2. A.A. Telke, J.-M. Rolain / International Journal of Antimicrobial Agents 46 (2015) 648–652 649
plasmid pBR322 were obtained from Sigma–Aldrich (Saint-
Quentin-Fallavier, France). Plasmid pZE21 MCS-1 was received
from the laboratory of Prof. Morten O. Sommer (Technical Univer-
sity of Denmark, Lyngby, Denmark). All strains were maintained on
Luria–Bertani (LB) and Mueller–Hinton agar at 37 ◦C. Kanamycin
(Sigma–Aldrich) was dissolved in sterile water at 50 mg/L and
was then further diluted in selective medium for plasmid stabil-
ity. Colistin sulphate was obtained from Sigma-Aldrich and a stock
solution was prepared in sterile water as per European Committee
on Antimicrobial Susceptibility Testing (EUCAST) guidelines. EZ1
DNA Investigator and Plasmid Spin Miniprep Kits were obtained
from QIAGEN (Courtaboeuf Cedex, France). End-ItTM DNA End-
Repair and Fast-LinkTM DNA Ligation Kits were obtained from
Epicenter® (Tebu-bio, Le Perray-en-Yvelines, France). Colistin Etest
strips were from bioMérieux (Marcy-l’Étoile, France). Restriction
enzymes were obtained from New England Biolabs (Hitchin, UK).
2.2. Genomic expression library construction
A genomic expression library was constructed by extracting
genomic DNA (gDNA) from S. algae MARS 14 using an EZ1 DNA
Investigator Kit (QIAGEN) followed by shearing the gDNA with a
Covaris S220 instrument (Covaris, Inc., Brighton, UK) in order to
obtain on average 3.0 kb DNA fragments. The size of the sheared
DNA was verified using an Agilent Bioanalyzer (Agilent Technolo-
gies, Palo Alto, CA) and DNA 7500 Kit (Supplementary Fig. S1).
Analyses were performed according to guidelines provided by Agi-
lent Technologies. The sheared DNA fragments were end-repaired
and ligated into a high-copy-number pZE21 MCS-1 expression plas-
mid and were then electrotransformed into E. coli TOP10 as the
expression host. A pZE21 MCS-1 plasmid possesses a kanamycin
resistance marker and a PLtetO-1 promoter [6]. The library was
titred by plating out 1 L and 10 L volumes of recovered cells onto
LB agar plates containing 50 mg/L kanamycin. Insert size distribu-
tion was estimated by gel electrophoresis of colony PCR products
obtained by amplifying the insert using pZE21 MCS-1 forward and
reverse primers (Supplementary Table S1).
Supplementary material related to this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ijantimicag.2015.
09.001.
The total size of the genomic expression library was determined
by multiplying average PCR-based insert size by the number of
CFU. The transformation mixture was enriched by growing the
cells in LB broth containing kanamycin (50 mg/L), followed by cre-
ation of glycerol stocks stored at -70 ◦C before processing. A 100 L
volume of stock library sample was placed on LB agar medium
containing kanamycin (50 mg/L) and colistin (8 mg/L) and was incu-
bated at 37 ◦C for 18–24 h. Resistant clones containing unique DNA
inserts were amplified by PCR and were sequenced using Sanger
sequencing technology (Applied Biosystems 3130xl Genetic Ana-
lyzer; Applied Biosystems, Carlsbad, CA).
2.3. Construction of ethanolamine phosphotransferase (EptA)
protein-expressing E. coli
The plasmid was extracted from colistin-resistant clones
obtained after screening the above genomic expression library
and was used as template for PCR using EptA-Forward and
EptA-Reverse primers (Supplementary Table S1) to obtain a
full-length eptA gene with PLtetO-1 promoter. The latter was ligated
at the EcoRV site of pBR322 and was then transformed into one-shot
electrocompetent E. coli TOP10. The EptA protein-expressing plas-
mid construct is shown in Supplementary Fig. S2. Transformants
were further confirmed by plasmid extraction and restriction map-
ping followed by PCR amplification of the ligated gene and DNA
sequencing.
Supplementary material related to this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ijantimicag.2015.
09.001.
2.4. RNA isolation and reverse transcription PCR (RT-PCR)
Total cellular RNA was isolated using an RNAprotect® Bacteria
Reagent Kit (QIAGEN) according to the manufacturer’s instructions.
