Core Oligosaccharide Portion of Lipopolysaccharide Plays Important Roles in Multiple Antibiotic Resistance in Escherichia coli

ABSTRACT

Gram-negative bacteria are intrinsically resistant to antibiotics due to the presence of the cell envelope, but the mechanisms of this resistance are still not fully understood. In this study, a series of mutants that lack one or more major components associated with the cell envelope were constructed from Escherichia coli K-12 W3110. WJW02 can only synthesize Kdo₂-lipid A, which lacks the core oligosaccharide portion of lipopolysaccharide (LPS). WJW04, WJW07, and WJW08 were constructed from WJW02 by deleting the gene clusters relevant to the biosynthesis of exopolysaccharide, flagella, and fimbriae, respectively. WJW09, WJW010, and WJW011 cells cannot synthesize exopolysaccharide (EPS), flagella, and fimbria, respectively. Compared to the wild type (W3110), mutants WJW02, WJW04, WJW07, and WJW08 cells showed decreased resistance to more than 10 different antibacterial drugs, but the mutants WJW09, WJW010, and WJW011 did not. This indicates that the core oligosaccharide portion of lipopolysaccharide plays an important role in multiple antibiotic resistance in E. coli and that the first heptose in the core oligosaccharide portion is critical. Furthermore, the removal of the core oligosaccharide of LPS leads to influences on cell wall morphology, cell phenotypes, porins, efflux systems, and response behaviors to antibiotic stimulation. The results demonstrate the important role of lipopolysaccharide in the antibiotic resistance of Gram-negative bacteria.

INTRODUCTION

Antibiotic resistance is one of the biggest threats to global public health and is becoming a growing concern for human and animal health due to the considerable spreading of antibiotic-resistant microorganisms (1). Gram-negative bacteria are intrinsically resistant to various antibiotics due to the presence of the outer membrane (OM) (2). This antibiotic resistance mechanism is different from those caused by either mutation or acquisition of genetic loci. The OM is the first important barrier to prevent antibiotics from entering bacterial cells. Lipopolysaccharide (LPS) (3), exopolysaccharide (EPS), flagella, and fimbriae are important macromolecules associated with OM (4). In Escherichia coli K-12, colanic acid (CA) is a major EPS (5, 6), and LPS molecules bind the divalent metal cations to form a well-ordered polyelectrolyte barrier (7). Biofilms are surface-adhering structures that are produced by bacteria in response to stress (8–10); they contain polysaccharides, lipids, nucleic acids, and protein entities like flagella, fimbriae, and OM vesicles (OMVs) (11, 12). The release of EPS through OMVs increases coaggregation of cells in the biofilms. OMVs contain pathogen-associated molecular patterns such as LPS, flagella, and porins. Some antibiotics can trigger OMV formation by causing cell envelope stress, inducing the SOS response or inhibiting cell wall biosynthesis (13), but OMVs usually contain enzymes that can inactivate antibiotics (14). E. coli OMVs can sequester colistin (polymyxin E [PE]) and degrade the antimicrobial peptide melittin, and addition of OMVs to bacterial cultures could protect bacteria from polymyxin and melittin (15). Flagellar proteins also contribute to the production of OMVs from E. coli W3110 (16). Biofilms or OMVs play important roles in antibiotic resistance (17), and LPS, EPS, flagella, and fimbriae are important components for biofilm or OMV formation (4, 8), and therefore it would be interesting to see the individual contribution of these molecules to the antibiotic resistance of bacteria.

LPS provides a permeation barrier against the entry of many antibiotics (18, 19). To establish a permeability barrier, millions of LPS molecules must be properly assembled on the OM (20). The inner core of LPS in Salmonella enterica subsp. enterica serovar Typhimurium plays important role in antibiotic resistance, especially against hydrophobic antibiotics (21–23); heptose-deficient LPS mutants are more sensitive to nafcillin (NF), rifamycin (Rif), erythromycin (EM), bacitracin, vancomycin (Van), novobiocin (Nov), kanamycin (Kan), and cloxacillin (21–23). How the heptose groups of LPS in E. coli K-12 strains influence antibiotic resistance remains unclear. The integrity of LPS molecule is important for maintaining normal function of porins and bacterial phenotypes. The truncation of LPS can influence the formation of porins (24), which plays a major role in resistance to some antibiotics, particularly β-lactams (25, 26). The specific interactions between the trimeric porins of Enterobacteriaceae and LPS are stabilized by calcium ions (27). However, how structural changes in LPS influence the expression of outer membrane porins (OMPs) needs to be further investigated.

