Mycobacterium leprae Phenolglycolipid-1 Expressed by Engineered M. bovis BCG Modulates Early Interaction with Human Phagocytes
The species-specific phenolic glycolipid 1 (PGL-1) is suspected to play a critical role in the pathogenesis of leprosy, a chronic disease of the skin and peripheral nerves caused by Mycobacterium leprae. Based on studies using the purified compound, PGL-1 was proposed to mediate the tropism of M. leprae for the nervous system and to modulate host immune responses. However, deciphering the biological function of this glycolipid has been hampered by the inability to grow M. leprae in vitro and to genetically engineer this bacterium. Here, we identified the M. leprae genes required for the biosynthesis of the species-specific saccharidic domain of PGL-1 and reprogrammed seven enzymatic steps in M. bovis BCG to make it synthesize and display PGL-1 in the context of an M. leprae-like cell envelope. This recombinant strain provides us with a unique tool to address the key questions of the contribution of PGL-1 in the infection process and to study the underlying molecular mechanisms. We found that PGL-1 production endowed recombinant BCG with an increased capacity to exploit complement receptor 3 (CR3) for efficient invasion of human macrophages and evasion of inflammatory responses. PGL-1 production also promoted bacterial uptake by human dendritic cells and dampened their infection-induced maturation. Our results therefore suggest that M. leprae produces PGL-1 for immune-silent invasion of host phagocytic cells.
Phenolglycolipid-1 Expressed by
BCG Modulates Early Interaction
with Human Phagoc ytes
, Catherine Astarie-Dequeker
, Caroline Demangel
, Wladimir Malaga
, Nadine Honore
, Nana Fatimath Bello
, Esther Perez
, Christoph e Guilhot
1 CNRS, IPBS (Institut de Pharmacologie et de Biologie Structurale), Toulouse, France, 2 Universite
de Toulouse, UPS, IPBS, Toulouse, France, 3 Institut Pasteur, Unite
e, Paris, France
The species-specific phenolic glycolipid 1 (PGL-1) is suspected to play a critical role in the pathogenesis of leprosy, a chronic
disease of the skin and peripheral nerves caused by Mycobacterium leprae. Based on studies using the purified compound,
PGL-1 was proposed to mediate the tropism of M. leprae for the nervous system and to modulate host immune responses.
However, deciphering the biological function of this glycolipid has been hampered by the inability to grow M. leprae in vitro
and to genetically engineer this bacterium. Here, we identified the M. leprae genes required for the biosynthesis of the
species-specific saccharidic domain of PGL-1 and reprogrammed seven enzymatic steps in M. bovis BCG to make it
synthesize and display PGL-1 in the context of an M. leprae-like cell envelope. This recombinant strain provides us with a
unique tool to address the key questions of the contribution of PGL-1 in the infection process and to study the underlying
molecular mechanisms. We found that PGL-1 production endowed recombinant BCG with an increased capacity to exploit
complement receptor 3 (CR3) for efficient invasion of human macrophages and evasion of inflammatory responses. PGL-1
production also promoted bacterial uptake by human dendritic cells and dampened their infection-induced maturation. Our
results therefore suggest that M. leprae produces PGL-1 for immune-silent invasion of host phagocytic cells.
Citation: Tabouret G, Astarie-Dequeker C, Demangel C, Malaga W, Constant P, et al. (2010) Mycobacte rium leprae Phenolglycolipid-1 Expressed by Engineered M.
bovis BCG Modulates Early Interaction with Human Phagocytes. PLoS Pathog 6(10): e1001159. doi:10.1371/journal.ppat.1001159
Editor: Sabine Ehrt, Weill Cornell Medical College, United States of America
Received April 29, 2010; Accepted September 23, 2010; Published October 21, 2010
Copyright: ß 2010 Tabouret et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded by the Centre National de la Recherche Scientifique (CNRS), and the Agence National e de la Recherche grant 06-MIME-032-02.
The NMR spectrometers were financed by the CNRS, the University Paul Sabatier, the Re
es and the European Structural Funds (FEDER). The
funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: Christophe.Guilhot@ipbs.fr
¤a Current address: INRA UMR 1225, IHAP, Ecole Nationale Ve
rinaire, Toulouse, France
¤b Current address: Laboratory of Molecular Microbiology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland,
United States of America
¤c Current address: Diseases of the Developing World (DDW), GlaxoSmithKline I+D, Tres Cantos, Madrid, Spain
Leprosy is a chronic human disease of the skin and peripheral
nerves caused by the intracellular pathogen Mycobacterium leprae.
Although control programs led by governmental and charitable
agencies have reduced the number of patients from ,10–15
million to less than 1 million over the last 10 years , the level of
new cases persists at , 250,000 per year in 2008 . Therefore, in
order to eradicate this disease, it is essential to assist multi-drug
therapy programs with additional control strategies.
Lepromatous leprosy is the most severe manifestat ion of the
disease and is characterized by po or cellular r esponses and
uncontrolled proliferation of the bacil li throughout the skin.
Lesions contain macrophage s filled with bacteria, but few T
lymphocytes and no organized granulomas , sugg esting that
M. leprae evades host immune recognition. Despite the early
discovery of M. leprae in 1873, b oth the biology of this bacterium
and the molecular basis of its pathogenicity remain obscure.
Functional studies have been hampered by the incapacity to
cultivate the leprosy bacillus in v itro and by its extremely slow
growth in animal models (doubling time o f , 14 days). Among
the molecules suspected t o be critical for the p athogenesis of
leprosy is the phenolic glycolipid 1 (PGL-1), a compound
produced in large quantities by M. leprae in vivo [4 ]. P GL-1
consists of a lipid core formed by a long-chain b-diol, which
occurs naturally as a diester of polymethyl-branched fatty acids.
This core is v-terminated by an aromatic nucleus that is
glycosylated by a trisaccharide, which is highly specific of M.
leprae. In contrast, the lipid core, called phenolphthiocerol
dimycocerosates, is conserved in other mycobacterial species
like M. t uberc ulos is and M. bovis, wher e it is linked to different
species-specific saccharidic groups .
PGL-1 has attracted a lot of interest because it might represent a
key virulence factor of M. leprae. Indeed, this compound is located
at the outermost surface of M. leprae and therefore is ideally
positioned to interact with host cell components. The trisacchari-
dic portion of PGL-1 was proposed to promote invasion of
Schwann cells via binding to the G domain of the a 2 chain of
PLoS Pathogens | www.plospathogens.org 1 October 2010 | Volume 6 | Issue 10 | e1001159
laminin-2 in the basal lamina, and may thus be responsible for the
unique capacity of M. leprae to invade peripheral nerves [6,7].
However, the critical importance of this interaction has been
challenged by observations that mycobacteria unable to produce
PGL-1 exhibited similar binding capacities to laminin-2 and
Schwann cells [3,8]. Therefore, the question of whether PGL-1 is
the only determinant of M. leprae conferring tropism for peripheral
nerves is still open. Supporting its putative involvement in the
pathogenesis of the leprosy bacillus, Neill & Klebanoff have
proposed that PGL-1 may be involved in the protection against
oxygen radicals, as coating Staphylococcus aureus with purified PGL-1
or deacylated-PGL-1 increased its capacity to survive within
human monocyte-derived macrophages and to resist in vitro to
reactive oxygen species . Consistent with these results, microbial
glycolipids, including PGL-1, were found to be highly effective in
scavenging oxygen radicals . Whether endogenously expressed
PGL-1 protects mycobacteria from the bactericidal mechanisms of
host cells nevertheless remains to be established. Regarding the
modulation of the host immune response, another major aspect of
leprosy pathogenesis, several lines of evidence suggest that PGL-1
plays a critical role. First, PGL-1 purified from M. leprae was found
to bind the complement component C3, thereby potentially
promoting M. leprae uptake by phagocytes through complement
receptors without triggering a strong oxidative burst . Second,
exogenously added PGL-1 modulated the cytokine response of
human monocytes . Third, M. leprae induced a poor activation
and maturation of dendritic cells and dampened the T-cell
responses induced by infected dendritic cells [13,14]. This
inhibition was partially relieved by treatment of M. leprae-infected
cells with anti-PGL-1 antibodies . Together, these studies
suggested that PGL-1 is a major virulence factor of M. leprae.
However, the cellular and molecular mechanisms by which PGL-1
participates in the cross-talk between the pathogen and the host
cells remain unclear. Clearly, tools to address these questions were
To our best knowledge, the lipid constituents of the M. leprae cell
envelope are structurally almost identical to those of M. bovis BCG,
except PGL. Importantly, in contrast to M. leprae, M. bovis BCG
can be cultivated in vitro and molecular tools are available
to modify its genome. Therefore, M. bovis BCG reprogrammed
to synthesize PGL-1 constitutes an ideal surrogate organism to
investigate the physiological role of this molecule in M. leprae
Here, we have identified the M. leprae genes required for the
biosynthesis of the trisaccharidic domain of PGL-1 and we have
genetically engineered M. bovis BCG to make it synthesize and
export PGL-1. Using this recombinant strain, we studied the
impact of PGL-1 on the initial encounter of mycobacteria with
human phagocytes. We found that PGL-1 deviates the route of
mycobacterial entry into human macrophages and dendritic cells
to suppress the initiation of innate immune responses.
