The LptA Protein of Escherichia coli Is a Periplasmic Lipid A-binding Protein Involved in the Lipopolysaccharide Export Pathway*

The LptA protein of Escherichia coli has been implicated in the transport of lipopolysaccharide (LPS) from the inner membrane to the outer membrane. Here we provide evidence that LptA binds structurally diverse LPS substrates in vitro and demonstrate that it interacts specifically with the lipid A domain of LPS. These results are consistent with LptA playing a chaperone role in the transport of LPS across the periplasm and have implications for possible assembly models.

Lipopolysaccharide (LPS) 2 is a major component of the outer leaflet of the outer membrane (OM) of Gram-negative bacteria (1)(2)(3). Lipid A (Fig. 1A) functions as the hydrophobic membrane anchor of LPS and is linked via a core oligosaccharide to the long chain polysaccharide known as O-antigen (2). Two distinct pathways are involved in the biosynthesis of the lipid A-core and O-antigen (Fig. 2). The lipid A-core portion (known as "rough" LPS (R-LPS)) is synthesized by a conserved pathway involving proteins on the cytoplasmic surface of the inner membrane (IM) and is then transported across the IM by an essential ATP-binding cassette (ABC) transporter, MsbA (reviewed in Refs. [2][3][4][5][6]. An additional protein, YhjD, has recently been implicated in LPS export across the IM (7). Hence, this process is more complicated than initially thought. O-antigen is synthesized separately and is present in its completed form at the periplasmic face of the IM, attached to the carrier lipid (undecaprenyl diphosphate). These two major pathways converge with the ligation of O-antigen to lipid A-core at the periplasmic face of the IM, mediated by an integral membrane protein, WaaL (reviewed in Ref. 2). This generates "smooth" LPS (S-LPS). Finally, the completed LPS mole-cule is translocated to the cell surface. O-antigen attachment is not required for lipid A-core transport to the OM because there is heterogeneity in the LPS molecular species; both R-LPS and S-LPS are found on the cell surface. The smallest LPS derivative supporting viability in Escherichia coli is lipid IV A (7,8) (Fig.  1B). However, this requires mutations in either MsbA or the integral membrane protein of unknown function, YhjD, to suppress the normally lethal consequence of an incomplete lipid A (7).
Although the early phases of LPS biosynthesis are generally well known, the mechanisms of LPS transport and its insertion into the OM still remain unclear. Equally unknown is the extent of coordination and interplay with other trafficking events required for OM assembly (9,10). Recent studies have begun to shed light on the essential components of LPS transport. These include the conserved OM protein known as LptD (formerly Imp or OstA). LptD is essential in E. coli, and depletion of this protein results in abnormalities in OM assembly and increased OM density (evident in sucrose gradient centrifugation profiles), which are consistent with reduced lipid incorporation (11,12). LptD exists in a complex with LptE (formerly RlpB), an essential OM lipoprotein, the absence of which results in a phenotype resembling an LptD defect (13). These proteins may provide a mechanism to move the LPS molecules into and/or across the OM barrier.
Recently, five additional essential E. coli proteins (LptA, LptB, LptC (formerly YrbK), LptF (formerly YjgP), and LptG (formerly YjgQ)) have been implicated in LPS transport to the OM (14 -17). LptB is a cytoplasmic nucleotide-binding domain protein belonging to the ABC protein superfamily (15). LptF and LptG have been proposed as transmembrane domain proteins that participate with LptB in an ABC protein complex to extract LPS from the IM en route to the OM (17). LptC is a bitopic IM protein that has also been suggested to play a role in LPS extraction from the IM (14 -16). LptA, the subject of this investigation, is proposed to be a periplasmic protein (15). There are apparent similarities to the Lol proteins involved in export of OM lipoproteins. LolCDE is an ABC protein pump that releases nascent lipoproteins to a periplasmic chaperone, LolA (18). LolA transfers its cargo to LolB in the OM (19). In a comparable LPS scenario, LptBFG (and perhaps LptC) might release LPS molecules from the biosynthesis/ligation site in the IM into the translocation pathway. LptA is a candidate for the chaperone delivering nascent LPS molecules to LptDE in the OM. However, attempts to demonstrate LPS release in spheroplasts have been unsuccessful, despite using conditions where efficient lipoprotein release was detected (20). In an alternative translocation pathway, LPS molecules may be translocated to the OM via sites where the IM and OM come into apposition (20,21). A molecular scaffold approach is involved in the translocation of capsular polysaccharides in E. coli (22). Resolution of these possibilities requires insight from biochemical information for the translocation components, which is currently lacking.
