ElyC and Cyclic Enterobacterial Common Antigen Regulate Synthesis of Phosphoglyceride-Linked Enterobacterial Common Antigen.

The Gram-negative cell envelope is a complex structure delineating the cell from its environment. Recently, we found that enterobacterial common antigen (ECA) plays a role maintaining the outer membrane (OM) permeability barrier, which excludes toxic molecules including many antibiotics. ECA is a conserved carbohydrate found throughout Enterobacterales (e.g., Salmonella, Klebsiella, and Yersinia). There are two OM forms of ECA (phosphoglyceride-linked ECAPG and lipopolysaccharide-linked ECALPS) and one periplasmic form of ECA (cyclic ECACYC). ECAPG, found in the outer leaflet of the OM, consists of a linear ECA oligomer attached to phosphoglyceride through a phosphodiester linkage. The process through which ECAPG is produced from polymerized ECA is unknown. Therefore, we set out to identify genes interacting genetically with ECAPG biosynthesis in Escherichia coli K-12 using the competition between ECA and peptidoglycan biosynthesis. Through transposon-directed insertion sequencing, we identified an interaction between elyC and ECAPG biosynthesis. ElyC is an inner membrane protein previously shown to alter peptidoglycan biosynthesis rates. We found ΔelyC was lethal specifically in strains producing ECAPG without other ECA forms, suggesting ECAPG biosynthesis impairment or dysregulation. Further characterization suggested ElyC inhibits ECAPG synthesis in a posttranscriptional manner. Moreover, the full impact of ElyC on ECA levels requires the presence of ECACYC. Our data demonstrate ECACYC can regulate ECAPG synthesis in strains wild type for elyC. Overall, our data demonstrate ElyC and ECACYC act in a novel pathway that regulates the production of ECAPG, supporting a model in which ElyC provides feedback regulation of ECAPG production based on the periplasmic levels of ECACYC. IMPORTANCE Enterobacterial common antigen (ECA) is a conserved polysaccharide present on the surface of the outer membrane (OM) and in the periplasm of the many pathogenic bacteria belonging to Enterobacterales, including Klebsiella pneumoniae, Salmonella enterica, and Yersinia pestis. As the OM is a permeability barrier that excludes many antibiotics, synthesis pathways for OM molecules are promising targets for antimicrobial discovery. Here, we elucidated, in E. coli K-12, a new pathway for the regulation of biosynthesis of one cell surface form of ECA, ECAPG. In this pathway, an inner membrane protein, ElyC, and the periplasmic form of ECA, ECACYC, genetically interact to inhibit the synthesis of ECAPG, potentially through feedback regulation based on ECACYC levels. This is the first insight into the pathway responsible for synthesis of ECAPG and represents a potential target for weakening the OM permeability barrier. Furthermore, this pathway provides a tool for experimental manipulation of ECAPG levels.

INTRODUCTION

The Gram-negative envelope is a complex multilayered structure comprised of the outer membrane (OM), the inner membrane (IM), and the periplasm containing a thin peptidoglycan layer (1, 2). The lipid component of the OM consists of an outer leaflet containing mainly lipopolysaccharide (LPS) and an inner leaflet containing phospholipids. A highly compact hydrophobic layer and highly hydrophilic layer formed by LPS, as well as the presence of transenvelope efflux pumps, render the OM impermeable to both hydrophobic molecules and large hydrophilic molecules (1, 3, 4).

The surge of antibiotic resistance in Gram-negative bacteria, especially in Enterobacterales (e.g., Escherichia coli, Klebsiella pneumoniae, and Salmonella sp.) has led to classification of five groups of Enterobacterales as urgent or serious threats by the Centers for Disease Control and Prevention (USA) (5–8). However, the study of the Gram-negative envelope, and specifically OM biogenesis, has led to the discovery of several antimicrobials in recent years (reviewed in references 9 and 10). Several antimicrobials have been identified targeting essential pathways in OM biogenesis including LPS biogenesis, protein secretion, OM protein biogenesis, and lipoprotein biogenesis (11–20). The continued success of this approach requires greater understanding of cell envelope biogenesis.

Enterobacterial common antigen (ECA) is a carbohydrate antigen present in the outer leaflet of the OM and in the periplasm and is conserved throughout Enterobacterales (reviewed in reference 21). The function of this molecule has remained largely unknown, in part because the biosynthesis pathways for ECA, O-antigen, and peptidoglycan overlap and in part because there are three forms of ECA that cannot currently be genetically separated (see Fig. S1 in the supplemental material). In many Enterobacterales, deleting the first gene in ECA biosynthesis, wecA, not only prevents ECA biosynthesis but also prevents O-antigen biosynthesis and increases precursor availability for peptidoglycan biosynthesis (22–25). Deletion of downstream genes in ECA biosynthesis, such as wecE or wecF, leads to accumulation of intermediates in ECA biosynthesis, interfering with peptidoglycan biosynthesis, altering cell shape, increasing envelope permeability, and activating envelope stress response systems (26–30). Three forms of ECA, LPS-linked ECA (ECALPS), cyclic ECA (ECACYC), and phosphoglyceride-linked ECA (ECAPG), are made from polymerized ECA chains. As many of the genes responsible for the steps in ECA biosynthesis separating these molecules are unknown (see below), assigning functions to these separate forms remains difficult. Nevertheless, it has become clear that in Salmonella sp., ECA plays a role in acid and bile salt resistance (31, 32) and is necessary for pathogenesis in a mouse model (32–35). In addition, we have discovered a role for ECACYC in maintaining the OM permeability barrier in E. coli (36).

A schematic representation of the ECA and peptidoglycan biogenesis pathways. These pathways compete for isoprenoid carrier (Und-P) and UDP-GlcNAc as the substrates. To form ECA, successive sugars are added to Und-P, and the final repeat unit is flipped across the membrane, polymerized, and made into ECALPS, ECAPG, or ECACYC. To form peptidoglycan, amino acids are attached to N-acetylmuramic acid, this molecule is attached to Und-P, and the final sugar is added. Then, the completed repeat unit is flipped across the IM and inserted into the peptidoglycan layer. Both pathways release Und-PP, which is recycled to yield Und-P for further synthesis. Pathway details are available in the text. G, N-acetylglucosamine; Ma, N-acetyl-d-mannosaminuronic acid; Gt, 4-acetamido-4,6-dideoxy-d-galactose; M, N-acetylmuramic acid. Download FIG S1, JPG file, 1.9 MB.