The gene clusters relevant to the biosynthesis of LPS, CA, flagella, and fimbriae are distributed at different locations in the chromosome of E. coli (Fig. 1a). LPS consists of a polysaccharide and Kdo₂-lipid A, and Kdo₂-lipid A is the simplest LPS structure required for cell survival (Fig. 1b) (28, 29). The biosynthesis of polysaccharide portion of LPS involves 25 genes. Deletion of either gmhD or waaC can lead to the truncation of LPS to Kdo₂-lipid A in E. coli (Fig. 1c) (30). WaaC and WaaF are responsible for the addition of the first and second heptose to Kdo₂-lipid A (3). In E. coli K-12 strains, O-antigen repeats cannot be synthesized due to the mutation of wbbL (3); therefore, E. coli LPS usually only contain a core oligosaccharide, which includes 4 heptoses, 3 glucoses, and 1 galactose. In E. coli, the biosynthesis of CA involves 21 genes, the biosynthesis and assembly of flagella involves 3 gene clusters that include 57 genes, and the biosynthesis and assembly of type I fimbriae involves 9 genes. The truncation of LPS structure can influence the biosynthesis of flagella and CA. Deletion of waaC, waaF, or waaG led to the absence of flagella, and the deletion of waaF enhanced the biosynthesis of CA (31–33).

FIG 1.

Deleted gene clusters (a) and various lipopolysaccharide (LPS) structures (b and c) for different Escherichia coli mutants. (a) LPS mutant WJW02 was constructed by deleting the gene clusters for biosynthesis of LPS core (14 genes) and O-antigen (11 genes). WJW04 was constructed by deleting four gene clusters (53 genes) for biosynthesis of LPS (25 genes) and exopolysaccharide (EPS; 28 genes). WJW07 was constructed by deleting seven gene clusters (110 genes). WJW08 was constructed by deleting 8 clusters (119 genes). WJW09 is an EPS mutant constructed by deleting 28 genes involved in colonic acid biosynthesis and type IV EPS. WJW010 is a flagellum mutant lacking 57 genes relevant to flagellum biosynthesis. WJW011 is a fimbrial mutant lacking 9 genes relevant to the biosynthesis of type I fimbriae. (b) The whole LPS structure contains polysaccharide and Kdo₂-lipid A. In E. coli W3110, LPS lacks O-antigen. (c) Comparison of LPS structures synthesized in W3110, ΔwaaC, WJW00, WJW01, WJW02, and ΔwaaFwaaL-Q strains.

In this study, we constructed E. coli mutants from E. coli W3110 that were unable to synthesize LPS core oligosaccharide, CA, flagella, and/or type I fimbriae (Fig. 1a) and analyzed their cell characteristics and resistance to 26 antibacterial drugs. The results demonstrate that the core oligosaccharide portion of LPS—but not CA, flagella, or fimbriae—is important for antibiotic resistance. Further analysis of different LPS mutants showed that the first heptose is the most critical group for antibiotic resistance in E. coli. The results help us to further understand how OM-associated molecules influence bacterial antibiotic resistance and provide a theoretical basis for developing strategies to target Gram-negative infections.

RESULTS

The core oligosaccharide portion of LPS—but not EPS, flagella, or fimbriae—plays an important role in multiple antibiotic resistance in E. coli.

To study the influences on the antibiotic resistance of E. coli LPS, EPS, flagella, and fimbriae, the mutants WJW02 (deletion of 25 genes for LPS biosynthesis), WJW04 (deletion of 53 genes for LPS and EPS biosynthesis), WJW07 (deletion of 110 genes for LPS, EPS, and flagellum biosynthesis), WJW08 (deletion of 119 genes for LPS, EPS, flagellum, and fimbria biosynthesis), WJW09 (deletion of 28 genes for EPS biosynthesis), WJW010 (deletion of 57 genes for flagellum biosynthesis), and WJW011 (deletion of 9 genes for fimbria biosynthesis) were constructed (Fig. 1).

A total of 26 antibacterial drugs that target different molecules, including DNA, RNA, proteins, the cell wall, the cell membrane, and fatty acids, were chosen to study the resistance of different E. coli mutants to antibacterial drugs (Fig. 2). E. coli cells were grown in LB medium (10 g/liter tryptone, 5 g/liter yeast extract, and 10 g/liter NaCl), and the MICs of different antibacterial drugs against the strains were determined. Compared to the control W3110, E. coli mutants WJW02, WJW04, WJW07, and WJW08 containing truncated LPS showed decreased MICs for many drugs, but mutants WJW09, WJW10, and WJW011, which contained the whole LPS, showed similar MICs for many drugs (Fig. 2). The results suggest that the core oligosaccharide portion of LPS plays an important role in multiple antibiotic resistance in E. coli. Rif inhibits RNA polymerase, and the MIC of Rif against WJW02 decreased 32-fold (Fig. 2a). MICs of the DNA gyrase-targeting drugs Nov and enrofloxacin (EN) against WJW02 cells decreased 128- and 8-fold, respectively (Fig. 2a and c). MICs of protein synthesis-targeting EM, clarithromycin (Cla), and clindamycin (CL) against WJW02 cells decreased 16-, 32-, and 64-fold, respectively (Fig. 2a). MICs of gentamicin (GM), Kan, and chloramphenicol (CM) against WJW02 decreased 4-fold. For other aminoglycoside antibiotics, MICs of WJW02 also showed a 2-fold decrease. Against WJW04, WJW07, and WJW08 mutants, in which LPS were also truncated, most antibiotics showed similar MICs to those against WJW02.

