J Antimicrob Chemother 2018; 73: 3359–3367

doi:10.1093/jac/dky330 Advance Access publication 1 September 2018

Rapid detection and discrimination of chromosome- and MCR-plasmid-

mediated resistance to polymyxins by MALDI-TOF MS in
Escherichia coli: the MALDIxin test
1–4
Laurent Dortet *†, Remy A. Bonnin3,4, Ivana Pennisi1, Lauraine Gauthier2–4, Agnès B. Jousset2–4,
Laura Dabos , R. Christopher D. Furniss1, Despoina A. I. Mavridou1, Pierre Bogaerts5, Youri Glupczynski5,
3

Anais Potron4,6, Patrick Plesiat4,6, Racha Beyrouthy4,7, Frédéric Robin4,7, Richard Bonnet4,7, Thierry Naas2–4,

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

1 1
Alain Filloux and Gerald Larrouy-Maumus †

1
MRC Centre for Molecular Bacteriology and Infection, Department of Life Sciences, Faculty of Natural Sciences, Imperial College
London, London SW7 2AZ, UK; 2Department of Bacteriology-Hygiene, Bicêtre Hospital, Assistance Publique - Hôpitaux de Paris, Le
Kremlin-Bicêtre, France; 3EA7361 ‘Structure, dynamic, function and expression of broad spectrum b-lactamases’, Paris-Sud University,
Paris Saclay University, LabEx Lermit, Faculty of Medicine, Le Kremlin-Bicêtre, France; 4French National Reference Center for Antibiotic
Resistance, Le Kremlin-Bicêtre, France; 5Laboratory of Clinical Microbiology, Belgian National Reference Center for Monitoring
Antimicrobial Resistance in Gram-negative Bacteria, CHU UCL Namur, Yvoir, Belgium; 6Bacteriology Unit, University Hospital of
Besançon, Besançon, France; 7Bacteriology Unit, University Hospital of Clermont-Ferrand, Clermont-Ferrand, France

*Corresponding author. Hôpital de Bicêtre, Service de Bactériologie-Hygiène, 78 rue du Général Leclerc, 94270, Le Kremlin-Bicêtre, France.
Tel: !33 (0) 1 45 21 20 19; E-mail: [email protected] orcid.org/0000-0001-6596-7384
†These authors made an equal contribution to the article.

Received 21 February 2018; returned 10 April 2018; revised 28 June 2018; accepted 26 July 2018

Background: Polymyxins are currently considered a last-resort treatment for infections caused by MDR
Gram-negative bacteria. Recently, the emergence of carbapenemase-producing Enterobacteriaceae has
accelerated the use of polymyxins in the clinic, resulting in an increase in polymyxin-resistant bacteria.
Polymyxin resistance arises through modification of lipid A, such as the addition of phosphoethanolamine
(pETN). The underlying mechanisms involve numerous chromosome-encoded genes or, more worryingly,
a plasmid-encoded pETN transferase named MCR. Currently, detection of polymyxin resistance is difficult and
time consuming.
Objectives: To develop a rapid diagnostic test that can identify polymyxin resistance and at the same time dif-
ferentiate between chromosome- and plasmid-encoded resistances.
Methods: We developed a MALDI-TOF MS-based method, named the MALDIxin test, which allows the detection
of polymyxin resistance-related modifications to lipid A (i.e. pETN addition), on intact bacteria, in ,15 min.
Results: Using a characterized collection of polymyxin-susceptible and -resistant Escherichia coli, we demon-
strated that our method is able to identify polymyxin-resistant isolates in 15 min whilst simultaneously discrimi-
nating between chromosome- and plasmid-encoded resistance. We validated the MALDIxin test on different
media, using fresh and aged colonies and show that it successfully detects all MCR-1 producers in a blindly ana-
lysed set of carbapenemase-producing E. coli strains.
Conclusions: The MALDIxin test is an accurate, rapid, cost-effective and scalable method that represents a
major advance in the diagnosis of polymyxin resistance by directly assessing lipid A modifications in intact
bacteria.

Introduction rapid dissemination of MDR Gram-negative bacteria. In this con-

text, and because of the paucity of alternative therapeutic options,
The increasing prevalence of antibiotic resistance is a threat to glo- polymyxins (polymyxin B and colistin) have become the last-resort
bal human health whilst the pipeline of new antimicrobials is ex- therapy, especially for treating infections caused by carbapenem-
tremely limited. One of the most concerning developments is the resistant Enterobacteriaceae.1 These organisms are becoming

C The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
V
For permissions, please email: [email protected].
3359
Dortet et al.

increasingly prevalent worldwide, with many producing (disruptions, deletions, mutations) involved are not systematically
carbapenem-hydrolysing enzymes [carbapenemase-producing described or characterized. Regarding plasmid-encoded resist-
Enterobacteriaceae (CPE)].2 This phenomenon has accelerated the ance, five families of mcr genes have already been reported in
use of polymyxins in CPE-endemic countries such as Greece and Enterobacteriaceae.31–35 MCR-2, MCR-3, MCR-4 and MCR-5 share
Italy, and as a result increased resistance to polymyxins is now only 81%, 34%, 33% and 31% amino acid identity, respectively,
being reported.3–7 with MCR-1 (Figure S1, available as Supplementary data at JAC
In Enterobacteriaceae, acquired resistance to polymyxins arises Online). This diversity will inevitably lead to the failure of systemat-
from modifications to the drug target, i.e. LPS.8,9 These modifica- ic detection of polymyxin resistance as the available molecular
tions consist of addition(s) of cationic groups such as phosphoe- biology tools are dedicated to mcr-1 and/or mcr-2 detection
thanolamine (pETN) and/or 4-amino-L-arabinose (L-Ara4N) to the (Figure S2).31,32,34,35
lipid A part of the LPS. This results in decreased binding of the poly- To overcome these issues, we developed a cost-effective assay
myxin to the LPS due to electrostatic repulsion between the added based on MALDI-TOF technology, named the MALDIxin test, which

