Critical role of the RpoE stress response pathway in polymyxin resistance of Escherichia coli

Abstract

Objectives

Polymyxins, including colistin, are the drugs of last resort to treat MDR bacterial infections in humans. In-depth understanding of the molecular basis and regulation of polymyxin resistance would provide new therapeutic opportunities to combat increasing polymyxin resistance. Here we aimed to identify novel targets that are crucial for polymyxin resistance using Escherichia coli BL21(DE3), a unique colistin-resistant model strain.

Methods

BL21(DE3) was subjected to random transposon mutagenesis for screening colistin-susceptible mutants. The insertion sites of desired mutants were mapped; the key genes of interest were also inactivated in different strains to examine functional conservation. Specific genes in the known PmrAB and PhoPQ regulatory network were inactivated to examine crosstalk among different pathways. Lipid A species and membrane phospholipids were analysed by normal phase LC/MS.

Results

Among eight mutants with increased susceptibility to colistin, five mutants contained different mutations in three genes (rseP, degS and surA) that belong to the RpoE stress response pathway. Inactivation of rpoE, pmrB, eptA or pmrD led to significantly increased susceptibility to colistin; however, inactivation of phoQ or eptB did not change colistin MIC. RpoE mutation in different E. coli and Salmonella resistant strains all led to significant reduction in colistin MIC (16–32-fold). Inactivation of rpoE did not change the lipid A profile but significantly altered the phospholipid profile.

Conclusions

Inactivation of the important members of the RpoE regulon in polymyxin-resistant strains led to a drastic reduction in polymyxin MIC and an increase of lysophospholipids with no change in lipid A modifications.

Introduction

Polymyxins, such as colistin (also known as polymyxin E) and polymyxin B, are antibiotics that kill Gram-negative bacteria by binding to lipid A, the lipid anchor of LPS or lipooligosaccharide, subsequently disrupting both the outer and inner membranes. Polymyxins are used as drugs of last resort to treat MDR infections in humans.¹ However, polymyxin resistance has been increasingly appearing in various significant Gram-negative pathogens, such as Escherichia coli and Klebsiella pneumoniae, posing a serious threat to the clinical treatment of MDR pathogens.² Gram-negative bacteria have acquired strategies to resist killing by polymyxins.^3,4 The most common and conserved mechanism is LPS modification, such as charge-neutralization modifications of lipid A with L-4-aminoarabinose (Ara4N) and phosphoethanolamine (pEtN) via two-component regulatory systems (e.g. PhoPQ and PmrAB). Constitutive lipid A modifications due to mutations in phoPQ and pmrAB can cause acquired polymyxin resistance in Gram-negative pathogens.³ Alarmingly, the spread of the transmissible mcr-1, which can modify lipid A with pEtN, was recently discovered in both humans and animals, drawing worldwide attention and fear.⁵

Despite the significant progress made toward understanding of polymyxin resistance, various intriguing and unsolved questions still exist.^3,6 The bacterial outer membrane (OM) is a dynamic and evolving antibiotic barrier; in addition to associated proteins, the inner leaflet is composed of phospholipids (PLs) and the outer leaflet is composed of exclusively LPS or lipooligosaccharide.⁷ Recently, the PL-mediated maintenance of strict asymmetry of the OM lipid bilayer has emerged as an important but understudied colistin resistance mechanism.⁸ In addition, colistin was demonstrated to kill bacteria by targeting LPS in the inner membrane (IM) rather than the OM⁹ and the asymmetric feature of the IM lipid bilayer was just starting to be revealed.¹⁰ Therefore, in-depth examination of polymyxin resistance is highly warranted for the elucidation of complex mechanisms and the discovery of novel targets to combat increasing colistin resistance.

In our recent study, the widely used E. coli BL21(DE3) (designated as ‘BL21’ hereinafter), which is resistant to colistin (MIC = 16 mg/L), appeared to be an excellent model strain to study the molecular basis of polymyxin resistance in Gram-negative bacteria.¹¹ In BL21, a single nucleotide mutation (G361A) in pmrB, which results in a Glu-121-Lys substitution in PmrB, was required for lipid A modification and colistin resistance.¹¹ However, using a functional cloning approach, we demonstrated that specific contribution of the Glu-121-Lys mutation in PmrB to colistin resistance in BL21 completely relies on 3′-downstream region of pmrB.¹¹ Specifically, at least a 103 bp region downstream of pmrB is essential for the PmrB-mediated lipid A modifications and colistin resistance, which suggested a novel regulatory mechanism of PmrB-mediated polymyxin resistance in E. coli, such as that mediated through the RNA-related cis-element or small peptide.¹¹

In parallel with a recent study,¹¹ we used another functional genomics approach—random transposon mutagenesis—to examine the molecular basis of polymyxin resistance in BL21. Our findings reported here provide compelling evidence showing that the sigma E (σ^E) stress-response pathway is essential for polymyxin resistance in BL21 and other strains. However, inactivation of the key players in the σ^E pathway, such as the regulator RpoE, did not change the composition of modified lipid A species but significantly altered the PL profile, which indicated that the essential role of the RpoE stress-response pathway in polymyxin resistance is attributed to its functionality in concert with the existing lipid A modifications rather than due to its direct impact on lipid A modifications. In addition, our findings also suggested functional conservation of the RpoE system in polymyxin resistance in different Gram-negative bacteria.

Materials and methods

Bacterial strains, plasmids and culture conditions

The major bacterial strains and plasmids used in this study are listed in Table 1. E. coli and Salmonella strains were grown in LB broth (Difco) or Mueller–Hinton (MH) broth (Difco) with shaking (250 rpm) or on agar at 37°C overnight. When needed, culture media were supplemented with ampicillin (100 mg/L, Fisher Scientific), kanamycin (30 mg/L, Fisher Scientific), chloramphenicol (10 mg/L, Fisher Scientific), colistin sulphate (4 mg/L, ACROS), vancomycin (50 mg/L, Sigma), SDS (10 g/L, Fisher Scientific) or EDTA (0.5 mM, Fisher Scientific).

Table 1.