FIG 2.

Comparison of MICs of 26 antimicrobial drugs against different E. coli mutants. (a) MICs of different antibiotics that target DNA and protein biosynthesis. Rif targets the β-subunit (RpoB) of DNA-dependent RNA polymerase; Nov targets DNA gyrase; and aminoglycosides, macrolides, lincomycin, and others target protein synthesis. (b) MICs of antibiotics that target cell wall synthesis, including bacitracin, fusidic acid (Fus), and β-lactams. (c) MICs of polymyxins, including polymyxin B (PB) and PE, that target cell membrane; enrofloxacin (EN), which targets DNA; and triclosan, which targets fatty acid biosynthesis. (d) MICs against ΔwaaFwaaL-Q, ΔwaaC, WJW00, and WJW01 mutants, using WJW02 and W3110 as controls.

Compared with W3110, mutants WJW02, the WJW04, WJW07, and WJW08 mutants showed 4-fold decreased MICs to fusidic acid (Fus) and 2-fold decreased MICs to bacitracin (Fig. 2b). For most β-lactams except for ampicillin (Amp), WJW02 showed 2-fold increased MICs compared to those against W3110 (Fig. 2b). Compared with WJW02, the WJW04, WJW07, and WJW08 mutants showed 2-fold decreased MICs for Amp, amoxicillin (Amx), and cefoperazone (Cfp) (Fig. 2b), while WJW09 and WJW011 showed decreased MICs to Amp. WJW07 and WJW08 showed 4-fold decreased MICs to cefoxitin (Fox) compared to that against WJW02, and WJW010 and WJW011 showed 2-fold decreased MICs to Fox compared to that against W3110 (Fig. 2b). The results suggest that the core oligosaccharide portion of LPS may play roles in resistance to bacitracin and Fus, exopolysaccharide and fimbriae may play roles in Amp resistance, and flagella and fimbriae show important roles in Fox resistance. MICs of the cell membrane-targeting polymyxin B (PB) and PE drugs against WJW02 cells decreased 4- and 2-fold, respectively. The MIC of fatty acid biosynthesis-targeting drug triclosan against WJW02 decreased 8-fold (Fig. 2c).

The ΔwaaC, WJW00, and WJW01 mutants can only synthesize Kdo₂-lipid A, while the ΔwaaFwaaL-Q mutant (lacking genes waaF and waaLUZYROBSPGQ) can add a heptose to Kdo₂-lipid A, forming Hep-Kdo₂-lipid A. The ΔwaaC, WJW00, and WJW01 mutants showed the same MICs as WJW02, but the ΔwaaFwaaL-Q mutant showed higher MICs to various antibiotics than WJW02. Compared to WJW02, the ΔwaaFwaaL-Q mutant showed 32-, 8-, and 4-fold higher MICs to Nov, CL, and Fus, respectively (Fig. 2d). The results indicate that the first heptose group in LPS is most critical for the resistance of E. coli to various antibiotics.

Since in most situations, E. coli mutants WJW09, WJW10, and WJW011 showed similar MICs compared to those against the wild type, W3110, it was necessary to see if flagella, fimbriae, and EPS were produced under the conditions of MIC determination. To this end, E. coli W3110 cells were grown with the addition of Rif, Nov, CL, Kan, or PB at the concentration of 0.5× MIC and observed under microscopy. The flagella and fimbriae were observed around W3110 cells grown both with and without antibiotics, indicating that flagella and fimbriae were produced in W3110 cells grown under the conditions of MIC determination (Fig. 3a). Notably, the transcriptional levels of genes responsible for the flagella, fimbriae, and EPS were downregulated in WJW02 compared to W3110 as the control (Fig. 3b and c). These results suggest that LPS truncation can downregulate the expression of flagella, fimbriae, and EPS, but is not the factor responsible for the increased sensitivity of WJW02 cells to various antibiotics. The results further confirm that the core oligosaccharide portion of LPS plays more important roles in resistance to most antibiotics.

FIG 3.

Analysis of flagella and fimbria formation in E. coli under conditions of MIC determination. (a) E. coli W3110 cells grown in under conditions of MIC determination with the addition of Rif, Nov, CL, Kan, or PB were observed by transmission electron microscopy (TEM). (b) The transcriptional levels of genes responsible for the flagella and fimbriae in WJW02 were determined by transcriptome analysis, using W3110 as the control. (c) The transcriptional levels of genes responsible for EPS in WJW02 were determined by transcriptome analysis, using W3110 as the control.

LPS truncation influences key characteristics relevant to antibiotic resistance.

