Analysis of the Topology and Active-Site Residues of WbbF, a Putative O-Polysaccharide Synthase from Salmonella enterica Serovar Borreze

Glycosyltransferases play a critical role in the synthesis of a wide variety of bacterial polysaccharides. These include O-antigenic polysaccharides, which form the distal component of lipopolysaccharides and provide a protective barrier important for survival and host-pathogen interactions. Synthases are a subset of glycosyltransferases capable of coupled synthesis and export of glycans. Currently, the O:54 antigen of Salmonella enterica serovar Borreze involves the only example of an O-polysaccharide synthase, and its generation of a lipid-linked product differentiates it from classical synthases. Here, we explore features conserved in the O:54 enzyme and classical synthases to shed light on the structure and function of the unusual O:54 enzyme.

ABSTRACT

Bacterial lipopolysaccharides are major components and contributors to the integrity of Gram-negative outer membranes. The more conserved lipid A-core part of this complex glycolipid is synthesized separately from the hypervariable O-antigenic polysaccharide (OPS) part, and they are joined in the periplasm prior to translocation to the outer membrane. Three different biosynthesis strategies are recognized for OPS biosynthesis, and one, the synthase-dependent pathway, is currently confined to a single example: the O:54 antigen from Salmonella enterica serovar Borreze. Synthases are complex enzymes that have the capacity to both polymerize and export bacterial polysaccharides. Although synthases like cellulose synthase are widespread, they typically polymerize a glycan without employing a lipid-linked intermediate, unlike the O:54 synthase (WbbF), which produces an undecaprenol diphosphate-linked product. This raises questions about the overall similarity between WbbF and conventional synthases. In this study, we examine the topology of WbbF, revealing four membrane-spanning helices, compared to the eight in cellulose synthase. Molecular modeling of the glycosyltransferase domain of WbbF indicates a similar architecture, and site-directed mutagenesis confirmed that residues important for catalysis and processivity in cellulose synthase are conserved in WbbF and required for its activity. These findings indicate that the glycosyltransferase mechanism of WbbF and classic synthases are likely conserved despite the use of a lipid acceptor for chain extension by WbbF.

IMPORTANCE Glycosyltransferases play a critical role in the synthesis of a wide variety of bacterial polysaccharides. These include O-antigenic polysaccharides, which form the distal component of lipopolysaccharides and provide a protective barrier important for survival and host-pathogen interactions. Synthases are a subset of glycosyltransferases capable of coupled synthesis and export of glycans. Currently, the O:54 antigen of Salmonella enterica serovar Borreze involves the only example of an O-polysaccharide synthase, and its generation of a lipid-linked product differentiates it from classical synthases. Here, we explore features conserved in the O:54 enzyme and classical synthases to shed light on the structure and function of the unusual O:54 enzyme.

INTRODUCTION

The O-antigenic polysaccharides (OPS) of Gram-negative bacteria are part of a glycolipid known as lipopolysaccharide (LPS), which represents a major component of the outer membranes (OM) of most Gram-negative bacteria (1). The OPS extends outwards from the cell surface, forming a barrier that protects the bacteria from factors in the external environment, including elements of the host immune response (2). OPS is a hypervariable structure, and the repeat-unit structures define the O-antigen serotype in bacterial serological typing systems. In Salmonella, there are 46 recognized O-antigens (3). Three different strategies are used by bacteria to synthesize OPS, and these are distinguished by the process used for their export across the inner membrane (IM). All three pathways assemble OPS glycans using the lipid carrier C₅₅ undecaprenyl- phosphate (und-P) and ligate the finished product to LPS lipid A-core at the periplasmic face of the IM (2, 4). In the Wzy-dependent process, individual lipid-linked repeat units are delivered to a periplasmic polymerase by a representative of the multidrug/oligosaccharyl-lipid/polysaccharide (MOP) flippase (5). Some OPS are polymerized in the cytoplasm and are exported by a pathway-defining member of the ATP-binding cassette (ABC) transporter superfamily (4, 6). In contrast to these well-studied mechanisms, the synthase-dependent pathway is poorly understood, and the O:54 antigen of Salmonella enterica serovar Borreze is the only known OPS that follows this process.

The O:54 antigen is a homopolymer of N-acetylmannosamine (ManNAc) with alternating β-(1→3) and β-(1→4) linkages. The enzymes involved in its production are encoded by genes carried on a naturally occurring 6.9-kb ColE1-based mobilizable plasmid (7, 8). The plasmid is unique in possessing a functional O-antigen gene cluster comprising three genes: wbbE, wbbF, and mnaA (formerly rfbA, rfbB, and rfbC, respectively). Assembly of the O:54 repeating unit (Fig. 1) begins on the cytoplasmic face of the IM, with the transfer of N-acetylglucosamine-1-phosphate (GlcNAc-1-P) to und-P by WecA. WecA is a phosphoglycosyltransferase possessing UDP-GlcNAc:undecaprenylphosphate GlcNAc-1-P transferase activity and is encoded by the locus directing biosynthesis of enterobacterial common antigen (ECA) (9, 10). The WecA reaction is employed by examples from all three OPS assembly pathways (2). MnaA, a UDP-GlcNAc 2-epimerase, converts UDP-GlcNAc to UDP-ManNAc, the precursor needed for synthesis of the O:54 glycan, but the plasmid-borne mnaA is not essential for O:54 biosynthesis in members of the Enterobacteriaceae due to the existence of a functional homolog (wecB) in the ECA locus. Following precursor synthesis, a monofunctional glycosyltransferase (GT), WbbE, transfers the first ManNAc residue to und-diphosphate (PP)-GlcNAc, committing the intermediate to O:54 biosynthesis and creating an adaptor on which chain extension can occur (7). Next, WbbF is believed to polymerize the polysaccharide by forming the repeat domain of the OPS, with growth occurring at the nonreducing end (7). The absence of a dedicated transporter for the O:54 antigen led to the proposal that WbbF is sufficient for both polymerization and export of the nascent OPS. Such dual properties are found in a class of enzymes called synthases, leading to the name of the pathway.