Because LptA was implicated in the transport of LPS to the OM, we investigated its ability to bind to LPS. Using a variety of LPS substrates, we now report that LptA specifically binds to the lipid A domain of LPS and also is capable of binding to the lipid A precursor lipid IV A . To our knowledge, these are the first functional data for any of the OM or periplasmic components of the LPS export apparatus.

MATERIALS AND METHODS
Recombinant DNA Techniques-Plasmids, PCR amplification products, and DNA fragments were isolated using the Qiagen Spin Prep, Qiaquick PCR Purification, and Qiaquick Gel Extraction Kits (Qiagen), respectively. Custom oligonucleotide primers were obtained from Sigma. PCR reagents were purchased from Roche Applied Science. Restriction endonucleases, T4 DNA ligase, and Antarctic phosphatase were all purchased from New England Biolabs and used according to the manufacturer's instructions. Transformation was carried out by electroporation with Gene Pulser from Bio-Rad and methods described elsewhere (23). Chromosomal DNA was prepared using the InstaGene kit (Bio-Rad) according to the manufacturer's directions.
Construction of the lptA Expression Vector-The lptA gene was amplified from E. coli W3110 (K12 wild type, F Ϫ , Ϫ ) chromosomal DNA with primers MT1 (5Ј-GCGCGCGAATTC-ACCATGAAATTCAAAACAAAC-3Ј) and MT2 (5Ј-GCGCG-CTCTAGATCAGTGGTGGTGGTGGTGGTGATTACCCT-TCTTCTGTGC-3Ј). The reverse primer incorporates the sequence encoding a C-terminal hexahistine (His 6 ) tag. The PCR product was digested with the restriction enzymes EcoRI and XbaI (sites underlined) and ligated into the corresponding sites of pBAD24, a vector containing an arabinose-inducible promoter for controlled expression (24), to generate the plasmid pWQ291. The identity of the cloned fragment was confirmed by DNA sequencing.
Overexpression and Purification of LptA-E. coli TOP10 (F Ϫ mcrA ⌬(mrr-hsdRMS-mcrBC) 80lacZ⌬M15 ⌬lacX74 deoR recA1 araD139 ⌬(ara-leu)7697 galU galK rpsL Str r endA1 nupG) carrying the plasmid pWQ291 was grown at 37°C in M9 minimal medium (25) containing 100 mg/ml ampicillin, 0.2% glycerol, 20 g/ml casamino acids, and 0.4% glucose for 18 h. This culture was diluted 1:100 in fresh medium containing no glucose and grown until the optical density (A 600 ) of the culture reached 0.6. Expression of LptA-His 6 was induced for 1 h by adding L-arabinose to a final concentration of 0.02% (w/v). Cells were then harvested by centrifugation at 5,000 ϫ g for 10 min. The cell pellet was resuspended in Buffer A (20 mM NaH 2 PO 4 , pH 7.5, and 300 mM NaCl) containing 20 mM imidazole, and the cells were disrupted by three passages through a French press. Unbroken cells and cellular debris were removed by centrifugation (12,000 ϫ g for 30 min). The membrane fraction was removed from the cell-free lysate by ultracentrifugation (100,000 ϫ g for 60 min). The LptA-His 6 protein was then purified from the supernatant using HIS-Select nickel affinity gel (Ni-NTA; Sigma). The column was washed sequentially with Buffer A containing 20 mM (W1) and 75 mM (W2) imidazole, respectively. The protein was eluted in Buffer A containing 300 mM imidazole (E1), followed by a final wash with Buffer A containing 500 mM imidazole (E2). Elution fractions were monitored by 12% polyacrylamide gels by SDS-PAGE (26). The fractions containing LptA-His 6 were pooled, desalted using a PD-10 column (GE Healthcare), and eluted in Buffer B (20 mM Tris, pH 7.5, and 150 mM NaCl). The desalted protein was concentrated using a Vivaspin 15R column (10,000 molecular weight cut-off; Vivascience). Approximately 5 mg of protein

-lipid A and its intermediate, lipid IV A , in E. coli.
