Identification of living legionella pneumophila using species-specific metabolic lipopolysaccharide labeling

Article abstract of DOI:10.1002/anie.201309072

Legionella pneumophila is a pathogenic bacterium involved in regular outbreaks characterized by a relatively high fatality rate and an important societal impact. Frequent monitoring of the presence of this bacterium in environmental water samples is neces

Full text of DOI:10.1002/anie.201309072

Angewandte
Chemie
DOI: 10.1002/anie.201309072
Carbohydrates
Identification of Living Legionella pneumophila Using Species-
Specific Metabolic Lipopolysaccharide Labeling**
Jordi Mas Pons, Audrey Dumont, Grꢀgory Sautejeau, Emilie Fugier, Aurꢀlie Baron,
Sam Dukan,* and Boris Vauzeilles*
Dedicated to Professor Jean-Marie Beau on the occasion of his 65th birthday
Abstract: Legionella pneumophila is a pathogenic bacterium
involved in regular outbreaks characterized by a relatively high
fatality rate and an important societal impact. Frequent
monitoring of the presence of this bacterium in environmental
water samples is necessary to prevent these epidemic events, but
the traditional culture-based detection and identification
method requires up to 10 days. Reported herein is a method
allowing identification of Legionella pneumophila by meta-
bolic lipopolysaccharide labeling which targets, for the first
time, a precursor to monosaccharides that are specifically
present within the O-antigen of the bacterium. This new
approach allows easy detection of living Legionella pneumo-
phila, while other Legionella species are not labeled.
important impact of each individual epidemic event. Mon-
itoring of the presence of this bacterium in potential
reservoirs, including cooling towers and other water-contain-
ing systems, is the major strategy used to limit the occurrence
of outbreaks, and methods allowing fast and easy detection of
living L. pneumophila could have a major impact since
standard, culture-based methods can require up to 10 days.[4]
We have previously shown that metabolic glycan label-
ing,[5] a strategy in which a modified monosaccharide bearing
a reporter function is metabolically incorporated into surface
glycans, could be efficiently used to target bacterial lip-
opolysaccharides (LPS) without species specificity. In this first
study, an azido derivative of Kdo, a bacterial monosaccharide,
was incorporated into the LPS inner core of various Gram-
negative bacteria, and detected by copper-catalyzed click-
chemistry with an alkyne-modified fluorophore.[6] This strat-
egy can be efficiently used to detect the overall presence of
living Gram-negative bacteria. A much more impactful
challenge would be to use metabolic lipopolysaccharide
labeling not only to detect, but also to identify a living
pathogenic bacterium of interest in the same operation by
using an analogue of a monosaccharide which would be
specifically present within the O-antigen of this bacterium. To
the best of our knowledge, metabolic glycan labeling has not
yet been used for direct species identification. The work
presented here successfully addresses this challenge in the
case of L. pneumophila.
I
n the summer of 1976 an outbreak of an unknown disease
occurred at the annual convention of the American Legion in
Philadelphia.[1] A total number of 221 people were infected
and 34 died from the consequences of this infection. The
causative agent was later identified as a bacterium,[2] which
was given the name Legionella pneumophila.[3] Since then,
outbreaks of legionellosis have occurred on many occasions,
and numerous strains of pathogenic L. pneumophila have
been identified. A characteristic aspect of infection by
L. pneumophila is its high fatality rate, thus resulting in an
[*] Dr. J. Mas Pons,[+] G. Sautejeau, Dr. A. Baron, Dr. B. Vauzeilles
Centre de Recherche de Gif
Institut de Chimie des Substances Naturelles du CNRS
Avenue de la Terrasse, 91198 Gif-sur-Yvette (France)
E-mail: boris.vauzeilles@cnrs.fr
Dr. A. Dumont,[+] Dr. E. Fugier, Dr. S. Dukan
Aix Marseille Universitꢀ, Laboratoire de Chimie Bactꢀrienne (UMR
7283), Institut de Microbiologie de la Mꢀditerranꢀe (IMM)
CNRS, 31 Chemin Joseph Aiguier-13402 Marseille (France)
E-mail: sdukan@imm.cnrs.fr
The O-antigen of L. pneumophila serogroup 1, which is
prevalent among infected cases,[7,8] is composed of an a(2!4)
homopolysaccharidic repeat of 5-N-acetimidoyl-7-N-acetyl-
legionaminic acid (Leg5Am7Ac)[9] (Figure 1a). The biosyn-
thesis of Leg (Figure 1b) starts from UDP-N,N’-diacetylba-
cillosamine, which is transformed into 2,4-diacetamido-2,4,6-
trideoxy-d-mannopyranose (1) by the dual action of a hydro-
lyzing 2-epimerase. In the next step, the precursor 1 is directly
transformed into N,N’-diacetyllegionaminic acid (Leg5A-
c7Ac) by the action of an aldolase, in the presence of
phosphoenolpyruvate (PEP).[10] This event controls the
stereochemistry of the newly generated stereogenic center
at C4. Legionaminic acid is then activated in the form of
a cytidine monophosphate donor (CMP)-Leg5Ac7Ac. Fur-
ther transformations are believed to occur at a later stage.