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

cationic groups and the positively charged antibiotic. Addition of aims to detect polymyxin resistance using a single bacterial colony
pETN and L-Ara4N is associated with chromosome-encoded resist- in ,15 min. By using a large panel of E. coli strains, we also demon-
ance mechanisms, such as mutations in the PmrA/PmrB or strate that this technique can, at the same time, efficiently dis-
PhoP/PhoQ two-component systems, or through alterations to the criminate between chromosome- and plasmid-encoded (mcr-like
master regulator MgrB.8,9 More than 10 different genes have been genes) resistance mechanisms.
implicated in chromosome-encoded resistance to polymyxin.
However, these chromosomal mutations have a high fitness cost, Materials and methods
and cannot be transferred between organisms.9,10 Recently, it has
been reported that addition of pETN can also occur through the ex- Bacterial strains and plasmids
pression of a plasmid-encoded pETN transferase, named MCR-1.11 We used a collection of 87 E. coli strains including 41 polymyxin-resistant
Since its initial description, the mcr-1 gene has been identified isolates, of which 29 were MCR producers (18 MCR-1, 2 MCR-1.5, 3 MCR-2,
worldwide in Enterobacteriaceae (mostly Escherichia coli), recov- 2 MCR-3 and 4 MCR-5). The 46 polymyxin-susceptible E. coli strains were of
ered from both animal and human samples.8,12 The mcr-1 poly- various phenotypes, from WT to carbapenemase producers (Table S1). The
myxin resistance gene is thus a major threat due to its lack of (or MALDIxin test was also prospectively evaluated using a collection of 78 iso-
very low) fitness cost and its ability to be transferred between lates of carbapenemase-producing E. coli received during October and
strains and species.13,14 In order to contain the spread of poly- November 2016, from the French National Reference Centre (NRC) for
myxin resistance in Enterobacteriaceae, a test that enables both Antimicrobial Resistance (Table 1).
In addition to the above clinical strains, functional mcr-like genes
rapid and reliable detection of polymyxin resistance as well as dis-
were cloned into pDM1, an isopropyl b-D-thiogalactoside (IPTG)-inducible
crimination between plasmid- and chromosome-encoded resist-
TetR derivative of pACYC184.36 mcr-1, mcr-2 and mcr-5 were cloned into
ance is urgently needed. the SacI/XmaI sites of the vector, while for mcr-3 the NdeI/XmaI sites
Currently, broth microdilution (BMD) is the gold standard tech- were used. The resulting plasmids were transformed into E. coli DH5a and
nique for polymyxin susceptibility testing,12,15 generating results MC1000.37 Whenever these strains were used (during susceptibility test-
up to 24 h after bacterial isolation.16 This method has been recent- ing or for the MALDIxin test), 0.5 mM IPTG was added to the growth
ly chosen as the unique reference by the CLSI and by the medium; the pDM1 vector was used as a control.
EUCAST,17 which rules out methods classically used for determin-
ation of antimicrobial susceptibility, such as agar dilution, disc dif-
fusion and gradient diffusion (Etest). Additionally, because of high
Susceptibility testing
rates of false susceptibility observed when using automated sys- MICs were determined by BMD according to the guidelines of the CLSI and
tems (MicroScanV, VitekV 2 and BD PhoenixTM), results obtained
R R EUCAST joint subcommittee.17 Results were interpreted using EUCAST
breakpoints as updated in 2017.38
with these approaches would need to be confirmed by BMD.15,18
Recently, a biochemical test, the Rapid Polymyxin NP test, which
detects bacterial growth in the presence of a defined concentra- MALDIxin test
tion of a polymyxin, has been developed.19 This test is claimed to Bacterial strains were cultured on lysogeny broth (LB) agar and Mueller–
be rapid (within 2 h) and is easy to perform, but is not able to dis- Hinton (MH) agar. A single colony was resuspended in 200 lL of distilled
criminate between plasmid- and chromosome-encoded resist- water in a 1.5 mL microtube. The bacteria were then washed three times
ance to polymyxin; the latter requires an additional step, usually with 200 lL of double-distilled water and resuspended in 100 lL of double-
molecular detection (e.g. PCR) of mcr-like genes. However, fast de- distilled water. To perform MS analysis, a patented 2,5-dihydroxybenzoic
tection of mcr-positive isolates is one of the key issues in curbing acid (DHB) matrix was used at a final concentration of 10 mg/mL in chloro-
the dissemination of this plasmid-encoded resistance and facilitat- form/methanol 90:10 v/v. We loaded 0.4 lL of the bacterial solution onto
ing the use of any future MCR inhibitors,20,21 which may become the target and immediately overlaid it with 0.8 lL of the matrix solution.
Bacterial solution and matrix were mixed directly on the target by pipetting
essential for the treatment of patients infected with MDR and
and the mix was dried gently under a stream of air (,1 min). MALDI-TOF
MCR-producing Enterobacteriaceae.22–30
MS analysis was performed on a 4800 Proteomics Analyzer (Applied
Molecular biology (e.g. PCR or WGS) is widely used for the detec- Biosystems) using the reflectron mode. Samples were analysed by operat-
tion of many antimicrobial resistance mechanisms. Nonetheless, ing at 20 kV in the negative ion mode using an extraction delay time set at
this may not be the best option for the detection of polymyxin re- 20 ns. Typically, spectra from 500 to 2000 laser shots were summed to ob-
sistance. The chromosome-encoded genes associated with poly- tain the final spectrum. MS data were analysed using Data Explorer version
myxin resistance are numerous, and the gene modifications 4.9 (Applied Biosystems).