Key bacterial plasmids and strains used in this study

Plasmids or strains	Description	Source or reference
Plasmids
ȃpZE21	Cloning and expression vector, Kanr	12
ȃpAH	Kanr gene in pZE21 was replaced with Ampr gene	This study
ȃpUC19	Clone vector, Ampr	Invitrogen
ȃpKD3	Template plasmid of chloramphenicol resistant cassette for gene disruption, Ampr, Cmr	13
ȃpKD13	Template plasmid of Kanr cassette for gene disruption, Ampr, Kanr	13
ȃpSIM6	Heat-inducible red recombinase expression plasmid, with a temperature-sensitive origin of replication	14
ȃpCP20	Plasmid containing flipase, for the removal of FRT-flanked antibiotic-resistant cassettes, Ampr	13
ȃpDegS	The pAH plasmid bearing degS ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpSurA	The pAH plasmid bearing surA ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpUgd	The pAH plasmid bearing ugd ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpRseP	The pAH plasmid bearing rseP ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpPgm	The pAH plasmid bearing pgm ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpUC19_RpoE	The pUC19 plasmid bearing rpoE ORF cloned from E. coli BL21(DE3), Ampr	This study
ȃpZE21_RpoE	The pZE21 plasmid bearing rpoE ORF cloned from E. coli BL21(DE3), Kanr	This study
E. coli strains
ȃBL21(DE3)	F−  ompT hsdSB (rB−, mB−) gal dcm (DE3)	StrataGene
ȃTOP10	F−  mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 araD139 Δ(ara leu)7697 galU galK rpsL (Strr) endA1 nupG	Invitrogen
ȃWD101	Polymyxin-resistant strain with a constitutive mutation in PmrA	15
ȃMG1655	F−  lambda−  ilvG−  rfb-50 rph-1	16
ȃ1D2	BL21(DE3) derivative with transposon inserted in surA, Kanr	This study
ȃ5G12	BL21(DE3) derivative with transposon inserted in ugd, Kanr	This study
ȃ10E4	BL21(DE3) derivative with transposon inserted in degS (site I), Kanr	This study
ȃ15C10	BL21(DE3) derivative with transposon inserted in degS (site II), Kanr	This study
ȃ24G9	BL21(DE3) derivative with transposon inserted in degS (site III), Kanr	This study
ȃ13D1	BL21(DE3) derivative with transposon inserted in rseP, Kanr	This study
ȃ21E12	BL21(DE3) derivative with transposon inserted in pgm (site I), Kanr	This study
ȃ26G9	BL21(DE3) derivative with transposon inserted in pgm (site II), Kanr	This study
ȃJL1466	Isogenic RpoE mutant of BL21(DE3), Kanr	This study
ȃJL1489	BL21(DE3), ΔrpoE::FRT (Kanr cassette removed by pCP20)	This study
ȃJL1669	BL21(DE3). ΔrpoE::FRT/pZE21_RpoE	This study
ȃJL1412	Isogenic PhoQ mutant of BL21(DE3), Kanr	This study
ȃJL1436	Isogenic PmrB mutant of BL21(DE3), Kanr	This study
ȃJL1564	Isogenic EptA mutant of BL21(DE3), Cmr	This study
ȃJL1565	Isogenic EptB mutant of BL21(DE3), Kanr	This study
ȃJL1456	Isogenic PmrD mutant of BL21(DE3), Kanr	This study
ȃJL1568	Isogenic ArnT mutant of BL21(DE3), Cmr	This study
ȃJL1654	Isogenic RpoE mutant of WD101, Kanr	This study
ȃJL1652	Isogenic RpoE mutant of TOP10, Cmr	This study
ȃJL1760	Isogenic RpoE mutant of MG1655, Kanr	This study
ȃJL1659	Isogenic RpoE mutant of WD101, Kanr cassette removed by pCP20	This study
ȃJL1662	WD101, ΔrpoE::FRT/pZE21_RpoE	This study
ȃJL1242	TOP10 strain carrying pZE21_MCR-1 plasmid (TOP10/pZE21_MCR-1), Colr and Kanr	17
ȃJL1655	Isogenic RpoE mutant of JL1652 (TOP10, ΔrpoE::cat/pZE21_MCR-1), Kanr	This study
ȃJL1761	MG1655, ΔrpoE::Kan/pUC19_RpoE	This study
Salmonella Typhimurium
ȃJSG435	Polymyxin-resistant strain with a constitutive mutation (pmrA505) in PmrA	18
ȃJL1648	Isogenic RpoE mutant of S. Typhimurium JSG435	This study

Kan^r, kanamycin resistant; Amp^r, ampicillin resistant; Cm^r, chloramphenicol resistant; Str^r, streptomycin resistant; Col^r, colistin resistant.

Random transposon mutagenesis

The BL21 strain was subjected to in vivo random transposon mutagenesis using the EZ-Tn5™ TnP Transposome™ Kit (Lucigen) and the mutant libraries were screened for increased sensitivity to colistin using a standardized protocol.^19,20 Briefly, 1 μL of EZ-Tn5 TnP transposome complex containing 20 ng of transposon was used to electroporate BL21 competent cells, which were prepared by using the standard method.^13,21 The kanamycin-resistant transformants were individually picked and inoculated into 96-well microplates containing LB broth supplemented with 30 mg/L kanamycin. Following 16 h of incubation at 37°C, cultures of mutants were replicated into microtitre plates containing LB broth supplemented with 30 mg/L kanamycin and 4 mg/L colistin. Those mutants that could not grow in colistin-containing media were selected from the initial plates and subjected to a second screening to confirm increased sensitivities to colistin by using different concentrations of colistin (from 0.5 to 4 mg/L). The specific transposon insertion sites of selected mutants were determined by directly sequencing the genomic DNA as described previously.^19,20

Antimicrobial susceptibility test

The susceptibilities of E. coli strains to colistin sulphate were determined by a standard microtitre broth dilution method with an inoculum of 10⁶ cfu/mL as previously described.^22,23 MIC of colistin was determined by the lowest concentration of the antimicrobial showing complete inhibition of bacterial growth after 18 h incubation at 37°C.

Site-directed mutagenesis

Lambda-red-based homologous recombination technology using the pSIM6 vector (gifted by Dr Donald Court)^14,21 was used to knock out target genes in BL21 and other strains. Mutational fragments encompassing a FRT-kan-FRT or FRT-cat-FRT cassette with 50 nt homologous arms immediately flanking each targeted region (primers in Table 2) were amplified using pKD13 or pKD3 as template plasmids,¹³ subsequently treated with DpnI (New England Biolabs) and purified using the QIAquick PCR Purification Kit (QIAGEN). The purified mutational fragments were electroporated into 50 µL of heat-shock-induced electrocompetent cells (containing pSIM6) using MicroPulser Electroporation Apparatus (Bio-Rad) and a 0.1 cm gapped electroporation cuvette (Bio-Rad) with the EC1 programme (1.8 kV). Recombinants were selected for kanamycin resistance (30 mg/L) at 32°C for 1–2 days, and then streaked onto LB plates and incubated at elevated temperature (37°C) to remove pSIM6. The mutation was further verified by PCR using flanking primers (Table 2) and transposon internal primers (K1 or C1, Table 2).¹³

Table 2.

