Interdependence of Shigella flexneri O Antigen and Enterobacterial Common Antigen Biosynthetic Pathways

ABSTRACT

Outer membrane (OM) polysaccharides allow bacteria to resist harsh environmental conditions and antimicrobial agents, traffic to and persist in pathogenic niches, and evade immune responses. Shigella flexneri has two OM polysaccharide populations, being enterobacterial common antigen (ECA) and lipopolysaccharide (LPS) O antigen (Oag); both are polymerized into chains by separate homologs of the Wzy-dependent pathway. The two polysaccharide pathways, along with peptidoglycan (PG) biosynthesis, compete for the universal biosynthetic membrane anchor, undecaprenyl phosphate (Und-P), as the finite pool of available Und-P is critical in all three cell wall biosynthetic pathways. Interactions between the two OM polysaccharide pathways have been proposed in the past where, through the use of mutants in both pathways, various perturbations have been observed. Here, we show for the first time that mutations in one of the two OM polysaccharide pathways can affect each other, dependent on where the mutation lies along the pathway, while the second pathway remains genetically intact. We then expand on this and show that the mutations also affect PG biosynthesis pathways and provide data which supports that the classical mutant phenotypes of cell wall mutants are due to a lack of available Und-P. Our work here provides another layer in understanding the complex intricacies of the cell wall biosynthetic pathways and demonstrates their interdependence on Und-P, the universal biosynthetic membrane anchor.

IMPORTANCE Bacterial outer membrane polysaccharides play key roles in a range of bacterial activities from homeostasis to virulence. Two such OM polysaccharide populations are ECA and LPS Oag, which are synthesized by separate homologs of the Wzy-dependent pathway. Both ECA and LPS Oag biosynthesis join with PG biosynthesis to form the cell wall biosynthetic pathways, which all are interdependent on the availability of Und-P for proper function. Our data show the direct effects of cell wall pathway mutations affecting all related pathways when they themselves remain genetically unchanged. This work furthers our understanding of the complexities and interdependence of the three cell wall pathways.

INTRODUCTION

Shigella flexneri is a Gram-negative bacterium which expresses two distinct populations of outer membrane (OM)-bound polysaccharides—lipopolysaccharide (LPS) and enterobacterial common antigen (ECA). Composed of three distinct domains that include lipid A, the proximal membrane anchor, the inner and outer core sugars, and a distal chain of O antigen (Oag) polysaccharides, LPS is known to play crucial roles in the viability, pathogenesis, and immune evasion of Gram-negative pathogens (1, 2). ECA exists in two membrane-associated forms, ECA_pg the most abundant, and ECA_lps, which occurs in some strains lacking O antigen, as well as a periplasmically restricted cyclic form, ECA_cyc (3). ECA is known to play roles in maintaining OM homeostasis and providing resistance to bile salt (4, 5). Peptidoglycan (PG) biosynthesis along with the two OM polysaccharide pathways form the three cell wall biosynthetic pathways whose polymerases belong to the shape, elongation, division, and sporulation (SEDS) protein family, and all rely on undecaprenyl phosphate (Und-P) as their biosynthetic lipid-linked anchor (6).

The S. flexneri Y serotype Oag comprises tetrasaccharide repeat units (RUs) which contain a single N-acetylglucosamine (GlcNAc) and three rhamnose (Rha) residues (7). ECA, universal in all Enterobacterales, comprises trisaccharide RUs containing N-acetylglucosamine (GlcNAc), N-acetyl-d-mannosaminuronic acid (ManNAcA), and 4-acetamido-4,6-dideoxy-d-galactose (Fuc4NAc) (8). Once completed, RUs from both pathways are translocated across the inner membrane (IM) and polymerized into linear chains by their separate homologs of the Wzy-dependent pathway (Fig. 1). The Wzy-dependent pathway is the most common bacterial polysaccharide biosynthetic pathway and consists of three proteins, Wzx translocase, Wzy polymerase, and Wzz polysaccharide copolymerase, which controls polymerization by Wzy (9). In S. flexneri, the proteins associated with the Wzy-dependent pathway for ECA are denoted as WzxE, WzyE, and WzzE, and for Oag these are termed WzxB, WzyB, and WzzB.

FIG 1.

OM polysaccharide biosynthesis in Shigella flexneri. Mutations in the genes post-generation of ECA and Oag lipid II lead to the accumulation of lipid-linked intermediates and Und-P sequestration. Once translocated across the IM, Und-P is released by either Wzy or the final glycosyltransferase from each pathway. Undecaprenyl is first generated through de novo synthesis and then acted upon by UppS to yield Und-PP (29). UppP then removes a phosphate group to yield Und-P, releasing Und-P on the periplasmic side of the IM (30). Und-P is translocated back across the IM to the cytoplasmic side, where then it can be acted on by MraY or WecA, which commits Und-P to either PG or OM polysaccharide biosynthesis, respectively, until it is released from the pathways (32). WecA yields Und-PP-GlcNAc, which is a common lipid-linked intermediate for both ECA and Oag biosynthesis. It is acted upon by WecG or RfbF to yield ECA lipid-II or Oag lipid-II, committing the Und-P moiety to ECA or Oag biosynthesis, respectively (8, 48). Addition of sugar residues by WecF and RfbG yields ECA lipid-III, Oag lipid-III, and lipid-IV, completing the ECA and Oag RU (8, 48). RUs are then acted upon by separate homologs of the Wzy-dependent pathway (Wzx flippase, Wzy polymerase, and Wzz polymerase copolymerase, denoted WzxB, WzyB, WzzB and WzxE, WzyE, WzzE for Oag and ECA biosynthesis, respectively), at which point Und-PP is predominately released from the OM polysaccharide pathways (9). Ligation of the complete polysaccharide chain onto the final lipid carrier molecule also releases Und-PP from these pathways. A similar series of biosynthetic steps also occurs in PG biosynthesis, where MurJ and FtsW release PG RUs from Und-PP (49). Und-PP is again acted upon by UppP to yield Und-P, which is available for PG or OM polysaccharide biosynthesis once again (30). Oag, O antigen; X, LPS core-sugars; Und-PP, undecaprenyl pyrophosphate. Symbol nomenclature for glycans adapted from 50 and 51.