FIG 1.

Synthase-dependent pathway for O:54 antigen biosynthesis. (1) WecA primes und-P for synthesis through the transfer of GlcNAc-1-P. MnaA, an epimerase, converts UDP-GlcNAc to UDP-ManNAc, the precursor needed for synthesis. (2) WbbE transfers the initial ManNAc residue to und-PP-GlcNAc, creating the adaptor region. (3) WbbF is the proposed synthase, which polymerizes the O:54 polysaccharide by sequential addition of ManNAc residues and exports the product to the periplasmic face of the IM, where ligation to the lipid A-core occurs.

Members of the synthase family are found in bacteria and eukaryotes and are capable of the coordinated synthesis and export of a variety of important biopolymers, such as cellulose, chitin, and hyaluronan (11). The GT modules of synthases are typically assigned to the GT2 family of inverting GTs in the CAZy database (http://www.cazy.org/) and share sequence motifs which correlate with the processivity of these enzymes (12–14). Cellulose synthase from Rhodobacter sphaeroides is the only example with a solved X-ray crystal structure and offers an influential prototype (12). While cellulose synthesis involves several proteins in a heterocomplex, the BcsAB subcomplex is sufficient for in vitro synthesis (15). BcsB is a periplasmic protein possessing a C-terminal transmembrane helix (TMH) that anchors the protein to the IM, while BcsA is an integral membrane protein possessing the catalytic site. The GT module in BcsA adopts a GT-A fold, characteristic of GT2 enzymes, and it lies between four amino-terminal and four carboxy-terminal TMHs, which form the narrow cellulose translocation channel that is occupied by a nascent glucan chain in the solved structure (12). Subsequent structures revealed a stepwise synthetic system, with a ratcheting mechanism to extrude the translocating cellulose chain (16).

Biosynthesis of cellulose, chitin, and hyaluronan occurs without the involvement of lipid-linked intermediates (12, 13, 17). In contrast, WecA-derived und-PP-linked intermediates initiate O:54 biosynthesis, and the final translocated product is transferred to lipid A-core at the periplasmic face of the IM by a ligase that requires an undecaprenyl-linked glycan as the glycosylation donor (8). This suggests fundamental differences between the activity of WbbF and that of other characterized synthases, yet WbbF does share some sequence features with the known synthases. To begin to address the structural and functional relationships between these enzymes, we investigated the membrane topology of WbbF and the functional importance of sequence motifs shared by WbbF and the characterized synthases.

RESULTS AND DISCUSSION

WbbF topology.

WbbF shares 40% similarity with the prototypic synthase, BcsA, and possesses conserved synthase motifs (Fig. 2). The conserved motifs and catalytic residues are discussed in detail below. To gain insight into the organization of the conserved motifs and the overall structure of WbbF, a structural model of WbbF was generated using the BcsA structure (PDB ID 4HG6). Despite having a low sequence identity with BcsA (24%), the model for WbbF produced by the Phyre2 server (18) showed 87% coverage of the full-length protein (Fig. 3). The model predicts a large cytoplasmic GT domain (see below), as expected for an enzyme whose donor substrate is a nucleotide diphosphosugar. Based on this prediction, the two proteins differ substantially in the number of TMHs: eight from the BcsA crystal structure, and three predicted in WbbF. Chitin synthase (CHS), its ortholog NodC, and hyaluronan synthase (HAS) from Saccharomyces cerevisiae, Sinorhizobium meliloti, and Streptococcus pyogenes, respectively, possess only three or four TMHs compared to the eight in BcsA (12, 13, 19). An earlier hydrophobic cluster analysis predicted four TMHs located between residues 11 and 40, 325 and 340, 385 and 406, and 416 and 438 and a PhoA active fusion at residue 368 (7). The Phyre2 (18) model contains one less TMH because residues 367 and 409, which are predicted to contain a TMH, were not modeled due to sequence variation between WbbF and BcsA. However, both predictions differ from results obtained with the Consensus Constrained TOPology (CCTOP) server, which integrates the output of 11 different topology servers to generate a reliable model (20, 21). CCTOP predicted a five-TMH topology with a reliability of 88% with a periplasmic N terminus, a cytoplasmic C terminus, and TMHs at positions 12 to 36, 290 to 305, 316 to 340, 385 to 405, and 418 to 438. To resolve the contradictory outputs from different methods, the topology was investigated experimentally using a cysteine-labeling strategy.

FIG 2.

Multiple sequence and secondary structure alignment of WbbF (GenBank accession no. AAC98402), NodC (GenBank accession no. WP_014531649), and BcsA. The predicted WbbF secondary structure (modeled by Phyre2 [18]) and that determined from the solved BcsA crystal structure (PDB ID 4HG6) are displayed above and below the alignment, respectively. α-Helices, β-strands, strict β-turns, and strict α-turns are represented by squiggles, arrows, TT, and TTT, respectively. Proposed catalytic residues, which were replaced with alanine, are indicated by a triangle, while the conserved DXD, (E/D)DX, and Q(Q/R)XRW motifs are highlighted in red. The alignment was generated with ClustalW (46) and displayed using ESPript (47).

FIG 3.

