Trapped translocation intermediates establish the route for export of capsular polysaccharides across Escherichia coli outer membranes

Significance

Exported long-chain polysaccharides protect the bacterial cell in harsh environments and play multifactorial roles in pathogenesis. These long-chain polysaccharides can take the form of secreted polymers (exopolysaccharides), or cell-associated capsular polysaccharides (CPSs) that form an organized surface layer called the capsule. Processes involved in the synthesis and export of CPSs are potential targets for new therapeutic strategies. Independent of their structures, these polymers rely on a family of outer membrane polysaccharide export (OPX) proteins for assembly on the cell surface. Combining structural information with site-specific in vivo photo–cross-linking provides direct evidence that the lumen of octameric complexes of Wza, a prototype OPX protein, is the export conduit for CPS.

Abstract

The outer membrane (OM) of Gram-negative bacteria is designed to exclude potentially harmful molecules. This property presents a challenge for bacteria that must secrete proteins and large glycoconjugates to grow, divide, and persist. Proteins involved in trafficking such molecules have been identified, but their precise roles are often unresolved due to the difficulty in capturing “snapshots” during the export pathway. Wza is the prototype for the large family of OM polysaccharide export proteins. In Escherichia coli, Wza is essential for the assembly of a capsule, a protective surface coat composed of long-chain polysaccharides. Wza creates an octameric α-helical channel spanning the OM, but the bulk of the protein exists as a large periplasmic structure enclosing an extensive lumen. Residues within the lumen of Wza were targeted for site-specific incorporation of the UV photo–cross-linkable unnatural amino acid p-benzoyl-l-phenylalanine. Using this in vivo photo–cross-linking strategy, we were able to trap polysaccharide translocation intermediates within the lumen of Wza, providing the first unequivocal evidence to our knowledge that nascent capsular polysaccharide chains exit the cell through the Wza portal.

The presence of an outer membrane (OM) is the defining characteristic of Gram-negative bacteria. The OM is an effective barrier against transport of polar molecules, including some antimicrobial compounds (1). However, bacteria must import nutrients and export large molecules, such as proteins, glycolipids, and polysaccharides, across the OM. This trafficking requires dedicated proteins and occurs without compromising the barrier properties of the OM. These processes are distant from the energy sources available in the cytosol and cytoplasmic membrane. Although machinery required for some of these processes has been identified, molecular details of the mechanisms are often poorly understood, in part because of the difficulty of capturing “snapshots” along the transit pathways.

Many bacteria produce and export long-chain (∼10⁵–10⁶ Da) polysaccharides in the form of secreted polymers (exopolysaccharides), or cell-associated capsular polysaccharides (CPSs) that envelop the cell in a hydrophilic layer known as the capsule. These polymers often offer protection from unpredictable and hostile environments. As a result, a polysaccharide coat is recognized as a key contributor to the pathogenic success of many bacterial species in humans, livestock, and plants (2). However, polysaccharides also participate in symbiotic relationships between bacteria and plants (3), and are important contributors to the formation of microbial communities known as biofilms (4). In addition to their roles in the producing bacteria, certain bacterial extracellular polysaccharides have been exploited for their unusual gelling properties in the food and pharmaceutical sectors (5).

Despite the diversity of known polysaccharide structures, their assembly follows a limited range of conserved strategies. Two of these strategies are represented in the capsules of Escherichia coli. These CPS assembly pathways are fundamentally different in terms of the mechanism and membrane topology of the polymerization process, but they share a common strategy for polymer export across the OM (6). Both pathways involve a representative of a family of outer membrane polysaccharide export (OPX) proteins, whose members are widespread in bacteria with different lifestyles that depend on capsular or secreted exopolysaccharides (7). The Wza lipoprotein from the E. coli group 1 capsule system is the prototypical OPX protein (8, 9). Wza forms a ∼340-kDa SDS-resistant octamer that spans the OM and traverses the periplasm, where it docks with an oligomeric inner membrane protein, Wzc (9–11). Together, these proteins form a contiguous molecular scaffold that spans the cell envelope (12). It has been hypothesized that the Wza–Wzc complex generates the export pathway for CPS through the OM (6) (Fig. 1), coupling chain growth to OM transit.