To understand the antibiotic resistance mechanism caused by different OM-associated molecules, the cell growth, motility, hydrophobicity, OM permeability, and biofilm formation of different E. coli mutants were analyzed (Fig. 4). For the growth patterns, none of the mutants showed obvious changes compared to the wild type, W3110. Mutants WJW07 and WJW08 showed slightly higher growth rates during the log phase, which may be due to the lack of flagella (34) (Fig. 4a). Compared to W3110, WJW02 showed sharply decreased cell motility, WJW010 showed no motility, WJW011 showed slightly decreased motility, and WJW09 showed slightly increased motility (Fig. 4b). Comparing to W3110, WJW02 cells showed 111% increased hydrophobicity (Fig. 4c), 4.71-fold increased OM permeability (Fig. 4d), and 83% decreased biofilm formation (Fig. 4e); WJW09 showed 49% increased hydrophobicity (Fig. 4c); WJW010 showed slightly increased biofilm formation (Fig. 4e); and WJW011 alone showed 80% decreased biofilm formation (Fig. 4e). WJW09, WJW010, and WJW011 cells showed similar permeability to that of the wild type, W3110 (Fig. 4d). Indeed, permeability is a well-known critical factor for antibiotic resistance (24, 25). These results further confirmed that the truncated LPS leads to stronger hydrophobicity and permeability, which facilitate the diffusion of hydrophobic antibiotics into cells (21, 22).

FIG 4.

Cell phenotype analysis for different E. coli mutants. (a) Growth curves of different E. coli mutants grown in LB medium at 37°C and 200 rpm rotation. The experiments were replicated three times, and error bars show the standard deviation from the mean of the three independent experiments. (b) Motility of cells on 0.3% agar LB plates. (c) Hydrophobicity of cells. (d) Outer membrane (OM) permeability of different E. coli cells. (e) Biofilm formation ability of different E. coli cells. (f) Relative transcriptional levels of ompF, ompC, and ompA in E. coli W3110 and WJW02. (g) SDS-PAGE analysis of outer membrane porins (OMPs) extracted and purified from E. coli W3110, WJW02, WJW07, WJW08, WJW04, WJW09, WJW010, and WJW011.

WJW02 showed increased resistance trends to β-lactams. In E. coli, β-lactams could be taken into the cells though the major porins OmpF and OmpC (35). The influence of LPS on porin folding has been widely studied for decades (24–27), and a recent study revealed specific interactions between the trimeric porins and LPS as a trimeric porin-LPS complex that is stabilized by calcium ions (27). However, the influence of the absence of the core oligosaccharide portion of LPS, together with the absence of CA, flagella, and/or fimbriae, on outer membrane porins (OMPs) is not known. Therefore, the OMPs in various E. coli mutants were investigated. WJW02 synthesized much less OmpF and OmpC, but similar levels of OmpA, compared to W3110 (Fig. 4g). The real-time PCR analysis also showed that the relative transcriptional levels of ompF and ompC in WJW02 decreased 92.2% and 40.8%, respectively (Fig. 4f). The results confirm that the truncated LPS is important for the biosynthesis of OmpC and OmpF (24–27) and indicate that the core oligosaccharide portion of LPS is important for OMPs at both RNA transcription and protein expression levels. WJW04, WJW07, and WJW08 synthesized slightly more OmpF and OmpC than WJW02, but much less than W3110, while WJW09, WJW010, and WJW011 showed similar amounts of OMPs to W3110 (Fig. 4g). The decreased levels of OmpF and OmpC in WJW02 would slow down β-lactam uptake, resulting in the increased resistance trends for β-lactams.

LPS truncation could influence the transcriptional levels of genes related to antibiotic resistance in E. coli.

To better understand the influence on E. coli cells of removal of the core oligosaccharide portion of LPS, we performed transcriptomic analysis of WJW02, using W3110 as the control. Expression of genes involved in drug efflux or resistance changed after LPS truncation (Fig. 5a). The efflux genes mdtIJ, emrE, yeeO, acrA and the resistance gene blr were upregulated, but yddA and prmD were downregulated. The 2^5.7-fold upregulation of blr, responsible for β-lactam resistance, is consistent with the increased MICs of β-lactams against E. coli WJW02. YddA is an ATP-binding cassette (ABC)-type multidrug exporter (36), and PmrD is responsible for resistance to colistin (37). The downregulated pmrD in WJW02 indicates that the core oligosaccharide portion of LPS, in addition to lipid A, may also participate in regulating PB and PE resistance. Moreover, genes relevant to osmotic stress were upregulated in WJW02 (Fig. 5b), suggesting that E. coli cells have to activate the transcription of these genes to struggle again osmotic stress when LPS is deeply truncated. The results further highlight the important role of the LPS core oligosaccharide on multiple antibiotic resistance.

FIG 5.