Identification of candidate genes for the synthesis of the
saccharidic domain of PGL-1
The structure of the mai n PGL produced by M. bovis BCG
consists of phenolphthiocerol dimycocerosates glycosylated at the
v-terminus by a 2-O-methylrhamnose (Figure 1A). The lipid core
is structu rally identical to that of the P GL-1 from M. leprae,butin
PGL-1 the saccharidi c domain is 3,6-di-O-Me-Glcp (b1-.4) 2,3
di-O-Me-Rhap (a1-.2) 3-O -Me-Rhap (a1- linked to phenol ring)
(Figure 1A) . Therefore, to reprogram the PGL bios ynthesis
pathways of M. bovis BCG to produce PGL-1, we needed (i) to
prevent the methylatio n at position 2 of the first rhamnosyl
residue, (ii) to provide M. bovis BCG with the enzymes require d
for methylation at po sition 3 of the first rhamnose, and for
synthesis and transf er of the terminal disaccharide on position 2.
The Rv2959c methyltransferase responsible for methylation of
posit ion 2 of the first rhamnosyl residue in the PGL of M.
tuberculosis has been identified . By analogy wi th M. tuberculosis,
inactivation of this gene in M. bovis BCG was expected to result in
the production of unmethylated PGL, the starting point of our
repro gramming process. The next step was to identify the M.
leprae genes required for the transf er of the te rminal disaccharide
and methylation at the defined positions of the carbohydrate
extension. We re asoned that six enzymes would be required: two
glycosyltransferases and four methyltransferases, assuming that
the same enzyme methylates position 3 on the both rhamnosyl
units . Having shown previously that genes involved in the
biosynthesis and translocation of lipids in mycobacteria are
usually clustered in the genome , we perf ormed bioinformatic
analyses of the M. leprae genome using the followin g cri teria:
genes encoding proteins with similarities to known glycosyl- or
methyltransferases, clustering of these genes within the M. leprae
genome, and proximity to or thologs of known PGL biosynthetic
genes. U sing this strategy, we identified 6 candidates for the
methylation and transfer of the two terminal res idues and for the
methylation of the first rhamnosyl residue: ML0128 and ML2348
encoding proteins with similarities to gl ycosyl transf erases, and
ML0126, ML0127, ML234 6c and ML2347 encoding proteins with
similarities to methyltr ansferases (Figures 1B and 1C) [18,19].
The six candidate genes were clustered on two genome regions
containing orthologs of genes involved in the fo rmation of M.
tuberculosis PGL (PGL-tb) and the related phthiocerol dimycocer-
osates (Figure 1C). In M. tuberculosis , these genes map to a
single locus that appears to be divided in M. leprae.Sequence
similarities between the protei ns encoded by the candidate genes
and other enzymes allowed us to assign them a putative fun ction
Mycobacterium leprae, the causative agent of leprosy, is a
chronic human disease responsible for irreversible periph-
eral nerve damage and deformities. Lepromatous leprosy,
the most severe form of the disease, is accompanied by T-
cell unresponsiveness, suggesting that M. leprae has
evolved strategies to modulate host immune responses.
However, the molecular mechanisms of M. leprae infection
remain poorly understood, mainly because this bacterium
has been to date impossible to grow in vitro. The present
study reports an innovative approach to study the
contribution of a phenolic glycolipid (PGL-1) specific of
M. leprae in the cross-talk of the pathogen with host cells.
We reprogrammed a biosynthetic pathway in a surrogate
host, M. bovis BCG, to make it synthesize and display PGL-1
in the context of a mycobacterial envelope. Using this
novel microbial tool, we found that PGL-1 production
enhances the cellular invasiveness of BCG and promotes
the entry via complement receptor 3-mediated phagocy-
tosis. Bacterial uptake via this route was associated with
reduced inflammatory responses in infected human
macrophages. In addition, we showed that PGL-1 produc-
tion inhibited the infection-induced maturation of human
dendritic cells. Our findings thus provide new insights into
the contribution and molecular mechanisms of action of
PGL-1 in leprosy pathogenesis.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 2 October 2010 | Volume 6 | Issue 10 | e1001159
Figure 1. Identification of the genes involved in the formation of the saccharidic domain of PGL-1 from
. (A) Structure of the
PGL from M. leprae, M. tuberculosis and M. bovis and role of the various enzymes from M. tuberculosis in the formation of the saccharidic domain of
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 3 October 2010 | Volume 6 | Issue 10 | e1001159
Construction of recombinant BCG producing PGL-1
To reprogram the PGL biosynthesis pathway in M. bovis BCG,
we first disrupted the Rv2959c ortholog by allelic exchange .
One clone exhibiting the expected PCR profile for a BCG
DRv2959c::km mutant was retained for further studies (Figure S1).
The kanamycin cassette used in this construct, flanked by two res
sites from transposon cd, was removed after transient expression of
the transposon cd resolvase from plasmid pWM19  to
generate the unmarked BCG DRv2959c (Figure S1). The lipids
produced by this mutant strain were analyzed by thin layer
chromatography (TLC) (Figure 2A). As expected, the spot
corresponding to PGL-bovis was no longer detectable and a
new, more polar, glycolipid (product 1) was observed. Matrix-
assisted laser desorption-ionisation time-of-flight (MALDI-TOF)
mass spectrometry analyses of purified product 1 gave a series of
pseudomolecular ions (M+Na)
with a major peak at 1516 amu,
i.e. 14 mass units lower than those of the usual PGL from wild-type
(WT) M. bovis BCG . Therefore, we concluded that this
compound corresponded to the expected unmethylated rhamno-
Next, a DNA fragment encompassing the ML0126, ML0127
and ML0128 genes was inserted into the mycobacterial vector
pMIP12H  to yield plasmid pBNF03 (Figure S1). In parallel, a
second DNA fragment carrying ML2346c, ML2347 and ML2348
genes was inserted into the integrative vector pMV361  to give
pWM76 (Figure S1). These two constructs were transferred
independently or simultaneously into the BCG DRv2959c mutant.
Lipids were extracted from BCG DRv2959c:pBNF03, BCG
DRv2959c::pWM76 and BCG DRv2959c:pBNF03::pWM76 and
analyzed by thin-layer chromatography (TLC) (Figure 2A). In the
case of BCG DRv2959c::pWM76, no new glycolipid was detected.
In sharp contrast, a new glycolipid, product 2, exhibiting higher
mobility was detected in extracts from M. bovis BCG
DRv2959c:pBNF03. MALDI-TOF mass spectrometry analysis of
product 2 gave a series of pseudomolecular ions (M+Na)
at m/z 1704 consistent with the addition of a deoxyhexosyl residue
Figure 2. Construction of a recombinant BCG producing the PGL-1 of
. (A) TLC analysis of lipids extracted from various
recombinant BCG strains. Lipid extracts dissolved in CHCl
were run in CHCl
OH (95:5, v/v). Glycolipids were visualized by spraying the plates
with 0.2% anthrone (w/v) in concentrated H
. The various glycolipids analyzed by mass spectrometry and
H NMR are numbered. (B) MALDI-TOF
mass spectrum of purified PGL-1 produced by r-BCG PGL-1. (C)
H NMR spectra of native PGL-1 (black spectrum) from M. leprae and recombinant PGL-
1 (red spectrum) from r-BCG PGL-1. Inserts correspond to enlargement of parts of spectra that are relevant for structure determination. (D) Structures
of the glycolipids produced by the various recombinant BCG strains after deletion of Rv2959c gene and transfer of 3 or 6 M. leprae genes.
PGL-tb. In M. leprae,p,p9 =4;n,n9 = 16–20; m = 17 ; R1 = 2CH
; R = common lipid core. (B) Candidate proteins for the formation of
the terminal disaccharide of PGL-1 and proposed enzymatic function. (C) DIM+PGL loci in M. tuberculosis and in M. leprae. Orthologs are linked by
dashed lines. Known functions of the encoded proteins are indicated above the open-reading frames.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 4 October 2010 | Volume 6 | Issue 10 | e1001159
and three O-Methyl groups to product 1 produced by BCG
DRv2959c. These results strongly supported the hypothesis that
ML0126, ML0127 and ML0128 genes are involved in the transfer
of the second rhamnosyl unit and the methylation of the first two
sugar residues of PGL-1 in M. leprae. When the two plasmids
pBNF03 and pWM76 were transferred into BCG DRv2959c,at
least three glycolipids were detected (Figure 2A). The two
quantitatively minor compounds exhibited the same mobility than
the PGL-1 intermediates observed in BCG DRv2959c and BCG
DRv2959c: pBNF03. When analyzed by MALDI-TOF mass
spectrometry, the most abundant compound, product 3, showed
a series of pseudomolecular ions (M+Na)
with a major peak at m/z
1894 in agreement with the addition of a di-O-Me-hexosyl unit
(190 uma) to the PGL-1 intermediate, product 2.