The polysaccharide chains of ECA consist of linear repeat units, each unit made of three sugars: GlcNAc (N-acetylglucosamine), ManNAcA (N-acetyl-d-mannosaminuronic acid), and Fuc4NAc (4-acetamido-4,6-dideoxy-d-galactose) (37, 38). Biosynthesis of ECA is initiated by attachment of GlcNAc-1-phosphate to the isoprenoid carrier, undecaprenyl-phosphate (Und-P), followed by the addition of the two remaining sugars (Fig. S1) (39–42). Und-P is a universal lipid carrier and is required for the biosynthesis of O-antigen, peptidoglycan, and capsular polysaccharides, as well as ECA (43–47). WzxE flips the complete ECA repeat unit linked to Und-P across the IM to the periplasmic face (48), and WzyE polymerizes ECA chains (49). The number of repeat units in the polymerized ECA molecule (chain length) is controlled by WzzE (50). The operon responsible for synthesis of ECA, the wec operon, contains the genes responsible for the steps in ECA biogenesis resulting in a polymerized ECA molecule attached to Und-PP located on the outer leaflet of the IM (39, 41, 49).

The steps through which the three forms of ECA are made from this precursor are less well understood (21). ECALPS is produced when WaaL, the O-antigen ligase, attaches ECA to the core polysaccharide of LPS (43, 51). ECALPS is presumably transported to the cell surface by the Lpt system responsible for transporting LPS to the cell surface (52). The second form, ECACYC, a cyclic carbohydrate, remains in the periplasm (29). It is generally made with precise chain length (4 repeat units in E. coli K-12), and WzzE is necessary for its synthesis (29, 53, 54). The final form, ECAPG, is a linear ECA chain linked to diacylglycerol through a phosphodiester bond (55). The mechanism through which ECAPG is formed and transported to the cell surface is completely unknown (21, 56, 57). This lack of knowledge impairs genetic studies of ECA function as mutants cannot be made that synthesize ECALPS and ECACYC in the absence of ECAPG.

Therefore, we set out to identify factors genetically interacting with the biosynthesis of ECAPG. We took advantage of the competition for substrates between the peptidoglycan and ECA biosynthesis pathways to find factors interacting with ECAPG biosynthesis (Fig. S1). Using transposon-directed insertion sequencing (TraDIS), we identified ElyC as a factor interacting with ECAPG biosynthesis. ElyC is an IM protein with two transmembrane domains and a large C-terminal globular DUF218 domain that resides in the periplasm (58, 59). Paradis-Bleau et al. found a ΔelyC mutant displays severe growth defects at low temperatures (22°C) and high frequency of cell lysis due to decreased peptidoglycan synthesis (28). They suggested that ElyC regulates the allocation of Und-P between synthesis pathways in E. coli.

Here, we have explored the role of ElyC in ECAPG biosynthesis. Our data demonstrate that ElyC posttranscriptionally regulates the synthesis of ECAPG, greatly inhibiting its synthesis during normal growth. Furthermore, we observed that ElyC only had its full effect on ECAPG synthesis in the presence of WzzE, suggesting that WzzE or ECACYC is involved in this regulatory pathway. In fact, we found that ElyC and ECACYC act together to regulate ECAPG biosynthesis. Our data demonstrate that the effect of ElyC and ECACYC on ECA levels is specific to ECAPG and not a result of allocation of Und-P between biosynthesis pathways. Overall, we have deciphered a novel pathway through which ElyC and ECACYC regulate ECAPG biosynthesis, providing insight into the elusive function of ElyC and demonstrating that ECACYC plays roles both in maintaining the OM permeability barrier and in regulating ECA biosynthesis.

RESULTS

Identification of candidate genes interacting with ECAPG biosynthesis.

There are genes known to be necessary for the biosynthesis of ECACYC and of ECALPS specifically: ECACYC synthesis requires wzzE, while ECALPS synthesis requires waaL (29, 36, 43). However, the genes and reactions responsible for producing ECAPG, by transferring the ECA polymer from Und-PP to form phosphoglyceride-linked ECAPG, and for its surface exposure are unknown (55–57). Therefore, we set out to identify factors involved in ECAPG biosynthesis, utilizing interactions between ECA and peptidoglycan biosynthesis.

The ECA and peptidoglycan pathways compete for Und-P and UDP-GlcNAc as depicted in Fig. S1 in the supplemental material (25, 60, 61). Although deletion of wecA causes generally mild phenotypes, deletion of later genes in ECA biosynthesis (e.g., wecB, wecG, wecF, or wzxE) causes the accumulation of ECA intermediates sequestering Und-P, disrupting peptidoglycan biosynthesis and resulting in increased permeability defects, cell shape defects, and envelope stress response activation (26–30, 62, 63). In fact, deletion of wzyE, the ECA polymerase, or all the flippases capable of flipping lipid IIIECA across the IM is lethal (29). Thus, we hypothesized that, in a strain making only ECAPG, disruption or dysregulation of the next step in ECA biosynthesis (transfer of polymerized ECA from Und-PP) would also be highly unfavorable due to sequestration of Und-P inhibiting peptidoglycan biosynthesis.

Therefore, we have used TraDIS (transposon-directed insertion sequencing) (64) to compare the favorability of gene disruptions in a mutant which makes ECAPG but not the other forms of ECA (ΔwzzE ΔwaaL) with an isogenic mutant that does not make ECA (ΔwecA-wzzE ΔwaaL) and with wild-type E. coli K-12 MG1655. For this approach, we generated high-density transposon libraries in each of these strains and performed Illumina sequencing of the transposon junctions in the initial pooled libraries, as well as after 10 generations of growth in liquid culture. The statistical properties of each data set were similar (Table S1). We then compared the transposon junction reads per gene between the three strains (Fig. 1A). To confirm we could detect changes in essentiality due to sequestration of Und-P, we analyzed the transposon junction reads in wzyE. In strains producing ECA, wzyE is essential; however, wzyE becomes nonessential when ECA synthesis is disrupted at an earlier step as accumulation of lipid IIIECA is prevented (29). In the ECAPG-only strain and wild-type MG1655, we detected very few transposon insertion reads in wzyE; however, we observed similar levels of insertions to nearby genes in the strain without ECA (Fig. 1B).

FIG 1

Screening for candidate genes interacting with ECAPG biosynthesis. (A) TraDIS was used to identify genes for which disruption was unfavorable in cells making ECAPG without the other forms of ECA. Scatterplots of transposon junction reads per gene are shown comparing the ECAPG strain (ΔwzzE ΔwaaL) with an isogenic strain without ECA (ΔwecA ΔwzzE ΔwaaL) and with wild-type MG1655. Results are shown following initial growth on plates and after 10 additional generations of growth in liquid medium. Putative ECAPG biosynthesis genes are shown in yellow, elyC is shown in green, wzyE is shown in purple, and ynbB is shown in red. Genes deleted in one of the strains are shown in cyan. (B) Histograms of transposon insertion reads in wzyE and adjacent genes are shown as a control for detection of changes in essentiality based on Und-P availability. Transposon insertions are observed in wzyE only in the strain without ECA. (C) Histograms of transposon insertion reads in elyC and adjacent genes. Transposon insertions are observed in the strain without ECA and the wild-type strain but not in the strain making only ECAPG, suggesting essentiality of elyC in the ECAPG-only strain.