Fig. 1.

Model of K30 CPS biosynthesis and export. The cartoon shows the proposed activities of a hypothetical biosynthetic complex carrying out coordinated synthesis and export of serotype K30 group 1 capsule in E. coli (model reviewed in ref. 6). Individual repeat units of the polymer are assembled by a series of enzymes on a lipid [undecaprenol diphosphate (und-PP)] acceptor on the inner leaflet of the inner membrane. The und-PP–linked repeat units are “flipped” across the inner membrane by an integral membrane protein (Wzx) and polymerized into long-chain polysaccharides by the periplasmic domain of protein Wzy in the periplasm. Export of the polymer to the cell surface requires formation of a transenvelope complex between the tetrameric inner membrane Wzc and the octameric OM Wza complex. The phosphorylation state of Wzc is controlled by the protein tyosine phosphatase (Wzb) and is necessary for capsule synthesis. The R1–R4 annotations on Wza refer to domains identified in the mutagenesis strategy.

The Wza octamer spans the OM with a novel α-helical barrel possessing a minimum diameter of ∼14 Å (13). Although open to the external medium, the purified Wza complex is sealed in the periplasm with a tyrosine ring at the base of the structure (13–15). The Wza monomer is composed of four domains arranged parallel to the eightfold symmetry axis of the complex, creating four distinct rings (rings R1–R4) in the octamer (Fig. 1 and Fig. 2A). Although the periplasmic R1 domain is closed in the purified protein (13, 15), cryo-EM suggests conformational changes occur in the R1 and R2 region of Wza upon its interaction with Wzc (12). Mutants lacking wza or wzc produce no CPS and are unable to synthesize detectable intracellular polymer (16, 17), suggesting a feedback process in which synthesis and export are coupled. This has made it impossible to capitalize on deletion mutants to study export events in isolation. As a result, there remains no direct experimental evidence that polysaccharides actually access the central lumen of the OPX complexes, a critical requirement if these proteins do provide the means for transit through the OM.

Fig. 2.

Incorporation of pBpa within the lumen of octameric Wza does not affect assembly and function of select residues. (A) Structures of the Wza multimer (Protein Data Bank ID code 2J58; surface representation) and monomer (cartoon representation) highlight the locations probed by the incorporation of the photo–cross-linkable unnatural amino acid pBpa. A cross-section of the octameric structure is shown to visualize residues in the lumen. The mutated residues are colored as follows: yellow, mutants could not restore CPS production in E. coli CWG281 Δwza; blue, mutants did restore CPS production but did not form detectable intermolecular cross-links upon UV treatment; red, mutants rescued CPS production in the Δwza mutant strain and generated new molecular species upon UV treatment. The Wza monomer is composed of four domains that form distinct rings in the octameric complex (labeled R1–R4). (B) Clipped view of Asn181 (Upper) and the corresponding structural model (modeling details are provided in Materials and Methods) for the pBpa mutation at residue 181 (Lower) looking “down” into the lumen of the Wza multimer. (C) Synthesis of full-length Wza in the mutants is dependent on inclusion of pBpa in the growth medium. (D) CPS production in CWG281 Δwza harboring the various mutant proteins was examined in the presence (+) or absence (−) of pBpa. CPS was detected by immunoblotting of cell lysates with K30-specific antiserum. The surface location of the CPS was confirmed by resistance (R) or sensitivity (S) of bacteria to K30 CPS-specific bacteriophage, which requires the CPS as a receptor (36). (E) Functional Wza mutants were expressed in the presence of pBpa and either UV-treated (+) or incubated on ice (−). Cells were solubilized in sample buffer, heated to dissociate the Wza octamer, and probed in immunoblots with Wza-specific antibodies. The control lysate (C) is WT cells solubilized in sample buffer without heating and shows the expected mixture of monomers and multimers. The asterisk indicates the weak intensity of Wza adducts for A180. Mo, monomer; Mu, Wza octamer.