These results suggested that the six proteins encoded by genes
ML0126, ML0127, ML0128, ML2346c, ML2347 and ML2348 are
sufficient to produce the specific saccharidic domain of PGL-1.
The six genes were grouped in a single integrative plasmid, named
pWM122 (Figure S1). This plasmid was transferred into BCG
DRv2959 to yield r-BCG PGL-1. PGL-1 production by the
recombinant strain was confirmed by TLC analysis, MALDI-TOF
mass spectrometry (Figure 2B) and NMR spectroscopy analysis
(Figure 2C). When analyzed by mass spectrometry, product 3
purified from r-BCG PGL-1 showed a series of pseudomolecular
with a major peak at m/z 1894 in agreement with
the expected structure (Figure 2B). The characterization of the
saccharidic part was achieved by NMR spectroscopy analysis,
using PGL-1 from M. leprae as reference. The two
spectra were super imposable (Figure 2C). All the signals
unambiguously reflecting the presence of phenolphthiocerol
dimycocerosates were seen in the spectrum of product 3: proton
resonances of p-substituted phenolic group (signals g, h at 6.95 and
7.14 ppm), of methine of the esterified b-glycol (a, 4.85 ppm), of
methyl substituents of polymethyl-branched fatty acids (e, 0.8–
1 ppm; f, 1.15 ppm), of methoxyl and methine groups on the
phthiocerol (b, 3.33 ppm and c, 2.85 ppm). The presence of the
three de-shielded anomeric protons confirmed the presence of a
trisaccharidyl part in product 3. The signals i, i
5.43 ppm, 5.12 ppm and 4.42 ppm were assigned respectively, to
the resonances of anomeric protons of 3-O-Me rhamnosyl, 2,3-di-
O-Me rhamnosyl and 3,6-di-O-Me glucosyl residues . In
addition, five singlets were observed in the region of sugar-linked
methoxyl (OMe) proton resonances at 3.35–3.7 ppm whose
chemical shift values were identical to those found for PGL-1.
All these results identified product 3 as PGL-1 and demonstrated
the role of the six transferred genes from M. leprae in the formation
of the saccharidic domain of PGL-1 (Figure 2D).
Characterization of the recombinant BCG strain
We first compared the amounts of PGL produced by WT BCG
and r-BCG PGL-1 in liquid culture. Each strain was cultured to
exponential phase in liquid medium and PGL were labeled for 24h
C] propionate, a precursor known to be incorporated in
methyl-branched fatty acids containing lipids, such as PGL.
Analysis of the labeled lipids by TLC showed that both strains
produced comparable amounts of PGL, with PGL-1 accounting
for approximately 20% of the total PGL in r-BCG PGL-1 after
24h. As M. leprae cannot be cultivated in vitro, we compared the
amounts of PGL-1 produced by r-BCG PGL-1 and M. leprae by
analyzing on TLC similar quantities of total lipids extracted from
in vitro grown r-BCG PGL-1 and WT M. leprae obtained from
infected armadillos. The amount of PGL-1 found in r-BCG PGL-1
was approximately 2-fold lower than that found in M. leprae.As
observed in Figure 2A and in the labeling experiments (data not
shown), several biosynthetic intermediates were found in lipid
extracts of r-BCG PGL-1. Interestingly, some of these intermedi-
ates were also found in M. leprae extracts but in lower quantities
. One possible explanation for the occurrence of significant
amounts of biosynthetic intermediates in the recombinant BCG
strain may reside in the fact that M. leprae genes were not optimally
expressed in M. bovis BCG, possibly due to their lower GC content.
Another explanation might be the different growth conditions used
(in vivo in infected armadillos for M. leprae and in vitro for r-BCG
PGL-1 and WT BCG). Indeed, we observed that the use of Sauton
medium instead of 7H9 to grow r-BCG PGL-1, led to production
of higher proportion of PGL-1 (data not shown).
Having modified seven enzymatic steps in BCG, we next
evaluated whether this metabolic reprogramming interfered with
some basic microbiological properties of BCG. No difference in
colony morphology or colony size could be detected in r-BCG
PGL-1 following growth on Petri plates, compared to the WT
control (data not shown). Moreover, the growth curves of both
strains in liquid medium were super-imposable during the 3-weeks
observation period (Figure 3A). Since PGL-1 was proposed to
confer protection against reactive oxygen species, we also
compared the viability of WT BCG and r-BCG PGL-1 exposed
for 24 h to increasing concentrations of hydrogen peroxide (H
or sodium nitrite (NaNO
) (at pH 5.5) (Figure 3). Although H
efficiently reduced bacterial viability at concentra-
tions higher than 6 mM and 2.5 mM respectively, PGL-1
production did not modify the cell resistance to reactive oxygen
or nitrogen intermediates (Figure 3B–C).
Together, these results suggested that basic microbiological
properties of BCG such as colony morphology, growth rates or
stress resistance were not affected by the metabolic reprogramming.
PGL-1 production augments BCG infectivity and
intracellular growth in human macrophages
We then used r-BCG PGL-1 to investigate the role of PGL-1 in
host cell infection. For this purpose, fluorescent forms of the WT
and recombinant BCG strains were constructed by transferring a
replicative plasmid carrying the gfp gene under the control of a
mycobacterial promoter. We first compared r-BCG PGL-1 and
the parental BCG strain for their capacity to invade human
monocyte-derived macrophages (hMDM), or human dendritic
cells (hDC), as these cell populations play major roles in the
initiation and regulation of inflammatory responses. Strikingly, the
number of hMDM infected by r-BCG PGL-1 was increased by
3066% compared to that infected by parental BCG after 2 hours
of interaction under non-opsonic conditions. This difference was
observed for all the multiplicity of infection (MOI 10 to 1) tested
(Figure 4A). Moreover, the number of intracellular r-BCG PGL-1
was increased by 70% compared to WT BCG (3.2+/20.22 versus
1.85+/20.5 bacilli/cell) at the analyzed MOI 10:1 (Figure 4B).
Opsonisation of the bacilli by pre-incubation with human serum
markedly augmented the phagocytosis of both strains by hMDM
to reach 20569% and 21868% for WT BCG and r-BCG PGL-1
respectively at MOI 10 (when normalized to 100% for BCG under
non-opsonic condition). However, it abolished the difference
previously observed between WT BCG and r-BCG PGL-1.
Similar observations were made with hDC, e.g. enhanced uptake of
r-BCG PGL-1 compared to WT BCG under non-opsonic
conditions (Figure S2). As for hMDM infection, addition of
human serum abolished the difference between WT BCG and r-
BCG PGL-1 (data not shown).
To determine if this effect was directly related to the presence of
PGL-1 at the surface of r-BCG PGL-1, purified PGL-1 or PGL-
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 5 October 2010 | Volume 6 | Issue 10 | e1001159
bovis were adsorbed onto WT BCG and the invasion efficiency of
coated and uncoated strains were compared (Figure 4C). Adsorp-
tion of purified PGL-1 onto WT BCG increased its capacity to
invade hMDM in a dose-dependent manner when compared to
the uncoated strain (Figure 4C). In contrast, the coating of WT
BCG with PGL-bovis had no significant effect on bacterial
internalization (Figure 4C). Together, these results clearly
established that, in the absence of opsonin, surface-exposed
PGL-1 significantly enhances the bacterial infectivity.
Having shown that PGL-1 promotes host cell invasion, we then
examined whether WT BCG and r-BCG PGL-1 differed in their
capacity to multiply within hMDM. The intracellular loads of WT
BCG and r-BCG PGL-1 were evaluated over a 8-day period by
counting the intracellular colony forming units (cfus) at various
time-points post-infection. As depicted in Figure 4D, the number
of cfus was higher for r-BCG PGL-1 at every time point due to the
enhanced invasion efficiency. In addition, the intracellular growth
of r-BCG PGL-1 was superior to that of WT BCG with a two-fold
and four-fold higher cfu count at 4 days and 8 days post-infection,
To determine if the growth advantage of r-BCG PGL-1 was
associated with altered phagosomal maturation toward fusion with
lysosomes, bacilli-containing phagosomes of hMDM infected with
either WT BCG and r-BCG PGL-1 were compared for their
acquisition of maturation markers. No difference in phagosome
staining with lysotracker, v-ATPase, or CD63 could be detected
between WT BCG and r-BCG PGL-1 at 24h and 96h, indicating
that the maturation of phagosomes containing either WT BCG or
r-BCG PGL-1 was not dramatically changed (Figure 4E and data
not shown). However, at 2h post-infection, the number of CD63-
positive phagosomes was significantly higher in cells infected with
WT BCG, compared to cells infected with r-BCG PGL-1
(Figure 4E). This difference was not retained at later time points,
suggesting that, although the phagosome maturation was not
affected by the occurrence of PGL-1 at the surface of
mycobacteria, the initial bacilli-containing vacuole was not exactly
the same for WT BCG and r-BCG PGL-1.