Descriptive statistics for TraDIS. Download Table S1, PDF file, 0.08 MB.

To identify genes possibly involved in ECAPG biosynthesis, we defined a set of criteria for genes putatively essential only when ECAPG is made without the other forms of ECA (Table S2). These genes had less than 200 reads in the ECAPG-only library under both growth conditions and had at least a 1-standard-deviation decrease in the ECAPG-only strain compared to the other two strains under both growth conditions. In addition, we limited our analysis to genes not known to be essential in wild-type E. coli K-12 that make proteins targeted to either the IM or the periplasm (65–67). We identified five genes that fit these criteria: elyC, ynbB, ymiB, lapA, and yoaI (Table S2). From these hits, we confirmed the data in the literature that ynbB was not necessary for the synthesis of ECAPG (68; unpublished data).

Possible genes involved in ECAPG biogenesis. Download Table S2, PDF file, 0.1 MB.

In this paper, we focus on elyC (Fig. 1C), which encodes an inner membrane protein. Previous work has shown that a ΔelyC mutant lyses at room temperature (22°C) due to a peptidoglycan synthesis defect but grows well at 37°C (28). The authors hypothesized this defect is due to competition between peptidoglycan and the ECA biosynthesis pathway, particularly at the step of allocation of Und-P. Data have also suggested that an ΔelyC mutant may have a periplasmic protein-folding defect (69) and may experience increased oxidative stress at low temperature (22°C) (70). The experiments described here were performed at 37°C.

elyC is essential in a strain producing only ECAPG.

To determine whether elyC was essential in a strain producing only ECAPG, we performed genetic linkage-disruption experiments. In these experiments, a Tn10 marker genetically linked to a deletion in the gene of interest is transduced into strains, selecting for the presence of Tn10. Based on the size of DNA packaged by the P1vir phage, the gene deletion is cotransduced with a calculable frequency (71). If there is selection against the deletion of the gene (i.e., the gene is essential), the cotransduction frequency observed will decrease. We first measured linkage between zbj-7230::Tn10 and ΔelyC::kan. These two markers were approximately 53% linked in wild-type MG1655 and in ΔwaaL and ΔwzzE single mutants (Table 1). However, in a ΔwzzE ΔwaaL double mutant producing only ECAPG, we observed only 1% linkage, demonstrating strong linkage disruption (Table 1). The linkage is restored in a complemented strain. We observed similar linkage disruption when transducing metE3074::Tn10 linked to ΔwzzE::kan into a ΔwaaL ΔelyC mutant (Table 1). These data confirm that elyC is essential when ECAPG is made in isolation but not in strains making two or more forms of ECA.

TABLE 1

elyC is essential in strain making only ECAPG

Donor	Recipient	Recipient form(s) of ECA	N a	P1vir cotransduction frequencyb
zbj-7230::Tn10 ΔelyC::kan (AM769)	MG1655	ECACYC, ECALPS, ECAPG	300	52.7%
	ΔwzzE (AM365)	ECALPS, ECAPG	300	54.3%
	ΔwaaL (AM366)	ECACYC, ECAPG	300	53.3%
	ΔwzzE ΔwaaL (AM395)	ECAPG	300	1.0%
	ΔwzzE ΔwaaL pBAD33-elyCc (AM1159)	ECAPG	306	62.1%
metE-3074::Tn10 ΔwzzE::kan (AM766)	MG1655	ECACYC, ECALPS, ECAPG	300	25.0%
	ΔelyC (AM743)	ECACYC, ECALPS, ECAPG	300	17.3%
	ΔwaaL (AM366)	ECACYC, ECAPG	300	24.3%
	ΔwaaL ΔelyC (AM745)	ECACYC, ECAPG	300	3.0%
thd::Tn10 ΔwaaL::kan (AM735)	MG1655	ECACYC, ECALPS, ECAPG	295	79.0%
	ΔelyC (AM743)	ECACYC, ECALPS, ECAPG	300	72.3%
	ΔwzzE (AM365)	ECALPS, ECAPG	300	76.0%
	ΔwzzE ΔelyC (AM744)	ECALPS, ECAPG	232	61.2%

The indicated number of transductants were analyzed. Transductants were harvested from three separate transductions.

P1vir was used to transduce the indicated markers into the indicated strain. Cotransduction frequency was determined by selecting the transductants for the presence of Tn10 and calculating the percentage of colonies containing the gene deletion.

Expression from complementing plasmid was induced with 0.2% arabinose.

Interestingly, we observed only slight linkage disruption when transducing thd::Tn10 ΔwaaL::kan into a ΔwzzE ΔelyC strain (Table 1). We confirmed these results by rebuilding the strains from wild-type MG1655 and with two different alleles of ΔelyC (Table S3) and through direct transduction of the ΔwaaL::kan allele. Although the triple deletion mutant, ΔwzzE ΔelyC ΔwaaL::kan thd::Tn10, could be built with ΔwaaL as the last deletion, the triple mutant colony size was extremely small compared to ΔwzzE ΔelyC thd::Tn10 colonies (Fig. S2). These data suggest that, although elyC is essential in a strain producing only ECAPG, its function is somehow modified in a ΔwzzE strain allowing survival when waaL is deleted last (see below).

Loss of waaL impedes growth in a ΔwzzE ΔelyC background. ΔwzzE ΔelyC cells were transduced with a P1vir lysate from a tdh::Tn10 ΔwaaL::kan strain, selecting for the presence of the Tn10. Representative ΔwzzE ΔelyC tdh::Tn10 colonies are marked with black arrows. Representative ΔwzzE ΔelyC ΔwaaL::kan tdh::Tn10 colonies are marked with blue arrows. The presence of ΔwaaL::kan was confirmed through kanamycin resistance. The triple-deletion cells showed greatly reduced growth compared to the isogenic controls. Download FIG S2, JPG file, 1.3 MB.

Linkage disruption with deletion of waaL Table S3, PDF file, 0.1 MB.

Deletion of elyC increases ECAPG levels.