We used an in vivo cross-linking strategy in an attempt to trap the nascent CPS within the lumen of the Wza octamer. The method exploits site-specific incorporation of the UV photo–cross-linkable unnatural amino acid p-benzoyl-l-phenylalanine (pBpa) (18), using an orthogonal aminoacyl-tRNA synthetase/tRNA pair evolved from Methanococcus jannaschii to reassign the amber codon (TAG) (19–22). In the context of macromolecular export, this strategy has been used successfully to identify the molecular surface involved in the interaction of SecA and SecY of the SecYEG translocon (23) and to identify the binding sites for LPSs within the β-jellyroll fold of LptA and LptC proteins from the LPS transport pathway (24). Successful incorporation of pBpa within the lumen of Wza revealed novel Wza-specific adducts that were dependent on synthesis of K30 capsule. Wza-specific adducts purified from the OM were cross-linked to K30 polysaccharide, providing the first direct evidence, to our knowledge, that CPS is extruded through the Wza channel to reach the cell surface.

Results

Site-Specific Incorporation of pBpa Within the Lumen of Wza.

The 2.26-Å crystal structure of the Wza octamer (13) was used to select 12 luminal surface-exposed amino acids located at different positions within the Wza octamer for replacement with pBpa (Fig. 2 A and B). These sites reflect different luminal diameters in the Wza crystal structure (Table S1). The resulting mutant Wza proteins were coexpressed with the mutant tRNA synthetase/tRNA pair in the absence or presence of pBpa, and samples were probed with Wza-specific antibodies (Fig. S1). As expected, no full-length Wza monomers were detected for any of the mutants in the absence of pBpa (Fig. 2C), although some constructs yielded truncated peptides that reacted with the anti-Wza antibodies (Fig. S1), which we attribute to degradation of certain truncated forms. All of these mutants would terminate before the α-helical transmembrane R4 domain. Nine of the 12 mutant proteins restored capsule production, based on sensitivity to a K30 capsule-specific bacteriophage and to K30 CPS-specific Western immunoblotting (Fig. 2D). Modeling of the three nonfunctional mutants did not reveal any obvious block in the lumen, although substitutions at these positions did have significant structural constraints. However, similar constraints were also evident for the mutation at Q356, which participated effectively in CPS assembly. The precise reason for the loss of function was not pursued further, but incorporation of the bulky pBpa group presumably affects one or more of Wza folding, assembly into an octamer, or activity.

In Vivo Cross-Linking Identified Wza-Specific Adducts.

The nine functional pBpa-Wza variants were subjected to UV photo–cross-linking using established protocols. Cell samples were then solubilized in SDS/PAGE buffer at 100 °C. Native Wza behaves as an SDS-stable octamer at room temperature but dissociates into a monomer at 100 °C (16). Three constructs (with pBpa replacements at N181, A186, and D207) gave robust signals reflecting novel adducts in immunoblots probed with anti-Wza antibodies, whereas the replacement at A180 produced only a small amount of adduct (Fig. 2E). The cross-linked products could reflect either protein/protein or protein/polysaccharide adducts. As a first step in differentiating between these possibilities, we examined whether adduct formation was dependent upon the production of K30 CPS. A ΔgalE mutant (deficient in UDP-glucose 4-epimerase) allows conditional production of galactose-containing glycans, including the K30 CPS; these glycans are only made if galactose is added to the culture medium (25, 26) (Fig. 3A). Exploitation of this system is dependent on Wza targeting and assembly in the OM of CWG1170 (ΔgalE) being indistinguishable under permissive or nonpermissive conditions (Fig. 3B). The profiles of novel UV-dependent adducts formed by the A180, N181, A186, and D207 Wza mutants in the ΔgalE background in the presence of galactose were consistent with those seen in the parent GalE⁺ strain. No adducts were produced in the absence of galactose (Fig. 4), confirming a requirement for active CPS synthesis.

Fig. 3.