RNA samples were treated with RNase-free DNase I (Ambion, Fos-
ter City, CA) to digest residual chromosomal DNA and were purified
with an RNeasy Kit (QIAGEN) prior to spectrophotometric quan-
tification at 260 nm. The housekeeping gene recA was used as an
internal control for Shewanella strains, and rspL for E. coli. Com-
parison with the recA and rspL genes allowed calculation of the
fold change in expression by the comparative threshold cycle (CT)
method. Colistin-susceptible S. putrefaciens [minimum inhibitory
concentration (MIC) = 0.19 mg/L] was used for comparative analy-
sis of gene expression. The primers used in this study are listed in
Supplementary Table S1.
2.5. Preparation of lipid A samples for matrix-assisted laser
desorption/ionisation time-of-flight mass spectrometry
(MALDI-TOF/MS) analysis
Lipid A samples were prepared as described previously [7] with
slight modification: colistin-resistant and -susceptible E. coli cells
were harvested from overnight cultures grown in 30 mL of LB
medium (pH 7.0). After overnight growth at 37 ◦C with aeration, the
cells were harvested and were re-suspended in 1 mL of lysis buffer
(iNtRON Biotechnology, Kyungki-Do, South Korea) and vortexed
vigorously, then 200 L of chloroform was added and the samples
were vortexed vigorously and incubated for 5 min at room temper-
ature. The phases were separated by centrifugation at 13,000 rpm
for 10 min and the upper phase was transferred to a new tube.
The upper phase was dried in a Vacufuge® plus apparatus (Eppen-
dorf, Hamburg, Germany) and was dissolved in 500 L of hydrolysis
buffer (pH 4.5) containing 12.5 mM sodium acetate and 1% sodium
dodecyl sulphate (SDS). For the release of lipid A from LPS, samples
were boiled for 1 h at 100 ◦C, dried in a Vacufuge® plus and were
re-suspended in a mixture of 100 L of water and 500 L of acidi-
fied ethanol (made by mixing 100 L of 4 M HCl with 20 mL of 95%
ethanol). The pellet was harvested by centrifugation at 2060 × g
for 10 min, washed with 500 L of 95% ethanol and centrifuged
again at 2060 × g for 10 min. The washing steps were repeated
to completely remove SDS. The pellet was dried at room tem-
perature for 5 min and lipid A was dissolved by the addition of
100 L of chloroform and methanol (3:1) and was used for MALDI-
TOF/MS. MALDI-TOF/MS analyses of lipid A were performed with
a MALDI-TOF Biotpyer (Bruker Daltonics). Analyses were carried
out in reflector mode at a mass range of m/z 1500–3000, with an
accelerating voltage of 20 kV and a delay time of 300 ns. The instru-
ment was externally calibrated. 2,5-Dihydroxybenzoic was used as
a matrix.
2.6. Minimum inhibitory concentration determination
MICs were determined in triplicate for each bacterial strain by
Etest and broth microdilution methods. Etest was performed using
a colistin Etest strip according to the manufacturer’s instructions
on Mueller–Hinton agar plates with a 0.5 McFarland inoculum
in sterile water and the results were interpreted as indicated
by EUCAST guidelines (Breakpoint tables for interpretation of
MICs and zone diameters. Version 4, 2014; http://www.eucast.
org/fileadmin/src/media/PDFs/EUCAST files/Breakpoint tables/
Breakpoint table v 4.0.pdf). The broth microdilution method used
a 100 L volume of a two-fold serial dilution of colistin in LB broth
3. 650 A.A. Telke, J.-M. Rolain / International Journal of Antimicrobial Agents 46 (2015) 648–652
(pH 7.2) dispensed in 96-well microtitre plates. Bacteria at a final
concentration of 5 × 105 CFU/mL in sterile water were added to
each well. MICs were recorded as the lowest concentration of
antibiotic that did not allow visible bacterial growth after 20 h
incubation at 37 ◦C under shaking conditions (750 rpm).