After confirming elyC’s essentiality in a strain making only ECAPG, we asked what the effect of the ΔelyC mutation was on surface exposure of ECA, as a ΔwaaL strain without ECAPG should not have ECA on its surface. We used a dot blot as a qualitative method (72) to detect surface-exposed ECAPG and ECALPS. ECACYC is not surface exposed. Surface-exposed ECA was detected in all ΔelyC strains including the ΔwaaL ΔelyC strain (Fig. 2A). In fact, the surface-exposed ECA levels appeared higher in the ΔelyC and ΔwaaL ΔelyC strains than in the wild-type or ΔwzzE ΔelyC strains. These observations suggested that there might be an increase in a non-ECALPS, surface-exposed species of ECA in the ΔelyC mutant. As dot blots are not ideal for determining quantitative changes, we performed ECA immunoblot analyses to detect the charged forms of ECA (ECAPG and ECALPS). ECACYC is not charged and cannot be observed through immunoblot analysis. We found a very large increase in linear ECA levels in both the ΔelyC and ΔwaaL ΔelyC strains (Fig. 2B; compare lanes 5 and 6 with lanes 1 and 4). Similar to the dot blot results, there was much less of an increase in ECA levels in the ΔwzzE ΔelyC strain (lane 7). These results suggest, when elyC is deleted, there is a large increase in a species of ECA which is neither ECALPS nor ECACYC.

FIG 2

Deletion of elyC increases levels of ECAPG. (A) The surface exposure of ECAPG and ECALPS was detected through dot blot assay. Whole cells or a whole-cell lysate was probed for ECA, BamD, or RcsF. BamD acted as a negative control for surface exposure, while RcsF acted as a positive control for surface exposure. ΔwecA served as a negative control for the presence of ECA. Surface-exposed ECA was detected in all ΔelyC strains including the ΔwaaL ΔelyC strain, suggesting ECAPG is present on the cell surface in these strains. (B) Immunoblotting was performed to examine ECA levels and chain length. A very large increase in ECA levels was observed in ΔelyC and ΔwaaL ΔelyC strains, but less of an increase was observed in the ΔwzzE ΔelyC strain. The nonspecific “X” band serves as a loading control. ΔwecA serves as a negative control for the presence of ECA. (C) ECALPS quantification was performed in indicated strains by WGA staining. Data are shown as fluorescence relative to OD600. The ΔwaaL and ΔwaaL ΔelyC strains serve as negative controls for the presence of ECALPS. There was a small but significant increase in ECALPS levels in the ΔelyC and ΔwzzE ΔelyC strains compared to their parent strains. Data are shown as the mean from three biological replicates ± standard error of the mean (SEM). *, P < 0.05 by the nonparametric Mann-Whitney test compared to elyC+ parent strain.

Thus, we sought to determine ECALPS levels. Wheat germ agglutinin (WGA) is a lectin protein used to detect glycans (β-GlcNAc or sialic acid multimers) in prokaryotes and eukaryotes (73–78). A beta-linked GlcNAc is present in the glycosidic bond that attaches ECA to LPS to form ECALPS, but this bond is absent in ECAPG. Therefore, we have found cell surface staining of MG1655 with WGA labels only ECALPS, providing a specific assay for this ECA species (Fig. S3A). Thus, we assayed WGA staining of elyC mutant cells and found deletion of elyC caused only a slight increase in the amount of ECALPS. This increase was similar between the ΔelyC and ΔwzzE ΔelyC strains (Fig. 2C). There are two possible explanations for the smaller increase in ECALPS levels than linear ECA levels: (i) ElyC plays a role in ECA biosynthesis that is specific to ECAPG or (ii) ECALPS levels are limited by availability of WaaL. Therefore, we overexpressed waaL in the wild-type and ΔelyC strains and assayed levels of ECALPS. Although we observed an increase in ECALPS levels when waaL was overexpressed in a wild-type strain, we did not see an increase in ECALPS level in the ΔelyC strain (Fig. S4A), demonstrating that ΔelyC is epistatic to waaL expression. These data suggest that the effect of ElyC is specific to ECAPG biosynthesis. Overall, the immunoblot, dot blot, and WGA staining experiments demonstrate that ElyC plays a role in ECAPG biosynthesis that leads to a large increase in the levels of a non-ECALPS species with deletion of elyC.

WGA staining and elyC overexpression validation. (A) Live cells of the indicated genotypes were stained with WGA conjugated to the Alexa Fluor 488 fluorophore, and their relative fluorescence was measured. No WGA sample denotes mock-stained wild-type control cells. Staining was observed in the wild-type strain and a ΔwzzE strain lacking ECACYC. However, no staining was observed in the ΔwecA, ΔwaaL, and ΔwzzE ΔwaaL strains that lack ECALPS. Data are shown as the mean from three biological replicates ± SEM. (B) ECALPS quantification was performed in a Lac− strain (MC4100 AraR/−) overexpressing elyC from the IPTG-inducible pCA24N vector. Overexpressing elyC decreased ECALPS levels. pV sample contains the strain with empty pCA24N treated with 100 μM IPTG. Data are shown as the mean from three biological replicates ± SEM. (C) ECAPG levels were assayed in the ΔwzzE ΔwaaL double mutant with empty pBAD33 and pBAD33-elyC induced by arabinose. elyC overexpression resulted in decreased ECAPG levels. The nonspecific “X” band serves as a loading control. (D) ECALPS levels were measured in MG1655 and the ΔwzzE mutant strain in the presence of empty pBAD33 and pBAD33-elyC induced by arabinose or repressed by α-d-fucose. ECALPS levels decreased with elyC overexpression in both strains. Representative data are shown as fold values relative to the vector control under the same induction conditions. Download FIG S3, JPG file, 0.06 MB.

Overexpression of waaL and murA. (A) waaL was overexpressed from the IPTG-inducible pCA24N vector in wild-type and ΔelyC strains, and ECALPS levels were measured using WGA staining. In wild-type cells, increasing overexpression of waaL causes significant increases in ECALPS levels. In ΔelyC cells, increasing waaL overexpression does not increase ECALPS levels, demonstrating that ΔelyC is epistatic to waaL overexpression for ECALPS levels. Data are shown as the mean from three biological replicates ± SEM. *, P < 0.05 by the Mann-Whitney test. (B) Linear ECA levels were assayed by immunoblotting in MG1655 strains harboring empty pCA24N (pV), pCA24N-elyC, or pCA24N-murA plasmids induced with different IPTG concentrations (0, 10, 25, and 100 μM IPTG). The 100 μM IPTG sample was omitted for pCA24N-elyC due to toxicity. Both murA and elyC overexpression decrease linear ECA levels. Download FIG S4, JPG file, 0.06 MB.

ElyC posttranscriptionally regulates the production of ECAPG and ECA overall.

There are two possible models to explain the increase in ECA observed when elyC is deleted. First, ElyC may decrease ECAPG levels by inhibiting the production of ECAPG. Second, there might be more than one step necessary to produce ECAPG and ElyC is responsible for a later step leading to the accumulation of a biosynthetic intermediate that is not distinguishable from ECAPG on an immunoblot. To differentiate between these models, we determined the effect of overexpressing elyC on ECA levels. In the first model, overexpression of ElyC should decrease ECAPG levels, while, in the second model, overexpression of ElyC should either not effect or increase production of ECAPG, depending on the rate-limiting step in synthesis.