Assembly of Wza multimers in the OM is not dependent on CPS synthesis. A system for conditional production of CPS was established by introducing a ΔgalE mutation, where synthesis is dependent on addition of galactose to the growth medium. (A) Western immunoblot of K30 CPS in E. coli E69 and derivatives in the absence (−) or presence (+) of galactose (Gal). Cell lysates were probed in an immunoblot with K30-specific antibodies. (B) Wza production and assembly in the OM were not dependent on the production of K30 CPS. Cell lysates and sarkosyl-insoluble OM fractions [in the presence (+) or absence (−) of galactose] were solubilized in SDS/PAGE sample buffer without heating, separated by SDS/PAGE, and immunoblotted with Wza-specific antibodies.

Fig. 4.

Novel Wza adducts formed by in vivo photo–cross-linking are dependent on CPS production. Lysates of CWG1171 (Δwza ΔgalE) expressing pBpa mutants were grown in the presence or absence of galactose and probed in immunoblots with Wza-specific antibodies. The control lysate is from WT cells solubilized in PAGE sample buffer without heating and shows the expected mixture of monomers and multimers. All other samples were heated to 100 °C to dissociate the multimers.

Trapping K30-Specific CPS Within the Lumen of Wza.

The adducts shown in Fig. 4 could represent protein–polysaccharide complexes or, alternatively, could result from protein/protein cross-linking that is dependent on a particular conformation(s) occurring only during active CPS production. Unfortunately, the presence of K30 CPS in these adducts could not be directly confirmed in whole-cell lysates by immunoblotting with anti-CPS antibodies due to the masking presence of CPS migrating to the same region of the PAGE gel. To generate definitive evidence for the existence of CPS/protein cross-links, C-terminal His₆-tagged variants of the N181 and D207 mutations were created and the UV cross-linking reaction was scaled up to facilitate purification of the products. The N181 and D207 mutants were selected for these experiments because they exhibited different adduct patterns in initial analyses (Figs. 2 and 4). The mutant proteins were purified in parallel with the WT Wza octamer from cells with and without UV treatment, following our published protocol (27); sarkosyl (N-lauroylsarcosine)-insoluble OM fractions were solubilized in 3-(N,N-dimethylmyristyl-ammonio)propanesulfonate (SB3-14) to release Wza and other OM proteins. The Wza-His₆ proteins were then purified using cobalt-affinity resin (Fig. S2). As expected, purified preparations of the N181 and D207 mutants contained SDS-stable multimers at room temperature. After heating at 100 °C, most of the larger protein complexes dissociated into monomers, but the anticipated adducts remained in protein isolated (only) from UV-irradiated cells (Fig. S2). The properties of the Wza-His₆ variants were qualitatively comparable to those of nontagged Wza mutants (Figs. 2 and 4) with respect to in vivo photo–cross-linking (Fig. S2). The patterns of adducts seen in the large-scale experiments were slightly different in terms of the relative intensities of the various bands, and additional low-intensity bands were recognized that were only seen in the cell lysates with higher loading, or overexposure. The two mutants share some bands with similar migrations on SDS/PAGE but differ in the relative intensities of those bands (Fig. 5A). We attribute these differences to subtle changes in experimental conditions resulting from scaling up the protocol, but profiles from replicates of each condition are consistent. CPS-specific antibodies reacted with the adducts in immunoblots, but, crucially, no reaction was detected with protein samples prepared in an identical manner from WT Wza-His₆, or from cells expressing the mutant proteins that were not UV-irradiated (Fig. 5B). The profiles of adducts revealed with anti-K30 antibodies were similar in the two mutants. However, the apparent differences in reactivity of certain adduct bands to anti-Wza and anti-K30 antibodies, as well as the presence of one Wza band lacking reactivity with K30 CPS antibodies, led us to question whether the adducts were a mixture of Wza/CPS and CPS-dependent Wza/protein cross-linked products. Consequently, the purified proteins were subjected to analysis by MS. Peptides from purified proteins were separated and analyzed by liquid chromatography (LC)-MS. Peptides corresponding to Wza were detected as expected (Fig. S3), but no additional proteins were identified in the UV-treated samples, indicating that the adducts are not formed between Wza and any other protein. The MS experiments did not identify peptides with covalently bound CPS, but it is unclear whether these large and heterogeneous products would ionize under the MS conditions.

