Identification of the Shigella flexneri Wzy Domain Modulating Wzz_pHS-2 Interaction and Detection of the Wzy/Wzz/Oag Complex

ABSTRACT

Shigella flexneri implements the Wzy-dependent pathway to biosynthesize the O antigen (Oag) component of its surface lipopolysaccharide. The inner membrane polymerase Wzy_SF catalyzes the repeat addition of undecaprenol-diphosphate-linked Oag (Und-PP-RUs) to produce a polysaccharide, the length of which is tightly regulated by two competing copolymerase proteins, Wzz_SF (short-type Oag; 10 to 17 RUs) and Wzz_pHS-2 (very-long-type Oag; >90 RUs). The nature of the interaction between Wzy_SF and Wzz_SF/Wzz_pHS-2 in Oag polymerization remains poorly characterized, with the majority of the literature characterizing the individual protein constituents of the Wzy-dependent pathway. Here, we report instead a major investigation into the specific binding interactions of Wzy_SF with its copolymerase counterparts. For the first time, a region of Wzy_SF that forms a unique binding site for Wzz_pHS-2 has been identified. Specifically, this work has elucidated key Wzy_SF moieties at the N- and C-terminal domains (NTD and CTD) that form an intramolecular pocket modulating the Wzz_pHS-2 interaction. Novel copurification data highlight that disruption of residues within this NTD-CTD pocket impairs the interaction with Wzz_pHS-2 without affecting Wzz_SF binding, thereby specifically disrupting polymerization of longer polysaccharide chains. This study provides a novel understanding of the molecular interaction of Wzy_SF with Wzz_SF/Wzz_pHS-2 in the Wzy-dependent pathway and, furthermore, detects the Wzy/Wzz/Und-PP-Oag complex for the first time. Beyond S. flexneri, this work may be extended to provide insight into the interactions between protein homologues expressed by related species, especially members of Enterobacteriaceae, that produce dual Oag chain length determinants.

IMPORTANCE Shigella flexneri is a pathogen causing significant morbidity and mortality, predominantly devastating the pediatric age group in developing countries. A major virulence factor contributing to S. flexneri pathogenesis is its surface lipopolysaccharide, which is comprised of three domains: lipid A, core oligosaccharide, and O antigen (Oag). The Wzy-dependent pathway is the most common biosynthetic mechanism implemented for Oag biosynthesis by Gram-negative bacteria, including S. flexneri. The nature of the interaction between the polymerase, Wzy_SF, and the polysaccharide copolymerases, Wzz_SF and Wzz_pHS-2, in Oag polymerization is poorly characterized. This study investigates the molecular interplay between Wzy_SF and its copolymerases, deciphering key interactions in the Wzy-dependent pathway that may be extended beyond S. flexneri, providing insight into Oag biosynthesis in Gram-negative bacteria.

INTRODUCTION

Bacteria biosynthesize an array of complex surface-associated polysaccharides that enable them to thrive and persist in distinct infectious niches (1, 2). Shigella flexneri produces a mixed population of surface glycolipids, with a large proportion comprised of lipopolysaccharide (LPS) and phosphatidylglycerol-linked enterobacterial common antigen (PG-ECA) (3). These polysaccharides aid S. flexneri’s pathogenesis by facilitating host cell surface adhesion, conferring resistance to antimicrobial agents, such as bile salts and acids encountered en route to the infectious niche, and modulating immune responses, such as complement activation (4, 5).

LPS is composed of three distinct domains. Located at the proximal end, lipid A forms the exterior leaflet of the outer membrane (OM) and serves as the anchor for LPS molecules. This is then followed by the oligosaccharide core, which can be further subdivided into inner and outer sugar entities. The remaining distal domain is composed of the highly variable O antigen (Oag) polysaccharide (6, 7). It is the high variability of the Oag component that forms the premise for the serotyping of many Gram-negative bacteria, including S. flexneri, which has over 19 confirmed serotypes classified by the composition and structure of its Oag antigenic determinants (8).

The Wzy-dependent pathway is the most common biosynthetic mechanism implemented for polysaccharide production by Gram-negative bacteria (3, 9). S. flexneri utilizes this system for the biosynthesis of the LPS Oag component. In brief, Oag oligosaccharide repeat unit (RU) synthesis begins in the cytoplasm, where the glycosyltransferase WecA adds an N-acetylglucosamine (GlcNAc) residue to the inner membrane (IM)-associated carrier lipid undecaprenol phosphate (Und-P). Subsequently, rhamnosyl transferases add three sequential rhamnose (Rha) residues to the undecaprenol-pyrophosphate (Und-PP)-bound GlcNAc residue, completing the Oag tetrasaccharide RU. The flippase, Wzx_SF, then translocates the Und-PP-RU across to the periplasmic side of the IM, enabling RU polymerization by the polymerase Wzy_SF (7, 10, 11). In S. flexneri, the length of the Oag chain synthesized is modulated by two distinct polysaccharide copolymerase 1 (PCP1) proteins, Wzz_SF and Wzz_pHS-2, which, in combination, regulate the production of a bimodal distribution of Oag chain lengths (12). The chromosomally encoded Wzz_SF regulates the production of LPS with short-type (S-type) Oag chains of approximately 10 to 17 RUs (4), whereas Wzz_pHS-2, encoded on the small plasmid pHS-2, regulates the production of LPS molecules with a very-long-type (VL-type) Oag chain of >90 RUs (13). Upon termination of Oag polymerization, the full-length polysaccharide is ligated by WaaL onto the lipid A core component [produced via the 3-deoxy-d-manno-octulosonic acid (Kdo)₂-lipid A biosynthetic pathway] before the full LPS molecule is exported to the cell surface by the Lpt system (14, 15).

The length of S. flexneri Oag is key in providing different survival advantages for the pathogen (16). Namely, S-Oag chains have been shown to be critical for resistance to bactericidal protein colicin E2 (4) and to enable the exposure of key surface proteins, such as IcsA, which modulates actin-based motility during the S. flexneri intracellular lifestyle (17). Conversely, the VL-Oag chains have been reported to confer resistance to complement (5). Overall, S. flexneri requires a combination and appropriate regulation of both Oag chain lengths for optimal virulence (12, 18).

The nature of interactions between Wzy_SF and the PCP1 proteins Wzz_SF and Wzz_pHS-2 in Oag polymerization is poorly understood. Wzy_SF is a 43.7-kDa, polytopic, integral IM protein, with 12 transmembrane (TM) segments forming 5 cytoplasmic loops (CLs) and 6 periplasmic loops (PLs) (Fig. 1) (19). Wzy_SF possesses two large PLs, PL3 and PL5, both of which contain the functional RX₁₅G motif (Fig. 1) that is believed to be involved in polymerization of the growing Oag chain (10, 20, 21). Similarly, Wzz_SF (37 kDa) and Wzz_pHS-2 (40 kDa) are both periplasm-exposed proteins, associated with the IM through two TM regions located at their N- and C termini (13, 22). Like all members of the PCP1 family, Wzz_SF and Wzz_pHS-2 share remarkable structural homology while maintaining very little amino acid sequence identity (23, 24). Structural studies of PCP1 proteins from various bacterial backgrounds, including Escherichia coli (22, 25) and Salmonella enterica serovar Typhimurium (26), have demonstrated a tendency for these proteins to form homooligomeric “bell-shaped” protomer structures. While debate remains in regard to the exact number of monomer units forming the structures, consensus suggests that octameric oligomers are the most entropically stable, favorable quaternary protein structure (23, 27, 28).