2.7. Bioinformatic analysis
A codon code aligner was used to assemble the DNA sequence
reads. The assembled DNA sequences were blasted (on 20 January
2015) against bacterial genomes in the Integrated Microbial
Genomes (IMG) database to retrieve the sequence information
including gene name and orthology [8]. Global nucleotide and
amino acid identity were determined using EMBOSS alignment
tools. Evolutionary history was inferred using the neighbour-
joining method, and evolutionary analyses were conducted in
MEGA6 [9].
3. Results
3.1. Strain Identification
MALDI-TOF analysis of the clinical isolate resulted in identifica-
tion of S. putrefaciens with a score value of 2.003 (suggesting secure
genus identification, probable species identification). To confirm
the bacterial species at the genetic level, the 16S rRNA gene was
amplified and the result was compared with the 16S rRNA sequence
of previously reported Shewanella spp. retrieved from the National
Center for Biotechnology Information (NCBI) GenBank. The global
16S rRNA sequence identity of the isolate was 99.2% with previously
reported S. algae JCM 21037. The 16S rRNA sequence was submitted
to the European Molecular Biology Laboratory (EMBL) bank with
strain name S. algae MARS 14 and accession no. LN795823.
3.2. Genomic expression library and determination of antibiotic
resistance
The average library insert size was found to be ca. 2900 bp. The
total library size was 1.30 × 108 bp. Five colistin-resistant clones
were obtained after functional screening. These were picked and
then further verified by growing them in fresh LB broth contain-
ing kanamycin (50 mg/L) and colistin (8 mg/L). All clones were able
to grow in the presence of the supplied colistin concentration at
which the control E. coli strain was unable to grow. A plasmid
was extracted from all clones and was used as a template for PCR
amplification of insert DNA sequences. Interestingly, after analy-
sis of insert DNA sequences from all clones, it was predicted that
all clones have a common DNA sequence or gene that is respon-
sible for conferring colistin resistance; this was further confirmed
by cloning (low-copy-number vector pBR322) and expressing the
predicted full-length gene in E. coli followed by analysis of the
colistin resistance pattern (Fig. 1). The predicted gene was deter-
mined to encode EptA belonging to the YhjW/YjdB/Yijp superfamily
according to amino acid sequence homology. It has 98.2% global
amino acid identity with a predicted metal-dependent hydrolase
protein from S. algae JCM 21037, a top hit from the IMG database
blast results. The full-length gene sequence was submitted to EMBL
bank with accession no. LN811438. The E. coli TOP10 transformant
harbouring the plasmid encoding EptA had a higher colistin MIC
compared with the control E. coli TOP10 strain (Fig. 1; Table 1).
3.3. Analysis of eptA transcription in colistin-resistant and
-susceptible strains
Total RNA was quantified by NanoDropTM (Thermo Scientific,
Wilmington, DE) and an equal amount of RNA sample was used for
Table 1
Minimum inhibitory concentrations (MICs) of Escherichia coli and Shewanella algae
MARS 14.
Strain MIC (mg/L)a
Etest BMD
pEptA
/E. coli TOP10b
4.0 14
pBR322/E. coli TOP10c
0.064 0.7
Shewanella algae MARS 14 4.0 12
BMD, broth microdilution.
a
European Committee on Antimicrobial Susceptibility Testing (EUCAST) clinical
breakpoints for Enterobacteriaceae family bacteria: sensitive, ≤2 mg/L; and resis-
tant, >2 mg/L.
b
pEptA
/E. coli TOP10, E. coli TOP10 transformant harbouring plasmid-encoded
ethanolamine phosphotransferase.
c
pBR322/E. coli Top10, E. coli TOP10 harbouring empty pBR322 plasmid.
RT-PCR-based analysis of gene expression. The expression level of
the eptA gene in S. algae MARS 14 was upregulated 27-fold com-
pared with the susceptible strain. Also, the eptA gene expression
level in colistin-resistant E. coli TOP10 was upregulated (31.86-fold
change) compared with susceptible E. coli TOP10.
3.4. Analysis of lipid A in colistin-resistant and -susceptible E. coli
strains
Samples of lipid A were prepared from colistin-resistant and
-susceptible E. coli strains and were analysed by MALDI-TOF/MS
(see Section 2.5). Analyses of the lipid A samples are shown in Sup-
plementary Fig. S3. The observed ions were consistent with the
normal E. coli lipid A molecules; namely, the bis-phosphorylated
hexa-acylated structure (m/z = 1797) and its hepta-acylated ver-
sion due to the addition of a palmitic acid residue (m/z = 2034).