Major primers used in this study

Primer	DNA sequence (5′−3′)a	Product size (bp)b	Target gene/region and function
pUC19-AmpR-F1	TGTCATGATAATAATGGTTTC	1105	The bla gene with its promoter region on pUC19
pUC19-AmpR-R1	ATAGAGCTCAAGGATCTTCACCTAGATC (SacI)
pgm-F	ATAGTCGACATGGCAATCCACAATCGTG (SalI)	1650	pgm ORF
Pgm-R	TTACGCGTTTTTCAGAACTTC
surA-F	ATAGTCGACATGAAGAACTGGAAAACGC (SalI)	1296	surA ORF
surA-R	TTAGTTGCTCAGGATTTTAAC
ugd-F	ATAGTCGACATGAAAATCACCATTTCCG (SalI)	1176	ugd ORF
ugd-R	TTAGTCGCTGCCAAAGAGA
degS-F	ATAGTCGACATGTTTGTGAAGCTCTTAC (SalI)	1077	degS ORF
degS-R	TTAATTGGTTGCCGGATATTC
rseP-F	ATAGTCGACATGCTGAGTTTTCTCTGGG (SalI)	1362	rseP ORF
rseP-R	TCATAACCGAGAGAAATCATT
K1	CAGTCATAGCCGAATAGCCT		Detection primer for Kanr cassette
C1	TTATACGCAAGGCGACAAGG		Detection primer for Cmr cassette
pZE-F	GAATTCATTAAAGAGGAGAAAGGT		Sequencing primer of pZE21 or pAH
pZE-R	TTTCGTTTTATTTGATGCCTCTAG		Sequencing primer of pZE21 or pAH
M13-F	TGTAAAACGACGGCCAGT		Sequencing primer of pUC19
M13-R	CAGGAAACAGCTATGAC		Sequencing primer of pUC19
RpoE(BL21DE3)_pKD13_F	CGTTTCGATAGCGCGTGGAAATTTGGTTTGGGGAGACTTTACCTCGGATGTGTAGGCTGGAGCTGCTTCG	1403	rpoE ORF, introducing Kanr cassette
RpoE(BL21DE3)_pKD13_R	TAATACCCTTATCCAGTATCCCGCTATCGTCAACGCCTGATAAGCGGTTGATTCCGGGGATCCGTCGACC
RpoE(BL21DE3)_pKD3_F	CGTTTCGATAGCGCGTGGAAATTTGGTTTGGGGAGACTTTACCTCGGATGGTGTAGGCTGGAGCTGCTTC	1114	rpoE ORF, introducing Cmr cassette
RpoE(BL21DE3)_pKD3_R	TAATACCCTTATCCAGTATCCCGCTATCGTCAACGCCTGATAAGCGGTTGCATATGAATATCCTCCTTAG
RpoE_F	CAAAAACAGATGCGTTACGG	837	Detection of rpoE mutation
RpoE_R	TCCCAGGTTTTCTGCATTTC
RpoE_(pZE21)_ClaI_F	CCATCGATCAAAAACAGATGCGTTACGG (ClaI)	855	The rpoE gene with its adjacent region, for cloning to pZE21
RpoE_(pZE21)_BamHI_R	GAAGGGATCCTCCCAGGTTTTCTGCATTTC (BamHI)
PhoQ(BL21DE3)_pKD13_F3	GTGATTACCACCGTTCGCGGCCAGGGCTATCTGTTCGAATTGCGCTGATGTGTAGGCTGGAGCTGCTTCG	1403	Mutational primer of PhoQ
PhoQ(BL21DE3)_pKD13_R3	TTAACGTAATGCGTGAAGTATGGACATATTTATTCATCTTTCGGCGTAGAATTCCGGGGATCCGTCGACC
PhoQ_F	TAATGGCAAAGTGGTGAGCA	1772	Detection of phoQ mutation
PhoQ_R	TTCTGCCAGTGACGTTCAAG
PmrB(BL21DE3)_pKD13_F3	GCTTTGGCTATATGCTGGTCGCGAATGAGGAAAACTAATTGAATCTGATGTGTAGGCTGGAGCTGCTTCG	1403	Mutational primer of PmrB
PmrB(BL21DE3)_pKD13_R3	TTCAGCGTGCTGGTGGTCAGCAGCTTTCTTTATATCTGGTTTGCCACGTAATTCCGGGGATCCGTCGACC
PmrB_F	AATGAACCCTCGACCAACAC	1376	Detection of pmrB mutation
PmrB_R	CGCTGTCTTATCAGGCCAAT
EptA(BL21DE3)_pKD3_F	TTTTGCTTTGCGAGCATATGCGCACTTTGTTCGATGGAAACACCGTGATGGTGTAGGCTGGAGCTGCTTC	1114	Mutational primer of EptA
EptA(BL21DE3)_pKD3_R	CAGCGTATCGTCTTCAACAATCAGAATTTTCATTCACTCACTCTCCTGCACATATGAATATCCTCCTTAG
EptA_F	GCCATGTATGCGCTGAATTA	1964	Detection of eptA mutation
EptA_R	ACCACCAGGCTGTAATGACC
EptB(BL21DE3)_pKD13_F	AGCCACTAAGCAGGGTGTTATCACCTGTTTGTCCAGGGTTTGTTTGCATGATTCCGGGGATCCGTCGACC	1403	Mutational primer of EptB
EptB(BL21DE3)_pKD13_R	TTTGATCGGCGAGAAAGTCAGCAGGCCGCTTAGTTAGCCGCTGCCTCTTTTGTAGGCTGGAGCTGCTTCG
EptB_F	GCACACTCTTTCCCCACACT	1907	Detection of eptB mutation
EptB_R	GCCCTCGTCAATCCCTTAAT
ArnT(BL21DE3)_pKD3_F	GGACGTGAAGGCTGGCTGGGTTGCCAACAAATTGCGGGTAGTCGCTGATGGTGTAGGCTGGAGCTGCTTC	1114	Mutational primer of ArnT
ArnT(BL21DE3)_pKD3_R	CAAGCTGGCAAAGACTAATGTTAGCCAGATCATTTGGGACGATACTGAATCATATGAATATCCTCCTTAG
ArnT_F	TTGCCCTTTAAGCGAACTGT	1915	Detection of arnT mutation
ArnT_R	ATGGCAAGACCAAGACAAGC
PmrD(BL21DE3)_pKD13_F	TTAATCTGTAAATTGATGTGAAAACTCTTAGCAAACAGGATAATGCAATGTGTAGGCTGGAGCTGCTTCG	1403	Mutational primer of ArnT
PmrD(BL21DE3)_pKD13_R	CTGATTTTCCTGCCCACGACAAAACAACGTTACTGAGTTTTCCCTGCCACATTCCGGGGATCCGTCGACC
PmrD_F	GTCAGGCGCTAAAAGAGTGG	583	Detection of pmrD mutation
PmrD_R	TTGCGCTAAGCAAAGCCTAT
RT-pmrB-F1	GTGCCGGACAGTCATTTTCT	101	Detect mRNA level of pmrB
RT-pmrB-R1	CTGGTCGAGCATGGTACTGA
RT-pmrD-F1	AAATGATCGCCGAAGTGAAG	131	Detect mRNA level of pmrD
RT-pmrD-R1	ATAACTGCTTGCCGAGAGGA
RT-EptA-F	CAGCGACTGGCAAATCT	157	Detect mRNA level of eptA
RT-EptA-R	TAGTTTCACGCGGGTAGC
RT-ArnT-F	TCAGCCAAGCCGCTATATTC	157	Detect mRNA level of arnT
RT-ArnT-R	ATCACCGCTGACAAATCTCC
pldA(+647)Fwd	AGCTTAAAATCGGCTATCACCTC	105	Detect mRNA level of pldA
pldA(+751)Rev	TCGGGTAACTTAAGCCTAACTCC
RT-mlaA-F1	GTGCGAGTTCCGGTACAGAT	112	Detect mRNA level of mlaA
RT-mlaA-R1	ACCGGTCGAACAATATACGG
RT-LplT_F	TGCGGCAGCGAAGTTAGTTA	101	Detect mRNA level of lplT
RT-LplT_R	GCTCGTGTTGCAGGGAAAAA
RT-Aas_F	TGTTCCTTTATCCAAGCCCG	101	Detect mRNA level of Aas
RT-Aas_R	GTGACCGAGGAAAGTCGAGG
RT-GapA-F	TATGACTGGTCCGTCTAAAGACAA	202	Detect mRNA level of GapA (internal control)
RT-GapA-R	GGTTTTCTGAGTAGCGGTAGTAGC