We assayed ECA levels by immunoblotting in strains overexpressing elyC in a wild-type or ΔwzzE background. We utilized the pCA24N-elyC plasmid from the ASKA collection, which expresses elyC under a leaky, isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter (79). We observed that increasing elyC overexpression in a wild-type background greatly decreased ECA levels (Fig. 3A, lanes 1 to 4). In a ΔwzzE mutant, a smaller decrease in ECA levels was observed at high IPTG concentrations (lanes 5 to 10). As with our previous results, this suggests that the removal of wzzE changes ElyC’s effect on ECA levels. To assay the effect of elyC overexpression on ECALPS in the wild-type and ΔwzzE strains, we performed WGA staining and observed a decrease in ECALPS levels that was similar in the two backgrounds (Fig. 3B). We observed similar results when elyC was overexpressed in MC4100 (a Lac− strain) (Fig. S3B) and when elyC was overexpressed from a pBAD33 plasmid (Fig. S3C and D). Overall, these results demonstrate that overexpressing elyC decreases ECA levels and ElyC is responsible for inhibiting the production of ECAPG and, to some extent, ECA production overall.

FIG 3

ElyC regulates ECAPG production posttranscriptionally. (A to C) elyC was overexpressed from the IPTG-inducible pCA24N vector in the indicated strains. (A) Triangles indicate increasing overexpression of elyC. In wild-type cells, even low-level overexpression of elyC greatly decreased ECA levels; however, less of a decrease was observed in the ΔwzzE strain. The nonspecific “X” band serves as a loading control. “V” samples indicate strains with empty pCA24N induced with 100 μM IPTG. Lanes 2 to 4 are induced with 0, 10, and 25 μM IPTG, respectively. Lanes 6 to 10 are induced with 0, 10, 25, 50, and 100 μM IPTG, respectively. (B) ECALPS quantification was performed by WGA staining, indicated by fluorescence relative to OD600. The ΔwaaL strain acts as a negative control. Overexpression of elyC reduced ECALPS levels, but no difference was observed between the wild-type strain and the ΔwzzE strain. (C) A bacterial luciferase reporter was used to assay the activity of the P promoter, the promoter for the wec operon containing genes for ECA biosynthesis. Despite the decrease in ECA levels observed, overexpression of elyC did not decrease P activity. (D) P activity was assayed as in panel C. No increase in P activity was observed in the ΔelyC strain compared to the wild-type strain. Quantitative data are shown as the mean from three biological replicates ± SEM. *, P < 0.05 by the nonparametric Mann-Whitney test; ns, P > 0.05 by the Mann-Whitney test.

We then asked whether elyC affects ECA levels on a transcriptional or posttranscriptional level. We constructed a reporter by cloning the promoter region of the wec operon into a promoterless pJW15 vector that harbors a bacterial luciferase operon (Fig. S5) (80, 81). Using this reporter, we observed no consistent decrease in P activity with elyC overexpression, despite the decrease in ECA levels (Fig. 3C). We also checked the effect of elyC deletion on P activity and found no increase the P reporter activity in this strain (Fig. 3D). Overall, we observed no indication that ElyC regulates ECA levels in a transcriptional manner, making posttranscriptional regulation most likely.

Vector map of pJW15-P. The region of P promoter, which drives expression of the wec operon responsible for many steps in ECA biosynthesis, from −500 to +20 relative to the wecA translational start site was cloned into the pJW15 vector between the EcoRI and BamHI restriction sites. The P promoter drives the expression of the luxCDABE genes causing luminescence when the P promoter is active. Adapted from data from the work of Wong et al. (J. L. Wong, S. L. Vogt, and T. L. Raivio, Methods Mol Biol 966:337–357, 2013, https://doi.org/10.1007/978-1-62703-245-2_21). Download FIG S5, JPG file, 0.1 MB.

ECACYC acts with ElyC to regulate ECAPG production.

Throughout our experiments, we observed that the effect of ElyC on ECAPG levels was less in the absence of wzzE. This led us to ask whether WzzE or ECACYC was playing a role in the pathway through which ElyC regulated ECAPG levels. To differentiate between the effects of WzzE and ECACYC on this pathway, we utilized the previous observation that levels of linear ECA are very low in a ΔwaaL mutant (36). We have confirmed this effect: there is much less ECA detectable by immunoblotting in ΔwaaL cells than in wild-type cells or a ΔwzzE mutant (Fig. 4A; lane 4 compared to lanes 2 and 3). However, in the ΔwzzE ΔwaaL mutant the ECA levels return to near-wild-type levels (lane 5). Our initial explanation was that there was much more ECALPS than ECAPG present and that the excess ECA freed by removing ECALPS was funneled into ECACYC, which is not detectable by immunoblotting. However, this did not fit well with our observations of overexpressing elyC in a ΔwzzE background (Fig. 3A and B), where we observed a large decrease in ECALPS levels but a relatively small decrease in ECA levels overall. Therefore, we purified ECACYC and quantified the ECACYC levels through matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectroscopy (Fig. S6). We found no effect of ΔwaaL on cyclic ECA levels (Fig. 4B and Fig. S6). Thus, while the ECACYC levels remained constant, the total amount of ECA was decreased in the ΔwaaL strain. That a free pool of ECA caused by the removal of ECALPS leads to decreased linear ECA levels and steady ECACYC levels demonstrates that ECACYC plays a role in regulating ECAPG biosynthesis (Fig. 4C). Thus, ECACYC is involved in regulation of ECAPG production. Then, we measured P activity in ΔwecA, ΔwzzE, and ΔwaaL strains to determine whether the regulation is transcriptional or posttranscriptional. We found that any changes in reporter activity in the mutants did not correlate with the changes in the amounts of ECA observed (Fig. 4A and D). Thus, ECACYC has a role in controlling ECAPG production that appears to be through posttranscriptional regulation. Importantly, the levels of ECA are similar between the ΔelyC and ΔwaaL ΔelyC strains (Fig. 2B), and the effect of ElyC on ECAPG levels is much less in the absence of wzzE (Fig. 2B and Fig. 3A). These data demonstrate that ElyC and ECACYC act together in this regulatory pathway.