Cell wall biosynthesis commences on the cytoplasmic side of the IM where the glycosyltransferases WecA and MraY act on Und-P to yield the foundations of OM polysaccharide biosynthesis and PG biosynthesis, respectively (3, 10, 11). Once acted upon, each Und-P molecule is committed to either one of the three biosynthetic pathways until completed RUs are cleaved off by FtsW/RodA or WzyE, WzyB, WaaL and the unknown final ECA glycosyltransferase in the process of polysaccharide biosynthesis (6, 11, 12). Lastly, undecaprenyl pyrophosphate (Und-PP) is then recycled via dephosphorylation, primarily by UppP, to yield Und-P, which then re-enters the finite pool of available Und-P (Fig. 1) (13).

It is the requirement of Und-P in all of the three cell wall pathways which fundamentally interlinks them and causes their interdependence upon one another. The interdependence of each of the OM polysaccharide pathways with PG biosynthesis has been shown separately, where the majority of our current understandings originate from the research performed by Jorgenson et al., who showed that various mutations in the LPS Oag and ECA biosynthetic pathways of Escherichia coli can influence PG synthesis (14, 15). In addition to this, Marolda et al. showed that components of the two Wzy-dependent pathways could supplement one another, but only under very specific circumstances (16). Recently, Leo et al. demonstrated that the Wzz proteins of both OM polysaccharide pathways could complement each other and partially restore polysaccharide modal length control (17).

In this study, we demonstrate the interdependence of all three cell wall pathways with one another. Through the use of wzy mutants, we show for the first time that restricting the rate of Und-P recycling is sufficient to induce a classical mutant phenotype associated with cell wall mutants, resulting in an altered cellular morphology (14, 15). Additionally, we show the direct effects of sequestering intermediates from related OM polysaccharide biosynthetic pathways through the use of anti-ECA Western immunoblotting and LPS silver-stained gels. This study reveals another layer of complexity in investigating cell wall mutants and reveals the extent of the interdependence between the three cell wall pathways.

RESULTS

Investigating the biosynthetic effects of the ΔwzyE mutation.

Research has shown that the sequestering of undecaprenyl phosphate (Und-P) in the form of dead-end biosynthetic precursors of interrupted polysaccharide pathways leads to altered cellular morphology (14, 15). As the two major OM polysaccharides of Shigella flexneri, Oag and ECA, share the same initial biosynthetic precursor, Und-PP-GlcNAc (Fig. 1), the OM polysaccharide profile of S. flexneri PE860 ΔwzyE was analyzed using anti-ECA Western immunoblotting and silver-stained SDS-PAGE (Fig. 2a and b). The wzyE mutant was unable to polymerize any detectable ECA as seen by a lack of a ladder banding pattern (Fig. 2a, lane 2), and when complemented with pWzyE, the mutant was able to partially polymerize ECA, observed as an ECA ladder banding pattern (Fig. 2a, lane 4).

FIG 2.

Analysis of the PE860 ΔwzyE phenotype. (a) Anti-ECA Western immunoblot showing ECA banding profiles of PE860, an isogenic ΔwzyE mutant denoted as ΔwzyE, and that mutant harboring pBAD33 or pWzyE as indicated. Mid-exponential-phase cells were collected (1 × 10⁹ cells) and lysed in lysis buffer in the presence of proteinase K, and following SDS-PAGE and Western transfer, the membrane was probed with polyclonal rabbit anti-ECA antibodies. SeeBlue Plus2 prestained protein ladder (Invitrogen) was used as a molecular mass standard. (b) Analysis of LPS profiles of PE860, an isogenic ΔwzyE mutant denoted as ΔwzyE, and that mutant harboring pBAD33, pWzyE, pWKS30, pUppS, or pWecA as indicated. Samples were made as described above and electrophoresed on an SDS-15% (wt/vol) PAGE gel and silver stained. The numbers of Oag RUs are indicated on the left-hand side. (c) Analysis of polymerization of smooth Oag (S-Oag) via densitometry (Image Lab). The degree of polymerization of S-Oag is represented as the densitometry of Oag RUs from 10 to 17 as a percentage relative to the parent. The data represent 3 independent experiments with SEM shown, and significance was calculated with independent Student’s t tests comparing the ΔwzyE mutant with relevant strains. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant result. (d and e) Cellular measurements via phase-contrast microscopy. 10 µL of mid-exponential-phase culture was dried on a microscope slide, mounted with Moviol, and sealed. Cells were imaged and cell lengths measured. Scale bars = 1 µm. The data represent 150 individual cells with SEM shown, and significance was calculated with independent Student’s t tests comparing the PE860 ΔwzyE mutant with relevant strains. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; and ns, non-statistically significant result. The box covers the upper and lower quartiles of the data, with the bisecting bar indicating the median cell length, whereas the whiskers indicate the maximum and minimum cell lengths. (f) Analysis of growth of PE860, an isogenic ΔwzyE mutant denoted as ΔwzyE, and that mutant harboring pBAD33 or pWzyE as indicated. Overnight culture was subcultured 1/20 in a 96-well tray with OD₆₀₀ absorbance readings taken every 20 min. The data represent averages of time points from 3 independent biological replicates.

