversity of functions including pilus assembly and disassembly, protein export, DNA import and phage entry. Studies of T2SS proteins are therefore in principle also useful for increasing our understanding of the T4BP system, but there are considerable differences between the two systems as well, in particular regarding the Inner Membrane Platform. For instance, the T4BP system does not contain sequence homologs of GspM, GspC and neither of GspL (Filloux, 2004). However, a functional homolog of GspL appears to exist in the T4BP system (Crowther et al., 2005).
EpsL is a 40 kDa bitopic inner-membrane T2SS protein. Its cytoplasmic domain binds to the cytoplasmic “secretion ATPase” EpsE (Sandkvist et al., 2000). EpsL also interacts with the inner membrane proteins EpsM and EpsF and with the pseudopilin EpsJ (Douet et al., 2004; Sandkvist et al., 1995; Sandkvist et al., 1999) thus bridging the Inner Membrane Platform with the Pseudopilus of the T2SS. So far, structural information at the atomic level has been published for the C-terminal domains of EpsE (Robien et al., 2003), the periplasmic domain of EpsM (peri-EpsM) (Abendroth et al., 2004b), the cytoplasmic domain of EpsL (Abendroth et al., 2004a), all the five pseudopilins (Kohler et al., 2004; Korotkov and Hol, 2008; Lam et al., 2009; Yanez et al., 2008a; Yanez et al., 2008b), the cytoplasmic domain of EpsL in complex with the N-terminal domain of EpsE (Abendroth et al., 2005), the two N-terminal domains from Xanthomanas campestris GspE (Chen et al., 2005), the PDZ domain of EpsC (Korotkov et al., 2006), the first cytoplasmic domain of EpsF (Abendroth et al., 2009), and the three N-terminal domains of GspD (Korotkov et al., 2009).
Here, we describe the 2.3 Å crystal structure of the periplasmic domain of EpsL (peri-EpsL) from V. parahaemolyticus (Vp), an organism occurring in brackish water which can cause gastrointestinal illness in humans when ingested. This species was chosen because of the difficulties encountered in our studies on the homolog of V. cholerae. The amino acid sequence identity of peri-EpsL for these two species is 56 %. The sequence identity of Vp peri-EpsL with the peri-GspL homolog from ETEC is 29 %. Therefore, structural studies on peri-EpsL from V. parahaemolyticus are of direct relevance for understanding the functioning of the T2SS in several important bacterial pathogens (Figure 1).
The structure reported here for Vp peri-EpsL reveals a surprising topological similarity between peri-EpsL and the periplasmic domain of EpsM (peri-EpsM), another protein from the Inner Membrane Platform. Both domains are cyclic permutations of the “ferredoxin fold” containing two similar contiguous domains, observed for the first time in peri-EpsM. Despite adopting similar folds, peri-EpsM and peri-EpsL form entirely different dimers. The cytoplasmic domain of EpsL is dimeric as well (Abendroth et al., 2004a; Abendroth et al., 2005). Since EpsL interacts with EpsE, which is most likely hexameric, possible modes of assembling EpsE and EpsL in the Inner Membrane Complex of the T2SS are discussed.
The construct of V. parahaemolyticus peri-EpsL was created through PCR amplification from genomic DNA using the following primers:
anti sense: GGTTCTCGAGCTGACGCTTCAACAC
The PCR products were cloned in the pET21a(+) vector (Novagen) using NdeI and XhoI restriction sites.
Se-Met labeled V. parahaemolyticus peri-EpsL (Vp Per-EpsL), comprising residues 319-404 (see also Figure 2A), was prepared using methods as described (Van Duyne et al., 1993) by expression in BL21gold(DE3) E. coli cells. 20 ml of an overnight culture in Luria-Bertani broth (LB) containing 100 μl/ml carbenicillin were used after washing twice to inoculate a 1 l culture of M9 medium (50 mM Na / K-phosphate (pH 7.4), 0.5 g/l of sodium chloride, 1 g/l of ammonium chloride, 2 mM magnesium sulphate, 0.1 mM calcium chloride, 0.2% (w/v) glucose) containing 100 μg/ml of ampicillin. The cells were grown at 37°C. In order to inhibit the biosynthesis of methionine, thirty minutes before induction with 1 mM IPTG at A600=0.6 an amino acid mixture of 125 mg each lysine-HCl, threonine, phenylalanine, and 65 mg each leucine, valine, and selenomethionine was added per liter of broth. The cells were harvested after over-night induction at 20°C and pelleted by centrifugation at 6000×g for 15 minutes.