FIG 4

ECACYC is involved in feedback regulation of ECAPG levels. (A) The level of the linear forms of ECA was assayed in the indicated strains by immunoblotting. The ΔwecA strain serves as a negative control for ECA. The nonspecific “X” band serves as a loading control. Much less ECAPG is observed in the ΔwaaL strain compared to linear ECA levels in the wild-type strain. However, in the ΔwzzE ΔwaaL strain, ECAPG levels return to near-wild-type levels. (B) ECACYC levels were compared in ΔwecH::kan and ΔwaaL ΔwecH::kan strains by MALDI-TOF. WecH is responsible for nonstoichiometric acetylation of ECA. The levels of ECACYC are comparable between the strains. (C) Model for ECAPG and ECACYC levels with loss of ECALPS is shown. In wild-type cells, the three forms of ECA are produced at an appropriate ratio. When ΔwaaL is deleted and ECALPS is lost, the remaining ECA is funneled into ECAPG and ECACYC. However, ECACYC levels are maintained at a consistent level and ECAPG levels are decreased, suggesting ECACYC levels are measured to provide regulation of ECAPG production and ECA levels overall (red arrows). (D) Activity of the P promoter was assayed by luciferase assay in the indicated strains. P activity does not correlate with changes in ECA levels, suggesting that the effect of ECACYC on ECA levels is posttranscriptional. Quantitative data are shown as the mean from three biological replicates ± SEM. ns, P > 0.05 by the Mann-Whitney test.

MALDI-TOF for effect of ΔwaaL on ECACYC levels. The ΔwecH ΔwaaL strain was grown in minimal medium containing normal nitrogen (N14), while the control ΔwecH strain was grown in minimal medium containing heavy nitrogen (N15). Deletion of wecH prevents nonstoichiometric acylation of ECA. Equal amounts of cells were combined, and ECACYC was purified and subjected to MALDI-TOF. The respective peaks, ΔwecH ΔwaaL (m/z 2427) and ΔwecH (m/z 2439), corresponding to ECACYC are labeled. The three colors represent three biological replicates. The areas of the respective peaks in each biological replicate are shown in the inset table. Download FIG S6, JPG file, 1.1 MB.

Und-P allocation is not responsible for the effect of ElyC on ECA levels.

Previous reports have shown that the overexpression of the gene responsible for Und-P synthesis or the gene responsible for the first step in peptidoglycan biosynthesis relieved peptidoglycan stress in a ΔelyC strain (28, 82). This led to the suggestion that ElyC balances Und-P use between pathways (28). It is not possible for the effect of ElyC specifically on ECAPG to be caused by Und-P allocation. However, it is possible that Und-P allocation is responsible for the effect on total ECA levels.

Therefore, we asked whether the effect of ElyC on overall ECA biosynthesis was due to Und-P allocation. We compared the effect of overexpressed murA, the first committed step for peptidoglycan biosynthesis (61), with that of the overexpression of elyC to determine whether they phenocopy each other. Although overexpression of murA does cause a decrease in the abundance of ECALPS, it is much less than that caused by elyC (Fig. 5A). Both murA and elyC decrease levels of linear ECA observed through immunoblotting (Fig. S4B). This result suggests that increasing the competition for substrates may not be solely responsible for the effect of ElyC on ECA biosynthesis. If ElyC did act through balancing Und-P utilization, we further reasoned that overexpression of wecA, the first gene in ECA biosynthesis, would suppress the effect of elyC overexpression on ECA biosynthesis. To determine whether this was true, we overexpressed both wecA and elyC in the wild-type strain. We observed that overexpression of wecA increased production of ECALPS when elyC was not overexpressed (Fig. 5B, fucose samples). However, there was no increase in ECALPS levels with wecA overexpression when elyC was induced with low arabinose or high arabinose levels (Fig. 5B). Therefore, the effect of ElyC on total ECA levels is not through allocation of Und-P.

FIG 5

ElyC regulates ECA levels independently of Und-P availability. ECALPS levels were assayed by WGA staining. Data are shown as fluorescence relative to OD600. (A) elyC or murA, the first gene in peptidoglycan biosynthesis, was overexpressed from the IPTG-inducible pCA24N vector. Overexpression of murA increases the utilization of Und-P by the peptidoglycan biosynthesis pathway. The effect of murA overexpression on ECALPS levels was smaller than that of elyC. (B) elyC was overexpressed from the arabinose-inducible and fucose-repressible pBAD33(K) plasmid. wecA was overexpressed from the IPTG-inducible pCA24N vector. Overexpression of wecA increases the utilization of Und-P by the ECA biosynthesis pathway. wecA overexpression increases production of ECALPS in the absence of elyC overexpression. However, wecA overexpression does not suppress the decrease in ECALPS levels when elyC is overexpressed. Fuc, 0.05% fucose; LB, no inducer or repressor; low ara, 0.02% arabinose; high ara, 0.2% arabinose. Data are shown as the mean from three biological replicates ± SEM. *, P < 0.05 by the nonparametric Mann-Whitney test; ns, P > 0.05 by the Mann-Whitney test.

DISCUSSION

In this study, we have elucidated a novel pathway regulating the biosynthesis of ECAPG (Fig. 6). We have provided evidence that ElyC specifically regulates the biosynthesis of ECAPG by acting as an inhibitor under normal physiological conditions. Furthermore, we have revealed that full function of ElyC requires the presence of ECACYC and that ECACYC itself can regulate the level of ECAPG. The mechanism of regulation by both ElyC and ECACYC appears to be posttranscriptional.

FIG 6

Model for regulation of ECAPG biosynthesis. Our data demonstrate that ElyC and ECACYC are part of a pathway that regulates the levels of ECAPG and that ElyC can also regulate total ECA levels to a lesser extent. WzxE, WzyE, and WzzE flip completed ECA repeat units on Und-PP across the IM and polymerize the ECA molecule to a regulated chain length. We propose that ElyC and ECACYC act together to regulate the reaction that removes polymerized ECA from Und-PP and forms phosphoglyceride-linked ECAPG. “?” represents the unknown enzymes(s) responsible for this reaction. As ECACYC is present in the periplasm, its levels can be assessed by ElyC to provide feedback regulation. This assessment may involve a physical interaction.

A number of studies have shown that disruption of intermediate steps in ECA biogenesis leads to isoprenoid carrier stress and peptidoglycan synthesis defects due to the accumulation of ECA synthesis intermediates on Und-P (26–30). In fact, a recent study has biochemically confirmed the accumulation of these intermediates (25). Similar defects have been observed with the disruption of O-antigen or colanic acid biosynthesis (60, 83). It is clear that the stress on peptidoglycan synthesis increases the further down the ECA biosynthesis pathway that the disruption occurs (26, 29, 60). Thus, blocking the pathway after the first sugar is added to Und-P causes very little stress, while blocking the pathway after addition of the next sugar causes cell shape defects, stress response activation, and increased permeability, and blocking the pathway after the addition of the third sugar is lethal (26–29, 30).