3360
Rapid detection of colistin resistance by MALDI-TOF MS JAC
Table 1. Detection and characterization of polymyxin resistance using conventional techniques (MICs and PCR) and the MALDIxin test on 78 carbape-
nemase-producing E. coli strains received by the French NRC during October and November 2016

Polymyxin resistance MALDI-TOF results

Carbapenemase No. of isolates CST MIC (mg/L) mcr-1/-2 PCRa PRRE. coli interpretation

KPC-2 1 0.25 # 0 susceptible

KPC-3b 1 0.25 # 0 susceptible
NDM-1 3 0.25 # 0 susceptible
NDM-1 1 0.5 # 0 susceptible
NDM-1 1 4 ! (mcr-1) 1.73+0.23 plasmid-encoded resistance
NDM-1 1 4 ! (mcr-1) 2.34+0.12 plasmid-encoded resistance

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

NDM-5 2 0.25 # 0 susceptible
NDM-5 3 0.5 # 0 susceptible
VIM-1 1 0.5 # 0 susceptible
VIM-4 1 0.25 # 0 susceptible
OXA-48 27 0.25 # 0 susceptible
OXA-48 26 0.5 # 0 susceptible
OXA-48 1 1 # 0 susceptible
OXA-48b 1 4 ! (mcr-1) 1.29+0.63 plasmid-encoded resistance
OXA-181 4 0.25 # 0 susceptible
OXA-204 1 0.25 # 0 susceptible
OXA-244 1 0.25 # 0 susceptible
OXA-244 1 0.5 # 0 susceptible
OXA-181!NDM-5 1 0.25 # 0 susceptible

CST, colistin.
a
!, positive PCR for the gene indicated in parentheses; #, negative PCR.
b
Strains were isolated from the same patient and have been previously reported by Beyrouthy et al.42 as E. coli WI1 and E. coli WI2, respectively.

Statistical analysis In all polymyxin-resistant E. coli strains, an additional peak at

All experiments were carried out on three independent bacterial cultures m/z 1919.2 was observed (Figure 2b and c) independent of the
(on three different days). Data were compared two-by-two using the un- resistance mechanism involved (chromosome- or plasmid-
paired Welch’s t-test. P values ,0.05 were considered statistically different. encoded). This peak corresponds to the addition of one pETN moi-
ety to the phosphate group at position 1 of native lipid A (Figure 1,
right bottom panel), leading to an increase of 123 mass units com-
Results
pared with the mass assigned to native lipid A (Figure 2b and c).
We have previously reported a method for the direct detection of In the case of plasmid-encoded resistance (mcr-like genes in
lipid A from intact and untreated Gram-negative bacteria.39 In Enterobacteriaceae), a third peak at m/z 1821.2 was routinely
Enterobacteriaceae, polymyxin resistance is associated with the observed in addition to the peaks corresponding to the native lipid
addition of, at least, pETN on the phosphate group at position 40 or A (m/z 1796.2) and the pETN-modified lipid A (m/z 1919.2)
1 of lipid A (Figure 1). Accordingly, this modification should be (Figure 2c). This m/z 1821.2 peak was assigned to the addition of a
observed on MALDI-TOF spectra obtained from polymyxin- pETN moiety onto the phosphate group at position 40 of native lipid
resistant isolates as it corresponds to an increase of 123 mass units A with concomitant loss of the phosphate group at position 1
compared with the mass assigned to native lipid A (at m/z 1796 in (Figure 1, right upper panel), and appears to be a specific marker of
E. coli) (Figure 1). MCR-like enzymes. This observation is in line with the fact that the
dephosphorylation can only occur on the phosphate group at pos-
ition 1.40
To further support this observation, we analysed 73 E. coli isolates
Determination of polymyxin resistance in E. coli using including 46 polymyxin-susceptible strains with various antimicrobial
the MALDIxin test resistance phenotypes. Of these strains, four possess chromosome-
In polymyxin-susceptible E. coli strains the negative mass spec- encoded resistance to polymyxin and 23 are MCR producers (Table
trum scanned between m/z 1600 and 2200 was dominated by a S1). The intensity of the peaks corresponding to the native lipid A (m/
set of peaks assigned to bis-phosphorylated hexa-acyl lipid A z 1796.2), the pETN-modified lipid A (m/z 1919.2) and the specific
(Figure 2a). The major peak at m/z 1796.2 is known to correspond marker of MCR resistance (m/z 1821.2) were recorded from three in-
to the hexa-acyl diphosphoryl lipid A containing four C14:0 3-OH, dependent experiments. The ratio [termed polymyxin resistance
one C14:0 and one C12:0, and referred to as native lipid A (Figure 1, ratio (PRR) hereafter] of the sum of the intensities of the peaks asso-
left panel). ciated with modified lipid A (for E. coli peaks at m/z 1919.2 and m/z

3361
Dortet et al.

SUSCEPTIBLE RESISTANT

O O OH O
H 2N
O P O P O O
HO O
OH OH O H O
pETN NH
O NH OH
O O
O
O OH
O O OH
O

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

O OH
O
O P O O
HO O
OH O H O O
NH
NH O P OH

ke
O O H3C
O O OH H3C

-li
O 14 H3 C 14 H 3C
OH
O O OH CR H 3C
14 H3C 14
M
12
O 14
Modified Lipid A
(1821 m/z)
M
CR