The lipid A preparations from strains carrying EptA showed addi-
tional ions due to the addition of phosphoethanolamine (PEA) to
the bis-phosphorylated structure (m/z 1921; i.e. 1797 + 124) and
the hepta-acylated structure (m/z 2158; i.e. 2034 + 124).
Supplementary material related to this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ijantimicag.2015.
09.001.
4. Discussion
Bacteria of the genus Shewanella are ubiquitous marine
organisms known for their remarkable metabolic capabilities. She-
wanella putrefaciens, Shewanella haliotis and S. algae have been
reported to cause human infections [2]. Similar to S. algae, the
phylogenetically closely related species S. haliotis was found to be
resistant to polymyxin antibiotics [10]. PCR can only be used to
screen a sample for known genes; it is an ineffective method for
identifying novel resistance genes. Functional genomics screening
obviates this problem by identifying genes by their function in
an expression vector rather than by a specific sequence used for
PCR probing. Therefore, a functional genomics strategy was used
to decipher the colistin resistance mechanism in S. algae MARS
14. Colistin and polymyxin B act as potent antibacterial lipopep-
tides by disrupting the LPS structure in the outer membrane
of Gram-negative bacteria [3]. LPS is an essential component of
the outer monolayer of nearly all Gram-negative bacteria. LPS is
composed of a hydrophobic anchor known as lipid A, an inner
core oligosaccharide and a repeating O-antigen polysaccharide.
Biosynthesis of lipid A is largely conserved across Gram-negative
organisms. The sugar 3-deoxy-d-manno-octulosonic acid (Kdo) 3
is the first sugar added to lipid A, and laboratory strains of E. coli
cannot survive without synthesising the minimal LPS substructure
Kdo2-lipid A [11]. Many bacterial species, such as E. coli, add two
Kdo sugars with a single Kdo transferase, whereas Shewanella add a
4. A.A. Telke, J.-M. Rolain / International Journal of Antimicrobial Agents 46 (2015) 648–652 651
Fig. 1. Etest for determination of colistin resistance. pBR322/E. coli TOP10, Escherichia coli TOP10 with empty pBR322 plasmid; pEptA
/E. coli TOP10, E. coli TOP10 with
ethanolamine phosphotransferase protein-expressing plasmid construct. pBR322-based strains were maintained by addition of 100 mg/L ampicillin to the growth medium.
European Committee on Antimicrobial Susceptibility Testing (EUCAST) clinical breakpoints for Enterobacteriaceae family bacteria: sensitive, ≤2 mg/L; and resistant, >2 mg/L.
single Kdo that is then phosphorylated by a separate enzyme [11].
In strains of Shewanella, Kdo is further modified by converting the
C8 hydroxyl group to a primary amine [8-amino-3,8-dideoxy-d-
manno-octulosonic acid (Kdo8N)] [11]. Chromosomal deletion of
kdnA and kdnB genes, required for incorporation of Kdo8N into
lipid A present in LPS of Shewanella spp., resulted in increased sen-
sitivity to polymyxin as well as to bile salts, suggesting a role in
outer membrane LPS [11]. It was predicted that a single gene was
responsible for colistin resistance in all functionally selected clones.
A search of the Conserved Domains database (http://www.ncbi.
nlm.nih.gov/Structure/cdd/wrpsb.cgi; 20 January 2015) using the
amino acid sequence of this predicted gene as a query retrieved the
lipid A PEA transferase (also called as PmrC) and sulfatase domains.
Ethanolamine phosphotransferase (EptA or so-called PmrC) con-
sists of a hydrophobic transmembrane domain linked to a globular
C-terminal sulfatase domain and catalyses the addition of PEA
to the lipid A moiety of outer membrane LPS [12]. The amino
acid sequence of EptA, or so-called PmrC, was blasted against
the IMG database (tblastp) to look for closely related protein
sequences belonging to different bacterial genera. Several hits
for ethanolamine phosphotransferases belonging to pathogenic
Gram-negative bacteria were found with 40–45% global amino acid
identity with the query sequence.