Restriction sites are underlined in the primer sequence and the names are identified in parentheses.

The amplicon size using WT genomic DNA of BL21 (GenBank Accession #: AM946981) or pUC19 as templates.

Construction of the plasmid pAH

To construct an ampicillin-resistant derivative of pZE21, the kanamycin-resistant cassette in pZE21 was replaced by an ampicillin-resistant cassette. In brief, the ampicillin-resistant cassette was amplified with primer pUC19-AmpR-F1 and pUC19-AmpR-R1 (Table 2), resulting in a 1104 bp fragment containing an internal AatII site and an introduced SacI site. The ampicillin-resistant cassette was then digested with AatII/SacI, and ligated to the AatII/SacI-digested pZE21 backbone (1297 bp). The ligation mix was electroporated into TOP10. The plasmid (pAH) extracted from the ampicillin-resistant clone was sequenced to confirm there was no mutation.

Complementation in trans

The functional cloning vector pZE21^11,12,17,24, its ampicillin-resistant derivative plasmid pAH and the cloning vector pUC19 (Table 1) were used for complementation. Briefly, the selected genes were PCR-amplified with corresponding primers (listed in Table 2) and PfuUltra DNA polymerase (StrataGene) and using genomic DNA of BL21 as the template. The PCR products were digested with corresponding restriction enzymes (as indicated in Table 2) and ligated into same enzyme-digested pZE21, pAH or pUC19. These ligation mixes were transformed into corresponding mutants via electroporation. Plasmids were extracted from transformants and validated by Sanger sequencing with primers pZE-F and pZE-R, or M13-F and M13-R (Table 2).

Quantitative real-time RT–PCR (qRT–PCR)

Overnight-grown E. coli strains were inoculated into LB broth and grown to logarithmic phase. One millilitre of RNAprotect Bacteria Reagent (QIAGEN) was mixed immediately with 0.5 mL of bacterial suspension. Bacteria were pelleted and total RNAs were extracted using the RNeasy Mini Kit (QIAGEN). Residual genomic DNA was removed through DNase I (QIAGEN) treatment. qRT–PCR was performed using the Power SYBR Green RNA-to-CT One-Step Kit (Life Technologies) and following the procedure detailed in our previous publication.^11,23,25 Briefly, triplicate reactions in a volume of 20 μL were performed for each RNA sample. The difference in target gene transcription between WT and mutants were calculated using the −ΔΔCT method as described previously.^11,23,25,26 Expression of the GapA gene was used as an internal control.

Lipid A analysis

Lipid A was extracted from bacterial pellets using the Bligh–Dyer method.^27,28 Approximately 200 mL of logarithmic phase (OD₆₀₀ = 0.8–1) cell culture of each strain was used for lipid A extraction. Once extracted, lipid A samples were subjected to LC/electrospray ionization MS (LC/ESI-MS) analysis at Duke Medical Center as previously described.^29,30 Briefly, normal-phase LC-ESI-MS of the lipid extracts was performed using an Agilent 1200 Quaternary LC system coupled to a high-resolution TripleTOF5600 mass spectrometer (Sciex, Framingham, MA, USA). Chromatographic separation was performed on a Unison UK-Amino column (3 μm, 25 cm × 2 mm) (Imtakt USA, Portland, OR, USA). Lipids were eluted with mobile phase A, consisting of chloroform/methanol/aqueous ammonium hydroxide (800:195:5, v/v/v), mobile phase B, consisting of chloroform/methanol/water/aqueous ammonium hydroxide (600:340:50:5, v/v/v/v) and mobile phase C, consisting of chloroform/methanol/water/aqueous ammonium hydroxide (450:450:95:5, v/v/v/v), over a 40 min-long run, performed as follows: 100% mobile phase A was held isocratically for 2 min and then linearly increased to 100% mobile phase B over 14 min and held at 100% B for 11 min. The mobile phase composition was then changed to 100% mobile phase C over 3 min and held at 100% C for 3 min, and finally returned to 100% A over 0.5 min and held at 100% A for 5 min. The LC eluent (with a total flow rate of 300 µL/min) was introduced into the ESI source of the high-resolution TF5600 mass spectrometer. MS and MS/MS were performed in negative-ion mode, with the full-scan spectra being collected in the m/z 300–2000 range. The MS settings were as follows: ion spray voltage (IS) = −4500 V, curtain gas (CUR) = 20 psi, ion source gas 1 (GS1) = 20 psi, declustering potential (DP) = −55 V, and focusing potential (FP) = −150 V. Nitrogen was used as the collision gas for tandem MS (MS/MS) experiments. Data analysis was performed using Analyst TF1.5 software (Sciex, Framingham, MA, USA).