A, lipid A is a glucosamine-based lipid that serves as the hydrophobic anchor of LPS and is the bioactive component of the molecule associated with Gram-negative septic shock (6). The predominate hexa-acyl species with two Kdo residues from the inner core is shown here. B, lipid IV A is a lipid A precursor that lacks both Kdo residues, as well as the two fatty acyl chains attached to the distal glucosamine at the 2Ј and 3Ј position (6). Lipid IV A is the minimal lipid A structure that supports the growth of E. coli (7,8).
were obtained from 1 liter of culture. Protein concentration was determined by the bicinchoninic acid method (27), using bovine serum albumin as the standard.
Properties of LptA-The molecular weight of purified LptA-His 6 was determined by MALDI-TOF mass spectrometry performed at the Biological Mass Spectrometry Facility, University of Guelph. The oligomerization status of LptA was determined on a Superose 12 HR 10/30 column (Amersham Biosciences) equilibrated with Buffer B. Protein elution was monitored at a wavelength of 280 nm. The column was calibrated with Sigma molecular mass standards . Injections were carried out with a total of 200 g of LptA-His 6 in a final volume of 0.2 ml.
Preparation of Cellular Fractions-Typically, 200 ml of E. coli cultures were grown and harvested as described above. Cell-free extract, soluble (cytosol/periplasm) fraction, and washed membranes were prepared as described previously (28). All samples were prepared at 4°C and stored in aliquots at Ϫ20°C. The soluble and membrane fractions were solubilized in SDS-containing sample buffer by boiling at 100°C for 5 min and then separated on 12% SDS-polyacrylamide gels. Proteins were visualized by SimplyBlue SafeStain (Invitrogen) or by Western immunoblotting. For Western immunoblotting, proteins were transferred onto nitrocellulose membranes (Pall Life Sciences) and probed with Qiagen anti-pentahistidine mouse monoclonal antibodies according to the manufacturer's instructions. Colorimetric detection was used with a secondary goat anti-mouse antibody conjugated to alkaline phosphatase (Jackson ImmunoResearch Laboratories) and visualized with nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate.
LPS Preparation-S-LPS from E. coli serotype O9a was isolated using the hot water-phenol method (29). R-LPS from E. coli W3110 was isolated using the phenol-chloroform-light petroleum (PCP) method (30).
In Vitro LPS Binding Assay-The in vitro LPS binding assay was based on the approach described previously (36) with some slight modifications. Briefly, the assays were carried out in 500-l reactions in Buffer B containing 500 g of purified LptA-His 6 and either purified LPS (150 g) or 100,000 cpm of each radiolabeled lipid A compound (Kdo 2 -[4Ј-32 P]lipid A, Kdo 2 -[4Ј-32 P]lipid IV A , or [4Ј-32 P]lipid IV A ) at a specific activity of ϳ10 Ci/nmol. Each reaction was performed at room temperature for 1 h on a rotary shaker. Ni-NTA resin (250 l, washed in 1 ml of Buffer B) was added to the reaction mixture and incubated for another hour. Afterward, the reaction mixture was transferred to the micro bio-spin chromatography column (Bio-Rad), and the flow-through was collected (FT). The resin was then washed twice with 1.5 ml of Buffer B (W1-W2). LptA-His 6 was eluted with 500 l of Buffer B containing 300 mM imidazole (E1), and the elution step was repeated three more times (E2-E4). A final wash with 500 l of Buffer B containing 500 mM imidazole was employed to ensure all protein was eluted (E5). The radioactivity in the lipid A binding assay was detected by liquid scintillation counting. In addition, 20 l from each fraction collected was spotted onto silica gel 60 thin layer chromatography (TLC) plates (Merck), and the plates were dried under a cool air stream for 20 min. When [4Ј-32 P]lipid IV A was employed as the binding substrate, the lipid was separated using the solvent chloroform, pyridine, 88% formic acid, and water (50:50:16:5, v/v). For binding assays containing either Kdo 2 -[4Ј-32 P]lipid IV A or Kdo 2 -[4Ј-32 P]lipid A as the binding substrate, the TLC plates were developed in the solvent chloroform, pyridine, 88% formic acid, and water (30:70:16:10, v/v). After drying and overnight exposure of the plate to a PhosphorImager Screen, the product formation was detected and analyzed using a Bio-Rad Molecular Imager PhosphorImager equipped with Quantity One Software.