PGL-1 promotes the entry of r-BCG into human
macrophages via CR3-mediated phagocytosis
Our results established that r-BCG PGL-1 infected hMDM
more efficiently than WT BCG via a route leading to poor early
acquisition of CD63. Since complement receptor 3 (CR3) and
mannose receptor (MR) mediate the non-opsonic internalization
of several mycobacterial species, such as M. kansasii , M.
tuberculosis  or M. leprae , we assessed the possible
involvement of these receptors in the uptake of r-BCG PGL-1.
The effect of a pre-treatment with blocking antibodies raised
against human CR3 or MR on the differential uptake of WT BCG
and r-BCG PGL-1 by hMDM was evaluated. Anti-CR3 blocking
antibodies slightly, but not significantly, modulated the phagocy-
tosis of WT BCG (25%617% inhibition with anti-CR3)
(Figure 5A). Phagocytosis of WT BCG was not affected by the
anti-MR antibody (Figure 5A). In contrast, a marked inhibition of
r-BCG PGL-1 uptake by hMDM (54611% inhibition, p,0.01)
was observed following pre-incubation with an anti-CR3 antibody
(Figure 5A). CR3 blockade restored r-BCG PGL-1 phagocytosis
rates similar to those observed with WT BCG (Figure 5A). This
effect was specific of the anti-CR3 antibody since it was not
observed in the presence of the anti-MR or isotype control
antibodies. In the presence of fresh human serum, the uptake of r-
BCG PGL-1 was similar to that of WT BCG and superior to that
of non-opsonized bacteria (Figure 5B). Pre-treatment with an anti-
CR3 antibody reduced the uptake of both r-BCG PGL-1 and WT
BCG to a similar extent (Figure 5B). Together, these results
strongly suggested that PGL-1 expression confers on BCG the
capacity to exploit the CR3 pathway for hMDM invasion in non-
To investigate further this hypothesis, we evaluated the
differential capacity of WT BCG or r-BCG PGL-1 to infect
recombinant CHO cells expressing human CR3 (CHO-Mac1)
. Following overnight incubation with mycobacteria at MOI
(100:1), only 5 to 7% of control CHO cells had ingested at least
one bacterium under non-opsonic conditions and up to 10% in the
presence of fresh serum. No difference between the WT BCG and
r-BCG PGL-1 was observed. Expression of human CR3 by CHO
cells resulted in enhanced mycobacterial uptake, with up to
3467% of CHO-Mac1 cells infected with WT BCG. These results
showed that WT BCG may use to some extent the CR3 pathway
to invade phagocytes. However, as indicated by the poor
inhibition of WT BCG uptake by hMDM treated with anti-CR3
antibody, BCG preferentially employs other routes for macro-
phage invasion. Importantly, uptake of r-BCG PGL-1 by CR3-
expressing CHO was much more important than that of WT
BCG, irrespectively of the MOI (Figure 5C). In accordance with
our previous findings, opsonic conditions completely abolished the
difference between WT BCG and r-BCG PGL-1 (data not shown).
Figure 3. Microbiological properties of recombinant BCG. A) In vitro multiplication of WT BCG (blue line) and r-BCG PGL-1 (magenta line). 3-
week cultures of both BCG and r-BCG PGL-1 were diluted to final OD
= 0.001 in 7H9 broth containing ADC and 0.05% Tween 80 and incubated at
37uC. At the indicated time points, the OD
was measured. Resistance to H
(B) or NaNO
(C). Exponentially growing WT BCG and r-BCG PGL-1
were diluted 1:100 in 7H9 broth supplemented with ADC and containing various concentrations of H
(0, 3, 6 or 12 mM) and NaNO
(0, 2.5 or
5mM). For NaNO
, pH was adjusted to 5.5. After 24 h of incubation with ROI or RNI, serial dilutions were plated on 7H11 supplemented with OADC
and cfus were evaluated after 3 weeks of incubation at 37uC. The indicated values are means (+/2 SEM) of three independent experiments.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 6 October 2010 | Volume 6 | Issue 10 | e1001159
Collectively, these results demonstrated that PGL-1 improves
mycobacterial entry into hMDM via the CR3 pathway, a route
poorly accessible for WT BCG in the absence of opsonins.
PGL-1 production impairs the initiation of innate immune
We next examined whether PGL-1 production influenced the
innate responses of human phagocytes to mycobacterial infection.
This was first investigated by monitoring the effect of PGL-1 on
the activity of the transcription factor NF-kB, which controls the
expression of multiple inflammatory genes in hMDM and hDC.
We used a THP-1 cell line transfected with a reporter system
under the control of a promoter inducible by NF-kB. Infection of
THP-1 cells with WT BCG induced strong expression of the
reporter gene, indicative of potent activation of the NF-kB
pathway. After 16 h of incubation in the detection medium, the
nm values obtained for the WT BCG were 23%,
60%, 70% higher than for r-BCG PGL-1 at MOI 10:1, 1:1 and
Figure 4. Effect of PGL-1 production on hMDM infection. (A) Percentage of infected hMDM after 2 hours of contact with the mycobacterial
strains at various MOI. (B) Number of bacteria per hMDM after 2 hours of contact with WT BCG or r-BCG PGL-1 at MOI 10:1. (C) Percentage of infected
macrophages following contact with uncoated WT BCG (grey bar), WT BCG coated with PGL from M. bovis (black bars) or with PGL-1 from M. leprae
(white bars) at MOI 10:1. 100% corresponds to 26% of infected macrophages. (D) Intracellular multiplication of WT BCG and r-BCG PGL-1 in hMDM.
hMDM were infected for 2h at MOI 10:1 and the number of cfus was evaluated at various time points by plating serial dilutions of cell lysates on
mycobacterial solid medium. (E) Colocalization of WT BCG or r-BCG PGL-1 with CD63 at 2h, 24h and 96h post infection. Representative confocal
microscopy images of WT or r-BCG PGL-1 phagosomes within CD63-immunostained macrophages at 2h post-infection. White arrows indicated a
phagosome counted positive for CD63. Bar: 4
mM. The percentage of phagosomes positive for each marker was calculated by counting 100
phagosomes from at least 10 different fields in duplicate. The values are means 6 SEM of 4 independent experiments for panels A, and 2
independent experiments for panels B, C, D and E each performed in triplicate or duplicate for panel E. Black bars correspond to BCG control and
white bars to r-BCG PGL-1. The significance of differences between BCG control and r-BCG PGL-1 was evaluated: *, p,0.05 ;**, p,0.01; ***, p,0.001;
n.s., not significant.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 7 October 2010 | Volume 6 | Issue 10 | e1001159
1:10, respectively (Figure 6A). Therefore, the NF-kB response
triggered by r-BCG PGL-1 was significantly lower (p,0.01) than
that of WT BCG, whatever the MOI considered. Accordingly,
hMDM infected with r-BCG PGL-1 produced lower amounts of
the inflammatory cytokine TNF-a than BCG-infected controls
after 2 hours of infection, even though bacteria expressing PGL-1
were more efficiently internalized than the WT controls
(Figure 6B). Other cytokines, IL-12 (p40 and p70) and IL-10,
were also assayed. Both WT BCG and r-BCG PGL-1 induced the
production of poor levels of these cytokines both at 2h and 24h
post-infection, and no difference was observed between WT-BCG
and r-BCG PGL-1 infected hMDM (Figure S3).
To evaluate if the route of r-BCG PGL-1 entry into hMDM
could explain the defective TNF-a production by infected hMDM,
we examined the level of production of this cytokine when CR3-
mediated phagocytosis was blocked. Incubation of hMDM with an
Figure 5. Role of various receptors in the enhanced invasiveness of human macrophages by r-BCG PGL-1. (A) Effect of a pre-incubation
of hMDM with antibodies blocking either CR3 or MR, or irrelevant isotype controls, on the percentage of infected hMDM. (B) Effect of pre-incubation
of hMDM with human serum and a blocking anti-CR3 mAb, or an irrelevant isotype control antibody, on the percentage of infected cells. (C)
Percentage of infected CHO-Mac1 cells after an overnight contact with bacteria at various MOI. Data are presented as the percentage of phagocytosis
with respect to WT BCG under non-opsonic conditions (100%). 100% corresponded to 34% of infected cells for panel A and B, and 38% for panel C.
The values are means 6 SEM of 3 or 4 independent experiments. Black bars corresponded to BCG control and white bars to r-BCG PGL-1. The
significance of differences between BCG control and r-BCG PGL-1 was evaluated:**, p,0.01; ***, p,0.001; n.s., not significant.