LC/MS/MS analysis of PLs

The lipids were extracted from the cells grown in MH broth by using the standard acidic Bligh–Dyer method as described in our previous publication.³⁰ Approximately 100 mL of logarithmic phase (OD₆₀₀ = 0.8–1) cell culture of each strain was used for PL extraction. Once extracted, lipid A samples were subjected to LC/MS/MS analysis at Duke Medical Center as previously described.³¹ Briefly, normal phase LC was performed on an Agilent 1200 Quaternary LC system equipped with an Ascentis Silica HPLC column (5 µm, 25 cm × 2.1 mm; Sigma–Aldrich, St. Louis, MO, USA). Mobile phase A consisted of chloroform/methanol/aqueous ammonium hydroxide (800:195:5, v/v/v); mobile phase B consisted of chloroform/methanol/water/aqueous ammonium hydroxide (600:340:50:5, v/v/v/v); mobile phase C consisted of chloroform/methanol/water/aqueous ammonium hydroxide (450:450:95:5, v/v/v/v). The elution programme consisted of the following: 100% mobile phase A was held isocratically for 2 min and then linearly increased to 100% mobile phase B over 14 min and held at 100% B for 11 min. The LC gradient was then changed to 100% mobile phase C over 3 min and held at 100% C for 3 min, and finally returned to 100% A over 0.5 min and held at 100% A for 5 min. With a total flow rate of 300 L/min, the LC eluent was injected into the ion spray source of a TripleTOF 5600 quadrupole time-of-flight tandem mass spectrometer (AB SCIEX, Framingham, MA, USA). Instrumental settings for negative-ion ESI and MS/MS analysis of lipid species were as follows: IS = −4500 V; CUR = 20 psi; GS1 = 20 psi; DP = −55 V; and FP = −150 V. The MS/MS analysis used nitrogen as the collision gas. Data analysis was performed using the Analyst TF1.5 Software.

Statistical analysis

Statistical analyses were performed using GraphPad Prism (version 5.01). The non-parametric Wilcoxon test for paired data was chosen to compare target gene levels between WT and rpoE mutant. For all analyses, P < 0.05 was considered as statistically significant.

Results

Identification of polymyxin-sensitive mutants using random transposon mutagenesis

The in vivo random transposon mutagenesis system is efficient for BL21 with more than 10 000 kanamycin-resistant mutants generated by using 1 μL of EZ-Tn5 transposome. Of a total of 5745 randomly selected mutants, 29 mutants failed to grow in LB broth containing 4 mg/L colistin. The eight mutants with the most dramatic reduction in colistin MIC (MIC = 0.25–1 mg/L) when compared with BL21 WT (MIC = 16 mg/L, Table 3) were selected for further characterization. The insertion sites of the eight mutants were mapped by direct sequencing, and the inactivated genes are summarized in Table 3. Complementation of all the mutants restored colistin resistance to a similar level as observed in the parent BL21 strain (MIC = 8–16 mg/L, Table 3), indicating that the reduced colistin MIC phenotype of these mutants is directly linked to the mutated genes.

Table 3.

Phenotypic and genetic characteristics of the E. coli BL21 parent strain, its mutant derivatives, and complementation constructs

WT strain and mutants	Colistin MIC (mg/L)a	Gene inserted	Insertion siteb	Function
BL21	16 (N/A)	N/A	N/A	N/A
1D2	0.25 (16)	surA	465 (−)	Chaperon protein, σE regulon
5G12	1 (8)	ugd	497 (+)	UDP-glucose 6-dehydrogenase
10E4	0.25 (8)	degS	237 (−)	Serine protease, regulator for σE
15C10	0.25 (8)	degS	358 (−)	Serine protease regulator for σE
24G9	0.25 (8)	degS	274 (+)	Serine protease regulator for σE
13D1	0.25 (16)	rseP	1033 (+)	Zinc protease, regulator for σE
21E12	0.25 (16)	pgm	617 (−)	Phosphoglucomutase
26G9	1 (8)	pgm	982 (+)	Phosphoglucomutase

N/A, not applicable.

The MIC in parentheses is for the complementation construct carrying the corresponding full-length gene from BL21.

Position at corresponding gene in BL21 (genome accession #: AM946981). The symbol in parentheses indicates relative orientation of inserted kanamycin resistance gene with the inserted gene: +, same orientation; −, opposite orientation.

Interestingly, of the final eight mutants, five mutants with the lowest MIC for colistin (0.25 mg/L) had transposon insertion in three genes (degS, rseP and surA) that belong to the σ^E stress-response pathway (Table 3).³² In particular, three degS mutants identified from the same screening (10E4, 15C10 and 24G9) displayed three different transposon insertion sites (Table 3), further confirming the critical role of the σ^E stress-response pathway in colistin resistance. In response to stress resulting from cell envelope damage, DegS and RseP, the proteases anchored in the IM, can cleave RseA (the membrane-bound anti-σ^E factor) sequentially, leading to activation of RpoE (Figure 1).^33,34 SurA, a periplasmic chaperone required for the folding of OM proteins, is part of the σ^E regulon (Figure 1).³⁵ Since the findings from the random transposon mutagenesis study revealed multiple key components of the RpoE stress-response pathway that consistently contributed to colistin resistance in BL21, subsequently, we performed a series of molecular and lipidomics studies to understand the molecular basis of this newly identified polymyxin-resistance mechanism as described below.

Figure 1.