In Vitro Phospholipid Binding Assay-Phosphatidylethanolamine (PtdEtn; Sigma) was dissolved in chloroform/ methanol (4:1, v/v) and transferred to a tube where the solvent was evaporated under nitrogen gas. The lipids were dispersed in Buffer B by sonication for 1 min in a sonicator bath prior to incubating with purified LptA-His 6 at room temperature for 1 h on a rotary shaker. Ni-NTA resin (250 l, washed in 1 ml of Buffer B) was added to the reaction mixture and incubated for another hour. Afterward, the reaction mixture was transferred to the micro bio-spin chromatography column and processed as described above. A range of amounts of PtdEtn were tested (25, 50, 250, and 500 g) in standard 500-l reactions containing 500 g of LptA-His 6 . 20-l aliquots of each elution fraction from the Ni-NTA column were spotted onto a TLC plate. Chromatography was performed using chloroform, methanol, and water (65:25:4, v/v). After drying, PtdEtn was detected by spraying the plate with 10% sulfuric acid in ethanol and then charring.
LPS Detection-LPS was examined by SDS-PAGE using 4 -12% BisTris NuPAGE gels from Invitrogen and visualized by silver staining (37), according to the Hitchcock and Brown (38 method described previously.

LptA Is a Soluble Periplasmic Protein-Previous work by
Sperandeo et al. (15) proposed that LptA is a periplasmic protein with a predicted molecular mass of 20.13 kDa and showed that its expression is under the control of a ⌭ -dependent promoter. Primary sequence analyses of E. coli LptA using the ExPASy tools (39) reveal a putative N-terminal signal sequence (Met 1 -Ala 27 ), suggesting the protein is exported into the periplasm. In contrast, the LptA homolog in Neisseria meningitidis NMB0355 was recently reported to be located in the membrane fraction rather than in the soluble periplasmic fraction (9). Therefore, to evaluate the possible membrane association of LptA from E. coli, we constructed a His 6 -tagged version of LptA under the control of the arabinose-inducible araC promoter (plasmid pWQ291). From the cell-free lysate (Fig. 3A,  lane 2), the soluble and membrane fractions were separated with two centrifugation steps. Based on the profiles from the SimplyBlue SafeStain-treated SDS-polyacrylamide gel, LptA-His 6 is essentially confined to the soluble fractions (Fig. 3A,   lanes 3 and 5). Only a minor trace of LptA-His 6 was seen in the membrane fractions (Fig. 3A, lanes 4 and 6), and this was detectable only with the increased sensitivity of an anti-pentahistidine immunoblot.