O OH O
-li
Ch de
en

H3C O P O O
ke

H 3C HO O
ro d r
co

14 H 3C 14 H 3C OH O H O O O
m es

NH NH2
14 H3C NH O P O P O
os is

H 3C 14 O O
12 O O
om tan

14 OH OH
O OH
e- ce

O O OH pETN
O
Native Lipid A
(1796 m/z)

H3C H 3C
14 H3C 14 H 3C
14 H3C 14
Modified Lipid A
Plasmid-encoded H 3C
12 (1919 m/z)
resistance 14

Figure 1. Lipid A modifications caused by chromosome-encoded determinants or MCR enzymes that can be detected by the MALDIxin test.
The structure of the native E. coli lipid A is depicted on the left. Phosphate groups are circled in blue, dephosphorylation at position 1 is circled in green,
phosphoethanolamine (pETN) groups are circled in orange. Numbers at the end of each fatty acid chain indicate the number of carbon atoms.
The positions of the peaks of interest in the mass spectrum for each structure are also given. This figure appears in colour in the online version of JAC
and in black and white in the print version of JAC.

1821.2) to the intensity of the peak of native lipid A (for E. coli m/z chromosome-encoded (0.1 , PRRE. coli , 0.5) and plasmid-encoded
1796.2) allows accurate distinction between polymyxin-susceptible resistance (PRRE. coli .0.5).
and polymyxin-resistant isolates, but also discrimination between
chromosome-encoded and MCR-related (i.e. plasmid-encoded) re-
sistance to polymyxin. (Figure 2d). The PRR value for all susceptible Performance of the MALDIxin test on colonies grown on
E. coli strains was found to be 0 (Table S1 and Figure 2d). The ratio clinically relevant media
ranged from 0.109 to 0.481 (average of 0.210) for E. coli strains with We assessed the performance of the MALDIxin test using colonies
chromosome-encoded resistance to polymyxin, and from 0.533 to grown on media routinely used to test for antimicrobial susceptibil-
4.844 (average of 2.340) for all MCR producers (Table S1 and ity, i.e. MH agar. In addition, because of the rapid rise of MCR-1-
Figure 2d). The distribution of the PRRE. coli values of polymyxin- producing Enterobacteriaceae since 2016, it has been suggested
susceptible isolates compared with E. coli strains with chromosome- that polymyxin-supplemented media should be used to screen
encoded resistance to polymyxin and MCR-producers were all signifi- organisms isolated from infected patients.41 Accordingly, the
cantly different (P , 0.001, Welch’s t-test). Analysis of receiver oper- MALDIxin test was conducted on three E. coli strains (a WT strain, a
ating characteristic (ROC) curves allowed us to define two cut-off strain with chromosome-encoded resistance and an MCR-1-
values for PRRE. coli (0.1 and 0.5) that discriminate polymyxin- producing strain), grown on LB agar, MH agar and polymyxin
susceptible (PRRE. coli ,0.1) and polymyxin-resistant isolates B-supplemented MH agar at final concentrations of either 1 or
(PRRE. coli.0.1) and also allows further discrimination between 2 mg/L. As shown in Figure S3, similar spectra were obtained

3362
Rapid detection of colistin resistance by MALDI-TOF MS JAC
(a) (d) P < 0.0001
P < 0.0001
5
1796.2
Polymyxin
susceptible
4.5
Intensity

I1919 + I1821
4 PRRE. coli =
I1796

3.5

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

1600 1720 1840 1960 2080 2200
m/z
3

(b) 1796.2 Polymyxin

resistant 2.5

PRRE. coli
(chromosome)

2
Intensity

1919.2
1.5
+ 123

1600 1720 1840 1960 2080 2200 1

m/z
P < 0.0001
(c) 1821.2 Polymyxin 0.5
1796.2 resistant
1919.2 0.4
(mcr-1)
0.3
0.2
Intensity

+ 25
0.1
+ 123 0
Polymyxin Chromosome- Plasmid-encoded
susceptible encoded (mcr-like)
1600 1720 1840 1960 2080 2200
m/z
Polymyxin resistant

Figure 2. Results of the MALDIxin test on E. coli. Representative spectra of (a) a polymyxin-susceptible E. coli isolate; (b) a chromosome-encoded
polymyxin-resistant E. coli isolate; (c) a polymyxin-resistant E. coli isolate producing MCR-1. Peaks of interest are indicated. The peak at m/z 1796.2
(blue) corresponds to the native E. coli lipid A, the peak at m/z 1821.2 (green) corresponds to the addition of pETN on the phosphate group at position
40 of the native lipid A with concomitant loss of the phosphate group at position 1, and the peak at m/z 1919.2 (red) corresponds to the addition of
one pETN on the phosphate group at position 1 of the native lipid A. (d) Distribution of the polymyxin resistance ratios (PPRs) for the 79 tested E. coli
strains (Table S1). Three independent experiments were performed for each strain. Cut-off values for discrimination between polymyxin resistance
and polymyxin susceptibility (0.1) and for discrimination between chromosome- and MCR-encoded resistance to polymyxin (0.5) are indicated by a
red and a black dotted line, respectively. P values (unpaired Welch’s t-test) are indicated. This figure appears in colour in the online version of JAC and
in black and white in the print version of JAC.

irrespective of the growth medium, with the PRRE. coli being between Validation of the MALDIxin test on a prospective
2.89 and 5.69 for MCR-1-producing strains and between 0.21 and collection of carbapenemase-producing E. coli from the
0.37 for strains harbouring chromosome-encoded resistance. French NRC
To identify any impact of colony age, the MALDIxin test was
performed on fresh colonies (overnight incubation) grown on both In order to further validate the MALDIxin test, all carbapenemase-
LB and MH agar and colonies stored for 48 h, 1 week or 2 weeks producing E. coli received from the French NRC during October and
(agar plates were kept at 4 C). Similar spectra were obtained for November 2016 were blindly analysed in triplicate (three colonies
all tested colonies leading to no significant differences in their selected from the same MH agar plate). This collection of
PRRE. coli (Table S2). 78 carbapenemase-producing E. coli strains included 2 KPC