Figure 6. PGL-1 production impairs the initiation of innate immune responses. (A) Activation of the NF-kB pathway, as evaluated by the
THP-1 Blue cell line and assay of SEAP activity, following infection with WT BCG (black bars) or r-BCG PGL-1 (white bars) at various MOI. The mean
nm values for WT BCG were 1.13, 0.905, and 0.246 at MOI 10:1, 1:1 and 1:10 respectively whereas the value obtained for r-BCG PGL-1 were 0.87,
0.368 and 0.073 at the same MOI. The value used to normalize at 100% was 1.13 units. (B) TNF-a production by hMDM after 2 hours of post-infection.
One hundred percent corresponded to 1053861579 pg/ml. (C) TNF-a production by hMDM pre-incubated with mAbs directed against CR3 or with
irrelevant isotype control and infected during 2 hours at a MOI of 10:1. One hundred percent corresponded to 41016551 pg/ml. (D) Expression of
maturation markers at the surface of infected hDC, as analyzed by flow cytometry. Data are presented as the percentage of SEAP activity, TNF-a
production or MFI with respect to the WT BCG (100%). The values are means 6 SEM of 2 (panels A and C) and 4 independent experiments (panels B
and D). In panels C and D, grey bars corresponded to uninfected controls. Differences between BCG control and r-BCG PGL-1 were statistically
evaluated: *, p,0.05 ;**, p,0.01; ***, p,0.001; n.s., not significant.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 8 October 2010 | Volume 6 | Issue 10 | e1001159
anti-CR3 antibody, or an irrelevant isotype matched antibody did
not trigger any significant TNF-a secretion. As expected, the
presence of the control antibody did not affect the difference
observed between WT BCG and r-BCG PGL-1: infection of
hMDM with r-BCG PGL-1 resulted in defective TNF-a
production, compared to hMDM infected with WT BCG
(Figure 6C). In contrast, hMDM infected with BCG or r-BCG
PGL-1 in the presence a blocking anti-CR3 antibody produced
equivalent levels of TNF-a, demonstrating the critical role of CR3
in the down modulation of the inflammatory response induced by
r-BCG PGL-1 (Figure 6C).
In parallel, we evaluated by flow cytometry the effects of PGL-1
on the phenotypic maturation of hDC following infection with
WT BCG or r-BCG PGL-1. Here, only cells harboring fluorescent
bacilli were considered, and propidium iodide (PI)
excluded from the analysis. Notably, a reproducible inhibition of
maturation was observed in hDC infected with r-BCG PGL-1
compared to BCG-infected hDC, as witnessed by the reduced
surface expression of MHC class II, CD80, CD83 and CD40
From these results, we conclude that bacterial production of
PGL-1 suppresses the initiation of innate immune responses by
infected phagocytes. In the case of hMDM, the immunomodula-
tory effects of PGL-1 are due to the preferential use of the CR3
pathway for bacterial phagocytosis.
In this study, we successfully modified the PGL biosynthetic
pathway in BCG to generate a recombinant strain expressing
PGL-1, thereby circumventing the difficulties in growing and
genetically manipulating M. leprae. Like the native molecule in the
leprosy bacillus, BCG-expressed PGL-1 was located in the
outermost layer of the envelope (data not shown). Since both
species have otherwise very similar envelopes, our r-BCG PGL-1
strain represented an ideal surrogate of M. leprae for studying PGL-
1 interactions with host cells in a relevant biochemical and
structural context. We found that PGL-1 promoted bacterial entry
in phagocytes via CR3, a property not shared by the phenolic
glycolipids of other mycobacterial species such as M. bovis.
Importantly, deviation of the phagocytosis pathway resulted in
reduced innate immune responses and was associated with
improved intracellular multiplication. Our findings thus strongly
suggest that M. leprae has evolved PGL-1 production as a strategy
to escape innate immunity and establish long-term residence in the
The biosynthesis pathway of PGL involves more than 20
enzymatic steps. The steps required for the formation of the lipid
core are common to all mycobacterial species producing PGL, and
orthologs of the required genes could be identified by genome
comparisons. In contrast, the saccharide appendage of PGL is
species-specific. In the present study, we report the identification of
the genes of M. leprae that are necessary and sufficient for its
synthesis. Our bioinformatic analyses and the finding that the
transfer of three M. leprae genes (ML0126, ML0127 and ML0128)
leads to the production of a PGL-1 intermediate harboring 2,3 di-
O-Me-Rhap (a1-.2) 3-O-Me-Rhap domain, led us to propose
that: i) ML0128 is the rhamnosyl transferase involved in the
attachment of the second rhamnosyl residue on position 2 of the
first unit, ii) ML0127 is the methyltransferase involved in the
methylation of position 2 of the second rhamnosyl residue, iii) and
ML0126 is the enzyme responsible for methylation at position 3 of
the first and second sugar residues. With regard to the transfer and
modification of the terminal glucosyl unit, we concluded that
ML2348 is the glucosyltransferase and ML23246c and ML2347
are the methyltransferases required for the modification of the 3
and 6 positions.
We investigated the role of PGL-1 in the early steps of
mycobacterial interaction with host immune cells and found that
PGL-1 augments the capacity of recombinant BCG to invade
phagocytes, improves the multiplication of mycobacteria in
infected hMDM, and impairs the infection-induced inflammatory
responses. Uptake of BCG by macrophages occurs via various
receptors including CR3 [31,32]. Nevertheless, CR3 is poorly used
by human macrophages to internalize BCG and needs to be
activated, notably through cooperation with other cell surface
receptors, such as CD14/TLR2, for efficient phagocytosis .
Optimal use of this pathway thus requires the presence of serum,
or the addition of a lipopolysaccharide binding protein . In
agreement with these previous results, we found that uptake of
BCG by hMDM in the absence of serum was poorly inhibited
when the CR3 pathway was blocked. Increased uptake of r-BCG
PGL-1 by hMDM was only observed in non-opsonic conditions.
In addition, strong inhibition of r-BCG PGL-1 entry in hMDM
was observed following CR3 blockade. These results are consistent
with previous studies showing that M. leprae preferentially invades
human monocytes through the CR3 receptor in non-opsonic
conditions . Our findings suggest that, in the absence of
opsonins, PGL-1 interacts either with a co-receptor of the CR3
mediated phagocytosis pathway or more likely with CR3 itself.
This interaction might occur through the lectin site of the CR3
alpha chain which was shown to bind various sugar moieties .
The terminal disaccharide of PGL-1, which is missing in PGL-
bovis, may therefore be crucial for the interaction with CR3.
Engagement of CR3 has been reported to be associated either with
pro- or anti-inflammatory responses, depending on the ligand and
costimuli [34–36]. For instance, the fungus pathogen, Blastomyces
dermatitidis uses the CR3 phagocytosis pathway for TNF-a
suppression and immune evasion . Here we demonstrate that
PGL-1 production confers similar properties to mycobacteria, as
preferential phagocytosis of r-BCG PGL-1 via CR3 induced lower
inflammatory responses than those observed with BCG. With
regard to DC, M. leprae has been reported to inhibit their infection-
induced cell maturation and subsequent release of proinflamma-
tory cytokines by comparison to BCG, or M. tuberculosis . M.
-infected DC showed defective expression of major histo-
compatibility complex II expression and CD83 costimulatory
molecule, resulting in poor induction of CD4+ and CD8+ T cells
responses . Our observation that PGL-1, when expressed by
BCG, suppresses the maturation of hDC strongly suggests that
PGL-1 is responsible for the impaired maturation of M. leprae -
infected hDC. On the basis of our results using human
macrophages and dendritic cells, we propose that by promoting
phagocytosis of M. leprae bacilli via CR3, PGL-1 expression may
contribute to the defective cellular responses of multibacillary
lepromatous leprosy patients.
In conclusion, we developed in this study an innovative
approach to understand the role of PGL-1 in the leprosy
pathogenesis. This approach might be extended to the study of
PGL and lipids produced by other mycobacterial species. With
regard to PGL, our knowledge of the biosynthesis pathway of
PGL-1 and PGL-tb will largely facilitate the construction of BCG
expressing PGL of other human pathogens. For instance, the
saccharidic domain found in M. marinum and M.ulcerans, i.e. 3-O-
Me-Rhap (a1- linked to phenol ring) [38,39], corresponds to the
first residue of the carbohydrate domain of PGL-1. Therefore, the
microbial tools are now available to compare the biological
properties of the various PGL in the context of comparable and
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 9 October 2010 | Volume 6 | Issue 10 | e1001159
relevant mycobacterial cell envelopes. This information is crucial
to understand the specificities of the various mycobacterial
diseases, such as the different organ tropism or subversion of host
Materials and Method s
Bacterial strains and growth conditions
M. bovis BCG Pasteur 1173P2 was cultured in Middlebrook
7H9 broth (Invitrogen, Cergy-Pontoise, France) containing ADC
(0.2% dextrose, 0.5% bovine serum albumin fraction V, 0.0003%
beef catalase) and 0.05% Tween 80 and on solid Middlebrook
7H11 broth containing ADC and 0.005% oleic acid (OADC)
(Becton Dickinson, Sparks, USA). When required, kanamycin
(Km) and hygromycin (Hyg) were added to the medium at the
final concentration of 40
mg/ml and 50mg/ml respectively.