Structural comparison of BcsA and WbbF. The Phyre2 (18) model of WbbF was generated using the structure of BcsA (PDB ID 4HG6) as a template. Residues 9 to 368 and 410 to 459 were modeled. The horizontal lines represent the approximate location of the membrane bilayer that divides the cytoplasmic and periplasmic compartments. α-Helices are colored red, β-strands are yellow, and loops are green. The active sites of each protein are shown below the model. Known and putative active-site residues are shown as cyan sticks, with the conserved Q(Q/R)XRW (top helix) and T(E/D)DX (left helix) motifs are shown on the orange helices as sticks.

The native WbbF protein contains seven cysteine residues, so these were all replaced with alanine residues to facilitate the reintroduction of individual cysteine residues at specific reporter sites. In total, eight residues were replaced with cysteine to probe the number of TMHs. The changed residues are N2, A102, S285, V312, I348, S368, L413, and V447, and their positions are shown in the context of the four-TMH model in Fig. 4A. Each WbbF derivative contained a C-terminal FLAG epitope for protein detection, to ensure comparable levels of protein expression in whole-cell lysates (Fig. 4B). The wild-type (WT) and variant proteins consistently migrated as a doublet on SDS-PAGE. The predicted molecular weight of WbbF-FLAG is 54.5 kDa. The upper band (apparent molecular weight of ∼55.6 kDa calculated using Image Lab [Bio-Rad]) corresponds to the fully unfolded version. The lower band (∼47.3 kDa) may represent a partially unfolded form of WbbF or a degradation product. The denaturation step of WbbF sample preparation was performed at 37°C, because incubation at higher temperatures results in protein aggregation, rendering WbbF undetectable. A similar phenomenon has been experienced with other membrane proteins (22–24). To confirm that the numerous variant proteins retained function in O:54 biosynthesis and transport, cells expressing them were tested for surface-exposed immunofluorescence using rabbit polyclonal anti-O:54 antibodies (Fig. 4C). WbbF^C→A (cysteine-free) retained activity, as did each variant; as anticipated, control cells transformed with the vector alone gave no signal.

FIG 4.

(A) Topology model of WbbF^C→A (7). The conserved DXD, (E/D)DX, and Q(Q/R)XRW motifs are highlighted in orange. The residues which were replaced with cysteine and probed for OGM reactivity are highlighted in green. The initial model output was generated using Protter (48). (B) Western immunoblots of whole-cell lysates of E. coli Top10 harboring plasmid-encoded WbbF-FLAG and variants, within the O:54 cluster. Immunoblots were probed with the FLAG epitope-specific antibody to confirm that all proteins were expressed at comparable levels. (C) Immunofluorescence microscopy of S. enterica serovar Borreze and E. coli Top10 cells. Cells were fixed and probed with anti-O:54 antigen antibody and a fluorescent secondary antibody for detection.

The eight single-cysteine WbbF^X→C FLAG variants were examined for reactivity with a membrane-impermeant fluorescent reagent, Oregon Green 488 maleimide carboxylic acid (OGM) (25, 26). WbbF^C→A-FLAG (pWQ1026) provided the negative control in labeling experiments. The topology determination strategy was based on the reactivity of the single cysteine residues, which is dependent on whether they are located within the periplasm or cytoplasm. Cysteines in the periplasm (Fig. 5, lanes P) are expected to be labeled in intact cells, as OGM can penetrate the OM. In contrast, those within the lipid bilayer or cytoplasm remain inaccessible in the absence of cellular lysis to disrupt the IM barrier. The addition of OGM pre- and postlysis therefore distinguishes between periplasmic and cytoplasmic cysteines, respectively. Following OGM labeling, the cell membranes were collected, and WbbF-FLAG was extracted. Solubilized protein samples were enriched for WbbF-FLAG using magnetic beads specific for the FLAG epitope to remove as many contaminants as possible. However, attempts to purify WbbF using alternative methods were not successful due to low protein yields, typical of an integral membrane protein. The amounts of detergent-extracted WbbF variant proteins were comparable, as shown by protein staining in SDS-PAGE (Fig. 5, middle panel) and Western immunoblotting (Fig. 5, bottom panel), consistent with the data for whole-cell lysates (Fig. 4B).

FIG 5.

Site-directed fluorescent labeling of WbbF. The panels show SDS-PAGE and Western immunoblotting of DDM-solubilized WbbF-FLAG variants labeled with OGM. Top panel, SDS-polyacrylamide gel after exposure to UV light for detection of OGM-labeled residues. Asterisks indicate the location of WbbF-FLAG on the gel. Middle panel, the same gel as shown in the top panel, stained with SimplyBlue to confirm comparable protein loading. Bottom panel, Western immunoblot probed with anti-FLAG epitope antibodies, illustrating comparable expression of WbbF variants. Addition of OGM-labeled periplasmic (P) cysteine residues in unlysed cells, while addition postlysis labeled both periplasmic and cytoplasmic (P/C) cysteines.

Exposure of an SDS-polyacrylamide gel using UV light revealed a fluorescent WbbF-FLAG doublet in samples corresponding to OGM labeling in intact cells of N2C, I348, S368C, and V447C variants (Fig. 5, top panel). In contrast, the WbbF-FLAG doublet in A102C, S285C, V312C, and L413C was labeled only following lysis. These data are consistent with the originally proposed topology model (7), placing the N and C termini in the periplasm, the GT domain in the cytoplasm, and four domains spanning the membrane (Fig. 4A). Cysteine-labeling analysis was confined to determining the number of TMHs. No attempt was made to precisely define the borders of the TMHs by introducing additional cysteine residues, since this was not central to the goals, nor did it influence the conclusions of this study.

WbbF possesses catalytic core residues resembling known synthases.