3363
Dortet et al.

Day 0 Day 1 Day 2 Day 3 Day 4 1-3 weeks

Research workflow
Litres of 250 ml
Readout
bacterial Lipid Lipid A
culture extraction
MALDI-TOF
Routine workflow +
2h Polymyxin 16-24 h Plasmid
resistant
Readout
isolate – MICs
Polymyxin NP test Colistin-MAC Chromosome
(Dipicolinic acid

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

Patient 16-24 h 3-5 h inhibition)
samples

Colonies Standard Polymyxin DNA

susceptibility MICs extraction Readout
testing + PCR (mcr-1/–2) Gene(s)

Bacterial
identification
– +
MALDI-TOF Chromosome Plasmid
15min

MALDIxin test
Polymyxin susceptibility (R / S)
Readout
AND Lipid A
Plasmid- vs chromosome-
encoded resistance

Figure 3. Comparison of research (purple outline), routine clinical (green outline) and MALDIxin (blue outline) workflows for the detection of chromo-
some- and plasmid (MCR)-encoded resistance to polymyxin in Enterobacteriaceae. This figure appears in colour in the online version of JAC and in
black and white in the print version of JAC.

producers, 11 NDM producers, 2 VIM producers, 63 OXA-48-like become last-resort antimicrobials, largely used for the treatment
producers and 1 isolate producing two different carbapenemases of severe bacterial infections. In Enterobacteriaceae, resistance to
(Table 1). Among these isolates, two NDM-1 producers and one polymyxins results from alterations to lipid A, caused either by
OXA-48 producer were also found to be resistant to colistin, with adaptive chromosomal mutations or by acquired plasmid-
MICs of 4 mg/L.42 PCR analysis revealed that these three isolates encoded MCR enzymes. Irrespective of the resistance mechanism,
were positive for the mcr-1 gene (Table 1). When analysed using the detection of polymyxin resistance remains difficult and time
the MALDIxin test, all polymyxin-susceptible isolates (n " 75) consuming in clinical microbiology laboratories, and relies solely
showed a PRRE. coli of 0, corresponding to polymyxin susceptibility, on the determination of colistin or polymyxin B MICs using BMD.17
whilst for the three MCR-1-producing isolates a PRRE. coli .0.5 was To provide an accurate, reliable, rapid and cost-effective alterna-
obtained (1.73+0.23 and 2.34+0.12 for the two NDM-1-producers tive, we developed a diagnostic test based on MALDI-TOF MS for
and 1.29+0.63 for the OXA-48-producer), accurately classifying the detection of polymyxin resistance in Gram-negative bacteria.
them as resistant to polymyxin by a plasmid-encoded mechanism We call this method the MALDIxin test. After initial validation using
(Table 1). Therefore, in this prospective study, the MALDIxin test re- a well-characterized collection of polymyxin-susceptible and -re-
liably and rapidly identified all MCR-producing strains while con- sistant strains, we performed a prospective evaluation on
ventional methods would have needed 24–48 h with broth dilution carbapenemase-producing E. coli during which the MALDIxin test
method (16–24 h) and subsequent DNA extraction and PCR for accurately detected all MCR-1 producers. This result was obtained
mcr-1 and mcr-2 detection on resistant isolates (4 h). directly from isolated colonies within 15 min, whilst the confirm-
ation of colistin resistance and MCR production using standard
Discussion techniques would have required an additional 24–48 h (Figure 3).
An additional challenge when it comes to polymyxin resistance
Owing to the increasing incidence of carbapenemase-producing is the rapid dissemination of MCR-producing Enterobacteriaceae.
Enterobacteriaceae, polymyxins (colistin and polymyxin B) have Therefore, discrimination between chromosome- and plasmid-

3364
Rapid detection of colistin resistance by MALDI-TOF MS JAC
encoded resistance is important due to the higher potential for dis- microbiology laboratories (Figure 3). However, we need to note
semination of plasmid-encoded compared with chromosome- that routine use of the MALDIxin test will require switching the
encoded resistance mechanisms. In the future this distinction may MALDI-TOF MS machine to the negative ion mode, which is not
also be crucial in order to facilitate the use of MCR inhibitors, which currently used for bacterial identification; this is necessary due to
are being developed.20,21,43,44 Currently, this discrimination mostly the inherent negative charge of lipid A.
relies on molecular tests dedicated to the detection of mcr genes. Overall, the MALDIxin test is a rapid (15 min) and accurate diag-
These are based on real-time PCR,45–47 loop-mediated isothermal nostic tool (Figure 3), which represents a major advance in the de-
amplification48 and microarray techniques.49 These molecular tection of polymyxin resistance by directly assessing modification
assays are costly and, whilst they detect mcr-1 and/or mcr-2, they of lipid A, the cellular target of the polymyxins, in intact bacteria.
are not able to detect all members of the highly divergent and rap- The test has been validated for E. coli on different media
idly expanding mcr gene family (Figures S1 and S2). In fact, there (Figure S3), with fresh and aged colonies. Of note, polymyxin resist-
are growing numbers of reports describing new MCR families, all of ance is currently a much bigger clinical problem in Klebsiella pneu-