Transcriptome analysis of WJW02, using W3110 as a control. (a) Relative transcriptional levels (log₂R) of genes involved in multidrug efflux and antibiotic resistance. (b) Relative transcriptional levels (log₂R) of genes involved in osmotic stress.

LPS truncation influences cell morphology under stimulation by different antibiotics.

To understand how the LPS core oligosaccharide influences the cellular response to different antibiotics, the cell morphology changes of E. coli W3110 and WJW02 under stimulation by different antibiotics were analyzed (Fig. 6). According to their classification, six representative drugs—Rif, Nov, EM, CL, Kan, and PB—were chosen to stimulate E. coli W3110 and WJW02 cells.

FIG 6.

Cell and membrane morphology analysis by TEM and ultrathin-section TEM for E. coli W3110 and WJW02 cells under stimulation by different antibiotics. W3110-drug and WJW02-drug represent cells under the stimulation of the corresponding drug at 1× MIC. E. coli cells were grown in LB with or without drugs at 37°C and 200 rpm rotation for 30 min, and the cells were immediately collected at 4°C to prepare for TEM and ultrathin-section TEM analysis.

Transmission electron microscopy (TEM) images suggest that E. coli cells could struggle to survive under stimulation by different antibiotics. Under stimulation by Rif, which inhibits DNA transcription by binding to RNA polymerase (38), both W3110 and WJW02 cells showed a larger nucleic acid region; W3110 cells shrunk, while WJW02 cells showed cellular lysis, i.e., the cell membrane was still intact but its cytoplasm was lysed. Nov inhibits DNA gyrase by binding to the ATP-binding site (7). Under stimulation by Nov, both W3110 and WJW02 cells showed bright nucleic acid regions and formed many vesicles (Fig. 6, blue arrows) on the cytoplasmic side of the inner membrane; WJW02 cells became more expanded and more easily stained than W3110 cells. EM and CL are inhibitors of protein synthesis (39). Under stimulation by EM, W3110 cells showed a large amount of secretion around the OM, but WJW02 cells showed little secretion, and few WJW02 cells were lysed. Under stimulation by CL, the major difference between W3110 and WJW02 cells was that the former synthesized flagella. Under stimulation by Kan, many bulges on the cell surface and many patches in the cytoplasm near the inner membrane (Fig. 6, black arrows) were observed for both W3110 and WJW02 cells, possibly caused by aggregated proteins (40). PB is a cationic lipopeptide antibiotic targeting LPS (41); it has a similar chemical structure to that of colistin (PE), the last-line therapeutic agent against multidrug resistant Gram-negative bacteria (42). Under PB stimulation, W3110 cells formed many lumps on the surface, possibly because the insertion of the fatty acyl chain within the OM formed pore-like aggregates (42); WJW02 cells showed many vesicular secretions (Fig. 6, yellow arrows) but no lumps on the surface. The results indicate that LPS truncation affected the response of the cell envelope to different antibiotics. This further highlights the importance of the core oligosaccharide portion of LPS for multiple antibiotic resistance.

DISCUSSION

This study demonstrated that the core oligosaccharide portion of LPS plays important roles for antibiotic resistance in E. coli and that the first heptose of LPS is critical; however, CA, flagella, and fimbriae have no obvious impacts. Further study indicated that the phenotypic characteristics caused by the removal of core oligosaccharide portion of LPS, such as increased OM permeability and decreased biofilm formation, are closely related to antibiotic susceptibility. Notably, antibiotic stimulation experiments also suggest that removal of the core oligosaccharide portion of LPS influences the response behaviors of cell morphology to resist antibiotics, especially that of the cell wall. These results highlight the importance of the core oligosaccharide portion of LPS in multiple antibiotic resistance.

The effect of truncated LPS on cell morphology is unfavorable to antibiotic resistance in bacteria. Generally, when the integrity of the OM is damaged, E. coli cells activate sensing systems, leading to remodeling of the cell envelope to be more resistant to antibiotics (43). However, the removal of core oligosaccharide portion of LPS causes downregulation of yddA (responsible for the resistance against Rif, Nov, and EM) (36) and pmrD (responsible for the resistance against polymyxins) (37).

The phosphate diester bridges and salt bridges formed between the LPS are very important in forming a cell surface structure resistant to the penetration of antibiotics such as Nov, spiramycin (SP), and actinomycin D (3, 7, 23). The removal of polysaccharide from LPS in WJW02 led to weaker cellular protective barrier, and lack of the first heptose led to the weakest resistance against many antibiotics, suggesting that the first heptose of LPS is the most critical group for antibiotic resistance (22). The increased permeability of the OM in WJW02 is the main contributor to antibiotic susceptibility, especially against self-promoted uptake drugs such as Rif, Nov, EM, Kan, and PB (2). Other mutants lacking the core oligosaccharide portion of LPS also showed increased OM permeability, and the mutants lacking the first heptose of LPS showed the highest permeability (31). Biofilm formation could increase antibiotic resistance 1,000-fold higher than that of planktonic cells (44), but it is not clear whether the decreased biofilm formation of WJW02 cells contributes to its antibiotic sensitivity. It is true that WJW02 is most susceptible to antibiotics in a standard MIC assay, but in a standard MIC assay, bacteria are planktonic, growing in the log phase and not forming biofilms.