The Phyre2 (18) model predicts a large cytoplasmic GT domain with a GT-A fold containing two neighboring β/α/β domains (Fig. 3). Although WbbF shares a limited identity with the other synthases (BcsA, 24%; HAS, 21%; CHS, 19%; and its ortholog, NodC, 23%), it possesses conserved motifs shared by those enzymes in a comparable (predicted) structural context (Fig. 2). All known synthases contain a signature Q(Q/R)XRW motif in the GT domain, which correlates with the processivity of these enzymes (27). This motif lines the active site, with the Trp residue acting to stabilize the growing end of the polymer (the acceptor) via CH-π stacking interactions (11). Two additional catalytic GT motifs are also conserved: DXD and (E/D)DX. The Asp residues of the DXD motif are involved in coordination of Mg²⁺ (or Mn²⁺), which is crucial for the activity of some GTs (28). BcsA and WbbF also possess the TED motif, which overlaps with the (E/D)D from the (E/D)DX motif. These motifs are located on what is known as a “finger helix” and are oriented toward the nonreducing end of the glycan (28). In BcsA, the Thr forms hydrogen bonds with the 2′- or 3′-OH of the acceptor, while the Asp is proximal to the 4′-OH of the acceptor and therefore acts as a general base to catalyze its deprotonation (11). Additionally, an aromatic residue (Y149 in BcsA) at the entrance of the active site is speculated to interact with the uracil moiety of the UDP-sugar. This residue is critical for NodC and CHS activity (13). Aligning the sequences of WbbF, BcsA, and NodC identified four putative catalytic residues in WbbF, i.e., H62, D151, D244, and W284, which correspond to established catalytic residues in BcsA (Y149, D246, D343, and W383) and NodC (F58, D140, D241, and W281) (13, 28).

To validate the importance of these putative active-site residues in WbbF, each was replaced with alanine and the effect on O:54 antigen biosynthesis was examined. Each WbbF variant was expressed at levels comparable to that of wild-type WbbF (Fig. 6A). The activity of each WbbF variant was assessed by PAGE of whole-cell lysates and Western immunoblotting using O:54 antibodies (Fig. 6B), as well as immunofluorescence (Fig. 6C). Silver-stained gels label only the LPS-linked form of OPS (29), and O:54 LPS typically stains poorly due to (unknown) aspects of the OPS chemistry. In contrast, the Western blot reveals LPS-linked OPS as well as any unlinked (or unexported) biosynthetic intermediates. The D151A, D244A, and W284A variants were unable to support O:54 biosynthesis, confirming the importance of the DXD, TED, and Q(Q/R)XRW motifs in WbbF (Fig. 6B and C). Examination of permeabilized cells and the corresponding immunoblot revealed no evidence of internal O:54 antigen, ruling out unexpected uncoupled biosynthesis in the absence of export in any of the variants. The cells of these variants were typically smaller than those of the control. To rule out any second-site mutations in the other cloned O:54 genes (or elsewhere on the chromosome) that might confer an O:54-deficient phenotype, plasmids encoding each inactive variant were cotransformed with a plasmid carrying wild-type wbbF. As anticipated, the wild-type copy of wbbF restored both O:54 production and WT cell size. The cell size defects in these variants may result from sequestering initiated (but incomplete and unexported) und-PP-linked intermediates resulting from WbbE activity. Various cell size/shape defects have been observed in Escherichia coli cells accumulating pathway intermediates for other OPS and ECA (30). In contrast to the variants described above, the H62A variant still supported O:54 production. It is possible that the function of this residue is fulfilled by another nearby residue or that the interactions it makes with the donor are not essential for activity of WbbF. In summary, while some mutations eliminated O:54 biosynthesis, as predicted, none of the mutations examined here resulted in significant changes of OPS chain length distribution.

FIG 6.

Expression of WbbF variants and their effect on O:54 antigen production. (A) Western immunoblots of whole-cell lysates of E. coli Top10 harboring plasmid-encoded WbbF and variants. Immunoblots were probed with FLAG epitope-specific antibodies to confirm comparable expression of WbbF variants. WbbF-His₆ was utilized for complementation. (B) Silver-stained SDS-polyacrylamide gel (upper) and corresponding representative Western immunoblot (lower) of proteinase K-digested whole-cell lysates of S. enterica serovar Borreze and E. coli Top10 transformants harboring plasmid-encoded WbbF alone (WT WbbF) or the entire wb*_O:54 operon. The immunoblot was probed with anti-O:54 serum. The gel and immunoblot are representative of three biological replicates. (C) Fixed-cell immunofluorescence microscopy of nonpermeabilized and permeabilized S. enterica serovar Borreze and E. coli cells harboring the same plasmids as used in panel B. Cells were probed with anti-O:54 antigen antibody. Fluorescence observed in whole cells is due to the presence of O:54 antigen on the surface, while fluorescent permeabilized cells indicate total (including intracellular) polymer.

Implications for WbbF function.

The O:54 antigen is atypical, as it is the only known OPS in which synthesis is encoded by a naturally occurring ColE1-based mobilizable plasmid (7, 31). ColE1 plasmids can be transferred between bacteria in the presence of conjugative transfer functions provided by another plasmid (32, 33). These characteristics explain why the O:54 factor can be coexpressed with Salmonella isolates possessing additional chromosomally encoded OPS, which use a Wzx/Wzy-dependent pathway (31, 34, 35). The maintenance of two separate OPSs may be helped by the biosynthetic separation, but in one example, the isolate expresses only the chromosomally encoded OPS in the absence of the O:54 antigen. The expression of O:54 antigen can also be lost (presumably due to loss of the plasmid), while expression of chromosomally encoded OPS is retained. The coexpression of O:54 with other OPS types in Salmonella differs from other situations involving lateral transfer of OPS clusters in this species. There are examples of Salmonella isolates with two chromosomal OPS gene clusters, but in these cases, one cluster is no longer functional and only a single OPS is produced (3).