Diagram of key inner membrane components required for activation of the sigma factor RpoE (σ^E). This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Both RpoE and PmrAB regulatory systems are essential for the high-level colistin resistance in BL21

The most common and conserved mechanism of polymyxin resistance in E. coli is lipid A modification through two-component regulatory systems, such as PhoPQ and PmrAB. To obtain key information about the relationship between the σ^E stress-response pathway and the known PmrAB and PhoPQ regulatory network, we subsequently created a panel of isogenic BL21 mutants. As expected, mutagenesis of rpoE, the gene encoding the sigma factor σ^E for activation of the RpoE stress-response pathway (Figure 1) in BL21, led to significantly increased susceptibility to colistin (MIC = 0.5 mg/L). In addition, consistent with our recent study using functional cloning,¹¹ inactivation of PmrB also led to a dramatically reduced MIC of colistin (0.5 mg/L) (Table 4). Similarly, inactivation of anrT, eptA and pmrD also led to significantly reduced MIC (Table 4). Both pEtN transferase EptA³⁶ and 4-amino-4-deoxy-L-arabinose (L-Ara4N) transferase ArnT³⁷ transfer positively charged groups to lipid A, therefore masking the charge of its anionic phosphates and reducing the affinity of lipid A to positively charged polymyxins. PmrD is activated through the PhoPQ system, and serves as the connector between PhoPQ and PmrAB systems by preventing the deactivation of PmrA.³⁸ However, inactivation of PhoQ barely changed the MIC of colistin (Table 4), indicating that the PhoPQ system is not involved in the colistin resistance in BL21. In addition, inactivation of EptB, which is a pEtN transferase modifying LPS at the outer 3-deoxy-D-mannooctulosonic acid (Kdo) residue,³⁹ also did not change the MIC of colistin (Table 4). Together, this targeted mutagenesis study showed that the RpoE stress-response pathway is crucial for high-level polymyxin resistance by functioning in concert with the PmrAB-dependent regulatory system in BL21.

Table 4.

Contribution of PhoPQ and PmrAB regulon on colistin resistance in E. coli BL21

Strain name	Inactivated gene	Colistin MIC (mg/L)
BL21	N/A	16
JL1412	phoQ	8
JL1436	pmrB	0.5
JL1564	eptA	0.5
JL1565	eptB	16
JL1568	arnT	0.5
JL1456	pmrD	1

N/A, not applicable.

Effect of rpoE mutation on the susceptibility to other membrane-damaging agents

It has been well recognized that the major mode of action of polymyxin, a cationic peptide antibiotic, is to target the bacterial OM for disruption of membranes.⁷ Therefore, the initial interaction of polymyxin with bacterial OM likely activates the σ^E stress response, consequently triggering modification of appropriate target(s) for polymyxin resistance. To determine if the RpoE-mediated polymyxin resistance is a specific phenomenon for polymyxin or a general phenomenon for membrane-damaging agents, we first examined the susceptibility of BL21 and its isogenic rpoE mutant to sodium glycocholate, a detergent-like membrane-damaging bile salt that can trigger general stress response. We observed that both BL21 and its isogenic rpoE mutant displayed the same MIC of sodium glycocholate (20 g/L). Subsequently, we tested MICs of SDS and EDTA, the two well-established membrane stressors with different modes of action. Inactivation of RpoE did not change the SDS MIC (10 g/L) but led to a 4-fold MIC reduction in EDTA (8 mM) when compared with parent BL21. These MIC data were also confirmed by spotting of cultures on LB agar supplemented with the same membrane-damaging agents (1% SDS or 0.5 mM EDTA), as described by Goodall et al.⁴⁰ (Figure S1, available as Supplementary data at JAC Online). Together, these findings suggest that the RpoE-mediated polymyxin resistance is a phenomenon specific to polymyxin.

Inactivation of RpoE did not affect lipid A modifications

The specific Glu-121-Lys mutation in PmrB has been demonstrated to contribute directly to the high-level colistin resistance in BL21 by modifying lipid A, such as generating the Ara4N- and pEtN-modified lipid A species.¹¹ Thus, we initially speculated that inactivation of the key components in the RpoE stress-response pathway in BL21 would significantly change the lipid A profile. To test this, subsequently, we determined if inactivation of the RpoE stress-response pathway affected modification patterns of lipid A. Lipid A species were extracted from BL21 and its isogenic mutants with mutation in rpoE, degS, surA and rseP and then subjected to LC/ESI-MS analysis. As expected, various PmrAB-mediated lipid A modifications (addition of Ara4N and/or pEtN) were detected in BL21 (Figure 2, top panel). However, surprisingly, inactivation of rpoE, the key sigma factor of the RpoE stress-response pathway, did not abolish the peaks of various Ara4N- and pEtN-modified lipid A species (Figure 2, the second panel from top); this finding has been confirmed in three independent experiments. This finding is consistent with the observation that the expression (mRNA) levels of several key genes (pmrB, eptA, arnT and pmrD) in PmrAB regulons in the rpoE mutant were not decreased when compared with BL21 (data not shown). In addition, the isogenic degS, rseP and surA also displayed a similar lipid A profile as that observed in the parent BL21 strain (Figure 2, bottom three panels). Therefore, the critical role of the RpoE stress-response pathway in polymyxin resistance is attributed to its functionality in concert with the existing lipid A modifications rather than due to its direct impact on lipid A modifications.

Figure 2.

LC/ESI-MS of lipid A profile in BL21 WT strain and its isogenic mutants in RpoE stress-response pathway. The lipid A species (detected as [M−2H]²⁻ ions) with Ara4N modification (solid blue circle) and/or pEtN modification (solid orange circle) are shown as cartoons next to specific peaks. Side peaks are due to different acyl chain lengths (differing by C₂H₄, 28 Da). Representative chemical structures and the corresponding cartoons of hexa-acylated lipid A and its Ara4N and pEtN modifications are shown below figure panels.