The decreased OmpF in WJW02 cells leads to increased resistance to β-lactams (26). LPS truncation can influence the normal folding and assembly of OmpF (45), and Gram-negative trimeric porins have specific LPS binding sites that are essential for porin biogenesis (27). Our results confirm that deeply truncated LPS has negative impacts on the transcription and expression of OmpF and OmpC. In addition to the decreased expression of OmpF, the remarkably upregulated blr gene may be another factor in the increased resistance of WJW02 to β-lactams (35). Moreover, this study also demonstrated that the removal of CA, flagella, and fimbriae did not influence porins, including OmpF formation; more OmpF was produced in WJW04, WJW07, and WJW08 cells than in WJW02 cells. Therefore, only WJW02 shows increased resistance to β-lactams.

Removal of the core oligosaccharide portion of LPS influences the response manner of the cell envelope. For Rif and Nov, TEM analysis showed that the nucleic acid region in E. coli cells became bright, indicating that the entry of Rif or Nov facilitates an attack on the nuclear area (7, 38). Nov could bind the interface between the ATPase subunits and the transmembrane subunits of LPS transporter LptB₂FG to alter OM permeability (7). The vesicles observed in the TEM images of E. coli cells under stimulation by Nov may be related to such binding. For the protein-targeting drugs EM and CL, a previous report also indicated that LPS-defective mutants had higher susceptibility to many macrolide antibiotics (46). Aminoglycosides can bind to LPS molecule, leading to self-uptake across the OM (47); the removal of core oligosaccharide portion of LPS facilitates the membrane crossing of Kan and leads to the aggregation of cellular proteins. Bacterial resistance to polymyxins occurs via several mechanisms (42, 48–50), and bacterial cross-resistance exists against PB and colistin (42, 51). In E. coli, PB or colistin resistance is mainly caused by lipid A modification (41), including the attachment of phosphoethanolamine (PEtN) and 4-amino-4-deoxy-l-arabinose (l-Ara4N) to the phosphate groups of lipid A and the attachment of PEtN to Kdo₂-lipid A. Bacterial resistance to colistin may also be acquired through transposable genetic elements such as mcr, whose product can also add a PEtN group to lipid A (51, 52). The upregulation of the arnBCADTEF operon responsible for the addition of l-Ara4N is activated by PmrD (49). The pmrD gene was downregulated in WJW02, consistent with this strain’s increased sensitivity to PB. Under stimulation by PB, OM folding observed for W3110 cells but not for WJW02 cells might be related to PEtN and l-Ara4N modification of LPS. Lpt proteins for transporting LPS are potential targets of antimicrobial agents (50). Therefore, the bacterial cell envelope has important contributions to antibiotic resistance, but more detailed mechanisms need to be investigated.

MATERIALS AND METHODS

Construction of E. coli mutant strains.

Bacterial strains constructed in this study are listed in Table 1, and plasmids and primers used in this study are listed in Tables S1 and S2 in the supplemental material, respectively. E. coli K-12 W3110 was chosen because it is a typical strain for investigating the OM-associated molecules, including LPS, EPS, CA, outer membrane proteins, lipoproteins, flagella, and fimbriae. E. coli W3110 has been extensively studied for producing vaccine adjuvants (29, 53–56). WJW02 was constructed by site recombination (57), in which the LPS core gene cluster was deleted by an Flp/FLP recombination target (FRT) system, and the LPS O-antigen gene cluster was deleted using a Cre/lox mutant system. For deletions of the other genes clusters, the CRISPR-Cas9 system (58) was applied. Briefly, the plasmid pCas containing a cas9 gene was first introduced into the started strain (59). Overnight cultures of these resultant strains were inoculated in LB medium to an optical density at 600 nm (OD₆₀₀) of 0.04 supplemented with 10 mM arabinose and 50 mg/liter Kan for plasmid maintenance and grown at 30°C and 200 rpm until the OD₆₀₀ reached 0.6. For preparation of competent cells harboring pCas, 50-ml induced cultures were collected, washed three times with an ice-cold 10% glycerol-water solution, and then stored in 2 ml of identical glycerol solution at −70°C. Various targeting plasmids for the desired knockout genes were derived from the original pTargetF (pMB1 replicon; spectinomycin resistance) by replacing the native N₂₀ sequence upstream of the single guide RNA (sgRNA). In detail, the corresponding N₂₀ sequences of target cassettes were embedded into primers to amplify the whole pTargetF plasmid. The PCR products were digested with DpnI for template plasmid elimination and subsequently transformed into E. coli DH5α to generate circular plasmids using the inherent repair system of bacteria with an overlap terminus. The pTarget genes were verified by PCR with the primers pTarget-genes and pTargetF-R.