To target the Leg pathway for metabolic glycan labeling,
we embarked upon the synthesis of an azido derivative of 1,[11]
namely 6-azido-2,4-diacetamido-2,4,6-trideoxy-d-mannopyr-
anose (2, Figure 1c), as well as its less polar, monoacetylated
derivative 3, which we believed might enter more easily into
Dr. B. Vauzeilles
CNRS and Universitꢀ Paris-Sud, Laboratoire de Synthꢁse de
Biomolꢀcules, Institut de Chimie Molꢀculaire et des Matꢀriaux
d’Orsay
UMR 8182, 91405 Orsay (France)
[+] These authors contributed equally to this work.
[**] This work was partially supported by the CNRS, the Sociꢀtꢀ
d’Accꢀlꢀration du Transfert de Technologies SATT Sud Est, the
Rꢀgion Provence-Alpes-Cꢂte d’Azur, and the Fondation pour la
Recherche Mꢀdicale (FRM—convention DCM20121225768). We
thank Sophie Jarraud (CNRL) for generous gift of all Legionella
strains.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201309072.
Angew. Chem. Int. Ed. 2014, 53, 1275 –1278
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1275

.
Angewandte
Communications
Scheme 1. Reagents and conditions: a) para-methoxyphenol, BF3.Et2O,
CH2Cl2;[12] b) CH3ONa, CH3OH;[13] c) 2-aminoethyl diphenylborinate,
DIPEA, benzoyl chloride, CH3CN;[14] d) Tf2O, pyridine, CH2Cl2;[15]
e) Bu4NN3, toluene;[15] f) Pd(OH)2/C, H2, CH3OH;[15] g) Ac2O,
CH3OH;[15] h) TsCl, pyridine;[6] i) MsCl, pyridine; j) NaN3, DMF;[6]
k) CAN, CH3CN/H2O (3:1);[16] l) Ac2O, pyridine, CH2Cl2. CAN=ceric
ammonium nitrate, DIPEA=diisopropylethylamine, DMF=N,N-di-
methylformamide, Ms=methanesulfonyl, PMP=para-methoxyphenyl,
Tf =trifluoromethanesulfonyl, Ts=4-toluenesulfonyl.
steps),[15] the manno configuration of which was confirmed by
Figure 1. Structures of Leg and 4eLeg, the Leg pathway and molecules
used in this study. a) Structure of Legionaminic (Leg) and 4epiLegio-
naminic (4eLeg) acids. b) Leg pathway in L. pneumophila, and the
structure of intermediate 1. c) Structure of analogues 2 and 3, as well
as ligand 4 and biotine-alkyne 5.