Fig. 5.

In vivo photo–cross-linking traps K30 polysaccharide within the lumen of multimeric Wza. WT Wza and Wza mutants with pBpa incorporated at residue 181 (WzapBpa181) or residue 207 (WzapBpa207) were purified from UV-treated (+) or untreated (−) cells. Proteins were purified by affinity chromatography (Fig. S2) and solubilized in sample buffer. Western immunoblots were probed with either Wza-specific antibodies (A) or K30-specific antibodies (B). Solid circles indicate the corresponding bands for WzapBpa181, and asterisks indicate the corresponding bands for WzapBpa207.

Discussion

The results presented here demonstrate trapping of CPS translocation intermediates within the lumen of the multimeric Wza structure by site-specific cross-linking and provide the first definitive evidence to our knowledge that CPS is translocated through the channel of Wza to the cell surface. Wza belongs to a large family of OPX proteins (7), but it is the only representative for which detailed structural information exists. Although different members vary in overall length and primary structure, they share a conserved polysaccharide export sequence (Pfam 02563) motif and have similarities in predicted secondary structure, including a predicted C-terminal α-helical secondary structure consistent with the α-helical barrel of Wza. We therefore predict translocation of high molecular-weight polysaccharides through the lumen of other OPX family members. The in vivo photo–cross-linking strategy could be applied in a similar manner to other OM portals with solved structures to understand the export process better and to aid in the identification of translocation intermediates.

Novel Wza adducts were identified in four of the nine functional Wza mutants. Each mutant generated a series of CPS-dependent adducts of varying sizes. CPS-containing adducts were typically poorly recognized by anti-Wza antibodies, but the effect of the attached glycan on availability of the Wza epitopes is unknown. We are unable to determine if the protein–CPS complexes of varying sizes differ in the number of cross-linked monomers to a single glycan chain, or just in the naturally heterogeneous chain length of the trapped CPS. Such heterogeneity is clearly evident in the largest CPS-containing adducts formed by the N181 and D207 mutants (Fig. 5). However, some products of homogeneous size were also evident, suggesting an additional accumulation of synthetic intermediates of a particular size captured by the cross-linking process. In our working model for CPS synthesis and export, we envisage that the addition of new repeat units to the reducing terminus of the nascent chain (i.e., that closest to the inner membrane) may participate in driving export through the channel in a coupled process. There is currently no available strategy to determine the chain length of trapped CPS to discriminate between CPS intermediates and full-length polymer. However, the size must be interpreted with caution because the different properties of CPS and CPS–protein complexes may lead to significant changes in migration in SDS/PAGE.

The presence of Wza-containing bands lacking obvious reactivity with anti-K30 CPS was surprising. The possibility of adducts formed between Wza and another cell envelope protein was ruled out by MS; only Wza was detected. We considered the possibility that these represented Wza/Wza adducts promoted, in some way, by CPS synthesis. In this scenario, the different bands could involve varying numbers of monomers. However, in vivo cross-linking studies looking at the interaction of SecY and SecA and LolA and LolC identified protein/protein interactions with different electrophoretic mobilities depending on the cross-linking amino acid position, although there is no explanation for this behavior (23, 28). Although no supporting evidence was found for Wza/Wza adducts in an unbiased analysis of the MS-derived peptides by StavroX (29), we are unable to rule out the possibility that the size or properties of peptides derived from these adducts led to ineffective ionization.