Surprisingly, a decrease in the smooth Oag (S-Oag) polymerization was observed when comparing the LPS of the PE860 parent to that of the wzyE mutant (Fig. 2b, lanes 1 and 2), and the mutant phenotype was rescued to wild type when complemented with pWzyE (Fig. 2b, lane 4). This reduction in S-Oag polymerization was further quantified via densitometry and showed a 61% statistically significant decrease in abundance between the parent and the wzyE mutant, as well as between the wzyE mutant and the complemented mutant with pWzyE (P < 0.001). No statistically significant difference in Oag between PE860 and the wzyE mutant complemented with pWzyE was observed (Fig. 2c).

As blocking ECA production leads to an altered cellular morphology due to PG biosynthesis disruption (14), we investigated whether wzyE mutants also exhibited altered morphologies. Cellular lengths were measured by phase-contrast microscopy (as described in Materials and Methods), and a statistically significant difference was observed in cell size when comparing PE860 and the wzyE mutant, as well as the wzyE mutant complemented with pWzyE, which reverted back to the wild-type length (P < 0.001) (Fig. 2d and e).

This suggested that the theoretical lack of available Und-P in the wzyE mutants affected PG biosynthesis in a similar manner as previously described (14). Growth curves were then performed to investigate if the wzyE mutant exhibited altered growth, presumably due to the lack of available Und-P compared to the parent (Fig. 2f). The wzyE mutants showed reduced growth, reaching a lower optical density at 600 nm (OD₆₀₀) compared to the parent; however, this phenotype was rescued when complemented with pWzyE. As absorbance readings are dependent on particle size, a CFU count was performed and showed that in addition to having a smaller cell length, the wzyE mutant produced fewer CFU than PE860 at each time point investigated (see Fig. S1 in the supplemental material).

Investigating the biosynthetic effects of the ΔwzyB mutation.

Considering the pleiotropic phenotype observed for the wzyE mutant, we decided to investigate if the Oag biosynthetic pathway, through the use of a wzyB mutant, was similarly affected. First, the OM polysaccharide profile of the wzyB mutant was assessed using silver-stained SDS-PAGE, where as expected, the wzyB mutant produced LPS with a single Oag RU, seen as a single band (Fig. 3a, lane 2). Introduction of pWzyB was able to partially complement the wzyB mutant (Fig. 3a, lane 4).

FIG 3.

Analysis of the PE860 ΔwzyB phenotype. (a) Analysis of LPS profiles of PE860, an isogenic ΔwzyB mutant denoted as ΔwzyB, and that mutant harboring pBCKs+ or pWzyB as indicated. Mid-exponential-phase cells were collected (1 × 10⁹ cells) and lysed in lysis buffer in the presence of proteinase K. Samples were then electrophoresed on an SDS-15% (wt/vol) gel and silver stained. The numbers of Oag RUs are indicated on the left-hand side. (b) Anti-ECA Western immunoblot of PE860, an isogenic ΔwzyB mutant denoted as ΔwzyB, and that mutant harboring pBCKs+, pWzyB, pWKS30, pUppS, or pWecA as indicated. Samples were made as described above, and following SDS-PAGE and Western transfer, the membrane was probed with polyclonal rabbit anti-ECA antibodies. SeeBlue Plus2 prestained protein ladder (Invitrogen) was used as a molecular mass standard. (c) Analysis of polymerization of ECA via densitometry. The degree of polymerization of ECA is represented as the densitometry of ECA RUs as a percentage relative to the parent. The data represent 9 independent experiments with SEM shown, and significance was calculated with independent Student’s t tests comparing the ΔwzyB mutant with relevant strains. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant result. (d and e) Cellular measurements via phase-contrast microscopy. 10 µL of mid-exponential-phase culture was dried on a microscope slide, mounted with Moviol, and sealed. Cells were imaged and cell lengths measured. Scale bars = 1 µm. The data represent 150 individual cells with SEM shown, and significance was calculated with independent Student’s t tests comparing the ΔwzyE mutant with relevant strains. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant result. The box covers the upper and lower quartiles of the data, with the bisecting bar indicating the median cell length, whereas the whiskers indicate the maximum and minimum cell lengths. (f) Analysis of growth of PE860, an isogenic ΔwzyB mutant denoted as ΔwzyB, and that mutant harboring pBCKs+ or pWzyB as indicated. Overnight culture was subcultured 1/20 in a 96-well tray with OD₆₀₀ absorbance readings taken every 20 min. The data represent averages of time points from 3 independent biological replicates.

ECA immunoblotting showed a reduction of ECA banding intensity between PE860 and the wzyB mutant (Fig. 3b, lanes 1 and 2). When complemented with pWzyB, the wzyB mutant displayed an ECA banding pattern similar to that of the parent (Fig. 3b, lanes 2 and 4). We then investigated if similar to the wzyE mutant, the wzyB mutant also had altered cellular morphologies that could be rescued to wild-type morphology by complementation. Cell length measurements made by phase-contrast microscopy showed that the wzyB mutant cells were significantly shorter than those of the parent (P < 0.001). When complemented with pWzyB, their lengths reverted back to wild-type (P < 0.001) (Fig. 3d and e).

Growth was then assessed by growth curves which showed that, similar to the wzyE mutant (Fig. 2f), the wzyB mutant grew to a lower OD₆₀₀ than the parent (Fig. 3f). Complementing the wzyB mutant with pWzyB restored the growth of the wzyB mutant to near-parental levels (Fig. 3f). We likewise performed a CFU count on the wzyB mutant, which showed that as with the wzyE mutant, the wzyB mutant produced fewer CFU than PE860 at each time point investigated (Fig. S1).