FIG 1.

S. flexneri Wzy topology map, based on PhoA-LacZ fusion protein analyses (19). Wzy_SF contains four Cys residues at positions C13, C60, C116, and C193 (blue), with C13 and C60 believed to be disulfide linked (29). Green moieties represent the arginine residues situated at the start, and in between, the functional RX₁₅G motifs (red), which, if disrupted, result in partial or complete loss of Wzy-mediated Oag polymerization (10). Yellow moieties represent the amino acid region (aa 351 to 357) shown to impact Wzy polymerization of the very-long (VL)-Oag modal length by Leo et al. (30). Transmembrane alpha-helical regions are numbered 1 to 12. Periplasmic loops (PLs) and cytoplasmic loops (CLs) are numerically annotated. The topology map was generated using Protter (https://wlab.ethz.ch/protter/) (48). Adapted from references 10 and 29.

Previously, it has been reported that the second Oag copolymerase in S. flexneri, Wzz_pHS-2, competes with Wzz_SF to regulate the modal length of the polysaccharide (12). Studies by Carter et al. (12) illustrated that an increase in Wzz_SF expression and production of the S-Oag length resulted in a concurrent, proportional decrease in VL-Oag and vice versa, suggesting that these two PCP1 proteins were in direct competition, though the nature of the competition remained unclear. Recently, our group has shown differences in the molecular interactions between Wzy_SF/Wzz_SF and Wzy_SF/Wzz_pHS-2. Through in vitro assays using the thiol-reactive reagent polyethylene glycol-maleimide (mPEG) to probe the niche Wzy_SF conformation, we demonstrated that the interaction of Wzy_SF with Wzz_SF or Wzz_pHS-2 results in the polymerase undergoing unique, copolymerase-specific changes in its conformational arrangement (29). This result was subsequently corroborated with functional data, whereby it was demonstrated that disruption of the Wzy_SF N-terminal domain (NTD) cysteine moieties resulted in decreased VL-Oag biosynthesis but had no effect on the production of the S-Oag modal length. In an independent study, Leo et al. found the same effect on VL-Oag production through disruption of Wzy_SF amino acids 351 to 357 at the C-terminal domain (CTD) (30). These data formed the basis for the hypotheses that Wzz_SF and Wzz_pHS-2 bind at separate Wzy_SF binding sites and that the Wzy_SF NTD and CTD are arranged in close spatial proximity, together forming a unique interaction site for Wzz_pHS-2 (29).

This study aims to investigate these hypotheses and decipher the molecular interplay between Wzy_SF and the copolymerase proteins Wzz_SF and Wzz_pHS-2 through mutagenesis and native binding assays. In this study, we further investigated the effect of cysteine mutagenesis on Wzy_SF Oag polymerization. LPS profiles were first verified through silver-stained SDS-PAGE gels and independently confirmed via colicin E2/complement sensitivity assays, corroborating that disruption of cysteine moieties decreased the production of VL-Oag with no effect on S-Oag expression. High-expression vectors were then developed that enabled copurifications to be performed in the absence of a chemical cross-linker in order to investigate the native interaction of Wzy_SF and a cysteine-less Wzy_SF derivative (Wzy^CØ) with the PCP1 proteins. For the first time, reciprocal interactions between Wzy_SF/Wzz_SF and Wzy_SF/Wzz_pHS-2 were detected, with binding experiments revealing the importance of Wzy cysteine residues in the interaction with Wzz_pHS-2. In combination with the binding data, in silico structural analyses identified the key Wzy residues involved in forming an intramolecular NTD-CTD pocket that subsequently formed the site for Wzz_pHS-2 interaction. Finally, further copurification studies also revealed unprecedented detection of an interaction between native Wzy_SF, Wzz, and Und-PP-Oag, which is the first in vivo detection of the Wzy_SF-Wzz-Oag complex.

RESULTS

The impact of Wzy Cys-to-Ala mutations on Oag polymerization.

Wzy_SF (hereinafter referred to as Wzy) possesses four cysteine (Cys or C) moieties situated toward the polymerase NTD at residues 13, 60, 116, and 193 (Fig. 1). Previous work by our group (29) has illustrated that disruption of these Cys moieties (via replacement with alanine [Ala or A] residues) affects Wzy-mediated Oag biosynthesis. In this study, we first quantitatively corroborated the previously observed phenotypes via silver-stained LPS SDS-PAGE and densitometry for cells of S. flexneri serotype Y with either single (wzy, strain RM2608) or double (wzy wzz_SF::tet^r, strain RMA4337) mutations (Table 1). As previously observed, complementation of the wzy mutant with the cysteine-less Wzy derivative (Wzy^CØ) yielded 2-fold less VL-Oag than complementation with wild-type Wzy (Fig. S1A, lanes 3 and 5, and Fig. S1B in the supplemental material). Densitometry also substantiated that Cys replacement appeared to have no effect on the production of the S-Oag modal length (Fig. S1C). Importantly, in the wzy wzz_SF::tet^r mutant background, wherein competition of Wzz_SF and Wzz_pHS-2 for polymerase interaction is removed, the same 2-fold reduction in VL-Oag was again observed upon complementation with Wzy^CØ (Fig. S1A, lanes 4 and 6, and Fig. S1D). This supports the previous findings that disruption of Wzy Cys residues does not merely allow Wzz_SF to outcompete Wzz_pHS-2 for polymerase interaction but, rather, may directly disrupt the interaction between Wzy and Wzz_pHS-2. These results formed the premise for the subsequent binding experiments in this study.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
Strains
RMA2162 (PE860 derived)	S. flexneri Y serotype; lacks virulence plasmid (VP) and pHS-2 plasmid (which encodes WzzpHS-2)	Laboratory stock
RMA2163 (PE860 derived)	S. flexneri Y serotype wzzSF::kanr mutant, lacks VP and pHS-2	Laboratory stock
RMA5155 (PE860 derived)	S. flexneri Y serotype wzy mutant, lacks VP and pHS-2	Laboratory stock
RMA2608 (PE638 derived)	S. flexneri Y serotype wzy mutant, lacks VP, has pHS-2	Laboratory stock
RMA4337 (PE638 derived)	S. flexneri Y serotype wzy wzzSF::tetr double mutant, lacks VP, has pHS-2	Laboratory stock
RMA3325 (PE638 derived)	S. flexneri Y serotype wzy waaL::kanr double mutant, lacks VP, has pHS-2	Laboratory stock
Plasmids
pBCKs	pBluescript Ks II+, IPTG inducible, cmlr	Stratagene
pBCKs::wzy-3×FLAG	pBCKs encoding Wzy-3×FLAG	30
pBCKs::wzyCØ-3×FLAG	pBCKs encoding Wzy-3×FLAGC13A/C60A/C116A/C193A (referred to as WzyCØ in text)	29
pBAD18	Arabinose-inducible expression vector, cmlr	49
pBAD18::wzy-8×His	pBAD18 encoding Wzy-8×His (CTD)	29
pBAD18::wzyCØ-8×His	pBAD18 encoding WzyCØ-8×His (CTD)	29
pWSK29	IPTG-inducible expression vector, ampr	33
pWSK29::wzzSF-3×FLAG	pWSK29 encoding WzzSF-3×FLAG (NTD)	This study
pWSK29::wzzpHS-2-3×FLAG	pWSK29 encoding WzzpHS-2-3×FLAG (NTD)	This study
Background strains used for:	Plasmid(s)
Purification
RMA2163	pWSK29::wzzSF-3×FLAG	This study
RMA2163	pWSK29::wzzpHS-2-3×FLAG	This study
RMA2163	pWSK29	This study
RMA5155	pBAD18::wzy-8×His	This study
RMA5155	pBAD18::wzyCØ-8×His	This study
RMA5155	pBAD18	This study
RMA3325	pBAD18::wzy-8×His	This study
RMA3325	pBAD18::wzyCØ-8×His	This study
RMA3325	pBAD18	This study
Copurification
RMA2163	pBAD18::wzy-8×His + pWSK29::wzzSF-3×FLAG	This study
RMA2163	pBAD18::wzy-8×His + pWSK29::wzzpHS-2-3×FLAG	This study
RMA2163	pBAD18::wzyCØ-8×His + pWSK29::wzzSF-3×FLAG	This study
RMA2163	pBAD18::wzyCØ-8×His + pWSK29::wzzpHS-2-3×FLAG	This study
RMA2163	pBAD18 + pWSK29::wzzSF-3×FLAG	This study
RMA2163	pBAD18 + pWSK29::wzzpHS-2-3×FLAG	This study
RMA2163	pBAD18::wzy-8×His + pWSK29	This study
RMA2163	pBAD18::wzyCØ-8×His + pWSK29	This study