Amino acid sequence phylogenetic analysis of this gene showed
that this had closest similarity with EptA from Photobacterium
damselae subsp. piscicida DI21, Salmonella enterica enterica serovar
Cubana CFSAN002050, E. coli O157:H7 EDL933, E. coli O145:H28
RM12581 and Plesiomonas shigelloides serovar O1 302-73 (Sup-
plementary Fig. S4). Amongst these, a P. damselae subsp. piscicida
strain isolated from Italian aquaculture farms was reported to be
colistin-resistant at a concentration of 10 mg/L [13]. Previously, it
has been reported that the colistin-resistant phenotype would be
associated with increased expression of EptA or so-called PmrC
[14], which encodes the protein that adds PEA to lipid A. For
this reason, we studied the transcription levels of this EptA in
colistin-resistant and -susceptible strains. RT-PCR analysis showed
a significant increase in eptA gene expression in colistin-resistant
strains compared with colistin-susceptible strains. This suggests
that overexpression of predicted EptA resulted in a significant
increase in the MIC of colistin compared with the respective con-
trol. MALDI-TOF/MS analysis of E. coli lipid A confirmed that EptA
modified lipid A headgroups with PEA (Supplementary Fig. S3).
Similarly, Kim et al. overexpressed the chromosomally encoded
phosphoethanolamine transferase from virulent E. coli O157:H7
strain in an E. coli K-12-based expression host and observed addi-
tion of PEA to the lipid A moiety, followed by a slight increase in
resistance to polymyxin B [15]. Most Gram-negative bacteria build
up polymyxin resistance by modification of the outer membrane
LPS structure induced by the PmrA–PmrB regulatory system and
PhoP–PhoQ signal transduction network [3,16]. Salmonella enter-
ica serovar Typhimurium has been extensively studied to explore
its mechanism of resistance to polymyxin antibiotics. Salmonella
Typhimurium acquires polymyxin resistance by overexpression of
the ethanolamine phosphotransferase (or so-called pmrC) and pmrF
operons [17]. The two-component regulatory system PmrA/PmrB
controls expression of the pmrC and pmrF operons. Mutations in
this system can cause constitutive overexpression of pmrC and pmrF
operons, resulting in the extensive addition of PEA and 4-amino-4-
deoxy-l-arabinose to the lipid A moiety of outer membrane LPS [3].
Similar mechanisms were observed in polymyxin-resistant K. pneu-
moniae, E. coli and A. baumannii [3]. Overexpression of neisserial
phosphoethanolamine transferase (LptA) and of DsbA oxidoreduc-
tase in E. coli DH5␣ results in a 32-fold increase in polymyxin
resistance and a 26% increase in PEA in the lipid A moiety of LPS [18].
Overall, LPS-modifying enzymes such as phosphoethanolamine
transferase play a key role in protecting Gram-negative bacteria
from polymyxin antibiotics.
Supplementary material related to this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ijantimicag.2015.
09.001.
5. 652 A.A. Telke, J.-M. Rolain / International Journal of Antimicrobial Agents 46 (2015) 648–652
5. Conclusions
Functional genomics was successfully employed to decipher the
molecular support of resistance to colistin in S. algae MARS 14 that
was associated with outer membrane LPS structure modification
by addition of PEA through EptA activity.
Acknowledgments
The authors are very grateful to the laboratory of Prof. Morten
O. Sommer (Technical University of Denmark, Lyngby, Denmark)
for providing the pZE21 MCS-1 plasmid as well as to Linda Had-
jadj and Andre Barnaud for technical assistance. One of the authors
(AAT) is thankful to IHU Méditerranée Infection and Infectiopôle
Sud Foundations for a postdoctoral fellowship.
Funding: This work was partly funded by IHU Méditerranée
Infection.
Competing interests: None declared.
Ethical approval: Not required.
References
[1] Beleneva IA, Magarlamov TY, Eliseikina MG, Zhukova NV. Biochemical and
pathogenic properties of the natural isolate of Shewanella algae from Peter the
Great Bay, Sea of Japan. J Invertebr Pathol 2009;102:250–5.
[2] Vignier N, Barreau M, Olive C, Baubion E, Theodose R, Hochedez P, et al. Human
infection with Shewanella putrefaciens and S. algae: report of 16 cases in Mar-
tinique and review of the literature. Am J Trop Med Hyg 2013;89:151–6.