The pelleted cells were suspended in 30 ml of lysis buffer: 50 mM Tris-HCl (pH 8.0), 150 mM sodium chloride, and 0.1 mM PMSF. The cells were lysed by sonication and the debris was pelleted by centrifugation at 20,000×g for 20 minutes. The lysate was purified by immobilized metal affinity chromatography: 3ml of Ni-NTA (Qiagen) resin were equilibrated with 20 mM Tris-HCl (pH 8.0), 150 mM sodium chloride, 0.1 mM PMSF, and 2 mM tris (2-carboxyethyl)-phosphate (TCEP) and then incubated with the clarified lysate at 4°C for 1 hour. Following incubation, the column was washed with 30 ml of the same buffer containing 15 mM imidazole and the target protein was eluted with 15 ml of 150 mM imidazole-containing buffer. The pooled fractions were concentrated to 5 ml by ultrafiltration (10 kDa cut-off, Millipore). Gel-permeation chromatography was carried out in 20 mM Tris-HCl (pH 8.0), 150 mM sodium chloride, 0.1 mM PMSF, and 2 mM TCEP buffer with Superdex S-200 resin in a XK 60/16 system (Amersham-Pharmacia). Peak fractions were pooled and concentrated to 7.5-30 mg/ml. Protein concentrations were determined through the Bradford assay (BioRad) at 595 nm using bovine serum albumin as a reference.
Initial crystallization conditions were screened using commercial screens Wizard I and II (deCODE Biosystems), PEG-Ion, Index, and Salt RX (Hampton Research), using the sitting-drop method of vapor diffusion by mixing equal volumes (1-2 μl) of precipitant and protein at protein concentrations of 7.5-30 mg/ml. Initial hits (Index 19, 67, and 68) were refined and yielded the optimized crystallization conditions: (a) 1.2-1.4 M K/Na phosphate; (b) 25-30 % PEG 3350, pH 7.5-8.0, and 200 mM (NH4)2SO4, respectively. Crystals appeared from 5 days up to several weeks after setting up the drops. For vitrification of the crystals, two different protocols were established: (i) crystals were transferred directly into a buffer containing 30 % glycerol or sequentially transferred into buffers containing increasing amounts of glycerol (10 %, 20 %, and 30 %); and, (ii) crystals were transferred directly into a buffer containing 30 % ethylene glycol or sequentially transferred into buffers containing increasing amounts of ethylene glycol (10 %, 20 %, and 30 %). The crystals were then vitrified in liquid nitrogen.
Data of a single crystal were collected at a wavelength of 0.97924 Å on beam line 9.2 at SSRL (Stanford Synchrotron Radiation Laboratory). A fluorescence scan was carried out in order to optimize the wavelength for a strong anomalous signal. The data were indexed, integrated and scaled with the HKL2000 suite (Otwinowski and Minor, 1997). Se sites were searched for at 2.7 Å resolution with ShelxD (Schneider and Sheldrick, 2002), the correct hand was determined with ShelxE. The program Sharp (Bricogne et al., 2003) was used for the refinement of sites at 2.3 Å resolution and for phasing. For density modification using Solomon (Abrahams et al., 1996) and DM (Cowtan and Zhang, 1999) were employed. The good quality of the resulting experimental map allowed ARP/wARP (Perrakis et al., 1999) or Resolve (Terwilliger, 2000) to build ~150 residues. The model was completed manually in Coot (Emsley and Cowtan, 2004) and refined with Refmac5 (Murshudov et al., 1997). Seven TLS groups per chain were determined using the TLSMD server (Painter and Merritt, 2006). Tight NCS restraints were maintained throughout the refinement. Therefore, care was taken to assign the free reflections in thin resolution shells. Two elongated pieces of density were tentatively modeled as ethylene glycol since this was the cryo-protectant, three spherical densities were modeled as phosphates, most likely originating from the crystallization buffer.