The consequences of Wzy Cys substitution were further investigated through a series of assays characterizing susceptibility to various bactericidal agents. First, colicin sensitivity was assessed, to indirectly investigate the expression of S-Oag, which is the Oag modal length understood to be responsible for colicin resistance (4). Colicin E2, a bactericidal DNase that interacts with the OM surface receptor BtuB to gain intracellular access before cleaving DNA with undefined specificity, was employed (4). The MICs of colicin E2 were assessed in biological triplicates (Fig. S2A). As expected, both the S. flexneri wzy mutant (RMA2608) and the strain harboring the pBCKs empty-vector control exhibited high sensitivity to colicin-mediated killing, with MICs of less than 0.5 μg/mL. In contrast, the strains expressing the FLAG-tagged wild-type polymerase or the Wzy^CØ derivative both illustrated much higher colicin E2 resistance, with 3-fold and 2-fold greater MICs, respectively (Fig. S2A). Statistical analyses revealed no significant difference in colicin resistance between the strains expressing Wzy and Wzy^CØ, independently suggesting that these strains expressed comparable abundances of S-Oag.

Next, complement sensitivity assays were performed to indirectly investigate the expression of VL-Oag, which is the Oag modal length understood to be responsible for conferring complement resistance (5). Assays were performed using 10% (vol/vol) human serum over a period of 2 h. First, the S. flexneri wzy mutant (RMA2608) complemented with pBCKs vectors encoding FLAG-tagged Wzy or Wzy^CØ (or the pBCKs empty vector) was analyzed (Fig. 2A). Similar to the colicin E2 analyses, the wzy mutant and the strain harboring empty pBCKs both exhibited high susceptibility to complement-mediated killing, with an approximate 100-fold decrease in total survival after 120 min. This can be attributed to the complete absence of surface Oag, including VL-Oag, which is the modal length responsible for the inhibition of membrane attack complex (MAC) assembly (31, 32). In contrast, the strains expressing Wzy or Wzy^CØ, which are capable of S- and VL-Oag biosynthesis, demonstrated much greater resistance to complement. However, strikingly, the strain expressing the wild-type polymerase had significantly increased survival at each time point compared to the cysteine-less mutant, with 10-fold greater total survival after 2 h (120 min) (Fig. 2A). These results independently suggest that Wzy^CØ expression leads to compromised VL-Oag biosynthesis, which in turn increases the Wzy^CØ strain’s susceptibility to complement. This discrepancy was then corroborated in the S. flexneri wzy wzz_SF::tet^r mutant background, which eliminates the dual expression of S- and VL-Oag, enabling the exclusive analysis of VL-Oag (Fig. 2B). As a control, incubation with heat-inactivated serum demonstrated that all wzy (Fig. S2B) and wzy wzz_SF::tet^r (Fig. S2C) background strains experienced uncompromised survival (>100%) in the absence of active complement.

FIG 2.

Analysis of Wzy Cys mutagenesis via complement sensitivity. Survival of strains in the presence of 10% (vol/vol) human complement serum. (A) S. flexneri serotype Y wzy (RMA2608) (black), expressing Wzy (red), Wzy^CØ (blue), or pBCKs (gray) (Table 1). (B) S. flexneri serotype Y wzy wzz_SF::tet^r (RMA4337) (black), expressing Wzy (red), Wzy^CØ (blue), or pBCKs (gray) (Table 1). Assays were performed in biological triplicate, and statistical analysis was performed using two-way ANOVA with multiple comparisons. Data presented are the mean values ± standard errors of the means (SEM) (n = 3). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant.

Generation of high-expression purification constructs.

To investigate the effect of Cys mutagenesis on Wzy’s interaction with Wzz_SF and Wzz_pHS-2, binding assays were performed. However, in order to perform these assays, high-expression vectors were first generated for both copolymerase and polymerase proteins. For the expression of the copolymerases, new vectors were designed using the backbone pWSK29 (33). Constructs pWSK29::wzz_SF-3×FLAG and pWSK29::wzz_pHS-2-3×FLAG were generated, which, upon induction, ectopically expressed NTD 3×FLAG epitope-tagged Wzz_SF or Wzz_pHS-2, respectively. Using these vectors, the copolymerase proteins were purified via anti-DYKDDDDK G (anti-FLAG) resin affinity purification and successfully detected via anti-FLAG antibody Western immunoblotting (Fig. 3A, lanes 1 and 2). The functionality of the proteins expressed was also investigated via silver-stained SDS-PAGE gel analysis in a strain devoid of wzz_SF and naturally devoid of wzz_pHS-2 (RMA2163). The results illustrated that both pWSK29::wzz_SF-3×FLAG and pWSK29::wzz_pHS-2-3×FLAG produced functional protein products that were able to act as S- and VL-Oag chain length determinants, respectively (Fig. 3C, lanes 3 and 4). As expected, the pWSK29 empty-vector control yielded no purified protein (Fig. 3A, lane 3), and no Oag chain length regulation was detected (Fig. 3C, lane 5).

FIG 3.