Cross-linking studies with benzophenone photophores have defined a reactive distance of 3.1 Å, but reactivity is also strictly influenced by geometric constraints, requiring optimal positioning between the cross-linker carbonyl group and the target atoms (30). The five mutants that did not form cross-links cannot be simply explained by the internal diameter of the lumen at that point (Table S1), and they may instead lack the appropriate positioning. Interestingly, the reactive residues were all localized in or nearby ring R2 (Fig. 2A). Previous computational calculations (ΔG of dissociation) for the rings predicted that R2 may be destabilizing to the octameric complex (14). Furthermore, EM reconstructions of the Wza–Wzc complex under different pH conditions revealed a slightly wider R2 than that observed in the crystal structure of Wza alone, and this was attributed to the increased flexibility in this ring (14). The higher degree of flexibility of R2 may be related to our success with pBpa photo–cross-linking in this region. Previously, we predicted that the properties of R2 are important for entry of CPS into the lumen of the Wza octamer (Fig. 2A) from a polymerization process active at the periplasmic face of the cytoplasmic membrane. However, the formation of novel adducts in the A186 mutant, located between R1 and R2, suggests the entry point is in much closer proximity to Wzc. The predicted flexibility in this region could also facilitate conformational changes during CPS export, providing an opportunity for hypothetical Wza/Wza adducts.

Taken together, these results present the first direct evidence to our knowledge that CPS enters the lumen of the Wza octamer for translocation to the cell surface. Defects in capsule assembly and export render normally encapsulated pathogens susceptible to host defenses, so these processes represent potential targets for development of antibacterial compounds. Recently, an unusual cyclic octasaccharide was used to block an open ion-conducting Wza channel in lipid bilayer experiments (15). This open mutant was constructed by mutagenesis of amino acids on the periplasmic surface of the Wza channel. Addition of this octasaccharide to E. coli cultures caused a reduction in overall CPS production, resulting in sensitization of treated cells to serum-mediated killing (15), but the underlying physiological events have not been determined in vivo. The current studies provide an experimental platform to dissect the export process further and shed light on the mode of action of such inhibitors.

Materials and Methods

Bacterial Strains and Growth Conditions.

Information concerning the properties and construction of bacterial strains and plasmids is provided in Tables S2 and S3. Cultures were grown with aeration at 30 °C or 37 °C in LB broth (Invitrogen) containing appropriate antibiotics (25 μg⋅mL⁻¹ chloramphenicol, 100 μg⋅mL⁻¹ ampicillin, and 15 μg⋅mL⁻¹ gentamicin unless indicated otherwise). LB was supplemented with arabinose where indicated. To test capsule production in E. coli CWG1170 (ΔgalE) and CWG1171 (Δwza ΔgalE) mutants, cultures were grown in M9 minimal medium (31) supplemented with 0.2% glucose, or with 0.2% galactose plus 0.2% glucose (25).

In Vivo Photo–Cross-Linking.

WT Wza and Wza mutants were expressed in the presence of a mutant M. jannaschii tyrosyl-tRNA synthetase/tRNA pair encoded on plasmid pEVOL-pBpF [Addgene plasmid no. 31190 (19, 32)] to facilitate site-specific incorporation of the photo–cross-linkable amino acid pBpa (Bachem) into mutant proteins. The in vivo photo–cross-linking protocol was adapted from previously described procedures (33, 34). Briefly, overnight cultures were diluted to an A₆₀₀ of 0.1 into LB containing 1 mM pBpa (stock of 0.911 M in 1 M sodium hydroxide added to prewarmed media) and antibiotics at one-half of their working concentration (12.5 μg⋅mL⁻¹ chloramphenicol and 50 μg⋅mL⁻¹ ampicillin). At an A₆₀₀ of ∼0.4, Wza expression was induced with either 20 or 40 ng⋅mL⁻¹ anhydrotetracycline (Acros), and the aminoacyl tRNA synthetase/tRNA pair encoded by pEVOL-pBpF was induced with 0.02% arabinose. WT Wza and the majority of the mutants were induced with 20 ng⋅mL⁻¹ anhydrotetracycline, whereas N181, D207, and E286 were induced with 40 ng⋅mL⁻¹ anhydrotetracycline to generate equivalent amounts of Wza. Cells were grown to an A₆₀₀ of 1–1.2, and a 1-mL aliquot was removed and stored at −20 °C. The remaining culture was harvested, washed twice with cold PBS, and transferred to a single well of a six-well tissue culture plate on ice. The final suspension was ∼2.0 × 10⁹ cfu⋅mL⁻¹. Cell suspensions were irradiated for 5 min at a wavelength of 365 nm with a handheld UV lamp (115 V, 60 Hz, 0.2 A; Spectroline Model EN-180L) at a distance of 2.5 cm from the surface of the cell suspension. Cells were harvested by centrifugation and stored at −20 °C.