Based on the limited proteolysis results of the V. cholerae protein (See Supplementary Text), a construct of V. parahaemolyticus peri-EpsL spanning residues 319-404, which corresponds to residues 318-403 of V. cholerae peri-EpsL, was cloned and expressed as Se-Met protein (Figure 2A). Size exclusion chromatography of Vp peri-EpsL yields an molecular weight of approximately 17.1 kDa (Figure 2B) and dynamic light scattering data showed that peri-EpsL forms dimers of approximately 19-21 kDa in solution (Figure 2C). Since the molecular weight of a single chain of Vp peri-EpsL is 10.5 kDa this indicates that the protein forms dimers in solution. The crystals obtained allowed the structure to be solved by SeMet SAD methods. The structure of V. parahaemolyticus peri-EpsL has been refined at 2.3 Å resolution to an Rwork of 21.4 % and an Rfree of 25.6 % with good geometry (Table 1). There are two chains per asymmetric unit, with each chain consisting of residues Ser322 to Gln404 in the final structure, since the first three residues of the expressed protein was not represented by electron density in the maps obtained. Since density for Asp392 was weak in both chains, this residue is probably quite flexible and was not incorporated into the structure.
Each Vp peri-EpsL chain folds as a compact globular unit and consists of two α-helices and four β-strands forming an anti-parallel β-sheet (Figure 1, ,2D).2D). The secondary structures elements of the C-terminal periplasmic domain are labeled Cα1, Cα2 etc for the helices and CβA, CβB, etc for the strands in order to distinguish these secondary structure elements in the C-terminal periplasmic domain from helices α1, etc. and strands βA etc. in the cytoplasmic domain of EpsL (Abendroth et al., 2004a; Abendroth et al., 2005). The order of secondary structure elements is Cα1CβACβB- Cα2CβCCβD. Each of the αββ-units forms a compact sub-domain, with the two αββ-halves related by a pseudo twofold axis that runs vertical to the β-strands such that the two helices are on the same side of the sheet (Figure 2D). The fold of Vp peri-EpsL is an untypical variant of the ferredoxin fold. While the generic ferredoxin-fold also consists of four α-helices and two β-strands and also forms two αββ-units, the first of these αββ-units is discontinuous and made up from an N-terminal α-helix and two β-strands from the C-terminus of the polypeptide chain. In contrast, each αββ-unit in peri-EpsL is formed by a contiguous chain.
Although the two αββ-units in peri-EpsL adopt a similar fold they are very different in sequence. Using the INDONESIA program (Madsen et al., 2002) the two domains superimpose with an r.m.s. deviation of ~ 1.8 Å for Cα atoms, with only one out 37 aligned residues being identical.
Vp peri-EpsL forms dimers in the crystals, in agreement with the results of size exclusion chromatography and dynamic light scattering experiments (Figures 2B,C). The contacts between the two protomers in the asymmetric unit bury 872 Å2 solvent accessible surface. This is the result, however, of rather limited interactions between the α1-helices across the non-crystallographic twofold (not shown). An “A-A′ dimer” is formed by a crystallographic dyad and buries 1440 Å2 solvent accessible surface, a considerably larger amount than between the two subunits in the asymmetric unit. Two anti-parallel β-strands from the two protomers in the A-A′ dimer form an anti-parallel arrangement, resulting in an eight-stranded β-sheet in the dimer (Figure 2D). In addition to the polar interactions of residues in strand Cβ1 (Thr347-Gly353) with the equivalent strand Cβ1′ of the other subunit in the dimer, a large portion of the dimer interface is hydrophobic and involves residues Leu328 and Leu331 from helix Cα1, and Phe349 and Tyr351 from strand Cβ1. Also, Thr323 from helix Cα1 and residues Ile346, Thr347, Gly353, are involved in polar interactions between A and A′. The A-A′ dimer interface ΔG, as calculated by PISA (Krissinel and Henrick, 2007), is -10.8 kcal/mol which is much more favorable than the +0.5 kcal/mol calculated for the interaction between the A and B subunits in the asymmetric unit. The A-A′ dimer is therefore most likely the dimer corresponding to that observed in solution, and will hereafter be referred to as the “peri-EpsL dimer”, shown in Figure 2D.