The possibility that, if only one form of ECA could be made, disruption of a step in ECA biosynthesis past polymerization would also be lethal led us to our TraDIS approach to studying ECAPG biosynthesis. Interestingly, we observed the most robust phenotype from increased ECAPG biosynthesis rather than loss of ECAPG biosynthesis, elucidating another route to sequester Und-P. Under these conditions, the transfer of ECA from Und-PP to make ECAPG may become limiting, causing the buildup of polymerized ECA on Und-PP. Although we are analyzing other hits obtained in the TraDIS experiment, it is likely the phenotype of the ΔelyC mutant has a more substantial effect than the loss of ECAPG synthesis itself. In this case, our data would suggest that the Und-PP released by the polymerization of ECA combined with new synthesis of Und-P is sufficient for peptidoglycan synthesis. Recent advances in the biochemical analysis of ECA biosynthesis intermediates (25) will make this an interesting area for future investigation.

With the methods of detecting ECA we have employed, we cannot exclude the possibility that the increase in ECA we observe when deleting elyC is due to the accumulation of polymerized ECA on Und-P, suggesting ΔelyC disrupts ECAPG biosynthesis. In this model, overexpressing elyC would decrease the accumulation of ECA on Und-P, leading to lower levels of ECA observed. However, we do not feel this model fits our data well for the following reasons. (i) The ECA species observed in the immunoblot assay appears to be the major form of ECA present in the cells. To our knowledge, polymerized ECA attached to Und-P has never been detected when Und-P-linked ECA biosynthesis intermediates are investigated (25, 45). In fact, loss of the ability to make a full subunit of ECA was required to initially demonstrate that ECA is synthesized on Und-P (45). (ii) If overexpressing elyC increases conversion of Und-P-linked ECA to ECAPG, an amount of ECAPG should be produced equal to the amount of Und-P-linked ECA lost, leading to very little change in the amount of detectable linear ECA when elyC is overexpressed. (iii) At least some proportion of the increased ECA observed when elyC is deleted is surface exposed, and it is unlikely that there is a surface exposure mechanism for Und-P-linked ECA. (iv) Finally, accumulation of Und-P-linked ECA cannot explain the large decrease in linear ECA levels observed when waaL is deleted. Deletion of waaL prevents the production of ECALPS, removing one possible route for Und-P-linked ECA to be processed to the final forms of ECA; therefore, deletion of waaL would not be expected to decrease levels of Und-P-linked ECA.

The synthesis of ECA overall, and of ECAPG in particular, occurs at the IM, a location that is physically distant from the eventual localization of ECAPG at the cell surface (21). Thus, it would be difficult for the biosynthesis of ECAPG to be directly regulated based on the accumulation of ECAPG on the cell surface. ECACYC provides an ideal solution to this problem, allowing the biosynthesis of ECA to be assessed using a molecule located in the more accessible periplasm (29). As an IM protein with N-terminal transmembrane domains and a comparatively large periplasmic domain (58), ElyC is well situated to interact physically, as well as genetically, both with the ECA biosynthesis machinery and with ECACYC. Studies of ElyC’s physical interactions are ongoing in our lab.

Given their respective localizations, we favor a model in which ElyC provides feedback regulation for the reaction(s) producing ECAPG based on the levels of ECACYC present in the periplasm (Fig. 6). In this model, we speculate that ElyC undergoes constant transient interactions with ECACYC that control the activity or binding capability of ElyC. Thus, when ElyC and ECACYC interact, ElyC would become functional and inhibit the reaction producing ECAPG, possibly through direct interaction with the protein(s) responsible for synthesizing ECAPG. The inhibition could occur through alteration of activity or of degradation rates of the protein(s) producing ECAPG. The amount or time of interaction between ElyC and ECACYC would, therefore, control how much ECAPG is produced. Levels of ECACYC could in this way be constantly monitored to maintain appropriate levels of ECA production, while leaving ECACYC largely free to perform its functional roles in the cell.

In this model, it would be possible for ElyC and ECACYC to act alone. However, it is also possible that other members of the regulatory pathway exist. These pathway members would not be in our TraDIS data set if their loss did not cause a large increase in ECAPG biosynthesis (e.g., their effects are not inhibitory or they are redundant) or if the genes involved are essential due to their roles in other pathways. If there are other pathway members involved to transmit signals to the cytoplasm, the regulation of ECAPG biosynthesis could also occur by controlling the levels of the protein(s) responsible for ECAPG biosynthesis.

Previous work found deletion of elyC causes lysis of cells at room temperature in LB medium with 1% NaCl (LB Miller) (28). This lysis occurred due to a severe defect in peptidoglycan biosynthesis, which was attributed to allocation of Und-P between biosynthesis pathways. The peptidoglycan biosynthesis defect was not observed in cells grown at 37°C (28), the temperature at which our experiments were performed. Nevertheless, our results are consistent with the observation of isoprenoid stress inhibiting peptidoglycan synthesis in ΔelyC strains. Specifically, we have determined that this stress is due to the overproduction of ECAPG rather than to the initial allocation of Und-P for peptidoglycan biosynthesis. In fact, our data demonstrate that the effect of ElyC overexpression on total ECA levels is epistatic to the allocation of Und-P between biosynthesis pathways (Fig. 5). Thus, the effect of ElyC on Und-P availability for peptidoglycan synthesis is due to its role in regulating the synthesis of ECAPG.

Our data suggest three possible explanations for the more severe phenotype of elyC loss at lower temperatures. (i) Disruption or dysregulation of biosynthetic or transport pathways, such as protein secretion, can lead to cold-sensitive phenotypes due to slowing of the pathway at lower temperatures (84). Thus, growth at room temperature might slow ECA synthesis more than peptidoglycan synthesis, leading to increased sequestration of Und-P at lower temperatures. (ii) We have observed the chain length (number of repeat units per molecule) of ECA is less at 30°C than at 37°C (36). Thus, the same amount of ECA repeat units will be made into more final ECA molecules, increasing the amount of Und-P utilized for ECA synthesis at lower temperatures. (iii) Finally, ElyC may have an additional function at room temperature that also diverts Und-P from peptidoglycan synthesis that is not apparent at 37°C. Interestingly, Kouidmi et al. found an increase in periplasmic protein aggregation when ΔelyC cells were grown at room temperature that could be suppressed by overproduction of two periplasmic chaperones, DsbG and Spy, leading to restoration of peptidoglycan biosynthesis (69). These data may suggest an additional function for ElyC during growth at low temperatures.

The three forms of ECA are synthesized from a common precursor—polymerized ECA on Und-PP. Clearly, mechanisms are necessary to ensure the proper balance is maintained between the forms of ECA, both to support their proper functions and to avoid stress caused by dysregulation of biosynthetic pathways. WaaL, the O-antigen ligase, attaches both ECA and O-antigen to LPS (43, 85). In smooth strains that produce O-antigen, very little ECALPS is produced (43, 85, 86), suggesting that the availability of WaaL is a rate-limiting step in the production of ECALPS. Our results confirm that increasing the levels of WaaL causes more ECALPS to be produced. Our data further suggest the regulation of production of ECAPG and ECACYC is dependent on feedback regulation by ElyC based on ECACYC levels. The lesser effect of this regulatory pathway on ECALPS may suggest that ECALPS production is largely a by-product of O-antigen synthesis. In many Enterobacterales, O-antigen shares an initial GlcNAc residue with ECA (43, 87, 88), which may lead WaaL to be somewhat promiscuous. Nevertheless, surface exposure of ECALPS leads to production of ECA antibodies, the consequences of which have not been fully explored (21).