Confirmation of the expression and function of Wzz_SF/Wzz_pHS-2 and Wzy/Wzy^CØ from high-expression purification vectors. (A) Analysis of S. flexneri serotype Y wzz_SF::kan^r (RMA2163) transformed with pWSK29::wzz_SF-3×FLAG, pWSK29::wzz_pHS-2-3×FLAG, and pWSK29 (Table 1). Anti-FLAG antibody Western immunoblot of anti-DYKDDDDK G (FLAG) affinity resin-purified protein samples, separated through 12% (wt/vol) SDS-PAGE. The migration positions of the molecular-mass standard (kDa) are indicated on the left-hand side. Each lane was loaded with 10 μL of purified protein sample. (B) Analysis of S. flexneri serotype Y wzy (RMA5155) transformed with pBAD18::wzy-8×His, pBAD18::wzy^CØ-8×His, and pBAD18 (Table 1). Anti-His antibody Western immunoblot of immobilized metal affinity chromatography (IMAC)-purified protein samples, separated through 12% (wt/vol) SDS-PAGE. The migration positions of the molecular mass standard (kDa) are indicated on the left-hand side. Each lane was loaded with 10 μL of purified protein sample. (C) LPS was isolated from mid-exponential-phase whole-cell (WC) lysates of the wild-type (WT) strain (RMA2162, lane 1), the wzz_SF::kan^r strain (RMA2163, lane 2), and the complemented wzz_SF::kan^r derivatives (lanes 3 to 5) (Table 1). Samples were then separated via 15% (wt/vol) SDS-PAGE, and LPS visualized through silver staining. Approximately 2 × 10⁸ cells were loaded per lane. (D) LPS was isolated from mid-exponential-phase WC lysates of the WT strain (RMA2162, lane 1), the wzy strain (RMA5155, lane 2), and the complemented wzy derivatives (lanes 3 to 5) (Table 1). Samples were separated via 15% (wt/vol) SDS-PAGE, and LPS visualized through silver staining. Approximately 2 × 10⁸ cells were loaded per lane.

For the expression of polymerase proteins, vectors pBAD18::wzy-8×His and pBAD18::wzy^CØ-8×His were employed (29), which in trans overexpress Wzy and Wzy^CØ with a CTD 8×His epitope tag upon arabinose induction. His-tagged Wzy and Wzy^CØ were successfully purified via immobilized metal affinity chromatography (IMAC) and detected via anti-His antibody Western immunoblotting (Fig. 3B, lanes 1 and 2). The functionality of the expressed polymerases was also investigated via silver-stained SDS-PAGE in a strain devoid of wzy and naturally devoid of wzz_pHS-2 (RMA5155). The results show that both Wzy (Fig. 3D, lane 3) and Wzy^CØ (Fig. 3D, lane 4) are capable of Oag biosynthesis. Again, as expected, the pBAD18 empty-vector control yielded no purified protein (Fig. 3B, lane 3), and no Oag polymerization was detected (Fig. 3D, lane 5). Interestingly, while the total proportions of Oag produced by His-tagged Wzy and Wzy^CØ did not appear to differ, the distribution of the Oag did. The wild-type polymerase produced some Oag chains that were noticeably longer than those biosynthesized by Wzy^CØ (Fig. 3D, lane 3), and meanwhile, Wzy^CØ instead generated a greater proportion of S-Oag (Fig. 3D, lane 4). Given that both vectors overexpressed His-tagged Wzy and Wzy^CØ compared to endogenous cellular wzy levels, we speculated that the subsequent ratio of Wzy/Wzz_SF was too great, resulting in insufficient endogenous Wzz_SF to regulate Wzy-mediated Oag polymerization and the subsequent loss of modal-length control observed. In combination with the results from the previous silver-stained SDS-PAGE analyses (Fig. S1A), we also proposed that despite being equally overexpressed, Wzy^CØ had an impaired ability to biosynthesize longer Oag chain lengths and therefore favored polymerization of shorter Oag. This phenomenon was further investigated in this study, through binding assays interrogating Oag’s association with Wzy and Wzy^CØ.

Detection of native, reciprocal Wzy-PCP1 interactions.

Reciprocal copurifications of Wzy/Wzy^CØ with the PCP1 proteins were then performed without the use of a chemical cross-linker, to interrogate only native protein-protein interactions. The high-expression vectors generated as descried above (Table 1) were used to transform RMA2163, an S. flexneri serotype Y strain devoid of both wzz_SF and wzz_pHS-2 but still possessing the endogenous wzy. Transformed strains were grown, whole cells fractionated, and proteins independently purified via either IMAC or anti-FLAG affinity resin. After purification, samples were separated via electrophoresis, and proteins were detected through anti-His and anti-FLAG antibody Western immunoblotting.

First, samples obtained through IMAC purification were subjected to analysis. Anti-His antibody Western immunoblotting revealed that His-tagged Wzy and Wzy^CØ were both successfully purified at comparable levels from all copurification strains (Table 1 and Fig. 4A, lanes 1 and 2, 4 and 5, and 7 and 8), with the exception of those harboring the vector-only control (Fig. 4A, lanes 3 and 6). Anti-FLAG antibody Western immunoblotting of the same purified samples revealed that Wzy and Wzy^CØ were both copurified with comparable levels of Wzz_SF (Fig. 4A, lanes 1 and 2). Interestingly, the same immunoblot also revealed that while both Wzy and Wzy^CØ were also copurified with Wzz_pHS-2, there was a clear discrepancy in the binding of the VL-Oag copolymerase, with Wzy^CØ pulling down notably less Wzz_pHS-2 (Fig. 4A, lane 5) than the wild-type polymerase (Fig. 4A, lane 4). As expected, anti-FLAG antibody Western immunoblotting of the pWSK29 empty-vector control samples resulted in no detection of Wzz_SF or Wzz_pHS-2, ensuring that any detection of the copolymerases in IMAC-purified samples occurred exclusively as a result of native copurification with the polymerase (Fig. 4A, lanes 7 and 8). Additionally, the purification samples of the pBAD18 empty-vector control strains also had no detectable protein bands corresponding to FLAG-tagged copolymerase proteins (Fig. 4A, lanes 3 and 6).

FIG 4.

Reciprocal native protein copurifications of Wzy/Wzy^CØ and Wzz_SF/Wzz_pHS-2. Reciprocal copurifications of CTD 8×His-tagged Wzy/Wzy^CØ and NTD 3×FLAG-tagged Wzz_SF/Wzz_pHS-2 were performed in the absence of any chemical cross-linker (native) through IMAC (A) and anti-DYKDDDK G (FLAG) affinity resin (B) (see Materials and Methods). Following IMAC or anti-FLAG resin purification, protein samples were separated via 12% (wt/vol) SDS-PAGE and subjected to Western immunoblotting using both anti-His and anti-FLAG monoclonal mouse antibodies. S. flexneri serotype Y wzz_SF::kan^r (RMA2163) copurification strains are summarized in Table 1. Western immunoblot lanes are annotated numerically and with the proteins expressed by each copurification strain. The migration positions of the molecular mass standard (kDa) are indicated on the left-hand side. Each lane was loaded with 10 μL of purified protein sample. The reduced binding between Wzy^CØ and Wzz_pHS-2 is noted in the area surrounded by the red dashed box in each panel.