Several surface features of the Vp peri-EpsL dimer are quite striking. For instance, when viewed perpendicular along the twofold axis towards the β-sheet, a cluster of four positively charged residues is seen near the twofold, which is formed by two Lys 350 and two Arg 359 residues provided by two subunits (Figure 2E). The positive charge of these residues is conserved in the family of Vibrio EpsL sequences (Figure 1B) but not in the broader GspL family (Figure 1A). It could therefore be that these residues have a specific role in the T2SS of Vibrio species. A striking hydrophobic feature occurs on the other side of the Vp peri-EpsL dimer, where helices α1 are approaching each other (Figure 2E). The residues mainly responsible for this hydrophobic patch are Ala 326, Ala 332, Ala 333, Pro335 and Ala 336, from both subunits. Inspection of sequence alignments (Figure 1) shows that these residues are only poorly conserved, even among Vibrio species.
The interface residues in the Vp peri-EpsL dimer are, on the other hand, well conserved in the GspL family. For instance, Leu 328 is a Leu or Ile, once a Thr and once a Gly (Figure 1A). Leu 331 shows very little variability and is either a Leu or Met; Phe 349 is always a Phe, Ile or Leu; Tyr 351 either a Tyr, Phe or Trp; Ile 346 near the start of strand β-1 is a Ile, Leu, Val, or Met, and only once a Pro. Hence it seems likely that the Vp peri-EpsL dimer is conserved across the GspL family.
Searches in the PDB for structural homologs, using the DALI (Holm and Sander, 1993), SSM (Krissinel and Henrick, 2004) and INDONESIA (Madsen et al., 2002) search algorithms, revealed, most unexpectedly, that the periplasmic domain of EpsM (Abendroth et al., 2004b) is the closest structural homolog of peri-EpsL with a DALI Z-score of 7.0 and an r.m.s. deviation of 2.1 Å for 71 structurally equivalent residues (Figure 3). Peri-EpsM is the only protein in the search with the same permutation of the ferredoxin fold as observed in peri-EpsL. While the fold and permutation of secondary structure elements of peri-EpsL and peri-EpsM are conserved (Figure 3), the structure-based sequence alignment of these two periplasmic domains yields 13 identical amino acids out of 71 equivalent residues, or ~ 18 % identity.
Even though the folds of the peri-EpsL and peri-EpsM domains are similar, their dimer arrangements are unrelated (Figure 3). In peri-EpsM, the dimer-interface of 916 A2 as calculated by PISA is formed by extensive contacts between residues from helices α2 and α2′. Despite the close proximity in space, residues from strands β3 and β3′ in the interface, do not interact with one another leaving a cleft with a hydrophobic bottom in which an extra density from an as yet unknown molecule is clearly visible (Abendroth et al., 2004b). The peri-EpsL dimer interface is entirely different than that of the peri-EpsM dimer, with the former made up by residues from the N-terminal helices Cα1 and Cα1′ and the N-terminal strands CβA and CβA′, while the latter is made up by the C-terminal helices α2 and α2′ without involving β-strand interactions. (Figure 3).