Construction of plasmids and mutant strains of M. bovis
A 4.4kb DraI-NsiI fragment was recovered from cosmid B971
that contains a large portion of the M. leprae genome  and
inserted within plasmid pMIP12H  to yield plasmid pBNF03.
This plasmid carried open reading frames ML0126 to ML0128.
Plasmid pWM76 was generated by insertion of a 5.4 kb
Bst1107-XbaI fragment from cosmid L518 (containing genes
ML2346 to ML2348) between the AatII (previously blunt-ended)-
NheI restriction sites of vector pMV361 .
Plasmid pWM122 was constructed by insertion of two PCR
fragments between the NdeI and NheI sites of plasmid pMV361e,
a derivative of pMV361 containing the mycobacterial promoter
pBlaF*  . The first PCR fragment, containing genes
ML0126, ML0127 and ML0128, was obtained with primers 0126
(59-ATACATATGAGAGCAGCCGAAGCTTC-39) and 0128
plasmid pBNF03 as template DNA. The second PCR fragment,
containing genes ML2346, ML2347 and ML2348, was amplified
using primers 2346 (59-TATAAGCTTCAATCCAGCCGG-
GCGTGT-39) and 2348 (59-ATATCTAGACGTGTAGTGTC-
The mutant M. bovis BCG DRv2959c was constructed using the
strategy described by Bardarov et al. . Briefly, a PmeI
fragment, containing the Rv2959c gene disrupted by a kanamycin
cassette flanked by two res sites from transposon cd, was obtained
from plasmid pPET14  and inserted between the XbaI-SpeI
sites (made blunt) of cosmid pYUB854 . The resulting cosmid
was cut with PacI and ligated with the mycobacteriophage
phAE87 to form the recombinant mycobacteriophage phWM06.
Phage particules were then used to infect M. bovis BCG and allelic
exchange mutants were selected on 7H11 agar plates supplement-
ed with Km and OADC. Mutant clones were screened as
previously described  and one clone was selected for further
study. The unmarked mutant was generated following transient
expression of transposon cd resolvase from plasmid pWM19 .
One clone, PMM130, with an amplification pattern consistent
with the excision of the kanamycin cassette was retained for
further analysis (Figure S1). The various plasmids were transferred
in M. bovis BCG or PMM130 by electrotransformation and
transformants were selected on 7H11 agar plates supplemented
with OADC and Hyg.
The various M. bovis BCG recombinant strains were rendered
fluorescent by the transfer of plasmid pWM124, a derivative of the
mycobacterial plasmid pMIP12H allowing expression of gfp gene
from pblaF* promoter.
Biochemical analysis of M. bovis BCG recombinant strains
PGL produced by the various M. bovis BCG recombinant strains
were extracted and analyzed as previously described . For
quantification of the PGL production in WT BCG and r-BCG
PGL-1, each strain was cultured in 7H9 supplemented with ADC
and 0.05% Tween 80 to exponential growth phase and labeled
C] propionate (specific activity of
) for 24h. Lipids were extracted and analyzed as
previously described . To compare the amount of PGL-1
produced by r-BCG PGL-1 and M.leprae, 200 and 400
mg of total
lipids extracted from r-BCG PGL-1 grown 20 days in 7H9
supplemented with ADC or M. leprae recovered from infected
armadillos (kind gift from Dr P. J. Brennan and Dr J. S. Spencer
from Colorado State University, Fort Collins, CO, USA) were
spotted onto a silica gel 60 thin-layer chromatography (TLC) plate
(20620 cm, Merck). The TLC plate was run in CHCl
(95:5, v/v) and PGL were visualized by spraying the plates with a
0.2% anthrone solution in concentrated H
, followed by
heating. Lipids were quantified with a CAMAG TLC scanner
using the Win CATS v1.4.3 software.
Sensitivity to oxygen or nitrogen radicals
Exponentially growing bacteria were diluted 1:100 in fresh
liquid medium containing various concentrations of H
(0, 3, 6
or 12 mM) or NaNO
at pH 5.5 (0, 2.5 or 5 mM for NaNO
generate NO and NO
. After 24 h of incubation at 37uC with
the chemicals, serial dilutions were plated on 7H11 solid medium.
Cfus were counted after three weeks of incubation at 37uC. The
experiments were performed three times independently.
hMDM or hDC cultu res
Human blood samples, purchased from the Etablissement
Franc¸ais du Sang of Toulouse (France), were collected from fully
anonymous non-tuberculous control donors. Peripheral blood
mononuclear leukocytes and hMDM were obtained as previously
described . Briefly, peripheral blood monocytes were cultured
for 7 days on sterile glass coverslips in 24-well tissue culture plates
cells/well) containing RPMI 1640 (Gibco, Cergy Pontoise,
France) supplemented with 2 mM glutamine (Gibco) and 7% heat
inactivated human AB serum. The culture medium was renewed
on the third day. The hMDM were washed twice with fresh RPMI
medium before use.
For hDC, peripheral blood mononuclear cells were isolated
from whole blood by sedimentation over a Ficoll-Hypaque
gradient (GE Healthcare) and monocytes purified by negative
selection (Miltenyi Biotec). Immature DCs (iDCs) were prepared
from this CD14
fraction by culture in RPMI 1640 supplemented
with 1% human serum (DC medium), in the presence of 1,000 U/
ml GM-CSF (and 500–1,000 U/ml IL-4 (Peprotech) for 6 days.
DC maturation was monitored by flow cytometry using APC-
conjugated mouse anti-human CD83 (HB15e) or CD40 (5C3),
PE-conjugated mouse anti-human CD80 (L307.4) or HLA-DR
(G46-6), all from BD Biosciences. For infection studies, DCs were
then plated in 96 well plates at a density of 100,000 cells per 200
in DC medium.
CR3-transfected CHO-Mac1 cells are CHO cells stably
expressing human CR3. A subclone of CHO-Mac1 cells
expressing high levels of CD11b/CD18 was used in our
experiments. Cells were cultured in a-MEM supplemented with
10% heat-inactivated fetal bovine serum, L-glutamine and for
CR3-transfected CHO cells 0.1
mM methotrexate (Sigma, St
Louis). Prior to infection with mycobacteria, CR3 expression was
verified with a PE conjugated anti-CR3 mouse monoclonal
antibody (clone 2LPM) (Dako, Trappes, France) and analyzed
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 10 October 2010 | Volume 6 | Issue 10 | e1001159
by flow cytometry on a FACScalibur (Becton Dickinson) and cells
were seeded (4610
per well) over 12 mm diameter glass
coverslips in 24-wells plates and grown overnight at 37uC.
Immediately before infection, mycobacteria grown to exponen-
tial phase were pelleted at 5000 g for 10min, washed twice in PBS,
resuspended in serum free interaction medium (for experiments
performed under non-opsonic conditions) or pre-incubated in
fresh human AB serum for 30 minutes at 37uC (for experiments
under opsonic conditions). Mycobacteria clumps were dispersed
by drawing up and expelling the bacterial suspension 20 times
through a 25G needle attached to a 1-ml syringe. After a
clarification step, achieved by low speed centrifugation (200 g),
efficiency and reproducibility of dispersal were checked for each
strain by microscopic observation (magnification 6400). The
resulting suspension was diluted 1:10 in the interaction medium
(supplemented with 2% of fresh human AB serum for experiments
under opsonic conditions) and optical density was read at 600 nm
versus relevant blanks. For all infection experiments with the WT
and recombinant M. bovis BCG strains, we established that 0.1 OD
unit corresponded to 1610
bacilli (this was checked by colony
forming unit determination and confirmed by direct numeration
of bacteria under microscopic observation in a Thoma chamber.
Bacteria were further diluted in serum free interaction medium at
the desired multiplicity of infection (MOI), as stated in figure
legends. For each strain and each individual experiment, CFU
numbers were determined by plating serial dilutions of bacterial
suspension on 7H11 agar supplemented with 10% OADC.
When specified, bacilli were coated with purified PGL-1 or
PGL-bovis by suspending 5610
bacteria in 100 ml of 0.05% PGL
in petroleum ether as described previously . Control was
prepared by treating bacilli with solvent alone. The solvent was
evaporated off and the bacteria were resuspended in PBS.