Next, the same copurification strains were subjected to anti-FLAG affinity purification and analysis. Anti-FLAG antibody Western immunoblotting revealed that FLAG-tagged Wzz_SF and Wzz_pHS-2 were both successfully purified at comparable levels in all copurification strains (Fig. 4B, lanes 1 to 6), with the exception of those harboring the pWSK29 empty-vector control (Fig. 4B, lanes 7 and 8), which, as expected, yielded no bands corresponding to Wzz_SF and Wzz_pHS-2. Anti-His antibody Western immunoblotting of the same samples revealed that Wzz_SF and Wzz_pHS-2 were both copurified with a comparable level of wild-type Wzy (Fig. 4B, lanes 1 and 4). Importantly, the same immunoblot also revealed that, while both Wzz_SF and Wzz_pHS-2 were also copurified with Wzy^CØ, there was a clear discrepancy in the binding of the cysteine-less polymerase, with Wzz_pHS-2 pulling down notably less Wzy^CØ (Fig. 4B, lane 5) than Wzz_SF (Fig. 4B, lane 4). As expected, anti-His antibody Western immunoblotting of the pBAD18 empty-vector control samples resulted in no detection of Wzy or Wzy^CØ, confirming that any detection of polymerases in FLAG-purified samples occurred exclusively as a result of interaction with the copolymerases during purification (Fig. 4A, lanes 3 and 6). Additionally, the purification samples of the pWSK29 empty-vector control strains also had no detectable protein bands corresponding to His-tagged polymerase proteins (Fig. 4A, lanes 7 and 8).

Identification of a putative Wzy binding pocket modulating its interaction with Wzz_pHS-2.

As mentioned above, it has been reported that mutagenesis of S. flexneri Wzy amino acids (aa) 351 to 357, specifically the replacement of these residues with Ala moieties, results in a similar reduction in VL-Oag production (30) as was observed here and in our previous study (29). The same study also presented some preliminary binding studies, although not performed reciprocally, which suggested that Wzy’s interaction with Wzz_pHS-2 may be compromised through the replacement of this amino acid region (30). These data, in combination with the binding results from this study (Fig. 4), led to the speculation that the Wzy Cys residues, located at the Wzy NTD, may be in close proximity to and may potentially interact with aa 351 to 357 at the Wzy CTD. Subsequently, the available AlphaFold-predicted structure was analyzed to decipher the spatial arrangement of these moieties of interest (34). Cys residues 13, 60, and 116 (Fig. 5A, red) are located in Wzy alpha (α) helices α₁, α₃, and α₅, respectively, while the aa 351-to-357 region (Fig. 5A, blue) is situated in helix α₁₂ (Fig. 5A). Analysis of the Wzy 3-dimensional conformation illustrates that these helices are arranged in strikingly close proximity. Measurement of the intramolecular distances between the R groups of the Cys residues and aa 351 to 357 revealed small distances ranging from 3.4 to 16.6 Å (Fig. 5B). Further analysis of the in silico structure also revealed that C13, C60, and C116 and the majority of the aa 351-to-357 region also had side chains facing inward, forming a pocket whereby these R groups might be in favorable orientation for the formation of secondary intramolecular interactions (Fig. 5B). Overall, these results suggest that the Wzy NTD Cys moieties and CTD aa 351 to 357 may be in close spatial proximity, thus forming a Wzy pocket that facilitates interaction with Wzz_pHS-2.

FIG 5.

In silico-predicted protein structure of S. flexneri Wzy. (A) AlphaFold structure of S. flexneri Wzy, showing the locations of the Wzy NTD Cys residues (red) in α-helices 1, 3, and 5 and CTD aa 351 to 357 (blue) in α-helix 12. (B) Location of the Wzy interaction pocket, coordinated by C13, C60, C116, and aa 351 to 357. Atomic distances were measured using USFC Chimera software, specifically measuring the distances between cysteine thiol groups (-SH) and the R group of the uncharged amino acids T351, P352, M353, G354, I355, F356, and I357.

Detection of native in vivo Oag association with Wzy and the Wzy-Wzz-Oag complex.

Purifications to interrogate the association of the Und-PP-Oag substrate with Wzy or Wzy^CØ were also performed, again in the absence of a chemical cross-linker so as to investigate only native interaction. pBAD18::wzy-8×His, pBAD18::wzy^CØ-8×His, and pBAD18 (Table 1) were used to transform RMA3325, which is an S. flexneri serotype Y strain that is devoid of both wzy and waaL (but has wzz_SF and wzz_pHS-2) and, thus, accumulates lipid-linked oligosaccharide (LLO) repeat units in the periplasm. Transformed strains were grown, soluble membrane fractions prepared, and proteins purified through IMAC resin. Eluted proteins were then separated through electrophoresis, prior to anti-His and anti-MASFB antibody Western immunoblotting.

Anti-His antibody Western immunoblotting revealed that His-tagged Wzy and Wzy^CØ were both successfully purified at comparable levels in both strains (Fig. 6A, lanes 1 and 2), which was also corroborated by colloidal blue staining of the same samples (Fig. 6B, lanes 1 and 2). Furthermore, analysis of the same IMAC-purified samples through anti-Wzz_SF and anti-Wzz_pHS-2 antibody Western immunoblotting revealed that both Wzy and Wzy^CØ were also concurrently interacting with Wzz_SF (Fig. 6C, lanes 1 and 2) and Wzz_pHS-2 (Fig. 6E, lanes 1 and 2). Again, the interaction between Wzy^CØ and Wzz_pHS-2 was compromised, with the cysteine-less polymerase mutant pulling down less Wzz_pHS-2 than wild-type Wzy. Furthermore, using anti-MASFB antibody, which is a monoclonal antibody specific for the O antigen of S. flexneri (35), Western immunoblotting of the same samples revealed that Wzy and Wzy^CØ were also both copurified with two populations of natively interacting Und-PP-Oag (Fig. 6E) that were distinguishable by their lengths. Densitometric analyses showed that Wzy and Wzy^CØ appeared to be interacting with comparable abundances of Und-PP-Oag of the shorter chain length (population A) (Fig. 6F). In contrast, Wzy^CØ was observed to interact with less than 50% of the Und-PP-Oag chains of the longer length (population B) relative to the results for the wild-type polymerase (Fig. 6F). As a control, purification of the strain harboring the empty vector (pBAD18) yielded no detectable protein corresponding to Wzy/Wzy^CØ through anti-His antibody Western immunoblotting (Fig. 6A, lane 3) or colloidal blue staining (Fig. 6B, lane 3). Additionally, no PCP1 proteins (Fig. 6C and D, lanes 3) or Und-PP-Oag (Fig. 6E, lane 3) were detected through anti-Wzz_SF/anti-Wzz_pHS-2 antibody or anti-MASFB antibody immunoblotting, respectively, highlighting that any detection of PCP1 proteins or Oag was exclusively as a result of interaction in a complex with Wzy/Wzy^CØ during purification.

FIG 6.