Inactivation of RpoE changed the PL profile

Given the recently increased awareness of the role of PL-mediated membrane asymmetry in colistin resistance, we also examined the PL profile in BL21 and the selected isogenic mutants with mutation in the RpoE sigma factor or DegS, the initial membrane protease responding to the stress resulting from cell membrane damage (Figure 1). A panel of PLs were detected in all strains, including diacylglycerol (DAG), fatty acid (FA), phosphatidylglycerol (PG), cardiolipin (CL), lyso-phosphatidylglycerol (Lyso-PG), and lyso-phosphatidylethanolamine (Lyso-PE) (Figure 3). The most notable differences were found in FA, Lyso-PG and Lyso-PE, which were elevated in the rpoE and degS mutants when compared with those in BL21 (Figure 3). However, the mRNA levels of several key genes involved in membrane metabolism and recycling of PLs, such as those encoding membrane phospholipase PldA,⁴¹ PL retrograde transport MlaA, which is responsible for removing mislocalized PLs from the outer leaflet of the outer membrane,⁴² and lysophospholipid (LPL) transport and acylation machinery (PL repair system) LplT/Aas,⁴³ did not change significantly between BL21 and its isogenic rpoE mutant (data not shown). Consistent with this RT–PCR result, both BL21 and its isogenic rpoE mutant showed the same MIC (256 mg/L) for vancomycin, a large lipophilic antibiotic that is normally excluded by the LPS monolayer and not expected to traverse OM easily unless the integrity of the OM structure is severely impaired.^40,44

Figure 3.

LC/MS/MS analysis of PLs in the BL21 WT strain and its isogenic mutants in the RpoE stress-response pathway. Total negative ion chromatograms are shown. The lipids with significant increases in ΔrpoE and ΔdegS (FA, Lyso-PG, Lyso-PE) are highlighted in orange text above specific peaks.

Functional conservation of σ^E stress-response system in polymyxin resistance

RpoE is a prevalent and highly conserved stress-response factor in E. coli and other different Gram-negative bacteria.⁴⁵ Thus, subsequently, we also initiated pilot work to examine if the contribution of the σ^E stress-response pathway to polymyxin resistance in BL21 is a common theme in different E. coli strains and bacterial organisms. As shown in Table 5, inactivation of rpoE in E. coli WD101, which has a point mutation in PmrA,¹⁵ led to a 32-fold reduction in colistin MIC; complementation greatly restored its level to resistance. Mutation of rpoE in an E. coli strain harbouring mcr-1 also led to a 32-fold reduction in colistin MIC. A similar magnitude of reduction was also observed in S. enterica serotype Typhimurium JSG435, the strain carrying a point mutation in PmrA (Table 5).¹⁸ Notably, inactivation of rpoE in the colistin-susceptible E. coli TOP10 and MG1655 strains also led to significant reductions in colistin MIC (4- and 2-fold, respectively, Table 5), indicating that RpoE also contributes to the basal level of resistance (or intrinsic resistance) to colistin.

Table 5.

Inactivation of RpoE increased susceptibilities of different strains to colistin

Strain	Colistin MIC (mg/L)
	Parent	ΔrpoE	ΔrpoE/pRpoE
E. coli BL21	16	0.5	16
E. coli WD101	64	2	16
E. coli TOP10/pMCR-1	16	0.5	N/A
S. Typhimurium JSG435	32	2	N/A
E. coli TOP10	0.5	0.125	N/A
E. coli MG1655	0.5	0.25	0.5

N/A, not applicable.

Discussions

Polymyxins (e.g. colistin) have been recently re-introduced into clinical practice to treat infections caused by MDR Gram-negative pathogens, such as E. coli. Recent discovery of transmissible colistin resistance due to the plasmid-borne mcr-1 gene has caused worldwide attention and fear,⁵ further justifying in-depth mechanistic research so as to develop effective mitigation strategies. In Gram-negative bacteria, the OM is a robust permeability barrier due to its unique asymmetric lipid bilayer (Figure 4). However, polymyxin, a cationic peptide antibiotic, can still permeabilize the OM by targeting net anionic LPS, leading to cell death. A common and widely studied mechanism to resist polymyxin is LPS modification, such as charge-neutralization modifications of lipid A with Ara4N and pEtN via the two-component regulatory system (Figure 4). However, to date, many intriguing and unsolved questions regarding polymyxin resistance still exist, warranting the identification of new mechanisms.^3,4,6 In this study, with the aid of an excellent E. coli model strain BL21 together with random transposon mutagenesis, we identified a new target that is essential for colistin resistance, namely the sigma E (σ^E) stress-response pathway. Intriguingly, lack of changes in the composition of modified lipid A species in the isogenic RpoE pathway mutants of BL21 indicated that the RpoE stress-response pathway did not affect lipid A modifications, but contributed to polymyxin resistance through a novel unidentified mechanism that is in concert with the existing PmrAB-mediated lipid A modifications in BL21. The observation of increased FA, Lyso-PG and Lyso-PE in the rpoE and degS mutants (Figure 3) shed light on the understanding of the molecular basis of the new RpoE-mediated polymyxin resistance mechanisms. Given that inactivation of rpoE did not lead to increased susceptibility to SDS and vancomycin, the changes in the abundance of PLs in the outer leaflet of the OM may not be the major reason why the RpoE system contributes to polymyxin resistance. Instead, the RpoE stress-response system may strictly modulate the composition and relative abundance of different PLs in both OM and IM leaflets, consequently maintaining appropriate membrane asymmetry to confer polymyxin resistance. Despite a panel of critical molecular work performed, the current study did not reveal the molecular basis to explain how RpoE modulates specific PLs required for polymyxin resistance. To identify the specific key genes and/or network required for PL-mediated membrane asymmetry and polymyxin resistance, comprehensive studies, which include but are not limited to systemic transcriptome analysis of RpoE regulon, delicate gene manipulation, extensive microbiological and biochemical assays, and in-depth lipidomics analysis of membranes, are highly warranted in the future.

Figure 4.

Cartoon model of membrane asymmetry and polymyxin resistance mechanism in E. coli. OM lipids are highly asymmetric with PLs in the inner leaflet and LPS exclusively in the outer leaflet. Upon damage in the OM, the PLs can be mislocalized to the outer leaflet; such reduced asymmetry can be restored by two different systems (shown in the left section). The PLs present in the IM display great diversity with respect to composition, distribution and abundance. Upon completion of biosynthesis, LPS is inserted in the inner leaflet of IM and then flipped to the outer leaflet where lipid A can be modified by IM enzymes (e.g. ArnT and EptA). The modified LPS, which could be different with respect to lipid A species and abundance, contributes to polymyxin resistance. TCRS, two-component regulatory system.