This supported the hypothesis that the two major OM polysaccharide populations of S. flexneri, ECA and LPS Oag, are fundamentally linked at the biosynthetic level, where disruption of one of the biosynthetic pathways interferes with the other, most likely due to the sequestration of Und-P. Additionally, the results also suggested that PG biosynthesis was also affected, as Und-P is required in the biosynthesis of PG biosynthetic intermediates (Fig. 1).

Improving the cellular undecaprenyl pool rescues mutant wzy phenotypes.

To investigate if the pleiotropic wzy mutant phenotypes were due to a lack of available Und-P, wecA, which generates ECA/Oag lipid I and commits Und-P to OM polysaccharide biosynthesis in S. flexneri, and uppS, which is responsible for de novo synthesis of Und-PP, were ectopically expressed. The constructs pWKS30-UppS-HA and pWKS30-WecA-HA, denoted as pUppS and pWecA, were generated, transformed into the wzyE and wzyB mutants, and assessed to determine if either theoretically increasing or decreasing the cellular pool of Und-P through the expression of uppS either alleviated or exacerbated their pleiotropic phenotypes.

In the wzyE mutant, LPS analysis showed a statistically significant restoration in Oag banding intensity with either pUppS or pWecA (P < 0.0001 and P < 0.01, respectively) (Fig. 1c and 2b, lanes 2, 6, and 7). However, cell measurements showed that only complementation with pUppS, which would theoretically increase the cellular pool of Und-P, and not pWecA, was able to restore the wzyE mutant length back to the wild type (P < 0.0001 and nonsignificant, respectively) (Fig. 2d and e). Furthermore, growth curves showed that upon introduction of pUppS, the wzyE mutant was able reach a higher OD₆₀₀ of 0.7, whereas the introduction of pWecA resulted in the wzyE mutant reaching a lower OD₆₀₀ of 0.5 after 10 h of growth, with the wzyE mutant reaching an OD₆₀₀ of 0.6 (Fig. 4a).

FIG 4.

Effects of increasing or decreasing the cellular availability of Und-P on Δwzy mutant growth. (a and b) Analysis of growth of strains (a) PE860, an isogenic ΔwzyE mutant denoted as ΔwzyE, and that mutant harboring pWKS30, pWecA, or pUppS and (b) PE860, an isogenic ΔwzyB mutant denoted ΔwzyB, and that mutant harboring pWKS30, pWecA, or pUppS as indicated. Overnight cultures of the above-described strains were subcultured 1/20 in a 96-well tray with OD₆₀₀ absorbance readings taken every 20 min. The data represent averages of time points from 3 independent biological replicates.

Similar to the wzyE mutant, the wzyB mutant also showed partial rescue from its mutant wzy phenotypes upon introduction of pUppS and pWecA. The wzyB mutant displayed a statistically significant restoration in ECA banding intensity when transformed with either pUppS or pWecA (Fig. 3b, lanes 2, 6, and 7 and Fig. 3c). Cell measurements revealed that, similar to the wzyE mutant, the wzyB mutant length was only restored back to a wild-type length by pUppS, and not pWecA (P < 0.0001 and P < 0.001, respectively) (Fig. 3d and e). Growth analysis further showed that the addition of pUppS did not improve the growth of the wzyB mutant, and introduction of pWecA led to a decrease in growth relative to that of the wzyB mutant, reaching an OD₆₀₀ of 0.68 and 0.58 after 10 h of growth, respectively, compared to the wzyB mutant, reaching an OD₆₀₀ of 0.67 (Fig. 4b).

These results further supported the hypothesis that the three pathways are fundamentally linked through the common use of Und-P in their biosynthetic pathways. However, a plausible alternative explanation of the cause of the phenotypes was that the loss of OM ECA or LPS Oag was indirectly impacting the other unaffected polysaccharide pathway. To investigate this, we subsequently analyzed whether the theoretical reduction in available Und-P was the cause of the observed phenotypes or if the loss of the OM polysaccharides themselves was responsible.

The sequestration of Und-P and not the loss of OM polysaccharides is correlated with the wzy mutant phenotypes.

We investigated the impact of polysaccharide-specific mutations in additional genes involved in the biosynthesis of ECA and LPS Oag-specific biosynthetic intermediates, wecC and rmlD, respectively (Fig. 1). These two genes were selected because they are involved in the biosynthesis of ECA and Oag lipid II, respectively, and hence, once mutated, would prevent the formation of the substrates required by the second glycosyltransferases of each pathway, WecG and RfbF, which consequently would prevent the sequestration of Und-P. Additionally, we investigated the impact of a wecA mutant which is involved in the biosynthesis of lipid I, the common precursor for both OM polysaccharide biosynthetic pathways (Fig. 1).

The wecC mutant, as expected, was unable to produce ECA (Fig. 5b, lane 7); however, unlike the wzyE mutant (Fig. 5a, lane 3), it had an increase in Oag banding intensity compared to the parent PE860 (Fig. 5a, lane 7). The rmlD mutant showed a total loss of LPS Oag banding (Fig. 5a, lane 8) but, unlike the wzyB mutant (Fig. 5b, lane 5), showed an increased ECA banding intensity compared to PE860 (Fig. 5b, lane 8). As expected, the wecA mutant showed a total loss of both OM polysaccharides, seen as a lack of banding in both the ECA immunoblot and the SDS-PAGE silver stain gel, respectively (Fig. 5a and b, lane 9). The cellular morphologies of the wecC, rmlD, and wecA mutants were then assessed. This showed that the wecA, wecC, and rmlD mutants were significantly longer than the wzyE and wzyB mutants (Fig. 5c and d). Additionally, when comparing the parent with the wecC, rmlD, and wecA mutants, a statistically significant reduction in cell length (P < 0.0001) was observed, with no statistically significant difference in cell length observed when comparing the rmlD mutant with the wecA mutant (Fig. 5c and d). Growth analysis showed that the wecC mutant grew at a rate similar to the parental strain, reaching an OD₆₀₀ of 0.65 after 10 h of growth, whereas the rmlD and wecA mutants were only able to reach an OD₆₀₀ of 0.51 and 0.55, respectively, after 10 h of growth (Fig. 5e and f).