Detection of the association of native Oag with Wzy/Wzy^CØ and in vitro identification of the Wzy/Wzz/Und-PP-Oag complex. S. flexneri serotype Y wzy waaL::kan^r (RMA3325) strains harboring pBAD18::wzy-8×His, pBAD18::wzy^CØ-8×His or pBAD18 (summarized in Table 1) were subjected to IMAC purification. (A to E) Purified samples were then separated via 12% (wt/vol) SDS-PAGE and visualized via anti-His antibody Western immunoblotting (A), colloidal blue staining (B), or anti-Wzz_SF (C), anti-Wzz_pHS-2 (D), and anti-MASFB (E) antibody Western immunoblotting. Immunoblot lanes are annotated numerically and with the ectopic proteins expressed by each strain. The migration positions of the molecular-mass standard (kDa) are indicated on the left-hand side. Each lane was loaded with 10 μL of purified protein sample. Und-PP-Oag A and B populations are represented by the annotated boxes in panel E. (F) Densitometric analysis of the anti-MASFB antibody Western immunoblot. The amount of total Und-PP-Oag associated with Wzy or Wzy^CØ is represented as the relative percentage (%) of the amount of Und-PP-Oag associated with the wild-type polymerase. Densitometry was performed on populations A and B of Und-PP-Oag independently. Data presented are the mean values ± SEM (n = 3). Statistical analysis was performed using Student’s t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant.

DISCUSSION

The majority of research investigating the Wzy-dependent pathway focuses heavily on characterizing its individual protein constituents and elucidating the mechanism by which polysaccharide biosynthesis occurs (1, 3, 20). Here, we report instead the specific binding interactions of Wzy with its copolymerase counterparts. In this study, novel, native reciprocal interactions between Wzy and Wzz_SF/Wzz_pHS-2 were detected in the absence of a chemical cross-linker, which specifically highlighted the importance of Wzy Cys moieties in modulating its interaction with Wzz_pHS-2. These binding data, in combination with in silico structural analyses, revealed a site of interaction between Wzy and Wzz_pHS-2, namely, the Wzy residues involved in forming a unique Wzz_pHS-2 binding pocket. Additionally, for the first time, native interaction between Wzy, its substrate Oag, and the PCP1 proteins was detected, which is the first detection of the Wzy/Wzz/Und-PP-Oag complex.

Previous work by our group revealed that the replacement of Wzy NTD Cys residues affected Oag biosynthesis (29), and this formed the premise for subsequent analyses in this study. Initially, we corroborated the previously observed qualitative phenotypes, both via silver-stained LPS SDS-PAGE with densitometry and via colicin E2 and complement sensitivity assays. Our data cohesively demonstrated a statistically significant decrease in VL-Oag production by Wzy^CØ compared to the VL-Oag production by wild-type Wzy, with no apparent change in S-Oag production. The importance of Cys residues in both protein architecture and function is well known (36, 37). Disruption of Wzy Cys residues has been shown to alter the overall conformational arrangement of the polymerase (34), leading to speculation that, potentially, this new altered conformation favors interaction with Wzz_SF, subsequently resulting in Wzz_pHS-2 being outcompeted for polymerase interaction. However, analysis of Wzy^CØ- and Wzy-mediated Oag biosynthesis in the absence of Wzz_SF-polymerase competition revealed the same total decrease in the production of the VL-Oag modal length by Wzy^CØ. These data demonstrate that mutagenesis of Wzy NTD Cys residues does not merely allow Wzz_SF to outcompete Wzz_pHS-2 but, more likely, directly disrupts the Wzy-Wzz_pHS-2 interaction.

To directly investigate the interactions between Wzy and Wzz_SF/Wzz_pHS-2 while concurrently interrogating the effect of Cys mutagenesis on these polymerase-copolymerase interactions, pulldown binding assays were performed using wild-type Wzy and Wzy^CØ. The binding experiments involved native protein purification using newly generated, functional expression vectors and detection of target proteins through Western immunoblotting. Independent IMAC and anti-FLAG antibody reciprocal copurifications revealed that both Wzy and Wzy^CØ interacted successfully with comparable levels of Wzz_SF and that, meanwhile, Wzy^CØ interacted with notably less Wzz_pHS-2 than the wild-type polymerase did. Reciprocally, Wzz_SF and Wzz_pHS-2 both interacted with comparable levels of Wzy; however, Wzz_pHS-2 bound markedly less Wzy^CØ than Wzz_SF did. Overall, these data complement one another and, together, demonstrate that the interactions between Wzy/Wzz_SF and Wzy/Wzz_pHS-2 differ, with mutation of Wzy Cys moieties exclusively affecting the interaction with the VL-Oag copolymerase. Furthermore, these data support the model in which the two copolymerase proteins compete for Wzy interaction but bind the polymerase at unique, copolymerase-specific interaction sites (29, 30).

A recent study investigating the CTD of S. flexneri Wzy reported an LPS phenotype similar to that observed both in our previous work and in this study (29, 30). Specifically, the replacement of Wzy CTD residues 351 to 357 (T351, P352, M353, G354, I355, F356, and I357) with Ala moieties resulted in an approximately 50% reduction in VL-Oag, with no effect on S-Oag polymerization (30). Similarly, this decrease in VL-Oag production was true both in the presence and absence of Wzz_SF/Wzz_pHS-2 competition for Wzy interaction, with disruption from moieties P352A (the substitution of A for P at position 352), M353A, G354A, and I357A resulting in the most dramatic decreases in VL-Oag. The same study also presented some preliminary binding data, although not performed reciprocally, that suggested that the Wzy interaction with Wzz_pHS-2 might be compromised through replacement of residues aa 351 to 357 (30). These data, in combination with the results from this study, led to the speculation that Wzy may occupy a tertiary conformation with the NTD and CTD in close spatial proximity, together forming a unique Wzy binding site modulating Wzz_pHS-2 interaction (29).

The Wzy AlphaFold structure was subsequently analyzed to address this hypothesis and decipher the spatial arrangement of the NTD Cys moieties and CTD residues of interest (aa 351 to 357) (34). Measurement of the space between the amino acid R groups revealed only small distances among the Cys moieties (C13, C60, and C116), as well as, interestingly, small distances between the Cys residues and the CTD aa 351 to 357. Additionally, analysis of the in silico structure revealed that C13, C60, and C116 and the majority of aa 351 to 357 were not only close in proximity but also had inward-facing side chains, potentially forming a pocket between helices α₁, α₃, α₅, and α₁₂ whereby secondary intramolecular interactions, such as the known disulfide bond between C13 and C60 (29), were forged. The literature reports that α-helices have a significantly greater tendency to form hydrophobic-hydrophobic interactions than β-sheets do, with specifically short-to-medium range (3.8 and 9.5 Å) distances proving the most favorable for these interactions to occur (38, 39). Given that the Cys residues and aa 351 to 357 are comprised of uncharged amino acids and lie near or within this favorable distance range and that a disulfide bond between C13-C60 has already been experimentally verified (29), it is highly likely that the R groups of these moieties coordinate further secondary hydrophobic interactions between helices α₁, α₃, α₅, and α₁₂, giving rise to this NTD-CTD central pocket. In combination with the data showing that binding interrupts these key residues at the Wzy NTD (this study) and Wzy CTD (30), respectively, we suggest that this Wzy NTD-CTD pocket is a binding site for Wzz_pHS-2. Interruption of key residues within this region, at either protein terminus, would impair Wzy’s interaction with the VL-Oag copolymerase, accounting for the overall decrease in VL-Oag production initially observed via silver staining.