Phagocytosis was assessed as previously described . The role
of MR and CR3 in phagocytosis by hMDM was evaluated by
incubating cells with anti-CR3 (2LPM) or anti-MR monoclonal
antibodies at 10
mg/mL before infection. An irrelevant isotype-
matched antibody was used as control.
Colocalization experiments with Lysotracker, v-ATPase or
Maturation of phagosomes in hMDM was evaluated as
previously described . Recombinant WT BCG or r-BCG
PGL-1 expressing the gfp were used for these experiments. Briefly,
after infection, hMDM were washed and incubated with fresh
medium. LysoTracker Red labelling was performed by washing
hMDM at different time points after infection and incubating
them with the acidotropic dye (1:2000) in RPMI 1640 for 1 h.
Rinsed cells were fixed with 3.7% paraformaldehyde for 1 h. For
v-ATPase or CD63, mouse monoclonal Ab against human CD63
and rabbit polyclonal anti-serum against v-ATPase proton pump
were obtained from Caltag Laboratories (Burlingame, USA) and
from Synaptic Systems (Go¨ttingen, Germany), respectively.
Macrophages were fixed as described above, permeabilized by
incubation with 0.3% Triton X-100 for 10 minutes at room
temperature (RT), blocked by incubation with 0.3% BSA and
incubated with antiserum against v-ATPase (1/100) or mouse anti-
CD63 Ab (1:100) for 1 h at RT, and revealed with Rhodamine-
Red conjugated goat anti-rabbit or anti-mouse Ab. Colocalization
of WT BCG or r-BCG PGL-1 with the various maturation
markers was quantified with a Leica DM-RB fluorescence
microscope. Colocalization was determined as the fraction of
phagosomes with GFP fluorescence associated with LysoTracker,
v-ATPase or CD63 markers. For each marker, 100 phagosomes
from at least 10 different fields in duplicate in two independent
experiments for each time points were counted.
Quantification of NF-kB activity and TNF-a secretion in
The NF-kB activity resulting from cell stimulation with
mycobacteria was studied using the THP-1 Blue-CD14 cell line
(Invivogen, Toulouse, France). This cell line is a derivative of
THP-1 (human monocyte/macrophage cell line) that over-
expresses CD14 and is stably transfected with a reporter plasmid
expressing a secreted embryonic alkaline phosphatase (SEAP) gene
under the control of a promoter inducible by NF-kB and AP-1.
Cells were cultured according to the manufacturer’s instructions.
Bacteria were deposited in 96-wells plates at the indicated
concentrations in a volume of 20
ml and cells were added in
cells per well in the HEK-blue detection medium
(Invivogen) that contains a substrate for the SEAP and fetal calf
serum. Alkaline phosphatase activity, corresponding to NF-kB
activation, was measured after 16h by reading OD at 630 nm.
Positive controls lipomanan and lipopolysaccharide induced
consistent and relevant NF-kB activity. Unstimulated THP-1 cells
had a marginal NF-kB activity representing 2% of the stimulation
induced by WT-BCG. TNF-a secretion was assessed in superna-
tants with a quantitative ELISA test supplied by R&D (Abingdon,
UK), according to the recommendations of the manufacturer.
Data are presented as the mean 6 standard error of the mean
(SEM) or standard deviation (SD) of the indicated number of
independent experiments (n) performed in duplicates (phagocytosis
assays) or triplicates (NF-kB activity detection or TNF-a secretion
quantifications). The significance of differences was determined
with the non-parametric test of Wilcoxon for paired samples to
take into account the inter-donor variability in our analysis.
Figure S1 Construction of a recombinant M. bovis BCG strain
producing PGL-1. A) Schematic representation of the genetic
structure obtained during the construction of the M. bovis BCG
DRv2959c::km and M. bovis BCG DRv2959c::res recombinant
strains. The black box indicates the coding sequence. The light
gray region represents the fragment deleted during the construc-
tion of the knock-out mutant. The dark grey box corresponds to
the kanamycin (km) cassette flanked by the two res sites form
transposon cd. Binding sites of the primers used for PCR analysis
are indicated. B) PCR analysis of the M. bovis BCG DRv2959c::km
and M. bovis BCG DRv2959c::res recombinant strains. The
recombinant and parental strains were characterized by PCR
amplification using primers C (59-ATGTGGAGAATGCTCT-
GCGCC-39), D (59-ACGTTCTTCAGGTGGTTCCGG-39), E
(59-AACTCGCTCAGGATCTCCTGG-39), res1 (59-GCTCTA-
PCR was carried out in a final volume of 50
ml containing 1mM
primers, 2.5 units of GoTaq DNA polymerase, 10% Me
according to the recommendation of the manufacturer (Promega,
Charbonnieres, France). The amplification program consisted of 1
cycle of 2 min at 94uC followed by 30 cycles of 30s at 94uC, 30s at
60uC, 1min at 72uC and a final 10min at 72uC. kb, kilobases. C)
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 11 October 2010 | Volume 6 | Issue 10 | e1001159
Schematic representation of the genes carried by the various
plasmids used for the construction of the various recombinant M.
bovis BCG strains.
Found at: doi:10.1371/journal.ppat.1001159.s001 (1.11 MB EPS)
Figure S2 Effect of PGL-1 production on infection of hDC.
Percentage of infected hDC after 2 h of contact with the
mycobacterial strains at various MOI. The values are means 6
SEM of 4 independent experiments. Black bars correspond to WT
BCG and white bars to r-BCG PGL-1. For the four independent
experiments, 100% corresponded to 56.3%, 41.6%, 20.2% and
43.9% of infected hDC. The significance of differences between
BCG control and r-BCG PGL-1 was evaluated: *, p,0.05; n.s.,
Found at: doi:10.1371/journal.ppat.1001159.s002 (0.47 MB EPS)
Figure S3 Impact of PGL-1 on IL-10 and IL-12 production by
hMDM. hMDM were infected for 120 minutes with WT BCG
(black bars) or r-BCG PGL-1 (white bars) at MOI 10, washed and
further incubated in the presence of serum. At 2 h or 24 h post-
infection, the culture supernatant was removed and IL-10 (A), IL-
12p40 (B) or IL-12p70 (C) were assessed by ELISA. Values
represented the mean + SEM of 2 independent experiments each
performed in duplicate.
Found at: doi:10.1371/journal.ppat.1001159.s003 (0.68 MB EPS)
All biological materials extracted from M. leprae grown in armadillo as well
as antibodies against PGL-1 were kindly provided by Dr P. J. Brennan and
Dr J. S. Spencer from Colorado State University, Fort Collins, CO, USA
with support through the NIH/NIAID Leprosy Contract N01 AI-25469.
We are grateful to Dr I. Maridonneau-Parini (IPBS, CNRS, Toulouse,
France) for the gift of CHO-Mac1 cells. We thank F. Viala (IPBS, CNRS,
Toulouse, France) for technical assistance with figure preparation.
Conceived and designed the experiments: GT CD MD CG. Performed the
experiments: GT CAD CD WM PC AR NH NFB EP. Analyzed the data:
GT CAD CD PC CG. Wrote the paper: GT CD MD CG.
1. WHO (2004) WHO leprosy elimination project: status report 2003 Geneva,
Switzerland. pp 8–11.
2. WHO (2009) Global leprosy situation, 2009. Weekly epidemiological record 84:
3. Britton WJ, Lockwood DNJ (2004) Leprosy. The Lancet 363: 1209–1219.
4. Hunter SW, Brennan PJ (1981) A novel phenolic glycolipid from Mycobacterium
leprae possibly involved in immunogenicity and pathogenicity. J Bacteriol 147:
5. Daffe´ M, Lane´elle MA (1988) Distribution of Phthiocerol diester, phenolic
mycosides and related compounds in Mycobacteria. J Gen Microbiol 134:
6. Rambukkana A, Salzer JL, Yurchenco PD, Tuomanen E (1997) Neural
targeting of Mycobacterium leprae mediated by the G domain of the laminin-alpha2
chain. Cell 88: 811–821.
7. Ng V, Zanazzi G, Timpl R, Talts JF, Salzer JL, et al. (2000) Role of the cell wall
phenolic glycolipid-1 in the peripheral nerve predilection of Mycobacterium leprae.
Cell 103: 511–524.
8. Marques MAM, Antoˆnio VL, Sarno EN, Brennan PJ, Pessolani MCV (2001)
Binding of a2-laminins by pathogenic and non-pathogenic mycobacter ia and
adherence to Schwann cells. J Med Microbiol 50: 23–28.
9. Neill MA, Klebanoff SJ (1988) The effect of phenolic glycolipid-1 from
Mycobacterium leprae on the antimicrobial activity of human macrophages. J Exp
Med 167: 30–42.
10. Chan J, Fujiwara T, Brennan P, McNeil M, Turco SJ, et al. (1989) Microbial
glycolipids: possible virulence factors that scavenge oxygen radicals. Proc Natl
Acad Sci USA 86: 2453–2457.