As mentioned above, peri-EpsL folds into an unusual variant of the ferredoxin-fold. The ferredoxin-like fold is shared by an enormous variety of proteins (there are 44 superfamilies in the current SCOP database, see also (Abendroth et al., 2004b). It is most intriguing that the αββ-αββ permutation of the ferredoxin fold found in peri-EpsL has so far only been observed for the periplasmic domain of EpsM (Abendroth et al., 2004b), the protein that binds to EpsL in the type 2 secretion system (Johnson et al., 2006; Sandkvist et al., 1999). In these variants of the ferredoxin fold, a pseudo twofold axis relates the two halves. In the vast majority of the structures with a ferredoxin fold, the chain goes forth and back between the two halves, resulting in a first domain made up of two discontinuous segments of the polypeptide chain. In the ferredoxin fold of peri-EpsL and peri-EpsM, the two αββ units in the first and second domains are both continuous. In that sense, this variant is simpler than the common ferredoxin fold, and we propose therefore to call it the “fundamental ferredoxin fold”. The overall degree of sequence identity between peri-EpsL and peri-EpsM is ~ 18 % (Section 3.4) which might be a reflection of a common ancestor. In contrast, the cytoplasmic domains of these two bitopic inner membrane T2SS proteins have nothing in common whatsoever: the N-terminal domain of EpsM is only approximately 20 residues long, while the cytoplasmic domain of EpsL comprises about 240 residues and adopts a fold which is an unusual variant of the actin topology (Abendroth et al., 2004a; Abendroth et al., 2005).
Several investigations by the Sandkvist group have shed light on the interactions between EpsL and EpsM: (i) residues 84-99 of V. cholerae EpsM, forming the N-terminal helix α1, are required for stable interactions with V. cholerae EpsL (Johnson et al., 2007), and (ii) the region formed by residues 216-296 of EpsL interacts with EpsM (Sandkvist et al., 2000). The latter residues are not covered by our peri-EpsL structure (Figures 1, ,2A),2A), hence this structure cannot provide detailed information about the mode of interaction of these two periplasmic domains of the T2SS. Since the EpsL-contacting N-terminal helix of EpsM resides in the periplasm, it is likely that the EpsM-contacting residues of EpsL also occur in the periplasm. This most likely then comprises all or parts of the region from residue ~ 276, the end of the TM helix, to residue ~296, the end of region implied by biochemical studies (Sandkvist et al., 2000), which is located approximately 10 residues before the beginning of the peri-EpsL domain. Since both peri-EpsM and peri-EpsL form dimers (Figure 3), a global suggestion as to how the two periplasmic parts of EpsM and EpsL may be assembled is that the twofold axes of the two dimers coincide. This possibility is sketched very schematically in Figure 4 (to be discussed in more detail in the next Section) with the EpsL dimer in green colors and EpsM in shades of yellow. The full details of the interactions of EpsL and EpsM in the periplasm obviously still require further investigations.
The Vp peri-EpsL dimer is likely to be representative for the T2SS of other species given the high degree of conservation of interface residues observed in the V. parahaemolyticus dimer (Section 3.2) and the yeast two-hybrid studies on the periplasmic domain of Erwinia chrysanthemi GspL which indicated that this domain interacts with itself (Py et al., 1999). The observations that the periplasmic domain of EpsL forms a dimer in solution (Figures 2B,C) and in crystals (Figure 2D), and that the cytoplasmic domain of EpsL (cyto-EpsL) forms a dimer in crystals (Abendroth et al., 2004a; Abendroth et al., 2005) and the cytoplasmic domain of the homolog from E. chrysanthemi forms homo-multimers according to yeast-two hybrid studies (Py et al., 1999), suggest that also full-length EpsL forms dimers in the T2SS.
EpsL is known to interact in the cytosol with EpsE, a “secretion ATPase” associated with the Inner Membrane Platform of the T2SS (Abendroth et al., 2005; Sandkvist et al., 1995; Shiue et al., 2006). The major domains of EpsE are N1, N2, C1 and C2, where we ignore for simplicity the small but essential metal-binding CM domain (Possot and Pugsley, 1997; Robien et al., 2003). The precise multimeric nature of EpsE and its T2SS family members has been hard to establish. The crystal structure of V. cholerae EpsE missing the first domain, hereafter called also “ΔN1-EpsE”, revealed a 61 helical arrangement of subunits (Robien et al., 2003). Biochemical studies (Camberg and Sandkvist, 2005) provided evidence for hexamer formation in solution but hexamers represented only a small fraction of the total EpsE population. However, the monomer-hexamer equilibrium could obviously be greatly affected by interactions of GspE with lipids and partners in the T2SS. Interestingly, in X. campestris, GspE hexamer formation is influenced by the binding of ATP analogs (Shiue et al., 2006).