Finally, we investigated the impact of Cys mutagenesis and the subsequent disruption of the NTD-CTD region on the ability of Wzy to retain Und-PP-Oag during polymerization. IMAC purification of Wzy and Wzy^CØ and subsequent Western immunoblotting revealed that both polymerases successfully interacted with Und-PP-Oag; however, the quantitation of total associated Und-PP-Oag via densitometry revealed that Wzy^CØ pulled down less than 50% of the population B Und-PP-Oag relative to the amount pulled down by the wild-type polymerase, with no real change in the total associated Und-PP-Oag of population A. This suggests that disruption of Cys residues in that critical NTD-CTD site affected the ability of Wzy to associate with Oag, driving a preference for shorter Oag chain lengths. Western immunoblotting also revealed that Wzy and Wzy^CØ were simultaneously also interacting with Wzz_SF/Wzz_pHS-2, again corroborating a binding deficit between Wzy^CØ and Wzz_pHS-2. The concurrent detection of the native interaction of Wzy, the Und-PP-Oag intermediates, and the PCP1 proteins is the first reported in vivo detection of the Wzy/Wzz/Und-PP-Oag complex.

The deficit in Wzy-associated VL-Oag detected through anti-MASFB antibody Western immunoblotting (Fig. 7C), in combination with the silver-stained SDS-PAGE analyses, led to speculation that, beyond Wzz_SF and Wzz_pHS-2 interacting with Wzy at unique interaction sites, the means by which the S-Oag and VL-Oag modal lengths are achieved may differ. Robbins et al. performed the first study deriving the mechanism by which Oag extension occurs during polymerization, elucidating that individual Oag RUs are added at the reducing end of the nascent, growing polysaccharide (40). The regulator of Oag length was later identified as Wzz_SF, and it was found to modulate the production of the S-Oag modal length (41). Subsequently, the identification of a bimodal Oag distribution in certain enterobacterial species, such as S. flexneri, led to the identification of the VL-Oag chain length regulator, Wzz_pHS-2 (41–43). However, to date, the method by which Wzz_pHS-2 regulates the production of VL-Oag has always been assumed to be identical to that implemented by Wzz_SF to regulate the S-Oag length, despite the chain lengths being drastically different. Based on the results observed in this study, we propose a new putative model of VL-Oag polymerization and modal length control.

FIG 7.

Summary of S. flexneri PCP1-mediated Oag modal length control. A schematic summarizing S. flexneri Oag chain length regulation in the absence of both PCP1 proteins (Wzz_SF and Wzz_pHS-2) (A), in the presence of Wzz_SF (B), and in the presence of Wzz_pHS-2 (C). (A) S. flexneri polymerizes LPS Oag through the repeated addition (red arrow) of single Oag RUs (green) at the reducing end of the growing polymer, but in the absence of both PCP1 proteins, it experiences a total loss of modal length regulation (11). (B) In the presence of Wzz_SF, the same mechanism of Oag polymerization is implemented, with repeated additions (red arrow) of single repeat units (RUs) at the reducing polymer end; however, polysaccharide growth is terminated once the chain length has reached between 10 and 17 RUs. (C) In the presence of Wzz_pHS-2, some undecaprenol-diphosphate (Und-PP)-linked nascent Oag chains are produced that are longer than the final short (S)-Oag modal length, and these are subsequently fed back into Oag polymerization and implemented as the nucleophile to add to the reducing end of the growing polymer chain. Successive addition of multiple nascent chains results in the production of the very-long (VL)-Oag modal length (with over 90 RUs) in a faster and more energy-efficient manner than the repeated polymerization of individual Oag RUs. This hypothesis requires further experimentation to be consolidated.

It has been well characterized that in the absence of both PCP1 proteins, S. flexneri experiences a loss of chain length regulation, leading to a loss of Oag modal length (Fig. 7A) (11). Similarly, in the presence of exclusively Wzz_SF, we see the production of solely the S-Oag length (10 to 17 RUs) (Fig. 7B). This is widely accepted to occur through the Wzy-mediated addition of single Oag RUs to the reducing end of a growing polysaccharide chain, which either extends into the bell-shaped Wzz_SF oligomeric structure or associates on its exterior surface (26). Based on our data, we see that disruption of the Wzy Cys moieties and the subsequent disruption of the NTD-CTD domain modulating Wzz_pHS-2 interaction results in a reduction of VL-Oag through impaired Wzz_pHS-2 interaction. Concurrently, we also observe that this does not affect S-Oag biosynthesis or the association of the shorter Und-PP-Oag (population A) but does result in an effect on the production of the longer Und-PP-Oag lengths (population B). We propose that, potentially, these slightly longer nascent Oag chains serve as the nucleophilic intermediates for VL-Oag polymerization. Specifically, we speculate that the VL-Oag length is achieved through the addition of multiples of these longer Und-PP-Oag nascent chains at the reducing end of a growing Oag polymer, rather than the individual additions of >90 single Oag RUs (Fig. 7C); however, this hypothesis requires further experimentation to be consolidated.

Overall, we identify here for the first time a region of Wzy that likely forms a unique binding site for Wzz_pHS-2, elucidating the key Wzy NTD and CTD moieties that form an intramolecular pocket modulating the interaction. Novel copurification data highlight that disruption of residues within this NTD-CTD pocket impairs interaction with Wzz_pHS-2 without affecting Wzz_SF binding, in addition to interrupting the polymerization of longer polysaccharide chains. This study provides a novel understanding of the molecular interaction of Wzy with the PCP1 proteins in the Wzy-dependent pathway, detecting for the first time the Wzy/Wzz/Und-PP-Oag complex. This work may be extended beyond S. flexneri to provide insight into the interactions between homologous proteins expressed by related species, especially members of Enterobacteriaceae that produce dual Oag chain length determinants.

MATERIALS AND METHODS

Ethics statement.

The anti-Wzz_SF and anti-Wzz_pHS-2 antibodies were produced under the National Health and Medical Research Council’s Australian Code for the Care and Use of Animals for Scientific Purposes (44), and this study was approved by the University of Adelaide Animal Ethics Committee.

Bacterial strains, growth media, and growth conditions.

All bacterial background strains used in this study are summarized in Table 1. Strains were grown routinely in lysogeny broth (LB) (10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCl) or on LB agar (15 g/L agar). Strains harboring pBCKs and/or pWSK29 constructs were grown in LB at 37°C with aeration for 16 h, subcultured 1:20 into fresh broth with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for induction, and grown for 4 h. Strains harboring pBAD18 constructs were grown as described above but subcultured instead with 0.2% (wt/vol) arabinose for induction. Antibiotics were used at the following concentrations: 100 μg/mL ampicillin (Amp), 25 μg/mL chloramphenicol (Cml), 50 μg/mL kanamycin (Kan), and 10 μg/mL tetracycline (Tet).

DNA methods.

Plasmids used in this study are outlined in Table 1. Plasmid constructs were extracted from Escherichia coli DH5α storage strains using a QIAprep spin miniprep kit (Qiagen). The preparation of electrocompetent cells and the electroporation method implemented were as previously described (13).