TABLE 1.

Strains used in this study

Strain	Description	Source or reference
W3110	Wild-type Escherichia coli, F−, λ−	NEB
WJW00	Derived from W3110 by deleting gmhD; can only synthesize Kdo2-lipid A	30
WJW01	Derived from W3110 by deleting rfa operon (14 genes); can only synthesize Kdo2-lipid A	This study
WJW02	Derived from WJW01 by deleting wbbL-rfbB operon (11 genes); can only synthesize Kdo2-lipid A	This study
WJW03	Derived from WJW02 by deleting galF-wza operon (21 genes); cannot synthesize colanic acid	This study
WJW04	Derived from WJW03 by deleting yccC-ymcD operon (7 genes); cannot synthesize type IV EPS	This study
WJW05	Derived from WJW04 by deleting flhE-motA operon (13 genes); cannot synthesize flagella	This study
WJW06	Derived from WJW05 by deleting flagellar operon fliY-fliR (30 genes)	This study
WJW07	Derived from WJW06 by deleting flagellar operon flgN-flgL (14 genes)	This study
WJW08	Derived from WJW07 by deleting fimbrial operon fimB-fimH (9 genes)	This study
WJW09	Derived from W3110 by deleting EPS operons galF-wza and yccC-ymcD (28 genes)	This study
WJW010	Derived from W3110 by deleting flagellar operons flhE-motA, fliY-fliR, and flgN-flgL (57 genes)	This study
WJW011	Derived from WJW01 by deleting fimB-fimH operon (9 genes); cannot synthesize fimbria	This study
ΔwaaFwaaL-Q	Derived from W3110 by deleting waaF and waaL-waaQ operon (11 genes); can synthesize Hep-Kdo2-lipid A which contains a single heptose	This study
ΔwaaC	Derived from W3110 by deleting waaC; can only synthesize Kdo2-lipid A	31

The construction of gene knockout strains was carried out according to a protocol reported previously (58). For deletion of the galF-wza cluster, the plasmid pTargetF-galF-wza, containing an N₂₀ sequence for targeting the galF-wza locus, was extracted from overnight culture using a TIANprep Mini plasmid kit (Tiangen Biotech, Beijing, China). The editing template fragment with two homologous arms corresponding to the upstream and downstream regions of the galF-wza locus was obtained by overlap extension PCR. Then, pTargetF-galF-wza together with the template fragment was electroporated and cotransformed into the starting strain containing pCas. A 1-ml aliquot of cell culture concentrated by centrifugation at 10,000 × g for 2 min was plated on LB agar containing Kan (50 mg/liter) and spectinomycin (50 mg/liter) for counterselection. Finally, the galF-wza knockout strain was obtained using colony PCR with the verified primers (galF-wza-U-F and galF-wza-D-R), and both pTargetF-galF-wza and the temperature-sensitive plasmid pCas were subsequently cured by addition of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and incubation at 42°C with shaking, respectively. Other cluster deletions in this study were constructed similarly as described above.

Determination of MICs.

For MIC determination, a serial dilution method with 96-well plates was used (60). The bacterial strains were grown to the mid-log phase (OD₆₀₀ = 1.0) at 37°C and then diluted to an OD₆₀₀ of 0.01 with fresh LB medium. The concentrated antibiotic solutions were prepared using LB medium as the solvent. First, 200 μl of the concentrated antibiotic solution was put into the wells on the first row of the 96-well plates. Then, 100 μl of the concentrated antibiotic solution was transferred from the first row of wells to the second row of wells and mixed with 100 μl fresh LB medium. In turn, the antibiotic solution was diluted 2-fold each time it was transferred into the next row of wells. After the antibiotic solution was transferred into the last row of wells and mixed with 100 μl fresh LB medium, 100 μl of the solution was removed from the wells to keep the volume the same as that in other wells. Next, 5 μl bacterial culture was added to each well. The 96-well plates were covered with lids and incubated at 37°C for 18 h, and the OD₆₀₀ was determined by a microplate reader. The MIC value is determined as the minimum concentration of antibiotic that can inhibit bacterial cell growth. When it is difficult to determine the minimum concentration of antibiotic that can inhibit bacterial growth, the MIC value is determined as the concentration that can inhibit the growth of more than 80% of the bacterial cells. Three independent measurements were performed for each antibiotic concentration during the MIC determination for every strain.

Electron microscopy analysis.