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

which were discovered using a combination of conventional sus- moniae than in E. coli and MCR production is still a public health
ceptibility testing methods coupled with WGS.31,32,34,35 Despite issue that has not yet impaired human health. However, prelimin-
their divergent ancestral origins and their low identity (Figure S1), ary data indicate that it may be also applicable to other Gram-
all MCR enzymes possess a common enzymatic activity, the add- negative bacteria for which pETN addition is involved in polymyxin
ition of a pETN moiety to the lipid A of Gram-negative bacteria. The resistance. These include the clinically important organisms
MALDIxin test directly detects this signature and can therefore be Salmonella spp. and K. pneumoniae (which together with E. coli
used to identify any MCR-encoding strain. Most specifically, our make up .99.9% of MCR producers),12 as well as Acinetobacter
method identifies a specific signature (peak at m/z 1821) associ- baumannii, another emerging pathogen for which resistance to
ated with MCR production, leading to the accurate discrimination polymyxins is becoming critical.8,51 The combination of excellent
of chromosome- and plasmid-encoded resistance to polymyxins performance, cost-effectiveness and high-throughput scalability
in E. coli (Figures 1 and 2). are all desirable attributes that will allow further commercializa-
Recently, a simple phenotypic method, named Colistin-MAC, tion of the MALDIxin test. Finally, the test uses a technology that is
has been described for screening for MCR-1-mediated colistin re- already available in many clinical microbiology laboratories, thus
sistance.50 This method is based on colistin MIC reduction (8-fold allowing no-cost and hassle-free implementation.
reduction) in the presence of dipicolinic acid, when the polymyxin
resistance is caused by MCR-1 production. Although this method is
relatively cheap and easy to perform in a clinical microbiology la-
Funding
boratory, it requires the determination of additional MICs on
This work was partially funded by the University Paris-Sud, France. L. D.,
polymyxin-resistant isolates using the BMD method, which again R. A. B., L. G., A. B. J. and T. N. are members of the Laboratory of
results in an additional delay of at least 24 h compared with the Excellence in Research on Medication and Innovative Therapeutics
original molecular biology approach (Figure 3). The MALDIxin test (LERMIT) supported by a grant from the French National Research
is more efficient because of its ability to simultaneously detect Agency (ANR-10-LABX-33). L. D. was supported by funding from ‘the
polymyxin resistance and its genetic basis, in ,15 min (Figure 3). People Programme (Marie Skłodowska-Curie Actions) of the European
Compared with plasmid-borne mcr-1, the occurrence of chro- Union’s Horizon 2020 under REA grant agreement number [654909]’ and
mosomally encoded mcr-1 is extremely rare. Of note, despite their from ‘the University Paris Sud’. D. A. I. M. was supported by an ‘MRC
abilities to detect MCR producers, neither the MALDIxin test nor the Career Development Award (MR/M009505/1)’. G. L.-M. was supported by
currently available techniques (such as molecular assays dedi- an ‘MRC Confidence in Concept Award (RSPO_P54990)’.
cated to mcr detection or the Colistin-MAC) are able to distinguish
between plasmid- and chromosome-encoded mcr. Accordingly, a
chromosome-encoded resistance is usually assumed if the test Transparency declarations
result is negative independently of the technique used. L. Dortet, A. F. and G. L.-M. are co-inventors of the MALDIxin test for
Another advantage of the MALDIxin test is its amenability to which a patent has been filed by Imperial Innovations. The remaining
high-throughput use, which may prove invaluable for screening authors have none to declare.
large collections of strains such as those collected by the NRCs.
As this technique does not require specific sample preparation, a
Author contributions
large number of bacterial isolates can be loaded onto the same
MALDI-TOF target (between 48 and 96 positions are available on L. Dortet and G. L.-M. had full access to all of the data in the study, and take re-
the standard reusable MALDI target plates) and tested in the sponsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: L. Dortet and G. L.-M. Acquisition, analysis, or inter-
same run. Thus, consumable costs and technical handling time
pretation of data: all authors. Drafting of the manuscript: L. Dortet, G. L.-M. and
are drastically reduced compared with other techniques described A. F. Critical revision of the manuscript for important intellectual content:
for the detection of polymyxin resistance. From a One-Health per- L. Dortet, R. C. D. F., D. A. I. M., P. B., Y. G., P. P., R. A. B., T. N., A. F. and G. L.-M.
spective, the high-throughput potential of our test will be useful
for screening for MCR producers in the livestock industry, where
large numbers of mcr-positive isolates are being identified. Finally,
the MALDIxin test could easily be incorporated into routine work- Supplementary data
flows, because a MALDI-TOF MS-based approach for bacterial Figures S1 to S3 and Tables S1 and S2 appear as Supplementary data at
identification is now carried out routinely in many clinical JAC Online.

3365
Dortet et al.

19 Nordmann P, Jayol A, Poirel L. Rapid detection of polymyxin resistance in

References Enterobacteriaceae. Emerg Infect Dis 2016; 22: 1038–43.
1 Falagas ME, Kasiakou SK. Colistin: the revival of polymyxins for the man- 20 Daly SM, Sturge CR, Felder-Scott CF et al. MCR-1 inhibition with peptide-
agement of multidrug-resistant Gram-negative bacterial infections. Clin conjugated phosphorodiamidate morpholino oligomers restores sensitivity
Infect Dis 2005; 40: 1333–41. to polymyxin in Escherichia coli. MBio 2017; 8: e01315–17.
2 Nordmann P, Dortet L, Poirel L. Carbapenem resistance in 21 Wei P, Song G, Shi M et al. Substrate analog interaction with MCR-1 offers
Enterobacteriaceae: here is the storm! Trends Mol Med 2012; 18: 263–72. insight into the rising threat of the plasmid-mediated transferable colistin re-
3 Kontopidou F, Giamarellou H, Katerelos P et al. Infections caused by sistance. FASEB J 2017; 32: 1085–98.
carbapenem-resistant Klebsiella pneumoniae among patients in intensive 22 Delgado-Blas JF, Ovejero CM, Abadia PL et al. Coexistence of mcr-1 and
care units in Greece: a multi-centre study on clinical outcome and therapeutic blaNDM-1 in Escherichia coli from Venezuela. Antimicrob Agents Chemother
options. Clin Microbiol Infect 2014; 20: O117–23. 2016; 60: 6356–8.
4 Kontopidou F, Plachouras D, Papadomichelakis E et al. Colonization and in- 23 Di Pilato V, Arena F, Tascini C et al. mcr-1.2, a new mcr variant carried on a