Although synthases are rare in the context of OPS synthesis (and currently confined to the O:54 antigen), a variety of other polymers, such as cellulose, are constructed using these systems. The synthesis and translocation of cellulose have been studied in depth and have provided insight into how these processes are coupled. BcsA extends the cellulose polymer one glucose molecule at a time to the nonreducing terminus, with the newly added glucose acting as the acceptor in the reaction that follows (12). While the same direction of growth is shared in the biosynthesis of chitin and the O:54 antigen (7, 12, 13), only the O:54 product is assembled as an und-PP-linked intermediate, and this must be retained through export to provide a viable donor to glycosylate lipid A-core in the periplasm. As a result, some differences between the translocation pathways in cellulose synthase and WbbF are predicted.

Translocation of cellulose through the translocation channel is a result of conformational changes in BcsA. These changes are produced by a ratcheting mechanism, which involves movements of the finger helix and the “gating loop” (16). The finger helix interacts with the terminal glucose of the cellulose polymer, while the gating loop houses the FXVTXK motif and spans the entrance of the active site, controlling substrate access (28). Upon substrate binding, the finger helix shifts upwards and the gating loop transitions into the active site, pushing the polymer into the channel. Once the active site is empty, the gating loop retracts, and the finger helix shifts downwards. These movements are repeated for each subsequent binding of substrate and elongation. The finger helix and its motifs are conserved in WbbF. It is believed that movement of the finger helix is reliant on the gating loop. BcsA possesses an Ile upstream of the TED motif, which is speculated to cause steric clashes with the gating loop, in the absence of their coupled movements (16). This Ile is conserved among bacterial cellulose synthases, but in eukaryotic cellulose synthases and WbbF, the corresponding position is occupied by a Leu residue, which may play a similar role. Consistent with this possibility, an Ile→Leu replacement in BcsA retained 50% of wild-type activity (16). It is unclear whether the same gating loop structure exists in WbbF. The initial residue of the gating loop (Phe) aids in positioning the uracil moiety of the donor sugar (28) and is conserved in WbbF, but the portion of the sequence which aligns with the gating loop in BcsA could not be modeled in WbbF.

The cellulose synthase translocation channel is packed tightly against the cytoplasmic GT2 domain, and the Phyre2 model of WbbF reveals a similar structural arrangement (Fig. 3). Despite having fewer TMHs than cellulose synthase, and a portion of the sequence not being modeled, the model of WbbF reveals a putative translocation channel. It is feasible that these synthases exist as homodimers to facilitate formation of an adequate translocation channel with eight TMHs. Dimerization has been shown to occur in mammalian HAS, as well as HAS from Streptococcus equisimilis (SeHAS) (17, 36). The HAS enzymes from different Streptococcus species are highly conserved (∼70% sequence identity) (37). The homolog from S. pyogenes (SpHAS) also possesses four TMHs and a GT2 module, with conserved DXD, (E/D)DX, and Q(Q/R)XRW motifs (19). Nascent hyaluronan produced by SeHAS reconstituted in liposomes was inaccessible to a hyaluronan-degrading enzyme, indicating that polymer synthesis and translocation are spatially coupled events, resembling cellulose synthesis (17). Other bacterial transporters which export und-PP-linked oligo- and polysaccharides have been described. These include the MOP transporter (MurJ) for lipid II (38) and ABC transporters for both O-antigen (Wzm-Wzt) (39, 40) and N-linked glycan export (PglK) (41). In the ABC transporters, lateral gates allow the glycan part of the substrate to access the transporter lumen, while the undecaprenol lipid remains in the lipid bilayer, and this may also apply to MurJ. Whether dimerization offers an avenue to a similar strategy in WbbF awaits a solved structure. Unfortunately, we have been unable to produce sufficient amounts of purified WbbF to facilitate crystallization trials, and there are currently no homologs that can be pursued as alternatives. To date, WbbF remains the sole synthase which requires a lipid-linked acceptor.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The bacterial strains used in this study were E. coli Top10 [F⁻ mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80 lacZΔM15 ΔlacX74 deoR nupG recA1 araD139 Δ(ara-leu)7697 galU galK rpsL(Str^r) endA1] (Invitrogen) and S. enterica serovar Borreze (54:f,g,s:-¹; contains plasmids of 96, 4.5 and 2.3 MDa) (31). Bacteria were grown at 37°C in lysogeny broth (LB). When appropriate, media were supplemented with ampicillin (100 μg/ml), chloramphenicol (34 μg/ml), anhydrotetracycline (2.5 ng/ml), or l-arabinose (0.01% [wt/vol]).

Plasmids and DNA methods.

The plasmids used are listed in Table 1. A synthetic wb*_O54 operon was designed to include the three biosynthetic genes (wbbE, wbbF, and mnaA) in their native arrangement. For detection of WbbF, the nucleotide sequence encoding a FLAG tag epitope was introduced at the 3′ terminus of wbbF, and EagI and SacI restriction sites were added to either end of wbbF to facilitate its excision and cloning. The operon was synthesized by DNA 2.0 (Atum). Following synthesis, the synthetic wb*_O54 operon was amplified by PCR and cloned into pBAD24 using Gibson assembly as described by the manufacturer (New England Biolabs). Custom oligonucleotide primers used in this study were purchased from Sigma and are listed in Table 2. Plasmid DNA was purified using the PureLink quick plasmid miniprep kit (Invitrogen). Either KOD hot start (EMD Millipore) or PfuUltra high-fidelity (Agilent) DNA polymerase was used for amplification of DNA fragments and site-directed mutagenesis. Site-directed mutagenesis was performed using the QuikChange protocol (Agilent). DNA fragments from PCRs and restriction enzyme digests were purified using the PureLink PCR purification kit (Invitrogen). Restriction endonucleases and T4 DNA ligase (New England Biolabs) were used according to the manufacturer’s instructions. All constructs were verified by DNA sequencing (Advanced Analysis Center, University of Guelph).