To analyze cell morphology, the secretion, flagella, and fimbriae of E. coli W3110 under the conditions of MIC determination, the cells were collected at 2,000 × g for 3 min after 18 h of incubation surrounded by the antibiotics Nov, Rif, CL, Kan, or PB at 0.5× MIC. To analyze the changes of cell morphology and membrane structures of E. coli W3110 and WJW02 under stimulation by the antibiotics Nov, Rif, EM, CL, Kan, or PB, the cells were collected at 12,000 × g for 1 min after 30 min of stimulation. The cell samples were prepared for the electron microscopy observation according to methods described in our previous study (61). For TEM analysis, cells were fixed with 2.5% glutaraldehyde (pH 7.2). For ultrathin-section electron microscopy analysis, the cells were washed twice with phosphate-buffered saline (PBS; pH 7.4) and then fixed with 2.5% glutaraldehyde (pH 7.2).

Motility, surface hydrophobicity, OM permeability, and biofilm formation assays.

For motility assay, cultures grown in LB with an OD₆₀₀ of 1.0 were stabbed on LB with 0.3% agar plates and incubated at 37°C for 48 h. Motility halos were then measured.

The surface hydrophobicity, OM permeability, and biofilm formation ability of cells were determined according to the methods described in a previous study (30). Cells were grown in LB to an OD₆₀₀ of 1.0, culture was collected and resuspended with PBS to an OD₆₀₀ of 0.5, and suspensions were prepared for the determination of surface hydrophobicity and OM permeability. For surface hydrophobicity, a 1-ml resuspension was mixed with 800 μl of xylene and then incubated at room temperature for 3 h. The OD₆₀₀ of the aqueous phase after extraction with xylene was recorded as A, and the value of [(A₀ − A)/A₀] × 100 represents the hydrophobicity of the bacterial cells. OM permeability was measured by the fluorescence absorption of cells with N-phenyl naphthylamine (NPN). The 1.92-ml suspension was added to 80 μl of NPN, and the fluorescence of the mixture was immediately monitored by a spectrofluorometer (Hitachi, Tokyo, Japan). An excitation wavelength of 350 nm, an emission wavelength of 420 nm, and a slit of 7 nm were used for the experiment. Endpoint fluorescence was normalized as a fold change compared to the untreated sample (vehicle control). The fluorescence absorption per OD₆₀₀ value indicated the OM permeability. For biofilm formation ability analysis, conditions of standing incubation and shaking cultivation for cells in LB at 37°C for 72 h were both studied. The values represent means with standard deviation (SD) of three independent measurements.

Outer membrane protein extraction and identification.

OMPs were isolated as described previously (62) and analyzed by SDS-PAGE. E. coli cells were grown in LB medium at 37°C until the OD₅₉₅ reached 0.6, and then 15 ml of the culture was harvested by centrifugation at 5,000 × g for 10 min. The cell pellets were resuspended in 500 μl PBS (pH 7.2) and sonicated on ice until the solution became clear. The cell debris were eliminated by centrifugation at 12,000 × g for 2 min, and the supernatants were transferred to clean microcentrifuge tubes. The membrane fractions in the supernatants were collected by centrifugation at 12,000 × g for 1 h, and then resuspended in a 500-μl solution containing 2% Triton X-100 for 30 min. Next, the membrane fractions were collected again by centrifugation at 12,000 × g for 1 h, washed with PBS, and resuspended in 50 μl PBS (pH 7.4). The cellular protein samples and the extracted OM samples were subjected to SDS-PAGE analysis. The outer membrane proteins loaded on each lane were extracted from about 3 ml of E. coli cells with an OD of 0.6.

Transcriptomic analysis for different E. coli strains.

The whole-genome transcriptional analyses of E. coli strains W3110 and WJW02 were established according to a previously published method (61). Cultures of E. coli strains were diluted 100 times into 50 ml LB medium and grown at 37°C to the early exponential phase with an OD₆₀₀ of ∼1.3. The collected cells were washed once with PBS (pH 7.4) and stored with RNA storage solution.

The libraries were sequenced using an Illumina HiSeq 2000 instrument (BGI Shenzhen, China). P values are for differential gene expression tests. The calculated P value was Bonferroni corrected (63), taking a corrected P value of ≤0.05 as a threshold. Correction for false-positive errors (type I) and false-negative errors (type II) were performed using the false-discovery rate (FDR) method (64). The experiments and transcriptomics were performed by BGI Shenzhen, China.

Antibiotic stimulation.

The W3110 and WJW02 cells were cultured to an OD₆₀₀ of 0.5 in LB medium at 37°C and 200 rpm. Next, the antibiotic at 1× MIC was added to each culture; cultures continued growing for 30 min at 37°C and 200 rpm and were collected immediately at 4°C. For the antibiotics Rif, Nov, EM, CL, Kan, and PB, the added final concentrations for W3110 stimulation were 31.25, 1,000, 500, 2,000, 31.25, and 2 mg/liter, respectively; and for WJW02 stimulation, the added final concentrations were 0.98, 7.8, 31.25, 15.6, 7.8, and 0.5 mg/liter, respectively. Remarkably, the stimulated cultures should guarantee a survival rate of 60% to 80%.

Data availability.

Raw transcriptome data are available in the Sequence Read Archive (SRA) under accession number SRP160437.