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

fection by colistin-resistant Gram-negative bacteria in a cohort of critically ill transferable plasmid from a colistin-resistant KPC carbapenemase-
patients. Clin Microbiol Infect 2011; 17: E9–E11. producing Klebsiella pneumoniae strain of sequence type 512. Antimicrob
5 Neonakis IK, Samonis G, Messaritakis H et al. Resistance status and evolu- Agents Chemother 2016; 60: 5612–5.
tion trends of Klebsiella pneumoniae isolates in a university hospital in Greece: 24 Du H, Chen L, Tang YW et al. Emergence of the mcr-1 colistin resist-
ineffectiveness of carbapenems and increasing resistance to colistin. ance gene in carbapenem-resistant Enterobacteriaceae. Lancet Infect
Chemotherapy 2010; 56: 448–52. Dis 2016; 16: 287–8.
6 Rojas LJ, Salim M, Cober E et al. Colistin resistance in carbapenem-resistant 25 Falgenhauer L, Waezsada SE, Yao Y et al. Colistin resistance gene mcr-1
Klebsiella pneumoniae: laboratory detection and impact on mortality. Clin in extended-spectrum b-lactamase-producing and carbapenemase-
Infect Dis 2017; 64: 711–8. producing Gram-negative bacteria in Germany. Lancet Infect Dis 2016; 16:
7 Venditti C, Nisii C, D’Arezzo S et al. Letter to the Editor: surveillance of mcr-1 282–3.
and mcr-2 genes in carbapenem-resistant Klebsiella pneumoniae strains 26 Li A, Yang Y, Miao M et al. Complete sequences of mcr-1-harboring plas-
from an Italian hospital. Euro Surveill 2017; 22: pii"30604. mids from extended-spectrum-b-lactamase- and carbapenemase-
8 Jeannot K, Bolard A, Plesiat P. Resistance to polymyxins in Gram-negative producing Enterobacteriaceae. Antimicrob Agents Chemother 2016; 60:
organisms. Int J Antimicrob Agents 2017; 49: 526–35. 4351–4.
9 Olaitan AO, Morand S, Rolain JM. Mechanisms of polymyxin 27 Mulvey MR, Mataseje LF, Robertson J et al. Dissemination of the mcr-1 co-
resistance: acquired and intrinsic resistance in bacteria. Front Microbiol listin resistance gene. Lancet Infect Dis 2016; 16: 289–90.
2014; 5: 643. 28 Poirel L, Kieffer N, Liassine N et al. Plasmid-mediated carbapenem and
10 Choi MJ, Ko KS. Loss of hypermucoviscosity and increased fitness cost in colistin resistance in a clinical isolate of Escherichia coli. Lancet Infect Dis
colistin-resistant Klebsiella pneumoniae sequence type 23 strains. Antimicrob 2016; 16: 281.
Agents Chemother 2015; 59: 6763–73. 29 Yao X, Doi Y, Zeng L et al. Carbapenem-resistant and colistin-resistant
11 Liu YY, Wang Y, Walsh TR et al. Emergence of plasmid-mediated colistin Escherichia coli co-producing NDM-9 and MCR-1. Lancet Infect Dis 2016; 16:
resistance mechanism MCR-1 in animals and human beings in China: a 288–9.
microbiological and molecular biological study. Lancet Infect Dis 2016; 16: 30 Yu H, Qu F, Shan B et al. Detection of the mcr-1 colistin resistance gene in
161–8. carbapenem-resistant Enterobacteriaceae from different hospitals in China.
12 Poirel L, Jayol A, Nordmann P. Polymyxins: antibacterial activity, suscepti- Antimicrob Agents Chemother 2016; 60: 5033–5.
bility testing, and resistance mechanisms encoded by plasmids or chromo- 31 Borowiak M, Fischer J, Hammerl JA et al. Identification of a novel
somes. Clin Microbiol Rev 2017; 30: 557–96. transposon-associated phosphoethanolamine transferase gene, mcr-5, con-
13 Kong LH, Lei CW, Ma SZ et al. Various sequence types of Escherichia coli ferring colistin resistance in D-tartrate fermenting Salmonella enterica subsp.
isolates coharboring blaNDM-5 and mcr-1 genes from a commercial swine enterica serovar Paratyphi B. J Antimicrob Chemother 2017; 72: 3317–25.
farm in China. Antimicrob Agents Chemother 2017; 61: e02167-16. 32 Carattoli A, Villa L, Feudi C et al. Novel plasmid-mediated colistin resist-
14 Yang Q, Li M, Spiller OB et al. Balancing mcr-1 expression and bacterial ance mcr-4 gene in Salmonella and Escherichia coli, Italy 2013, Spain and
survival is a delicate equilibrium between essential cellular defence mecha- Belgium, 2015 to 2016. Euro Surveill 2017; 22: pii"30589.
nisms. Nat Commun 2017; 8: 2054. 33 Liu L, Feng Y, Zhang X et al. A new variant of mcr-3 in an extensively drug-
15 Jayol A, Nordmann P, Lehours P et al. Comparison of methods for detec- resistant Escherichia coli clinical isolate carrying mcr-1 and blaNDM-5.
tion of plasmid-mediated and chromosomally encoded colistin resistance in Antimicrob Agents Chemother 2017; 61: e01757–17.
Enterobacteriaceae. Clin Microbiol Infect 24: 175–9. 2017; 34 Xavier BB, Lammens C, Ruhal R et al. Identification of a novel plasmid-
16 Hindler JA, Humphries RM. Colistin MIC variability by method for contem- mediated colistin-resistance gene, mcr-2, in Escherichia coli, Belgium, June
porary clinical isolates of multidrug-resistant Gram-negative bacilli. J Clin 2016. Euro Surveill 2016; 21. doi:10.2807/1560-7917.ES.2016.21.27.30280.
Microbiol 2013; 51: 1678–84. 35 Yin W, Li H, Shen Y et al. Novel plasmid-mediated colistin resistance gene
17 Clinical and Laboratory Standard Institute (CLSI) and European Committee mcr-3 in Escherichia coli. MBio 2017; 8: e00543–17.
on Antimicrobial Susceptibility Testing (EUCAST) Joint Subcommittee. 36 Chang AC, Cohen SN. Construction and characterization of amplifiable
Recommendations for MIC Determination of Colistin (Polymyxin E) as multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid.
Recommended by the Joint CLSI-EUCAST Polymyxin Breakpoints Working Group. J Bacteriol 1978; 134: 1141–56.
http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/General_docume 37 Casadaban MJ, Cohen SN. Analysis of gene control signals by DNA fusion
nts/Recommendations_for_MIC_determination_of_colistin_March_2016.pdf. and cloning in Escherichia coli. J Mol Biol 1980; 138: 179–207.
18 Chew KL, La MV, Lin RTP et al. Colistin and polymyxin B susceptibility test- 38 European Committee on Antimicrobial Susceptibility Testing (EUCAST).
ing for carbapenem-resistant and mcr-positive Enterobacteriaceae: compari- Breakpoint Tables for Interpretation of MICS and Zone Diameters Version 7.1.
son of Sensititre, MicroScan, Vitek 2, and Etest with broth microdilution. J Clin http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Breakpoint_
Microbiol 2017; 55: 2609–16. tables/v_7.1_Breakpoint_Tables.pdf.