TABLE 1.

Plasmid summary

Plasmid	Description	Reference or source
pBAD24	Plasmid vector with l-arabinose-inducible promoter; Apr	49
pWQ572	pBAD24 derivative containing a chloramphenicol-resistance cassette and Ptet promoter; Cmr	25
pWQ799	Naturally occurring plasmid containing wb*O54 operon	8
pWQ203	pWQ572 derivative containing wb*O54 operon from pWQ799; Cmr	This study
pWQ1015	pWQ203 containing WbbFH62A	This study
pWQ1016	pWQ203 containing WbbFD151A	This study
pWQ1017	pWQ203 containing WbbFD244A	This study
pWQ1018	pWQ203 containing WbbFW284A	This study
pWQ1019	pBAD24 derivative encoding WbbF-His6; Apr	This study
pWQ1020	pBAD24 derivative containing synthetic wb*O54 operon encoding WbbF-FLAG	This study
pWQ1021	pSW1020 containing WbbFH62A-FLAG	This study
pWQ1022	pSW1020 containing WbbFD151A-FLAG	This study
pWQ1023	pSW1020 containing WbbFD244A-FLAG	This study
pWQ1024	pSW1020 containing WbbFW284A-FLAG	This study
pWQ1025	pWQ203 containing WbbFC→A	This study
pWQ1026	pSW1020 containing WbbFC→A -FLAG from pWQ1025	This study
pWQ1027	pSW1026 containing WbbFN2C-FLAG	This study
pWQ1028	pSW1026 containing WbbFA102C-FLAG	This study
pWQ1029	pSW1026 containing WbbFS285C-FLAG	This study
pWQ1030	pSW1026 containing WbbFV312C-FLAG	This study
pWQ1031	pSW1026 containing WbbFS368C-FLAG	This study
pWQ1032	pSW1026 containing WbbFL413C-FLAG	This study
pWQ1033	pSW1026 containing WbbFV447C-FLAG	This study

TABLE 2.