Since the discovery of ECACYC, there has been longstanding debate about whether ECACYC has a role in biosynthesis of the other forms of ECA or plays its own functional role within the cell. Previously, we demonstrated ECACYC plays a role in maintaining the OM permeability barrier (36). In our current work, we show that ECACYC also plays a role in regulating the synthesis of ECAPG. Thus, our work clearly indicates that ECACYC plays a dual role in the cell—both necessary for the proper function of the OM permeability barrier and involved in the regulation of ECA synthesis. This can be compared to classic biosynthetic pathways, such as those for amino acid synthesis, where the product of the pathway, useful in and of itself, also acts to allosterically regulate its own production, maintaining a constant pool of the biosynthetic product (89). Overall, the discovery of the ElyC-ECACYC pathway controlling ECAPG biosynthesis will provide a foothold in characterization of the mechanism of ECAPG biosynthesis, in understanding the regulation of ECA synthesis under changing environmental conditions, and in investigating both the functional and regulatory role of ECACYC.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Bacterial strains, plasmids, and primers used in this study are listed in Table S4 in the supplemental material. Strains were grown at 37°C in LB Lennox medium with the necessary antibiotics: kanamycin (25 mg/liter), chloramphenicol (20 mg/liter), and tetracycline (20 mg/liter), unless otherwise noted. IPTG at the indicated concentrations (0 to 100 μM) was used for overexpression from the pCA24N plasmid. l-Arabinose and α-d-fucose at the indicated concentrations were used to induce or repress the PBAD promoter, respectively. The deletion alleles were utilized from the Keio collection (65), unless otherwise indicated. New deletion alleles were constructed using λ-Red recombineering, as has been described previously (90). Mutants were made by P1vir transduction (91). Markerless deletion strains were generated by flipping out the kanamycin cassette with Flp recombinase as described previously (90).

Bacterial strains, plasmids, and primers. Download Table S4, PDF file, 0.2 MB.

Plasmids from the ASKA library were used for overexpression experiments (79). elyC with its native ribosome binding site (RBS) was cloned into pBAD33 (92) through HiFi Assembly (New England Biolabs [NEB]) per the manufacturer’s protocol using the pBAD33 and elyC (o/l pBAD33) primers (Table S4). Subsequently, the chloramphenicol resistance cassette was replaced with the kanamycin resistance cassette from pZS21 (93) using HiFi assembly and the pBAD33-elyC and kanR primers (Table S4). The promoter region of the wec operon was cloned from −500 to +20 in relation to the start codon of wecA upstream of the luxCDABE operon in the pJW15 vector (80, 81) using HiFi assembly and the Pwec and pNLP10/JW15 primers (Table S4).

TraDIS sample preparation and analysis.

Transposon mutant libraries were constructed from ΔwzzE ΔwaaL and ΔwecA ΔwzzE ΔwaaL strains by electroporation of the EZ-Tn5 Tnp transposome (Lucigen) as previously described (94). The library in wild-type MG1655 was previously described (94). About 306,000 and 186,000 individual colonies were pooled for the initial transposon library of the ΔwzzE ΔwaaL and ΔwecA ΔwzzE ΔwaaL strains, respectively. Liquid LB cultures were grown from the pooled libraries of mutants for 10 generations. DNA was extracted from the pooled libraries before and after growth in liquid medium using the DNeasy blood and tissue kit (Qiagen) according to the manufacturer’s instructions. Next, Illumina DNA fragment libraries were prepared using the TraDIS approach and sequenced on an Illumina HiSeq 2500 sequencer, as has been described previously (64, 94). Data were analyzed and mapped to the E. coli K-12 genome NC_000913.3 as has been described previously (94).

Analysis of ECA levels. (i) Immunoblot analysis for linear ECA levels.

ECA levels were assayed from overnight cultures, as previously described with slight modifications (36). Specifically, membranes were probed with a rabbit polyclonal anti-ECA antibody at a 1:30,000 dilution (a kind gift from Renato Morona at the University of Adelaide). Goat anti-rabbit secondary antibody conjugated to horseradish peroxidase (Prometheus) was used at a 1:100,000 dilution and detected using Prosignal Pico ECL (Prometheus) using Prosignal enhanced chemiluminescence (ECL) blotting film (Prometheus).

(ii) Dot blot for ECA surface exposure.

Surface-exposed ECA was detected using a dot blot assay as has been described previously with minor modifications (72). Specifically, the following antibodies were used for detection: anti-ECA (1:5,000), anti-BamD (1:5,000), and anti-RcsF (1:5,000) (95, 96).

(iii) WGA staining for ECALPS quantification.

Standard conditions for WGA staining of ECALPS were as follows. Two hundred fifty microliters of overnight culture was centrifuged for 3 min at 3,400 × g in round-bottom 96-well plates. After removing the supernatant, pelleted cells were washed with 200 μl of 1× phosphate-buffered saline (PBS). After washing, cells were resuspended in 200 μl of 1× PBS with a 1:100 volume of WGA-AF488 (Thermo Fisher Scientific) prepared per the manufacturer’s instructions. Samples were incubated in the dark at room temperature for 10 min. Then, cells were washed twice with 200 μl 1× PBS and resuspended in 110 μl 1× PBS. Next, 100 μl of each sample was aliquoted to a black-wall, clear-bottom 96-well plate where the optical density at 600 nm (OD600) and fluorescence at excitement (Ex.) 485 nm and emission (Em.) 519 nm was recorded using a BioTek Synergy H1 plate reader autogained based on sample fluorescence. Fluorescence relative to OD600 was calculated.

(iv) Quantification of ECACYC.

For ECACYC quantification, ECACYC was purified and subjected to matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry, and the relative abundance of ECACYC between samples was calculated as has been described previously (29, 36).

P reporter assay.

Overnight cultures of the indicated strains containing the pJW15-P plasmid were subcultured (1:100) into 100 μl of fresh LB broth in a black-wall, clear-bottom 96-well plate. The plate was sealed with a Breathe-Easy sealing membrane (Sigma), and the luminescence and OD600 of each strain were measured every 3 min for 6 h using a BioTek Synergy H1 plate reader as previously described (80, 81). Each biological replicate was performed in technical quadruplicate.

Data availability.

The sequencing data are available in the Sequence Read Archive database (SRA) (https://www.ncbi.nlm.nih.gov/sra, BioProject ID PRJNA763934).