The bacterial OM is a dynamic and evolving antibiotic barrier, with the inner leaflet composed of PLs and the outer leaflet composed of exclusively LPS or lipooligosaccharide.⁷ Membrane asymmetry is attributed to several key features of PLs and LPS including their composition, spatial distribution and abundance; however, significant knowledge gaps still exist, impeding depiction of a ‘high-resolution’ picture of membrane asymmetry in Gram-negative bacteria for polymyxin resistance. First, to maintain OM asymmetry and barrier integrity, most Gram-negative bacteria have evolved systems to remove PLs that are mislocalized to the outer leaflet, which include the PL retrograde transport system (e.g. Mla system)^8,46 and the conversion of PL to LPL by OM PL, followed by regeneration of PL using the LPL transport and acylation machinery in IM (Figure 4, left).⁴³ Thus, recently, the PL-mediated maintenance of the strict asymmetry of the OM lipid bilayer has emerged as an important but understudied colistin resistance mechanism.⁸ As discussed above, the findings from this study suggest that the RpoE system does not significantly affect abundance of total PLs in the outer leaflet of the OM. Second, most published cartoon models of the bacterial envelope simply use the same symbol to represent all PLs, which is misleading. In fact, membrane PLs are composed of various PLs with different levels of abundance, such as PE (∼70%), PG (∼25%), CL (∼4%) and PS (<0.1%) (Figure 4). The asymmetric feature of the IM lipid bilayer has just started to be revealed recently.¹⁰ Composition and relative abundance of various PLs in both the OM and IM are likely a critical contributing factor to maintain strict membrane asymmetry for polymyxin resistance. Finally, in terms of the mode of action of colistin, Sabnis et al.  ⁹ recently reported that colistin kills bacteria by targeting the LPS in both the OM and IM. It was also observed that the modified and unmodified LPS display significant asymmetry in OM and IM with respect to distribution and quantification.⁹ Based on these recent reports^8,9 and our findings reported here, we speculated that the σ^E stress-response pathway is crucial to maintain the strict lipid asymmetry of both the OM and IM for polymyxin resistance in E. coli. In particular, to date, how PL-mediated membrane asymmetry is regulated for colistin resistance is unknown. Intriguingly, our RT–PCR analysis indicated that inactivation of RpoE did not affect transcription of several key genes involved in PL metabolism and recycling, such as those encoding PldA,⁴¹ MlaA⁴² and LplT/Aas.⁴³ Therefore, we speculate that the RpoE system contributes to polymyxin resistance by using novel Mla/PldA-independent pathways to strictly modulate the composition and relative abundance of various PLs in membrane. This hypothesis, particularly with respect to the molecular basis of the accumulation of Lyso-PG and Lyso-PE in rpoE mutants, indicative of membrane destabilization,⁴⁷ needs to be examined in the future.

Our preliminary survey using representative colistin-resistant strains (Table 5) strongly suggested that the RpoE stress-response system likely contributes to clinical polymyxin resistance in diverse Gram-negative bacterial strains. RpoE is a sigma factor (σ^E) that is widely existed and conserved in the γ-proteobacteria.^45,48 It has been observed that the σ^E regulon includes a panel of genes involved in the synthesis, assembly and homeostasis of LPS in multiple organisms,⁴⁹ which potentially affects LPS-mediated membrane asymmetry and consequently contributes to polymyxin resistance. Despite the lack of clinical relevance and/or in-depth molecular validation, recent functional characterizations of RpoE also showed that inactivation of RpoE led to increased susceptibility to polymyxin B in different Gram-negative bacteria, including uropathogenic E. coli,⁵⁰  S. Typhimurium,^51,52  Vibrio parahaemolyticus  ⁵³ and Pseudomonas aeruginosa.⁵⁴ Polymyxin resistance has been increasingly appearing worldwide in various significant Gram-negative pathogens, such as E. coli, S. enterica, K. pneumoniae and P. aeruginosa,^55–62 posing a serious threat to the clinical treatment of MDR pathogens in humans. Therefore, future large-scale investigation using diverse clinical polymyxin-resistant pathogens is critically important for us to determine if the contribution of the σ^E stress-response pathway to polymyxin resistance is a common theme in various Gram-negative organisms.

Previously, RpoE and DegS were reported to be essential in an E. coli K-12 strain.^33,63,64 However, a systematic study confirmed they are not essential genes in an E. coli K-12 strain.⁶⁵ Therefore, they are probably conditional essential genes, i.e. only essential under certain conditions, such as high temperature or low salt. For example, RpoE is an important gene for bacterial growth at a high temperature (43°C).⁶⁴ The construction of rpoE mutants in our study was performed at 32°C, which may permit the formation of mutant colonies. Notably, the BL21 used in this study is not a K-12 strain. Furthermore, rpoE mutants were constructed in both K-12 (TOP10 and MG1655, Table 5) and non-K-12 strains.^50,66 Finally, complementation of specific mutants with degS or rpoE could restore the colistin resistance of corresponding mutants to the level of the parent strain (Tables 3 and 5), clearly indicating that the DegS and RpoE, rather than the possible suppressor mutations generated during mutant construction,^67,68 are indeed required for colistin resistance in BL21.

Together, the findings from this study strongly support that the RpoE stress-response pathway plays a critical role in high-level polymyxin resistance. Therefore, the inhibitors targeting the RpoE stress-response pathway are expected to increase susceptibility of bacterial pathogens to polymyxin. These inhibitors may be used in combination therapy to potentiate the efficacy of clinical polymyxin for the treatment of MDR infections. Recently, a small molecule, batimastat (also named BB-94), was characterized as an inhibitor directly acting on RseP, consequently blocking RpoE activation.⁶⁹ The discovery of BB-94 as an RseP inhibitor was utilized to better understand the RpoE stress-response pathway, as shown in a previous study.⁶⁹ However, the effectiveness of BB-94 in treating infectious diseases has not been demonstrated. Based on our discoveries reported here, we speculate that BB-94 can be used to assess the feasibility of using an RseP inhibitor to sensitize diverse colistin-resistant strains for combination therapy. This hypothesis needs to be examined in future studies.

Funding

This project was supported by AgResearch at The University of Tennessee. Atsushi Hinenoya was awarded the travel grant of the Overseas Research Scholar Program in Osaka Prefecture University. Fuzhou Xu was awarded the travel grant from the China Scholarship Council (201709110035). Ximin Zeng was partly supported by the United States Department of Agriculture National Institute of Food and Agriculture (Award No. 2018-67015-27475).

Transparency declarations

The authors have no conflicts of interest to declare.

Supplementary data

Figure S1 is available as Supplementary data at JAC Online.