LptA-His 6 was purified to near homogeneity by Ni-NTA affinity chromatography, and its identity was verified by Western immunoblotting with anti-pentahistidine monoclonal antibody (Fig. 3B). Both SDS-PAGE and Western immunoblot analyses of purified LptA-His 6 consistently revealed the presence of two protein species of different apparent molecular masses. However, MALDI-TOF mass spectrometry of the same sample protein revealed only one major ion peak at m/z 18234.66 atomic mass units, corresponding to a processed mature LptA-His 6 protein that lacks 27 residues (Met 1 -Ala 27 ) at the N terminus (predicted molecular mass ϭ 18189.36 Da (39)). This is consistent with the transport of a periplasmic protein across the IM. Gel filtration chromatography confirmed that purified LptA-His 6 is a monomer (data not shown). The additional protein band in the SDS-polyacrylamide gel appears to be a form of LptA-His 6 with altered electrophoretic migration characteristics. A similar observation was made for an O-antigenic polysaccharide (O-PS)-binding domain protein (Wzt) possessing extensive ␤-sheet structure (36). Secondary structure predictions (data not shown) suggest LptA also has high ␤-sheet content. These results corroborate and extend the previous proposal that E. coli LptA is a soluble periplasmic protein (15). LptA Binds to Both LPS and R-LPS In Vitro-By analogy to the Lol system, one potential role of LptA is a periplasmic chaperone protein. To investigate whether LptA can interact with LPS directly, LPS purified from E. coli serotype O9a was incubated with purified LptA-His 6 . The His 6 -tagged protein was then purified from the reaction mixture by Ni-NTA affinity chromatography resin and examined for the presence of bound LPS. As shown previously (36), purified LPS does not bind to Ni-NTA resin (Fig. 4A), and the classical ladder-like profile reflecting a mixture of R-LPS and S-LPS is evident in the flowthrough. In the presence of LptA-His 6 , a substantial amount of LPS is bound and coelutes with LptA-His 6 (Fig. 4B). No LPS binding occurs with irrelevant His 6 -tagged proteins (data not shown). LptA also binds S-LPS from other serotypes (data not shown), indicating that the structure of the O-PS plays no role in substrate recognition. E. coli K12 strains do not contain O-PS due to mutations in the O-PS biosynthesis locus (40). As demonstrated in Fig. 4C, LptA also binds R-LPS from W3110, thus further providing evidence that the O-PS is not required for substrate recognition.
LptA Specifically Binds to Lipid A-Next we wanted to investigate whether the core oligosaccharide domain of LPS was required for binding. In order to determine whether the core oligosaccharide domain is important for the interactions, Kdo 2 -[4Ј-32 P]lipid A (Fig. 1A) was synthesized in vitro and subjected to the same binding assay. In the absence of LptA-His 6 , the majority of the radiolabeled lipid was detected in the flow-through (FT) fraction (Fig. 5A). However, after incubation with LptA-His 6 the majority of 32 Plabeled lipid A coeluted with the protein (E1-4). These results indicated that the outer core oligosaccharide domain is not required in the binding of LPS by LptA (Fig. 5A). In addition, we generated Kdo 2 -[4Ј-32 P]lipid IV A and [4Ј-32 P]lipid IV A in vitro. These precursors lack the two fatty acyl chains attached to the distal glucosamine at the 2Ј and 3Ј positions (Fig. 1B). Both precursors were efficiently bound by LptA-His 6 (Fig. 5, B and C), indicating that glycosylation of the lipid A backbone and the number and asymmetric arrangement of fatty acyl chains of lipid A are not essential for binding. From the analysis of these truncated LPS precursors, LptA appears to bind to the lipid A domain of LPS, and efficient binding can be seen in vitro with as few as four fatty acyl chains.
LptA Does Not Bind to the Phospholipid PtdEtn-In E. coli, the predominate OM phospholipid is PtdEtn (41). Phospholipids are synthesized at the cytoplasmic side of the IM and are transported across the IM by an unknown mechanism in order to reach the OM (42). MsbA has been implicated in the transfer of both LPS and phospholipids across the IM (43); however, this finding is still controversial (7,44). Tefsen et al. (20) reported that newly synthesized phospholipids were not transported to the OM of spheroplasts, suggesting that a soluble periplasmic intermediate analogous to LolA may be required. Therefore, we wanted to investigate whether LptA can also bind to phospholipids. Purified E. coli PtdEtn was tested in the binding assay with purified LptA-His 6   did not bind to Ni-NTA resin (data not shown), and even after incubation with LptA-His 6, it was only detected in the flow-through (FT) fraction (Fig. 6). PtdEtn was also used in an attempt to act as an antagonist of LPS binding to LptA-His 6 , but no effect was observed (data not shown). The available data, therefore, suggest that regardless of the function(s) of MsbA, LptA is part of a periplasmic machinery specific for lipid A-based glycolipids.