3366
Rapid detection of colistin resistance by MALDI-TOF MS JAC
39 Larrouy-Maumus G, Clements A, Filloux A et al. Direct detection of lipid A 46 Chabou S, Leangapichart T, Okdah L et al. Real-time quantitative PCR
R
on intact Gram-negative bacteria by MALDI-TOF mass spectrometry. assay with TaqmanV probe for rapid detection of MCR-1 plasmid-mediated
J Microbiol Methods 2016; 120: 68–71. colistin resistance. New Microbes New Infect 2016; 13: 71–4.
R
40 Pieretti G, Cipolletti M, D’Alonzo D et al. A combined fermentative- 47 Dona V, Bernasconi OJ, Kasraian S et al. A SYBRV Green-based real-time
chemical approach for the scalable production of pure E. coli monophos- PCR method for improved detection of mcr-1-mediated colistin resistance in
phoryl lipid A. Appl Microbiol Biotechnol 2014; 98: 7781–91. human stool samples. J Glob Antimicrob Resist 2017; 9: 57–60.
41 Nordmann P, Jayol A, Poirel L. A universal culture medium for screening 48 Imirzalioglu C, Falgenhauer L, Schmiedel J et al. Evaluation of a loop-
polymyxin-resistant Gram-negative isolates. J Clin Microbiol 2016; 54: 1395–9. mediated isothermal amplification-based assay for the rapid detection of
42 Beyrouthy R, Robin F, Lessene A et al. MCR-1 and OXA-48 in vivo acquisi- plasmid-encoded colistin resistance gene mcr-1 in Enterobacteriaceae iso-
tion in KPC-producing Escherichia coli after colistin treatment. Antimicrob lates. Antimicrob Agents Chemother 2017; 61: e0232616.
Agents Chemother 2017; 61: e02540-16. 49 Bernasconi OJ, Principe L, Tinguely R et al. Evaluation of a new
43 Hinchliffe P, Yang QE, Portal E et al. Insights into the mechanistic basis of commercial microarray platform for the simultaneous detection of
b-lactamase and mcr-1/-2 genes in Enterobacteriaceae. J Clin Microbiol

Downloaded from https://academic.oup.com/jac/article/73/12/3359/5089139 by guest on 01 June 2021

plasmid-mediated colistin resistance from crystal structures of the catalytic
domain of MCR-1. Sci Rep 2017; 7: 39392. 2017; 55: 3138–41.
44 Zhou Y, Liu S, Wang T et al. Pterostilbene, a potential MCR-1 inhibitor 50 Coppi M, Cannatelli A, Antonelli A et al. A simple phenotypic method
that enhances the efficacy of polymyxin B. Antimicrob Agents Chemother for screening of MCR-1-mediated colistin resistance. Clin Microbiol Infect
2018; 62: e02146–17. 2018; 24: 201.e1–e3.
45 Bontron S, Poirel L, Nordmann P. Real-time PCR for detection of plasmid- 51 Qureshi ZA, Hittle LE, O’Hara JA et al. Colistin-resistant Acinetobacter
mediated polymyxin resistance (mcr-1) from cultured bacteria and stools. baumannii: beyond carbapenem resistance. Clin Infect Dis 2015; 60:
J Antimicrob Chemother 2016; 71: 2318–20. 1295–303.

3367