FIG 5.

The impact of wecA, wecC, and rmlD mutations and tunicamycin treatment on ECA, Oag, and growth. (a) Analysis of LPS profiles of strains PE860, treated with and without tunicamycin, PE860 ΔwzyE, denoted ΔwzyE, treated with and without tunicamycin, PE860 ΔwzyB, denoted as ΔwzyB, treated with and without tunicamycin, and PE860 mutants ΔwecC, ΔrmlD, and ΔwecA as indicated. Cells were grown to the mid-exponential phase and collected (1 × 10⁹ cells) and lysed in lysis buffer in the presence of proteinase K. Samples were then electrophoresed on an SDS-15% (wt/vol) gel and silver stained. Treatment with tunicamycin is denoted (+ Tnc). The numbers of Oag RUs are indicated on the left-hand side. (b) Anti-ECA Western immunoblot of the same strains as described above. Samples were made as described above, and following SDS-PAGE and Western transfer, the membrane was probed with polyclonal rabbit anti-ECA antibodies. SeeBlue Plus2 prestained protein ladder (Invitrogen) was used as a molecular mass standard. (c and d) Cellular length measurements via phase-contrast microscopy. 10 µL of mid-exponential-phase culture was dried onto a microscope slide, mounted with Moviol, and sealed. Cells were imaged and cell lengths measured. Scale bars = 1 µm. The data represent 150 individual cells with SEM shown, and significance was calculated with independent Student’s t tests comparing the relevant strains. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant result. The box covers the upper and lower quartiles of the data, with the bisecting bar indicating the median cell length, whereas the whiskers indicate the maximum and minimum cell lengths. Treatment with tunicamycin is denoted as (+ Tnc). (e and f) Analysis of growth of strains (e) PE860 and PE860 mutants ΔwzyE, ΔwecA, and ΔwecC and (f) PE860 and PE860 mutants ΔwzyB, ΔwecA, and ΔrmlD as indicated. Overnight cultures of the above-described strains were subcultured 1/20 in a 96-well tray with OD₆₀₀ absorbance readings taken every 20 min. The data represent averages of time points from 3 independent biological replicates.

Overall, the data supported the hypothesis that if one OM polysaccharide was absent, the reciprocal OM polysaccharide system could function without interference and that the cause of the wzy pleiotropic mutant phenotypes was likely not due to a lack in reciprocal OM polysaccharide (Fig. 1). Additionally, growth curve data suggested that there was an effect on the growth of strains lacking Oag, as the rmlD mutant had a similar growth profile as the wecA mutant, presumably due to the activation of the Rcs pathway (18).

Therefore, to further investigate if the lack of Und-P was the cause of the wzy mutant phenotypes, tunicamycin, an inhibitor of WecA (19), was used to treat both wzy mutants to prevent lipid I production (Fig. 1). Following treatment with tunicamycin, the parent and the mutants, wzyE or wzyB, showed no detectable OM polysaccharide production (Fig. 5a and b), as expected. However, microscopy analysis revealed that post-treatment with tunicamycin, the two wzy mutants showed a statistically significant restoration in cellular length (P < 0.0001) (Fig. 5c and d), which supported our hypothesis that the lack of available Und-P was most likely the cause of the pleiotropic phenotypes of the wzy mutants.

DISCUSSION

Jorgenson et al. demonstrated that E. coli K-12 mutants of the OM polysaccharide biosynthetic pathways LPS Oag and ECA displayed morphological abnormalities due to the sequestration of dead-end biosynthetic intermediates within the cell (14, 15). Here, we demonstrate that the two OM polysaccharide pathways of S. flexneri are intimately linked through their use of the universal biosynthetic membrane anchor, Und-P.

Our study demonstrates novel indirect cross talk between the two OM polysaccharide biosynthetic pathways, LPS Oag and ECA. Direct interactions between the two Wzy-dependent pathways has been previously speculated on, due to the structural similarities between the Wzz protomers from each pathway as well as molecular similarities in the substrates they translocate and polymerize (16, 20). Most recently it was shown that interchanging the transmembrane (TM) regions of WzzB and WzzE, most importantly TM2, was sufficient to restore polysaccharide modal length control in S. flexneri wzzB and wzzE mutants complemented with WzzB_ETM2 or WzzE_BTM2, respectively (17). Leo et al. further demonstrated that WzyB could natively pull down WzzE, as well as the chimeric Wzz proteins, which had only been demonstrated with WzyB and WzzB at the time of publication (17).

In our study, we have shown that the two OM polysaccharide pathways interact indirectly through perceived competition for Und-P, where for the first time, a drop in abundance of the reciprocal OM polysaccharide ECA and LPS Oag was observed via anti-ECA Western immunoblotting and LPS silver staining (Fig. 2b and 3b) when the wzyB/wzyE homologs, respectively, were deleted. The abundance of OM polysaccharide was restored upon complementation of the wzy mutants or via ectopic expression of pUppS or pWecA, which is in accordance with experimental observations from other OM biosynthetic mutants (14–16).