DISCUSSION
LPS and phospholipids are essential components of the OM in most Gram-negative bacteria, and their structure and biosynthesis are well known (2,7,42). However, little is known about the mechanisms of transport and assembly of both of these lipids in the OM (9,10). Identification and characterization of the inner membrane protein MsbA has provided insight into the way LPS is transported across the IM (5,7,45). However, it is still unclear how LPS is transported across the periplasm and inserted into the OM.
The data reported here suggest that LptA functions as an LPS-binding protein. By binding to lipid A, LptA may shield the hydrophobic moiety, facilitating LPS transfer in the aqueous environment of the periplasm. This is analogous to the Lol system for lipoprotein transport (9). LolA forms a ␤-barrel with an ␣-helical lid and possesses a hydrophobic cavity composed primarily of aromatic residues that protect the acyl chains of the bound lipoprotein (46). Secondary structure predictions for LptA reveal that this protein also contains extensive ␤-strands (47), and it is conceivable that LptA may share some structurefunction properties with LolA. Lipoproteins are transferred fromLolAtoLolB,theoutermembranecomponent,inanenergydependent process (19). LolB shares structural features with LolA. The observation that LptA shares sequence similarity to the conserved N-terminal periplasmic domain of LptD (48,49) is therefore intriguing. Despite the fact that a previous study (20) was unable to show LPS release from spheroplasts, all other elements bear striking similarity to the Lol system for lipoprotein export and assembly into the OM.
Early studies indicated that LPS could be transported to the OM by points of contact between the IM and the OM. Initial support for this model came from electron microscopy studies with conditional Salmonella mutants where newly synthesized LPS molecules appeared at specific connection sites between IM and OM (21). These sites have been referred to as "Bayer's junctions," (50) but their existence has been controversial as their detection was sensitive to the PtdEtn Solvent front Origin * FIGURE 6. LptA-His 6 does not bind PtdEtn. The ability of LptA-His 6 to bind to PtdEtn (500 g) was assessed by the standard Ni-NTA affinity chromatography-based binding assay. FT, flow-through; W1-W2, washes; E1-E5, elutions. A sample of each fraction was separated by TLC and PtdEtn was visualized as described under "Materials and Methods" (upper panel). LptA-His 6 protein in each elution fraction was also examined by SDS-PAGE and stained with Sim-plyBlue SafeStain (lower panel). The asterisk on the upper panel identifies the migration of imidazole. method used for electron microscopy (51,52). However, it is now clear that physical linkages between the membranes do occur in flagella basal bodies (53); drug efflux pumps (54); and type I (55), III (56), and IV (57) protein secretion (58) and capsule export machinery (22). More recent studies have supported the notion that the insertion sites are not random but instead follow a helical pattern (59). The discovery of a soluble periplasmic LPS-binding protein (LptA) does not necessarily preclude additional components in a molecular scaffold spanning the periplasm nor does it argue against specific sites of insertion, as these could be determined by the distribution and location of the outer membrane components LptD and LptE.
It is well established that Gram-negative bacteria can efficiently export LPS molecules with diverse sizes and structures to the outer membrane; these are seen clearly in LPS profiles resolved by SDS-PAGE. Meredith et al. (8) recently identified a viable E. coli mutant with an outer membrane composed of lipid IV A , an LPS derivative that lacks both Kdo sugars. This minimal structure is recognized and exported across the IM (7,8). An essential requirement for any periplasmic export component is, therefore, the ability to recognize diverse LPS structures, including the minimal essential LPS molecules. LptA meets those requirements. In addition, BLAST searches reveal that LptA homologues are widespread in Gram-negative bacterial genomes, indicative of a conserved pathway. The essential requirement for LptA for viability, its conserved nature, and its periplasmic location collectively make LptA an interesting candidate for the development of inhibitors active against Gramnegative pathogens.
Note Added in Proof-While this work was in review, the x-ray crystal structure of LptA was reported. The structure contains 16 antiparallel ␤-strands in a novel fold. Unfortunately, the mechanism by which LptA binds LPS and participates in its export is not resolved by the structure (Suits, M. D., Sperandeo, P., Dehó, G., Polissi, A., and Jia, Z. (2008) J. Mol. Biol. 380, 476 -488).