Purification of Wza-His₆ Cross-Linked Adducts.

Cultures expressing WT or mutant Wza with incorporated pBpa (treatment ±UV at 365 nm) were purified as previously described for multimeric Wza (27). Cultures (200 mL) were grown in the presence of 1 mM pBpa, and Wza expression was induced as described above. The cells were harvested, washed twice with PBS, and split into two equal parts. One-half of the sample was UV-treated for 5 min, and the remainder was incubated on ice. After UV treatment, both samples were centrifuged and the cells were resuspended in lysis buffer [20 mM sodium phosphate buffer (pH 7.5), 20 μg⋅mL⁻¹ DNase I, and 20 μg⋅mL⁻¹ RNase containing cOmplete EDTA-free protease inhibitor tablets (Roche)]. Cells were lysed by two passages through a French press at 10,000 psi, and unbroken cells were removed by centrifugation at 4,000 × g (cell lysate). Total membranes were collected from the cell-free lysate by centrifugation at 200,000 × g for 1 h at 4 °C. Inner membranes were selectively solubilized in 20 mM sodium phosphate buffer (pH 7.5), with 2% (wt/vol) sarkosyl (N-lauroylsarcosine; Sigma) (35) for 30 min at room temperature, and insoluble OMs were collected by centrifugation at 200,000 × g for 1 h at 18 °C. OM fractions were solubilized overnight at room temperature in 20 mM sodium phosphate buffer (pH 7.5), 50 mM NaCl, and 0.5% SB3-14 (Sigma), and insoluble aggregates were removed by centrifugation at 15,000 × g. Hexahistidine-tagged Wza proteins were purified on TALON (cobalt) metal affinity resin (Clontech) according to the manufacturer’s recommendations and eluted with 20 mM sodium phosphate buffer (pH 7.5), 300 mM NaCl, 0.05% SB3-14, and 250 mM imidazole.

Bacteriophage Sensitivity Assays.

The presence of K30 capsule on the cell surface was assessed using sensitivity to a K30 CPS-specific bacteriophage (36). Phage suspension (50 μL of bacteriophage at 10¹⁰ pfu⋅mL⁻¹) was spread over one-half of the surface of prewarmed LB agar plates (supplemented with 1 mM pBpa, 0.02% arabinose, and 40 ng⋅mL⁻¹ anhydrotetracycline), and 10 μL of overnight culture was cross-streaked and incubated for 4–6 h at 37 °C.

Building Structural Models for the Incorporation of pBpa.

For structural models, Wza side chains were changed to pBpa (code: PBF) using the PySwissSidechain plug-in (37) for PyMOL (PyMOL Molecular Graphics System Version 1.3; Schrödinger, LLC). Thirteen rotamers were available at position Asn181, of which only five showed no molecular collisions with the rest of the Wza oligomer. Rotamers with molecular collisions were removed from further consideration. The rotamer presented in Fig. 2B (blue) was chosen for display because it was the most similar in orientation to the side chain of the native asparagine residue. This rotamer leaves an opening diameter of ∼13 Å. The other four “noncolliding” rotamers would produce lumens with diameters of 12–14 Å.

MS.

LC-MS analyses were performed on purified C-terminal hexahistidine-tagged derivatives of the N181 and D207 mutants from control and UV-treated cells. Samples were analyzed on an Agilent 1200 HPLC liquid chromatograph interfaced with an Agilent UHD 6530 Q-TOF mass spectrometer at the Mass Spectrometry Facility of the Advanced Analysis Centre, University of Guelph. The mass-to-charge ratio was scanned across the range of 300–2,000 m/z in 4 GHz (extended dynamic range positive-ion auto MS/MS mode). Raw data files were loaded directly into PEAKS 7 software (Bioinformatics Solutions, Inc.), where the data were subjected to de novo sequencing and database searches. The MS data were also analyzed by StavroX (29) to search in an unbiased automated fashion for potential Wza/Wza adducts.