It is well documented that mutations affecting the ECA and Oag biosynthetic pathways across Enterobacteriales produce abnormal cellular morphologies (14, 15, 21–24). Specifically, in these studies, mutations in E. coli K-12 which led to biosynthetic dead-ends were used. Similar to the mutant phenotypes observed in E. coli K-12, the S. flexneri wzy mutants displayed altered cell morphology, albeit with a shorter cell length (Fig. 2d and e and 3d and e), unlike those previously observed mutants which produced elongated cells (14, 15).

A plausible explanation as to why the wzy mutants did not display the phenotypes observed by Jorgenson et al. is that instead of trapping Und-P in dead-end intermediates, wzy mutants only greatly restrict the recycling of Und-P from the lipid-linked intermediates as they produce OM polysaccharides which consist of a single RU (Fig. 1). This is seen in Fig. 5a, as when treated with tunicamycin, wzyB mutants lose their single RU and express an LPS profile similar to that of a wecA or rmlD mutant (Fig. 5a, lanes 8 and 9). This is because Wzys are not the only enzymes which release Und-P from the intermediates, but they do so alongside WaaL and the final unknown ECA ligase equivalent. As both WaaL and the putative ECA ligase are present in the cell, the complete sequestering of Und-P does not occur. Additionally, as stated above, the strains commonly used when these phenotypes were investigated are derived from E. coli K-12 (14, 15, 21–24).

Isolated in 1922, E. coli K-12 is a cycled, laboratory strain which historically was exposed to mutagens and unintentionally became rough and has existed as a rough strain for an extended period of time (25–28). As secondary mutations occur frequently in OM polysaccharide mutants (14, 15), it is plausible that studies of E. coli K-12 are unable to fully represent the intricacy of the OM biosynthetic pathways, which thus cautions direct comparisons of these OM pathways between rough and smooth LPS backgrounds such as that of S. flexneri.

The restriction of Und-P from related pathways is a common cause of the pleiotropic phenotypes seen in cell wall mutants and is frequently corrected by the supplementation of Und-PP via expression of undecaprenyl pyrophosphotase synthase (UppS) (14, 15). UppS synthesizes Und-PP in cells, which is then acted upon by UppP to yield Und-P, where its overexpression was shown to correct the abnormal morphologies of E. coli K-12 waaC, waaL, and wecE mutants (14, 15, 29, 30). This is consistent with our findings where the expression of pUppS was able restore wild-type ECA and LPS banding (Fig. 2b and c and 3b and c), reverting the wzy mutants back to wild-type length (Fig. 2d and e and 3d and e) as well as improving the growth of the wzyE mutant (Fig. 4a).

Overexpression of WecA is also a common tool to drive OM polysaccharide intermediates into dead-end pathways to exacerbate and investigate mutant phenotypes; however, distinctions cannot be made that the affects seen are due to the accumulation of intermediates or sequestering Und-P; rather, both affects are responsible (14–16, 31). The addition of pWecA did not decrease the cellular lengths of the wzy mutants (Fig. 2d and e and 3d and e). However, it did alter the growth characteristics of the cells, whereby they were unable to reach a similar OD₆₀₀ as the wzy mutants (Fig. 4a and b), which is presumably due to Und-P being sequestered away from PG synthesis. ECA and LPS Oag production was restored to near wild type, which was expected, as wecA drives the OM polysaccharide biosynthetic pathways forward (Fig. 2a and b and 3a and b) (32).

Additionally, we used tunicamycin to inhibit WecA and block OM polysaccharide biosynthesis. Tunicamycin is a known cell wall glycosyltrasferase inhibitor, preventing the transfer of GlcNAc and MurNAc (N-Acetylmuramic acid)-pentapeptides onto polyprenylphosphate lipid carriers (19, 33). Despite targeting MraY as well as WecA, the inhibitory concentration for WecA was shown to be 1,000 times less than that of MraY, allowing us to selectively inhibit WecA by using MraY subinhibitory concentrations (34). As expected, treatment with tunicamycin inhibited the production of ECA or Oag in the wzy mutants (Fig. 5a and b). Interestingly, the wzy mutants displayed longer cell lengths when treated with tunicamycin than when untreated, which supports our hypothesis that the wzy mutant phenotypes may be partially caused by a lack of available of Und-P for PG biosynthesis (Fig. 5c and d). Overall, as the results from the theoretical expression of UppS were sufficient to rescue the pleiotropic phenotypes of the wzy mutants and as a similar phenotypic rescue was observed when the wzy mutants were treated with tunicamycin, we believe the results support our hypothesis that the pleiotropic phenotypes of the wzy mutants were due to a lack of available Und-P, as it was sequestered in biosynthetic intermediates.

Lastly, we showed that not all OM polysaccharide mutations lead to the mutant phenotypes associated with cell wall mutants. The deletion of wecC and rmlD, which cause the loss of ECA and LPS Oag, respectively, appeared to increase the OM polysaccharide production of the other pathway compared to the parent (Fig. 5a and b). Length measurements also supported our hypothesis that both the wecC and rmlD mutants possibly did not sequester Und-P from PG synthesis, as each mutant possessed a statistically significantly longer cell length compared to the wzy mutants (P < 0.0001) (Fig. 5c and d). These results suggested that if one of the OM polysaccharide pathways is blocked in such a way that the cell does not theoretically sequester Und-P and/or biosynthetic intermediates, then Und-P can be funneled into the other unaffected OM polysaccharide pathway, leading to an increase in that OM polysaccharide population. Additionally, we found a reduction in the final OD₆₀₀ by the rmlD and wecA mutants compared to that of the wecC mutant (Fig. 5e and f) as well by the wzyB mutant expressing pUppS (Fig. 4b); we speculate that this can be explained by the induction of the Rcs and Cpx pathways.