1H NMR spectroscopy (J1,2 = 1.2 Hz;
J2,3 = 3.6 Hz; J3,4 =
10.0 Hz; J4,5 = 10.2 Hz). Conventional debenzoylation of 8
and reduction of the azido groups with H2 and Pd(OH)2/C was
followed by N-acetylation to give 9 in a high yield (82% over
3 steps).[15] The azido derivative 11 was obtained by selective
tosylation or mesylation in pyridine, with subsequent nucle-
ophilic substitution using sodium azide in dimethylform-
amide.[6]
In this strategy, mesylation and substitution gave better
results (50%) than the tosylation route (32%). The final
product 2 was obtained in a good yield (82%) from 11 by
deprotection of the anomeric position using cerium ammoni-
um nitrate.[16] Alternatively, the product 3 was prepared in
two steps from 11 in a respectable 71% yield by acetylation
and subsequent deprotection of the anomeric position using
the same reaction conditions as before.[16]
We then selected four different strains of L. pneumophila,
belonging to serogroup 1, including 1) a strain isolated from
one of the victims of the historical Philadelphia outbreak,[17]
2) the Paris strain, which was responsible for nosocomial
epidemics at the newly constructed, modern, and freshly
opened Georges Pompidou Hospital in Paris in 2000, and has
become endemic throughout Europe,[18] 3) the Lens strain,
which infected 86 people and killed 17 in the north of France
during the 2003–2004 winter and was therefore accountable
for the most important epidemics in this country. These
strains were grown in the presence of 2, and the incorporation
of the azido chemical reporter into LPS was monitored in
a subsequent step by using copper-catalyzed azide–alkyne
cycloaddition,[3] with copper sulfate, sodium ascorbate, TGTA
[4; a water-soluble tris(triazolyl) ligand for copper (I)],[19] and
the cell by passive transport, and be further transformed into
2 by the action of intracellular, nonspecific esterases. Once in
the cell, 2 might act as a precursor of an azido-labeled
analogue of legionaminic acid, and be further incorporated
within the O-antigen of the bacteria. A synthetic strategy
starting from d-galactose has therefore been developed to
access 2 and 3, and these products have been tested for
specific labeling of the LPS of living L. pneumophila
serogroup 1.
The compounds 2 and 3 were synthesized (Scheme 1)
using a method inspired by the synthesis of 1 by Tsvetkov and
co-workers.[15] The target compound, 6-azido-2,4-diaceta-
mido-2,4,6-trideoxy-d-mannose (2), was prepared in 11
steps from the commercially available b-d-galactose penta-
acetate with an overall yield of 17%, while 3-O-acetyl-6-
azido-2,4-diacetamido-2,4,6-trideoxy-d-mannose (3) was
obtained from the same starting material in 12 steps and
15% overall yield.
Glycosylation of p-methoxyphenol with b-d-galactose
pentaacetate in the presence of boron trifluoride etherate
gave 6 in a good 83% yield.[12] Zemplꢀn deacetylation using
sodium methoxide[13] and selective benzoylation using the
method developed by Lee and Taylor led to 7 (70% yield over
2 steps).[14] Conversion of 7 into the bis(triflate) derivative,
and its subsequent reaction with tetrabutylammonium azide
in toluene resulted into the bis(azido) compound 8 (89%; 2
1276
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1275 –1278

Angewandte
Chemie
acetyl-4-epi-legionaminic acid (4eLeg5Am7Ac),[20] with var-
ious degrees of 8-O-acetylation depending on the serogroup
(Figure 1a). Contrary to the Leg pathway, the 4eLeg biosyn-
thetic pathway has not been identified yet, but one could
speculate that it might involve similar intermediates. Strains
belonging to serogroups 3, 4, 5, 6, which represent together
with serogroup 1 between 68 and 85% of the L. pneumophila
present in the environment,[8] showed very bright membrane
labeling (Figure 3). This result tends to indicate that Leg and
4eLeg biosyntheses apparently share 1 as a common pre-
cursor, and our method allows clear and specific detection of
L. pneumophila strains that do not belong to serogroup 1.
Figure 2. Detection of metabolically incorporated 2 by various L. pneu-
mophila serogroup 1 strains as shown by CuI-catalyzed click reaction
with 5, and further visualization using an Alexa Fluor 488-IgG anti-
biotin antibody. Phase contrast and fluorescence images in the
presence (right panel) or absence of 2 (left panel). Scale bar=1 mm.
a biotine-alkyne probe (5) for 30 minutes (Figure 1c). Biotin
labeling was then visualized with a fluorescently labeled anti-
biotin antibody. All strains showed highly distinctive fluores-
cence on their membrane, indicative of an effective metabolic
incorporation of the chemical reporter (Figure 2). Those
results are in contrast with other bacteria such as Escherichia
coli and Pseudomonas aeruginosa, which both failed to show
any labeling under the same conditions (see Figure S1 in the
Supporting Information).