Colicin E2 sensitivity spot assay.

The colicin sensitivity spot assay was performed as previously described (4). Briefly, 5 × 10⁸ cells were spread onto 25-mL LB agar plates, which were then spotted with 5 μL of purified colicin E2 in Milli-Q (MQ) water in the following range of concentrations: 0.5, 1, 2, 4, 8, 16, 32, 64, 128, 256, and 512 μg/mL. Plates were incubated at 37°C for 16 h, and the MICs, here defined as the lowest colicin E2 concentration that generated a clear inhibition zone, were scored.

Complement sensitivity assay.

Complement sensitivity assays were performed as described previously (4, 31). Bacteria were grown and induced as described above, and 5 × 10⁸ cells were collected by centrifugation. Cells were then serially diluted in phosphate-buffered saline (PBS) to 1 × 10⁶ cells/mL in 1.5-mL Eppendorf tubes. Human serum (31) was subsequently added to the cell suspensions at a final concentration of 10% (vol/vol), and mixtures were incubated at 37°C without agitation. As a control, serum was heat inactivated by incubation at 56°C for 30 min. Viable cell counts were taken at 30-min intervals by plating 10 μL of cell suspensions on LB agar and are expressed as the percentages of the initial concentration (percent survival).

LPS SDS-PAGE and silver staining.

Bacteria were grown and induced as described above, and 1 × 10⁹ cells were harvested by centrifugation, resuspended in 2× lysis buffer (31), and heated at 100°C for 10 min prior to incubation with 0.5 mg/mL proteinase K (Sigma-Aldrich) for 2 h at 56°C. Electrophoresis using 15% SDS-PAGE gels was performed at 50 mA for 4 h. Silver staining of LPS was performed as described previously (31).

Protein SDS-PAGE.

Bacteria were grown and induced as defined above, and 5 × 10⁸ cells were harvested by centrifugation, resuspended in 2× sample buffer (45), and solubilized for 5 min at 37°C prior to SDS-PAGE for 1 h at 200 V using 12% polyacrylamide gels. Purified proteins were mixed 1:1 with 2× sample buffer (45).

Western immunoblotting.

Protein SDS-PAGE gels were transferred onto nitrocellulose membranes (Bio-Rad) for 1 h at 400 mA. Western transfers were then immunoblotted using primary monoclonal mouse anti-FLAG antibody (Sigma-Aldrich) at a 1:1,000 dilution, monoclonal mouse anti-His antibody (GenScript) at a 1:50,000 dilution, monoclonal mouse anti-MASFB antibody at a 1:500 dilution (35), polyclonal rabbit anti-Wzz_SF antibody at a 1:500 dilution (46), or polyclonal rabbit anti-Wzz_pHS-2 antibody at a 1:500 dilution (46). Primary antibodies were diluted in Tween-20 Tris-Buffered Saline (TTBS) containing 2.5% (wt/vol) skim milk. Western transfers were then immunoblotted with secondary goat anti-mouse horseradish peroxidase-conjugated antibody (KPL) used at a dilution of 1:30,000 in TTBS. Chemiluminescent reagent (Sigma-Aldrich) was then added to develop the Western immunoblot. Five microliters of SeeBlue plus2 prestained protein ladder (Invitrogen) was used as the standard for molecular mass.

Protein purification.

Cells were harvested from 1-L induced cultures by centrifugation (9,600 × g for 10 min at 4°C) (Avanti J-26 XPI centrifuge; Beckman Coulter) and resuspended in 1× sonication buffer (50 mM sodium phosphate, 500 mM NaCl, pH 7.0). Cell suspensions were then sonicated (Branson B15), unlysed cells and debris removed via low-speed centrifugation (3,500 rpm for 10 min at 4°C) (Labofuge 400 R centrifuge; Thermo Scientific), and the whole-membrane (WM) fraction collected by ultracentrifugation (250,000 × g for 1 h at 4°C) (Optima L-100 XP ultracentrifuge; Beckman Coulter). WM fractions were resuspended in 500 μL MQ water and 500 μL 2× solubilization buffer (50 mM sodium phosphate, 500 mM NaCl, 1% [wt/vol] DDM [N-dodecyl-β-d-maltopyranoside], pH 7.0) before overnight solubilization at 4°C with rotation. Soluble membrane fractions (SMFs) were then obtained via ultracentrifugation and incubated with either (i) Profinity IMAC Ni-charged resin (Bio-Rad) or (ii) anti-DYKDDDDK G (FLAG) affinity resin (GenScript). Eluted purified protein samples were resuspended 1:1 in 2× sample buffer and solubilized at 37°C for 5 min, and 10-μL amounts were loaded for SDS-PAGE.

IMAC Ni-charged resin purification.

SMFs were incubated for 2 h at room temperature (RT) with 100 μL of Profinity IMAC Ni-charged resin, preequilibrated using equilibration buffer (50 mM sodium phosphate, 500 mM NaCl, 0.004% [wt/vol] DDM, pH 7.0). Loaded His resin was then repeatedly washed using wash buffer (50 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, 0.004% [wt/vol] DDM, pH 7.0) before His-tagged protein was eluted into a final 200 μL of elution buffer (50 mM sodium phosphate, 500 mM NaCl, 500 mM imidazole, 0.004% [wt/vol] DDM, pH 7.0).

Anti-FLAG affinity resin purification.

SMFs were incubated for 2 h at RT with 100 μL of anti-DYKDDDDK G (FLAG) affinity resin, preequilibrated using equilibration buffer (50 mM Tris-HCl, 150 mM NaCl, 0.02% [wt/vol] DDM, pH 8.0). Loaded FLAG resin was then repeatedly washed using wash buffer (50 mM Tris-HCl, 150 mM NaCl, 0.02% [wt/vol] DDM, pH 8.0), before FLAG-tagged proteins were eluted into a final 200 μL of elution buffer (50 mM Tris-HCl, 150 mM NaCl, 0.02% [wt/vol] DDM, pH 8.0, 100 μg/mL FLAG peptide [Sigma-Aldrich]).

Colloidal blue SDS-PAGE staining.

Proteins separated by 12% (wt/vol) SDS-PAGE were visualized through incubation with brilliant blue G colloidal concentrate (Sigma-Aldrich) solution at room temperature with overnight shaking. Gels were destained with repeated washes of MQ water on the following day.

In silico structural analyses.

The AlphaFold (34) structure for S. flexneri Wzy (Rfc) (accession number P37784) was analyzed using the software UCSF Chimera version 1.14 (http://www.cgl.ucsf.edu/chimera/) (47). The distances between amino acid side chains were measured using the built-in “structure measurements” tool.

Statistical analyses.

Experiments in this study were performed in triplicate. Data are graphically represented as the mean value ± standard error of the mean (SEM), and statistical analyses were performed using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA). Student’s t test, one-way analysis of variance (ANOVA), and two-way ANOVA (both with multiple comparisons) were the statistical comparisons used to analyze the data as required. Results were considered significant where the P value was <0.05 (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, non-statistically significant).

Data availability.

All data are contained within the article.