11. Schlesinger LS, Horwitz MA (1991) Phenolic glycolipid-1 of Mycobacterium leprae
binds complement component C3 in serum and mediates phagocytosis by
human monocytes. J Exp Med 174: 1031–1038.
12. Silva CL, Faccioli LH (1993) Suppression of human monocyte cytokine release
by phenolic glycolipid-1 of Mycobacterium leprae. Int J Lepr Other Mycobact Dis
13. Hashimoto K, Maeda Y, Kimura H, Masuda A, Matsuoka M, et al. (2002)
Mycobacterium leprae infection in monocyte-derived dendritic cells and its influence
on antigen-presenting function. Infect Immun 70: 5167–5176.
14. Murray RA, Siddiqui MR, Mendillo M, Krahenbuhl J, Kaplan G (2007)
Mycobacterium leprae inhibits dendritic cell activation and maturation. J Immunol
15. Hunter SW, Fujiwara T, Brennan PJ (1982) Structure and antigenicity of the
major specific glycolipid antigen of Mycobacterium leprae. J Biol Chem 257:
16. Perez E, Constant P, Laval F, Lemassu A, Lane´elle MA, et al. (2004) Molecular
dissection of the ro le of two methyltransferases in the biosynthesis of
phenolglycolipids and phthiocerol dimycoserosate in the Mycobacterium tuberculosis
complex. J Biol Chem 279: 42584–45592.
17. Guilhot C, Daffe´ M (2008) Polyketides and polyketides-containing glycolipids of
Mycobacterium tuberculo sis: structure, biosynthesis and biological activities. In:
Kaufmann SHE, Rubin E, eds. Hanbook of tuberculosis Molecular biology and
biochemistry. Weinheim: Wiley-VCH Verlag GmbH & Co. pp 21–51.
18. Onwueme KC, Vos CJ, Zurita J, Ferraras JA, Quadri LEN (2005) The
dimycocerosate ester polyketide virulence factors of mycobacteria. Prog Lipid
Res 44: 259–302.
19. Perez E, Constant P, Lemassu A, Laval F, D affe M, et al. (2004)
Characterization of three glycosyltransferases involved in the biosynthesis of
the phenolic glycolipid antigens from the Mycobacterium tuberculosis complex. J Biol
Chem 279: 42574–42583.
20. Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, et al. (1998) Deciphering
the biology of Mycobacterium tuberculosis from the complete genome sequence.
Nature 393: 537–544.
21. Bardarov S, Kriakov J, Carriere C, Yu S, Vaamonde C, et al. (1997)
Conditionally replicating mycobacteriophages: a system for transposon delivery
to Mycobacterium tuberculosis. Proc Natl Acad Sci USA 94: 10961–10966.
22. Malaga W, Perez E, Guilhot C (2003) Production of unmarked mutations in
mycobacteria using site-specific recombination. FEMS Microbiol Lett 219:
23. Malaga W, Constant P, Euphrasie D, Cataldi A, Daffe´ M, et al. (2008)
Deciphering the genetic bases of the structural diversity of phenolic glycolipids in
strains of the Mycobacterium tuberculosis complex. J Biol Chem 283: 15177–
24. Constant P, Perez E, Malaga W, Lane´elle M-A, Saurel O, et al. (2002) Role of
pks15/1 gene in the biosynthesis of phenolglycolipids in the M. tuberculosis
complex: evidence that all strains synthesize glycosylated p-hydroxybenzoic
methyl esters and that strains devoid of phenolglycolipids harbour a frameshift
mutation in the pks15/1 gene. J Biol Chem 277: 38148–38158.
25. Stover CK, de la Cruz VF, Fuerst TR, Burlein JE, Benson LA, et al. (1991) New
use of BCG for recombinant vaccines. Nature 351: 456–460.
26. Daffe´M,Lane´elle M-A (1989) Diglycosyl phenol phthiocerol diester of
Mycobacterium leprae. Biochim Biophys Acta 1002: 333–337.
27. Le Cabec V, Cols C, Maridonneau-Parini I (2000) Nonopsonic phagocytosis of
zymosan and Mycobacterium kansasii by CR3 (CD11b/CD18) involves distinct
molecular determinants and is or is not coupled with NADPH oxidase
activation. Infect Immun 68: 4736–4745.
28. Cywes C, Godenir NL, Hoppe HC, Scholle RR, Steyn LM, et al. (1996)
Nonopsonic binding of Mycobacterium tuberculosis to human complement receptor
type 3 expressed in chinese hamster ovary cells. Infect Immun 64: 5373–5383.
29. Schlesinger LS, Horwitz MA (1990) Phagocytosis of leprosy bacilli is mediated
by complement receptor CR1 and CR3 on human monocytes and complement
component C3 in serum. J Clin Invest 85: 1304–1314.
30. Diamond MS, Garcia-Aguilar J, Bickfodt JK, Corbi AL, Springer TA (1993)
The I-domain is a major recognition site on leukocyte integrin Mac-1 (CD11b/
CD18) for four distinct adhesion ligands. J Cell Biol 120: 1031–1043.
31. Ernst JD (1998) Macrophage receptors for Mycobacterium tuberculosis. Infect
Immun 66: 1277–1281.
32. Sendide K, Reiner NE, Lee JS, Bourgoin S, Talal A, et al. (2005) Cross-talk
between CD14 and complement receptor 3 promotes phagocytosis of
mycobacteria: regulation by phosphatidylinositol 3-kinase and cytohesin-1.
J Immunol 174: 4210–4219.
33. Thornton BP, Vetvicka V, Pitman M, Goldman RC, Ross GD (1996) Analysis of
the sugar specificity and molecular location of the b-glucan-biding lectin site of
complement receptor type 3 (CD11b, CD18). J Immunol 156: 1235–1246.
34. Morelli AE, Larregina AT, Shufesky WJ, Zahorchak AF, Logar AJ, et al. (2003)
Internalization of cirvulating apoptotic cells by splenic marginal zone dendritic
cells: dependence on complement receptors and effect on cytokine production.
Blood 101: 611–620.
35. Medvedev AE, Flo T, Ingalls RR, Golenbock DT, Teti G, et al. (1998)
Involvment of CD14 and complement receptors CR3 and CR4 in nuclear
factor-kappa B activation and TNF production induced by lipopolysaccharide
and group B streptococcal cell walls. J Immunol 160: 4535–4542.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 12 October 2010 | Volume 6 | Issue 10 | e1001159
36. Marth T, Kelsall BL (1997) Regulation of interleukin-12 by complement
receptor 3 signalling. J Exp Med 185: 1987–1995.
37. Brandhorst T, Wu¨thrich M, Finkel-Jimenez B, Warner T, Klein B (2004)
Exploiting type 3 complement receptor for TNF-a suppression, immune evasion
and progressive pulmonary fungal infection. J Immunol 173: 7444–7453.
38. Ville´ C, Gastambide-Odier M (1970) Le 3-O-me´thyl-L-rhamnose, sucre du
mycoside G de Mycobacterium marinum. Carbohydr Res 12: 97–107.
39. Daffe´ M, Varnerot A, Vincent Le´vy-Fre´bault V (1992) The phenolic mycoside of
Mycobacterium ulcerans: structure and taxonomic implications. J Gen Microbiol
40. Eiglmeier K, Honore´ N, Woods SA, Caudron B, Cole ST (1993) Use of an
ordered cosmid library to deduce the genomic organization of Mycobacterium
leprae. Mol Microbiol 7: 197–206.
41. Le Dantec C, Winter N, Gicquel B, Vincent V, Picardeau M (2001) Genomic
sequence and transcriptional analysis of a 23-kb mycobacterial linear plasmid:
evidence for horizontal transfer and identification of plasmid maintenance
systems. J Bacteriol 183: 2157–2164.
42. Bardarov S, Bardarov Jr. S, Pavelka MS, Samdandamurthy V, Larsen M, et al.
(2002) Specialized transduction: an efficient method for generating marked and
unmarked targeted gene disruptions in Mycobacterium tuberculosis, M. bovis BCG
and M. smegmatis. Microbiology 148: 3007–3017.
43. Huet G, Constant P, Malaga W, Lane´elle M-A, Kremer K, et al. (2008) A lipid
profile typifies the Beijing strains of Mycobacterium tuberculosis. Identification of a
mutation responsi ble for a modification of the structures of phthiocerol
dimycocerosates and phenolic glycolipids. J Biol Chem 284: 27101–27113.
44. Astarie-Dequeker C, Le Guyader L, Malaga W, Seaphanh F-K, Chalut C, et al.
(2009) Phthiocerol dimycocerosates of M. tuberculosis participate in the invasion of
human macrophages by inducing changes in the lipid organization of the plasma
membrane. PLoS Pathogens 5: e1000289.
BCG Expressing PGL-1 Evades Host Innate Immunity
PLoS Pathogens | www.plospathogens.org 13 October 2010 | Volume 6 | Issue 10 | e1001159