This highly encouraging result prompted us to test the
specificity of our labeling strategy towards other strains of
Legionella, which did not belong to the pneumophila species,
and have therefore not been described to contain legiona-
minic acid within their LPS. A representative set of such
strains, containing L. gormanii, L. maceachernii, L. micdadei,
L. anisa, L. feeli, L. jordanis, L. tucsonensis, and L. bozemanii
were therefore subjected to the same labeling conditions, and
no membrane fluorescence was observed (see Figure S2),
a result consistent with the absence of Leg within their
lipopolysaccharides. The method is therefore able to effi-
ciently discriminate L. pneumophila serogroup 1 from other
Legionella not belonging to the pneumophila species.
We then evaluated the capacity of our method to label
L. pneumophila strains belonging to other serogroups. This is
an interesting point, since although serogroup 1 is found in
most infected cases, other serogroups are abundant in the
environment. The possibility to label these other serogroups
would therefore be an important result allowing a better
evaluation of the presence of L. pneumophila in a given
sample. Most of these serogroups have been shown to contain
another isomer of a 5-acetamidino-7-acetamido-3,5,7,9-tetra-
deoxynon-2-ulosonic acid, namely 5-N-acetimidoyl-7-N-
Figure 3. Detection of metabolically incorporated 2 by various L. pneu-
mophila strains belonging to serogroups other than serogroup 1 (same
conditions as used for Figure 2).
Interestingly, the only exception observed concerned
a
L. pneumophila strain belonging to serogroup
7
(Figure 3), a serogroup which is very poorly represented
both in infection cases and in the environment.[7,8] Serogroup
7 has been described to present a still unidentified isomer of
5-acetamidino-7-acetamido-3,5,7,9-tetradeoxynon-2-ulosonic
acid within its O-polysaccharide.[20] This observation is con-
sistent with the absence of labeling, and tends to indicate that
1 is not an intermediate in the corresponding biosynthetic
pathway. An isomer of 1 is most certainly involved as
a substrate for the aldolase.
Another set of experiments was performed using 3,
a monoacetylated derivative of 2. It was expected that this
less-polar esterified compound might enter the bacterial cell
more efficiently by passive transport, and could be further
Angew. Chem. Int. Ed. 2014, 53, 1275 –1278
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1277

.
Angewandte
Communications
[4] For discussions on recent alternative methods, see for example:
a) B. Cao, F. Yao, X. Liu, L. Feng, L. Wang, Appl. Environ.
Microbiol. 2013, 79, 6647 – 6654; b) B. Bedrina, S. Macian, I.
Solis, R. Fernandez-Lafuente, E. Baldrich, G. Rodriguez, BMC
Microbiol. 2013, 13, 88; c) S. Jarraud, G. Descours, C. Ginevra,
G. Lina, J. Etienne, Methods Mol. Biol. 2013, 954, 27 – 56.
[5] For some recent reviews, see: a) M. Grammel, H. C. Hang, Nat.
Chem. Biol. 2013, 9, 475 – 484; b) S. Gautam, T. J. Gniadek, T.
Kim, D. A. Spiegel, Trends Biotechnol. 2013, 31, 258 – 267;
c) P. V. Chang, C. R. Bertozzi, Chem. Commun. 2012, 48, 8864 –
8879.
[6] A. Dumont, A. Malleron, M. Awwad, S. Dukan, B. Vauzeilles,
Angew. Chem. 2012, 124, 3197 – 3200; Angew. Chem. Int. Ed.
2012, 51, 3143 – 3146.
[7] V. L. Yu et al., J. Infect. Dis. 2002, 186, 127 – 128.
[8] a) A. Doleans, H. Aurell, M. Reyrolle, G. Lina, J. Freney, F.
Vandenesch, J. Etienne, S. Jarraud, J. Clin. Microbiol. 2004, 42,
458 – 460; b) T. G. Harrison, B. Afshar, N. Doshi, N. K. Fry, J. V.
Lee, Eur. J. Clin. Microbiol. Infect. Dis. 2009, 28, 781 – 791.
[9] Y. A. Knirel, E. T. Rietschel, R. Marre, U. Zahringer, Eur. J.