[3] Olaitan AO, Morand S, Rolain JM. Mechanisms of polymyxin resistance:
acquired and intrinsic resistance in bacteria. Front Microbiol 2014;5:643.
[4] Kieffer N, Poirel L, Nordmann P, Madec JY, Haenni M. Emergence of colistin
resistance in Klebsiella pneumoniae from veterinary medicine. J Antimicrob
Chemother 2015;70:1265–7.
[5] Kamke J, Taylor MW, Schmitt S. Activity profiles for marine sponge-associated
bacteria obtained by 16S rRNA vs 16S rRNA gene comparisons. ISME J
2010;4:498–508.
[6] Lutz R, Bujard H. Independent and tight regulation of transcriptional units in
Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements.
Nucleic Acids Res 1997;25:1203–10.
[7] Lee H, Hsu FF, Turk J, Groisman EA. The PmrA-regulated pmrC gene medi-
ates phosphoethanolamine modification of lipid A and polymyxin resistance
in Salmonella enterica. J Bacteriol 2004;186:4124–33.
[8] Markowitz VM, Chen IM, Palaniappan K, Chu K, Szeto E, Pillay M, et al. IMG
4 version of the Integrated Microbial Genomes comparative analysis system.
Nucleic Acids Res 2014;42(Database issue):D560–7.
[9] Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: molec-
ular evolutionary genetics analysis version 6.0. Mol Biol Evol 2013;30:
2725–9.
[10] Kim D, Baik KS, Kim MS, Jung BM, Shin TS, Chung GH, et al. Shewanella haliotis
sp. nov., isolated from the gut microflora of abalone, Haliotis discus hannai. Int
J Syst Evol Microbiol 2007;57:2926–31.
[11] Gattis SG, Chung HS, Trent MS, Raetz CR. The origin of 8-amino-3,8-
dideoxy-d-manno-octulosonic acid (Kdo8N) in the lipopolysaccharide of She-
wanella oneidensis. J Biol Chem 2013;288:9216–25.
[12] Naessan CL, Egge-Jacobsen W, Heiniger RW, Wolfgang MC, Aas FE, Rohr A, et al.
Genetic and functional analyses of PptA, a phospho-form transferase targeting
type IV pili in Neisseria gonorrhoeae. J Bacteriol 2008;190:387–400.
[13] Lagana P, Caruso G, Minutoli E, Zaccone R, Santi D. Susceptibility to antibiotics
of Vibrio spp. and Photobacterium damsela ssp. piscicida strains isolated from
Italian aquaculture farms. New Microbiol 2011;34:53–63.
[14] Arroyo LA, Herrera CM, Fernandez L, Hankins JV, Trent MS, Hancock RE. The
pmrCAB operon mediates polymyxin resistance in Acinetobacter baumannii
ATCC 17978 and clinical isolates through phosphoethanolamine modification
of lipid A. Antimicrob Agents Chemother 2011;55:3743–51.
[15] Kim SH, Jia W, Parreira VR, Bishop RE, Gyles CL. Phosphoethanolamine substi-
tution in the lipid A of Escherichia coli O157:H7 and its association with PmrC.
Microbiology 2006;152:657–66.
[16] Olaitan AO, Diene SM, Kempf M, Berrazeg M, Bakour S, Gupta SK, et al. World-
wide emergence of colistin resistance in Klebsiella pneumoniae from healthy
humans and patients in Lao PDR, Thailand, Israel, Nigeria and France owing to
inactivation of the PhoP/PhoQ regulator mgrB: an epidemiological and molec-
ular study. Int J Antimicrob Agents 2014;44:500–7.
[17] Merighi M, Carroll-Portillo A, Septer AN, Bhatiya A, Gunn JS. Role of Salmonella
enterica serovar Typhimurium two-component system PreA/PreB in modulating
PmrA-regulated gene transcription. J Bacteriol 2006;188:141–9.
[18] Piek S, Wang Z, Ganguly J, Lakey AM, Bartley SN, Mowlaboccus S, et al. The
role of oxidoreductases in determining the function of the neisserial lipid A
phosphoethanolamine transferase required for resistance to polymyxin. PLOS
ONE 2014;9:e106513.