It is also useful to look at other secretion ATPases, related to EpsE but not from T2SSs, which have been studied crystallographically (Savvides, 2007). Closest in sequence to the N2-C1-C2 domains of EpsE are the secretion ATPases GspE from Archaeoglobus fulgidus (Yamagata and Tainer, 2007) and PilT from Aquifex aeolicus (Satyshur et al., 2007). GspE and PilT share 22 and 34 % sequence identity with the N2-C1-C2 domains of V. cholerae EpsE, respectively. These studies indicate that these non-T2SS homologs of EpsE can form hexameric arrangements with cyclic C6 symmetry, and also that these proteins can adopt multiple different conformations while remaining assembled as hexamers but with either C2 or C3 symmetry (Satyshur et al., 2007; Yamagata and Tainer, 2007). No structural information is available of partners of the secretion ATPases GspE and PilT, but for the T2SS three crystal structures involving EpsE's partner EpsL are presently available: (i) the cytoplasmic domain of EpsL by itself (Abendroth et al., 2004a); (ii) the same domain in complex with “N1-EpsE”, the first domain of EpsE (Abendroth et al., 2005); and, (iii) the periplasmic domain of EpsL reported in the current paper (Figure 2D). It is therefore of interest to look into the way in which a likely hexamer of the N2-C1-C2 domains of GspE may interact via their N1 domains with dimers of GspL in the Inner Membrane Platform of the T2SS. Additional key observations to be taken into account are:
Different options for combining a GspE hexamer with GspL dimers, while fulfilling the conditions that N1-GspE and cyto-GspL assemble as heterotetramers, that the TM helices of GspL cross the inner membrane, and that the periplasmic domains of GspL form dimers, include:
For option (iii) above, there are two variants possible for dimerization of full length GspL. In a first variant, the same two GspL chains form the periplasmic dimer and the cytoplasmic dimer, as depicted in Figure 4. Alternatively, the periplasmic GspL dimer might be formed by GspL chains from two different cytoplasmic GspL dimers. In both variants, six GspL chains are combined with six GspE chains giving a GspE6:GspL6 stoichiometry.
Although no dimer formation has been observed by cyto-EpsL domains in solution (Abendroth et al., 2004a; Abendroth et al., 2005), the fact that cyto-GspL interacts with itself in yeast two-hybrid studies and forms dimers in crystals (Abendroth et al., 2004a; Abendroth et al., 2005; Py et al., 1999), combined with the observations that peri-GspL forms dimers in yeast two-hybrid studies (Py et al., 1999), in solution (Figures 2B,C) and in crystals (Figure 2D), makes it likely that cyto-EpsL forms dimers in full-length EpsL. Of course, whether one of the above-sketched options, or still other arrangements of GspE and GspL subunits, actually occurs in the inner membrane of the assembled T2SS, still needs experimental confirmation. In addition, GspL interacts with other T2SS proteins, including GspM, GspJ and GspF (Douet et al., 2004; Michel et al., 1998; Py et al., 1999; Py et al., 2001; Sandkvist et al., 1995; Sandkvist et al., 1999; Sandkvist et al., 2000; Shiue et al., 2006), in ways which still have to be unraveled.
Appendix A. Supplementary Data: Supplementary data associated with this article can be found in the online version.
We thank Claudia Roach, Jaclyn DelaRosa, Stewart Turley, Konstantin Korotkov and Partha Sampathkumar for assistance with molecular biology, protein characterization and data collection, and Dr. Mark Strom and Rohinee Paranjpye, NOAA, Seattle for providing V. parahaemolyticus DNA. We are indebted to the support staff of beam line 9-2 of the SSRL for assistance during data collection. SSRL is supported by the Department of Energy and by NIH. This work was supported by grant AI34501 from the NIH to W.G.J.H and by the Howard Hughes Medical Institute (HHMI).
PDB Deposition: Coordinates and structure factors have been deposited in the PDB with PDB code 2W7V.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.