The Rcs and Cpx membrane stress response pathways are known to contribute to the phenotypes of cell wall mutants (35–38) and have been implicated in their pleiotropic phenotypes across various backgrounds—Serratia marcescens (21, 39), E. coli K-12 (14, 15, 40–42), and Proteus mirabilis (18, 43). It has been shown that the lack of Oag due to a waaL mutation in P. mirabilis was sufficient to induce the Rcs pathway (18, 43). Therefore, it is plausible that in S. flexneri, the lack of Oag itself can induce the Rcs pathway, leading to the observed reduced growth in the rmlD, wecA, and wzyB mutants, including the wzyB mutant when complemented with pUppS (Fig. 4b and 5f), Whereas it has been indicated specifically that the mutations which accumulate lipid-linked intermediates, and not the lack of ECA itself, stimulate the Rcs pathway (14). Therefore, as wecC mutants do not accumulate lipid-linked intermediates (Fig. 1), the Rcs pathway is not stimulated in the wecC mutant, allowing for growth comparable to that of the parent (Fig. 5e).

In conclusion, the results here reveal another layer of complexity in our understanding of the interactions between the cell wall biosynthetic pathways. The critical link between LPS, ECA, and PG biosynthesis lies in their combined reliance on Und-P as the cells’ universal membrane anchor (Fig. 1). As such, cell wall mutants derived from these pathways are likely to provide unclear mutant phenotypes, influenced by their related and unrelated pathways, such as Rcs and Cpx pathways. This also calls into question previously published data in which cell wall mutants were investigated, as alternative plausible explanations may now better explain the phenomena observed. These findings provide a greater understanding of the complexities and interlacings of cell wall biosynthetic pathways and provide an awareness of the possible entanglements which can occur when investigating cell wall mutants.

MATERIALS AND METHODS

Ethics statement.

The ECA antibodies were produced under the National Health and Medical Research Council Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and was approved by the University of Adelaide Animal Ethics Committee.

Bacterial strains, growth media, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. Bacteria were routinely grown at 37°C in lysogeny broth (LB) with aeration or on LB agar (LBA) (44). The antibiotics used were as follows: 50 µg kanamycin (Kan) mL⁻¹, 100 µg ampicillin (Amp) mL⁻¹, 25 µg chloramphenicol (Cml) mL⁻¹, and 10 ng tunicamycin mL⁻¹ with 3 µg polymyxin B nonapeptide mL⁻¹ (PBMN; Sigma). Strains carrying pBAD33, pBCKs+, or pWKS30 constructs requiring induction were grown in LB at 37°C with aeration for 16 h, subcultured (1/20) into fresh broth, and induced with either 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for pBCKs+ and pWKS30 constructs or 0.2% (wt/vol) l-arabinose for pBAD33 constructs. Cultures were grown for another 4 h.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Reference or Source
Strains
RMA2162	S. flexneri PE860 Y serotype; strain lacks virulence plasmid and pHS-2 plasmid	Laboratory stock
RMA2171	S. flexneri PE860 ΔrmlD	Laboratory stock
NMRM55	S. flexneri PE860 ΔwzyE	This study
NMRM339	S. flexneri PE860 ΔwzyB	This Study
NMRM345	S. flexneri PE860 ΔwecA	This Study
NMRM348	S. flexneri PE860 ΔwecC	This Study
Plasmids
pBAD33	Arabinose inducible, expression vector, Cmlr	52
pWzyE	pBAD33 encoding WzyE3xFLAG, Cmlr	This study
pBCKs+	pBluescript KS +, IPTG inducible, expression vector, Cmlr	Stratagene
pWzyB	pBCKs+ encoding WzyB3xFLAG, Cmlr	Laboratory stock
pWKS30	IPTG inducible, expression vector, Ampr	53
pUppS	pWKS30 encoding UppS-HA, Ampr	This study
pWecA	pWKS30 encoding WecA-HA, Ampr	This study
pWALDO-WzyE-GFP-His8	Cloning vector expressing WzyEEC-GFP-His8	54

DNA methods.

Plasmid constructs were purified from E. coli DH5α strains using a QIAprep Spin miniprep kit (Qiagen). Preparation of electrocompetent cells and electroporation methods were performed as described previously (45).

Chromosomal mutagenesis.

S. flexneri Y ΔwzyE, ΔwzyB, ΔwecC, and ΔwecA strains were generated using λ Red mutagenesis as described previously (46). Briefly, primers were designed (ΔwzyE:NM1/NM2, ΔwzyB:NM136/NM137, ΔwecC:NM151/NM152, and ΔwecA:NM134/NM135) to PCR amplify either a kanamycin or chloramphenicol resistance cassette flanked with 50 bp of the start or end of the coding regions for the respective genes (Table 2). The purified PCR fragments were then electroporated into the parent PE860 strain carrying pKD46 to generate the mutant strains. The antibiotic resistance cassettes were eliminated by the introduction of pCP20 (46).

TABLE 2.