Sequences of oligonucleotide primers

Primer	Sequence (5′→3′)a	Features
O54fwd	gatcggatccGCTGTATAGGAAATTAGGAAATTAG	Forward primer to amplify wb*O54 operon; BamHI restriction site (underlined)
O54rev	gatcctgcagTCACTTAATATCCTCCAAGGATAC	Reverse primer to amplify wb*O54 operon; PstI restriction site (underlined)
SW13	CTAATTCTTGTGGCTGCGGCTAATGAAGAGGCTGTGATTGGCTCAACAC	Forward primer used to generate WbbFH62A variant
SW14	CAATCACAGCCTCTTCATTAGCCGCAGCCACAAGAATTAGAAAGCGG	Reverse primer used to generate WbbFH62A variant
SW9	TATGATTTGGTCATGGTGTTGGCTGCCGACAATTTTGTTGATGCGAATATCCTTACT	Forward primer used to generate WbbFD151A variant
SW10	ATCAACAAAATTGTCGGCAGCCAACACCATGACCAAATCATAATTTTCTTTAACAGT	Reverse primer used to generate WbbFD151A variant
SW7	GTTTTAAATCTCTGACCGAGGCCATTGAACTGGAAATTGAAATTG	Forward primer used to generate WbbFD244A variant
SW8	CAATTTCAATTTCCAGTTCAATGGCCTCGGTCAGAGATTTAAAAC	Reverse primer used to generate WbbFD244A variant
SW11	CTCAAACAACGCTATCGCGCGTCAAAGGGACACTGGTATGT	Forward primer used to generate WbbFW284A variant
SW12	ACCAGTGTCCCTTTGACGCGCGATAGCGTTGTTTGAGGCTTATTC	Reverse primer used to generate WbbFW284A variant
WbbFf	gatcgaattcaccATGAATGATTATATAATTGACATAG	Forward primer for the amplification of wbbF; EcoRI restriction site (underlined)
WbbFr	gatcctgcagttaatgatgatgatgatgatgTTTGTGCTCTTCCTTTATTTTATTATGC	Reverse primer for the amplification of wbbF and introduction of a his6-tag; PstI restriction site (underlined)
SW35	cccgtttttttgggctagcaggaggaattcaccATGGGGCATCAGTTTACGGTTTGC	Forward primer to amplify synthetic wb*O54 operon adding homologous pBAD24 region for Gibson assembly; EcoRI restriction site (underlined)
SW36	ctcatccgccaaaacacccaagcttctgcagTCACTTAATATCCTCCA	Reverse primer to amplify synthetic wb*O54 operon adding homologous pBAD24 region for Gibson assembly; HindIII restriction site (underlined)
SW17	gatccggccgATGAATGATTATATAATTGACATAGTGG	Forward primer for the amplification of wbbF; EagI restriction site (underlined)
SW18	gatcgagctcTTTGTGCTCTTCCTTTATTTTATTATGC	Reverse primer for the amplification of wbbF; SacI restriction site (underlined)
KB1	GCTAAAAATAAAAAAGACTATCCTGACGCTCCTCCTGAAGCCCG	Forward primer used to generate WbbFC49A variant
KB2	CAAGAATTAGAAAGCGGGCTTCAGGAGGAGCGTCAGGATAGTC	Reverse primer used to generate WbbFC49A variant
KB3	CCACTGATCGGACAGGACTTATCGCTGATAGTCATGAAGTAAAG	Forward primer used to generate WbbFC102A variant
KB4	CCACATGCTTTACTTCATGACTATCAGCGATAAGTCCTGTCCG	Reverse primer used to generate WbbFC102A variant
KB5	GCCGGAAGCTATACAGGCGTATCTGGATGCTAAAAACTCAACAT	Forward primer used to generate WbbFC183A variant
KB6	GCCAAAAGAGAGAAGAGATGTTGAGTTTTTAGCATCCAGATACG	Reverse primer used to generate WbbFC183A variant
KB7	CTCAACATCTCTTCTCTCTTTTGGCTACGCTACATCATACTGGAT	Forward primer used to generate WbbFC195A variant
KB8	GGAAAAATCGATTCATCATCCAGTATGATGTAGCGTAGCCAAAAG	Reverse primer used to generate WbbFC195A variant
KB9	CTGATAAATACTGGAGGATTTGCTTTTAAATCTCTGACCGAGGA	Forward primer used to generate WbbFC237A variant
K10	CCAGTTCAATGTCCTCGGTCAGAGATTTAAAAGCAAATCCTCCA	Reverse primer used to generate WbbFC237A variant
c328af	GGGCCGTGCTTTGCAGGTTGCTATTATTTTCATCAATATCTTTC	Forward primer used to generate WbbFC328A variant
c328ar	GAAAGATATTGATGAAAATAATAGCAACCTGCAAAGCACGGCCC	Reverse primer used to generate WbbFC328A variant
c394af	CGTCACATTAATATCCATCGCTTATGGTATGCTGATTTTACC	Forward primer used to generate WbbFC394A variant
c394ar	GGTAAAATCAGCATACCATAAGCGATGGATATTAATGTGACG	Reverse primer used to generate WbbFC394A variant
SW115	GTGTTAACGGCCGATGTGTGATTATATAATTGACATAGTGGAATATGTTTTATATG	Forward primer used to generate cfWbbFN2C variant
SW116	CACTATGTCAATTATATAATCACACATCGGCCGTTAACACCTTAAAATA	Reverse primer used to generate cfWbbFN2C variant
SW41	GATCGGACAGGACTTATCTGTGATAGTCATGAAGTAAAGCATGTGGATAC	Forward primer used to generate cfWbbFA102C variant
SW42	GCTTTACTTCATGACTATCACAGATAAGTCCTGTCCGATCAGTGGA	Reverse primer used to generate cfWbbFA102C variant
SW25	ACAACGCTATCGCTGGTGTAAGGGACACTGGTATGTGGCTTTTAC	Forward primer used to generate cfWbbFS285C variant
SW26	CACATACCAGTGTCCCTTACACCAGCGATAGCGTTGTTTGAGGCTTA	Reverse primer used to generate cfWbbFS285C variant
SW27	TTGAGCGTAAGTGGAAATATTGTGATCAATTGTTATATCTGTTCTCTATGGGC	Forward primer used to generate cfWbbFV312C variant
SW28	GAGAACAGATATAACAATTGATCACAATATTTCCACTTACGCTCAACAAAGGTCAAC	Reverse primer used to generate cfWbbFV312C variant
SW110	GAAAATTACCATCCAGAGTGTGGAAATATTTCTACGGCGATAAAAGATC	Forward primer used to generate cfWbbFI348C variant
SW111	CGCCGTAGAAATATTTCCACACTCTGGATGGTAATTTTCTTTAAGAAGAC	Reverse primer used to generate cfWbbFI348C variant
SW29	CAATATGAGTTTTGCCGACTGTGTGAGTGCGCAGTTTAGCTCAATAAATTG	Forward primer used to generate cfWbbFS368C variant
SW30	CTAAACTGCGCACTCACACAGTCGGCAAAACTCATATTGGTCACAGTA	Reverse primer used to generate cfWbbFS368C variant
SW31	GGATGGATAAAGGTATTTTCTGTAATCCATTCAGGGTATTTTTTTCCGGTC	Forward primer used to generate cfWbbFL413C variant
SW32	GAAAAAAATACCCTGAATGGATTACAGAAAATACCTTTATCCATCCATGCACC	Reverse primer used to generate cfWbbFL413C variant
SW33	GGAAAAAACAGCATAAATGGTGTGTTACGCCGCATAATAAAATAAAGGAGG	Forward primer used to generate cfWbbFV447C variant
SW34	ATTTTATTATGCGGCGTAACACACCATTTATGCTGTTTTTTCCAACGAAAAAG	Reverse primer used to generate cfWbbFV447C variant

^aBoldface indicates a point mutation.

Detection of WbbF-FLAG and WbbF-His₆ in whole-cell lysates.

Overnight cultures were diluted 1:50 in 5 ml of LB medium and grown to an optical density at 600 nm (OD₆₀₀) of 0.5, at which point the culture was transferred to 18°C and allowed to grow until an OD₆₀₀ of 0.6 was reached. Protein expression was then induced using l-arabinose (0.01% [wt/vol]), and the culture was grown for an additional 16 h. The OD₆₀₀ of the culture was measured, and cells from a volume corresponding to 2 OD₆₀₀ units were collected by centrifugation (13,000 × g for 1 min). Ten microliters of cOmplete Mini EDTA-free protease inhibitor (Roche Applied Science) solution (1 tablet per 1 ml of water) and 25 μl of B-PER II (ThermoFisher) were added to the pellet. The pellet was then resuspended in 63 μl of water, and 1 μl of both rLysozyme (EMD Millipore) (diluted 1:10 in water) and Benzonase (Novagen) (diluted 1:5 in 50 mM Tris-HCl, 20 mM NaCl, 2 mM MgCl₂, pH 8) were added. The mixture was rocked at room temperature for 30 min before 100 μl of 2× Laemmli sample buffer (42) was added. Samples were incubated in Laemmli sample buffer at 37°C for 30 min prior to separation by SDS-PAGE using 10% resolving gels in Tris-glycine buffer. Proteins were visualized using SimplyBlue SafeStain (Life Technologies). Western immunoblotting was performed with FLAG and His₆ epitope-tagged proteins, which were transferred to nitrocellulose membranes and probed with anti-FLAG (Qiagen; diluted 1:1,000) or anti-His₅ (Qiagen; diluted 1:2,000) antibodies. The secondary antibody was horseradish peroxidase-conjugated goat anti-mouse antibody (Jackson ImmunoResearch; diluted 1:3,000), and chemiluminescence detection was accomplished using Luminata Crescendo substrate (EMD Millipore).