Biochem. 1994, 221, 239 – 245.
[10] P. A. Glaze, D. C. Watson, N. M. Young, M. E. Tanner, Bio-
chemistry 2008, 47, 3272 – 3282.
[11] For the pioneering use of a metabolic precursor to modify
mammalian glycans, see: H. Kayser, R. Zeitler, C. Kannicht, D.
Grunow, R. Nuck, W. Reutter, J. Biol. Chem. 1992, 267, 16934 –
16938.
[12] N. W. McGill, S. J. Williams, J. Org. Chem. 2009, 74, 9388 – 9398.
[13] G. Zemplꢀn, A. Kunz, Chem. Ber. 1923, 56, 1705 – 1710.
[14] D. Lee, M. S. Taylor, J. Am. Chem. Soc. 2011, 133, 3724 – 3727.
[15] Y. E. Tsvetkov, A. S. Shashkov, Y. A. Knirel, U. Zahringer,
Carbohydr. Res. 2001, 335, 221 – 243.
incorporated into the LPS after deacetylation to 2 by
nonspecific esterase activity within the cell. The availability
and extent of such an activity inside bacterial cells has been
the subject of recent debate within the scientific commun-
ity.[21] No labeling was observed under our conditions, even in
the case of the L. pneumophila strains that were efficiently
labeled by 2 (see Figure S3), suggesting that such an esterase
activity is not present at a sufficient level within the cell to
allow efficient production of 2 from 3 and further metabo-
lization and incorporation of 2 to a detectable proportion.
This work therefore leads to several interesting results:
1) The approach we have described appears as an efficient
strategy to specifically detect and identify living L. pneumo-
phila, a pathogenic bacterium of high sanitary and economic
impact, and requires about one day on laboratory samples. To
the best of our knowledge, this is the first time that metabolic
labeling has been applied to species identification, through
the use of a specific monosaccharide. 2) Interestingly, the
absence of labeling under our laboratory conditions with an
acetylated precursor strengthens the claim that nonspecific
esterase activity within such bacteria might not be sufficient
for the efficient liberation inside the cell of a previously
acetylated carbohydrate precursor,[21] an aspect that should be
evaluated when considering the use of this strategy for
metabolic labeling experiments, prodrug approaches, or other
biosynthetic applications. 3) The use of an analogue of a Leg
precursor allows labeling of relevant serogroups using the
same conditions. The overall success of this strategy offers
some mechanistic insight into the biosynthesis of Leg and
4eLeg as 1 appears to be a common intermediate in these two
metabolic pathways.
[16] M. C. Hewitt, P. H. Seeberger, J. Org. Chem. 2001, 66, 4233 –
4243.
[17] C. W. Moss, R. E. Weaver, S. B. Dees, W. B. Cherry, J. Clin.
Microbiol. 1977, 6, 140 – 143.
[18] H. Aurell, J. Etienne, F. Forey, M. Reyrolle, P. Girardo, P. Farge,
B. Decludt, C. Campese, F. Vandenesch, S. Jarraud, J. Clin.
Microbiol. 2003, 41, 3320 – 3322.
Received: October 17, 2013
Keywords: biosynthesis · carbohydrates · click chemistry ·
imaging agents · synthesis design
[19] A. Baron, Y. Bleriot, M. Sollogoub, B. Vauzeilles, Org. Biomol.
Chem. 2008, 6, 1898 – 1901.
[20] Y. A. Knirel, S. N. Senchenkova, N. A. Kocharova, A. S. Shash-
kov, J. H. Helbig, U. Zꢁhringer, Biochemistry 2001, 66, 1035 –
1041.
.
[1] D. W. Fraser et al., N. Engl. J. Med. 1977, 297, 1189 – 1197.
[2] J. E. McDade, C. C. Shepard, D. W. Fraser, T. R. Tsai, M. A.
Redus, W. R. Dowdle, N. Engl. J. Med. 1977, 297, 1197 – 1203.
[3] D. J. Brenner, A. G. Steigerwalt, J. E. McDade, Ann. Intern.
Med. 1979, 90, 656 – 658.
[21] A. K. Antonczak, Z. Simova, E. M. Tippmann, J. Biol. Chem.
2009, 284, 28795 – 28800.
1278
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 1275 –1278