DNA oligonucleotides used in this study

Primer	Oligonucleotide sequence (5′-3′)	Target
Construct generation specific primers
VL70	gtaacacccgggttgtttaactttaagaaggagactcg	pWALDOWzyE-GFP-His8
NM25	tcttcgcatgctcacttgtcatcgtcatccttgtagtcgatgtcatgatctttataatcaccgtcatggtctttgtagtctccttcaacctgcgtccgg	wzyE gene
NM122	cgatagaattcgtagggcttcagtgatatagtctgcgcc	uppS gene
NM123	attagtctagattcaagcgtaatctggaacatcgtatgggtaggctgtttcatcaccgggctc	uppS gene
NM128	ttagagaattcgggttcggaacggactttcccttc	wecA gene
NM129	attagtctagattcaagcgtaatctggaacatcgtatgggtatttggttaaattggggctgccacca	wecA gene
λ Red mutagenesis specific primers
NM1	caatcaactgtaagccacgcagcgtataggttggtgccgtggtgttgttattcattgatgggaattagccatggtcc	wzyE gene
NM2	tatctacaaggctggcagcgggcgttggcgattgccgccagggaggtcgcgtgtaggctggagctgcttc	wzyE gene
NM134	ggttatacttctgctaataattttctctgagagcatgcattgtgaatttagtgtaggctggagctgcttc	wecA gene
NM135	tttcccaggcattggttgtgtcatcacatcctcatttatttggttaaattatgggaattagccatggtcc	wecA gene
NM136	tgttataaaaattttatttatatttttcatattcgtaaggtgatgtttttgtgtaggctggagctgcttc	wecB gene
NM137	agtaataacctcacttctggagcaaaataaaggatcttaaaaatagggaaatgggaattagccatggtcc	wecB gene
NM151	aaaataatcggatatcactatgagttttgcgaccatttctgttatcggactgggttacatgtgtaggctggagctgcttc	wecC gene
NM152	ttttctcatcagcgccagactcctttggcatcgacgacatactgctgatatgggaattagccatggtcc	wecC gene

Generation of complementing constructs.

Constructs for ectopic protein expression were generated via PCR and restriction cloning. 141 Primers used for construct generation used in this study are listed in Table 2. Generation of pWzyE was performed by using primers (VL70/NM25) to PCR amplify a fragment of DNA containing wzyE with XmaI and SphI restriction enzyme sites from pWALDO-WzyE-GFP-His₈. The resulting wzyE fragment was digested with XmaI and SphI and subcloned into likewise-digested pBAD33 to give pBAD33-WzyE-3xFLAG. Generation of pUppS and pWecA was performed using primers NM122/NM123 and NM128/129 to PCR amplify a fragment of DNA from S. flexneri 2457T containing uppS or wecA, respectively, with EcoRI and XbaI restriction enzyme sites. The resulting DNA fragments were digested with EcoRI and XbaI and cloned into likewise-digested pWKS30 to give pWKS30-UppS-HA and pWKS30-WecA-HA. DNA sequencing was used to confirm that no mutations had been introduced during the PCR amplification and to ensure the presence of the C-terminal epitope tags.

LPS/ECA sample preparation.

Bacteria were grown and induced as described above before 1 × 10⁹ cells were collected by centrifugation (2,000 × g), resuspended in 2× lysis buffer (47), and heated at 100°C for 10 min before incubation with 2.5 mg/mL proteinase K (Sigma-Aldrich) for 2 h at 56°C (47).

LPS SDS-PAGE and silver staining.

LPS samples were heated at 100°C for 10 min before being loaded and electrophoresed on SDS-15% PAGE gels for 13 h at 12 mA. Silver staining of LPS was performed as described previously (47).

ECA PAGE and Western immunoblotting.

ECA samples were heated at 100°C for 10 min before being loaded and electrophoresed on SDS-15% PAGE gels at 200 V for 1 h. SDS-PAGE gels were transferred onto nitrocellulose membranes (Bio-Rad) at 400 mA for 1 h. Membranes were then blocked with 5% (wt/vol) skim milk in Tris-Tween buffer saline (TTBS) (45) followed by overnight incubation with polyclonal rabbit anti-ECA antibodies at a 1:500 dilution in 2.5% (wt/vol) skim milk in TTBS. Detection was performed with goat anti-rabbit horseradish peroxidase-conjugated antibodies (KLP) and chemiluminescence reagent (Sigma). Then, 5 µl of SeeBlue Plus2 prestained protein ladder (Invitrogen) was used as a molecular mass standard.

Measuring OM polysaccharide abundance.

Densitometry was performed on three biological replicates from silver-stained SDS-PAGE gels for LPS Oag quantification and anti-ECA Western immunoblots. The degree of polymerization was calculated by normalizing the densitometry of S-Oag and ECA molecules to the parent, where the relative abundance of each sample to the parent was presented as a scatterplot.

Growth curves.

Bacteria were grown for 16 h as described above, and 1 × 10⁷ cells were collected via centrifugation and resuspended in 1 mL of LB. Cells were then subcultured (1/10) into 135 µL of fresh LB medium in a 96-well tray. The tray was then incubated at 37°C for 10 h with aeration. OD₆₀₀ absorbance readings were taken every 20 min (BioTek PowerWave XS2).

CFU/counting.

Bacteria were grown for 16 h as described above. Cells were normalized and subcultured into 10 mL of LB. Then, 20 µL of cell culture was taken at 0, 2, 4, 6, and 8 h and serially diluted 1:10 with phosphate-buffered saline (PBS) prior to spotting 10 µL of cell suspension from the range of 10⁻⁵ to 10⁻⁸ in triplicate onto LBA plates. Plates were incubated overnight (O/N) at 37°C, and colonies were counted the next day.

Microscopy.

Bacteria were grown as described above, and 10 µL amounts of cultures were spotted onto glass slides. The cultures were then allowed to dry, followed by mounting with Moviol (Calbiochem) and sealing with nail polish. Cells were observed via phase-contrast microscopy (Olympus IX70) under a 100× oil lens objective (44). Images of cells were acquired, and cellular lengths were then manually measured using Metamorph 7.5.6.

Statistical analysis.

Independent Student’s t tests were performed on triplicate experimental data values using the statistical analysis tool GraphPad Prism 9. Graphs were plotted with the standard error of the mean (SEM), and statistical significance was displayed as the following: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, non-statistically significant.