Site-directed fluorescence labeling.

Labeling of WbbF variants with Oregon Green 488 maleimide carboxylic acid (OGM; Fisher Scientific) was based on a protocol described elsewhere (26) and modified by Larue et al. (25). Overnight cultures were used to inoculate 200 ml of LB medium (1:100) supplemented with ampicillin. WbbF variants were expressed as described above. Cells were collected by centrifugation at 5,000 × g for 10 min at 4°C and resuspended in 10 ml of 25 mM sodium phosphate buffer containing 250 mM NaCl, pH 7.5. Two 3-ml aliquots of the cell suspension were removed. In aliquot 1, periplasmic cysteine residues were modified with 40 μM OGM during a 15-min incubation at room temperature. Following labeling, the reaction was quenched with 1 mM β-mercaptoethanol. Aliquot 2 remained untreated. Aliquots 1 and 2 were centrifuged at 5,000 × g for 10 min at 4°C and washed twice. The pellets were then resuspended in 2 ml of buffer containing 5 mM EDTA, 100 μg/ml lysozyme, and 20% (wt/vol) sucrose and incubated for 15 min at room temperature. Cells were disrupted with 18 ml of cold water containing cOmplete Mini EDTA-free protease inhibitor (Roche Applied Science). A 10 μM concentration of OGM was added to aliquot 2, and both aliquots were sonicated for 30 s (10 s on, 10 s off) prior to a 15-min incubation at room temperature. In aliquot 2, all available cysteines are labeled with OGM. The reaction was again quenched with 1 mM β-mercaptoethanol. Unbroken cells and debris were removed from both aliquots by centrifugation at 15,000 × g for 15 min at 4°C. Cell-free lysates were centrifuged at 100,000 × g for 1 h at 4°C to collect membranes, which were solubilized overnight at 4°C in 1 ml of buffer containing 1% N-dodecyl-β-d-maltopyranoside (DDM) (Sigma). The remaining insoluble material was removed by centrifugation at 100,000 × g for 1 h at 4°C, and WbbF-FLAG was purified from 500 μl solubilized membranes using 25 μl of anti-FLAG M2 magnetic beads (Sigma) in accordance with the manufacturer’s batch protocol. Proteins were eluted using 20 μl of 1× Laemmli sample buffer. Samples were warmed for 30 min at 37°C before 10-μl samples were analyzed by SDS-PAGE using 10% resolving gels. OGM was visualized by exposing the gels to UV light using a Bio-Rad Gel Doc. Protein visualization and Western immunoblotting were carried out as described above.

Immunofluorescence microscopy.

To assess O:54 production and export, immunofluorescence microscopy was performed with fixed cells, in accordance with the protocol described by Clarke et al. (43), to assess O:54 production and export. In brief, overnight cultures were diluted 1:50 in 5 ml of LB medium supplemented with the appropriate antibiotics, anhydrotetracycline and/or l-arabinose. Cultures were grown at 37°C until an OD₆₀₀ of 0.5 was reached. One OD₆₀₀ unit of cells was collected by centrifugation, resuspended in 5% (vol/vol) formaldehyde, and incubated at 4°C for 16 h. Fixed cells were collected, washed twice with 1 ml of phosphate-buffered saline (PBS), and finally resuspended in 100 μl of PBS. Ten-microliter samples were added in duplicate to the wells of a poly-l-lysine-coated glass slide and incubated for 10 min at room temperature. Cells in one well for each sample were permeabilized using 10 μl of 0.5 mg/ml lysozyme (in 25 mM Tris-HCl, 10 mM EDTA, pH 8) and 0.1% Triton X-100 (in PBS) solutions. Following the addition of each solution, the slide was incubated for 15 min. The wells were then blocked for 15 min with 1% (wt/vol; in PBS) bovine serum albumin (BSA). After washing, the slides were incubated with anti-O:54 antiserum (Statens Serum Institut, Denmark; diluted 1:100 in 1% BSA) at room temperature for 30 min and washed. The slides were then treated with rhodamine red-conjugated goat anti-rabbit antibody (Jackson Immunoresearch; diluted 1:50 in 1% BSA) and washed again. The slides were mounted in Vectashield (Vector Laboratories) and viewed using a Zeiss Axiovert 200 microscope with a 100× lens objective. The images were processed using Volocity software (PerkinElmer).

LPS analysis.

Cells and transformants were grown as described above. Cells from one OD₆₀₀ unit were collected by centrifugation (13,000 × g for 1 min), and whole-cell lysates were digested by using proteinase K (44). Samples (10 μl) were separated on 12% resolving gels in Tris-glycine buffer (42). Silver staining was then performed by following the protocol described by Tsai and Frasch to visualize LPS (45). O:54 antigen was also detected by immunoblotting separated samples, which were transferred to a nitrocellulose membrane (Protran; GE Healthcare) at 200 mA for 60 min in 25 mM Tris, 150 mM glycine, 20% (vol/vol) methanol. The membrane was probed with anti-O:54 serum (Statens Serum Institut, Denmark; diluted 1:1,000), followed by alkaline phosphatase-conjugated goat anti-rabbit secondary antibody (Cedarlane Laboratories; diluted 1:3,000). OPS was detected using nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Roche Applied Science).