Surface-Exposed Glycopeptidolipids of Mycobacterium smegmatis Specifically Inhibit the Phagocytosis of Mycobacteria by Human Macrophages. Identification of a Novel Family of Glycopeptidolipids.

Phagocytosis by macrophages represents the early step of the mycobacterial infection. It is governed both by the nature of the host receptors used and the ligands exposed on the bacteria. The outermost molecules of the non-pathogenic Mycobacterium smegmatis were extracted by a mechanical treatment and found to specifically and dose-dependently inhibit the phagocytosis of both M. smegmatis and the opportunistic pathogen M. kansasii by human macrophages derived from monocytes. The inhibitory activity was attributed to surface lipids because i) it is recovered only in the organic solvent phase of a water/chloroform partition of the surface-exposed material and ii) it is dramatically reduced by alkaline hydrolysis, but not by a protease treatment. Fractionation of surface lipids by adsorption chromatography indicated that the major inhibitory compounds consisted of phospholipids and glycopeptidolipids (GPLs). Mass spectrometry and nuclear magnetic resonance spectroscopy analyses, combined with chemical degradation methods, demonstrated the existence of a novel family of GPLs that consists of a core composed of the long-chain tripeptidyl aminoalcohol with a di- O -acetyl-6-deoxytalosyl unit substituting the allo -threoninyl residue and a 2-succinyl-3,4-di- O -CH 3 -rhamnosyl unit linked to the alaninol end of the molecules. These compounds, as well as diglycosylated GPLs at the alaninol end and de- O -acylated GPLs, but not the non-serovar specific di- O -acetylated GPLs, inhibited the phagocytosis of M. smegmatis and M. avium by human macrophages at a few nanomolar concentration without affecting the rate of zymosan internalization. At micromolar concentrations, the native GPLs also inhibit the uptake of both M. tuberculosis and M. kansasii . De- O -acylation experiments established the critical roles of both the succinyl and acetyl substituents.

Therefore, elucidation of the mechanisms involved in the interaction between macrophages and mycobacteria could help to develop new pharmacological strategies to prevent macrophage infection.
Pathogenic mycobacteria have the ability to persist in macrophages (5) that are usually meant to kill microorganisms. This successful parasitism involves mycobacterial strategies to protect themselves from potent host anti-microbial processes, such as restriction of lysosomal fusion with their phagosomes (6)(7)(8). The early process of recognition and internalization of mycobacteria in macrophages is essential in the outcome of infection. In human macrophages live or heat-killed bacteria are indistinguishably refractory to fusion with lysosomes (10). In sharp contrast, when mycobacteria were serum opsonized, their phagocytosis was associated with an oxidative response (9) and a maturation of phagosomes towards a fusion with lysosomes (6,11,12). This led us to propose that, during the initial steps of the infection, mycobacteria have developed a common strategy, which consists in the use of receptors of non-opsonic phagocytosis uncoupled to bactericidal responses. In agreement with this proposal, we recently demonstrated that the mannose receptor, which efficiently participates in binding and internalization of pathogenic and non-pathogenic mycobacteria, was not coupled to bactericidal functions in human macrophages (9). Similarly, binding and internalization of the opportunistic pathogen M. kansasii through the complement receptor type 3 (CR3) did not activate the NADPH-oxidase in macrophage cell lines (13).
The mycobacterial envelope is composed of a plasma membrane surrounded by a complex cell wall, which in turn is recovered by a superficial layer composed of proteins, carbohydrates and, to a lesser extent, lipids (14)(15)(16). This outermost structure, also called capsule in the case of pathogenic species, represents a privileged interface between bacilli and their host cells. Some of its components have been implicated in the interaction with macrophages (17). The major carbohydrate constituent of the capsule, a glycogen-like glucan (14), has been shown to inhibit the binding of M. tuberculosis to CR3-expressing CHO cells 7 We extracted mycobacterial surface-exposed compounds by gentle surface abrasion with glass beads (15) and tested their role in the interactions between mycobacteria and MDMs to identify the most active bacterial partners of the phagocytic process. As we have demonstrated that pathogenic and non-pathogenic strains use the same phagocytic receptors to infect cells (9,10,22,23), this work was performed with the non-pathogenic species, M.
smegmatis. Using this approach, we demonstrated that surface exposed C-type GPLs inhibited the non-opsonic phagocytosis of GPL-containing mycobacteria and to a lower extent those of other mycobacteria.  Hewlett-Packard 5890 series II gas chromatograph, fitted with an OV1 fused-silica capillary column (12 m x 0,30 mm) and connected to a Hewlett-Packard 5989X mass spectrometer in Electron-Impact (EI) mode with an ionization potential of 70 eV. The temperature programs used were 100 °C (delay 3 min) to 290° C at 8 °C/min or isotherm at 40 °C for alditol acetates or short acids methyl esters analysis respectively.

Bacterial Strains and Growth Conditions
One-and two-dimensional 1 H-NMR spectra were recorded on a Bruker AMX-500 spectrometer using standard pulse sequences available in the Bruker software. The chemical shifts were expressed in parts per million relative to acetone as internal standard (δ H 2.22).

Isolation and Culture of Human Macrophages Derived from Monocytes (MDMs) -
Human peripheral blood monocytes were isolated as previously described (9) and cultured on sterile glass coverslips in 24-well tissue culture plates (5x10 5 cells/well) containing RPMI medium with 10% heat-inactivated FCS and antibiotics, for 6 to 7 days at 37°C in 5% CO 2 .
The culture medium was renewed at the third day. Before use, MDMs were washed twice with fresh RPMI medium and equilibrated for 20 min at 37°C in 5% CO 2 .

Infection of Adherent Macrophages and Phagocytosis Assay -When specified, MDMs
were pretreated for 15 min at 37°C with either the crude SEMs, the different phases obtained after phase partition or the purified lipid fractions resuspended in sterile apyrogenic water and sonicated for 10 min. All dilutions were performed in RPMI medium. MDMs were then put in contact with bacilli for 1 hr and washed twice with fresh medium to remove unbound particles. Phagocytosis of FITC-stained bacteria was determined as previously described (22 were permeabilized in methanol for 6 min at -20°C and washed in PBS containing 0.1% Tween 20. Phagosomes containing particles were stained with a lysosomal membrane marker, CD63, revealed with a fluorescent-conjugated secondary Ab as previously described (9).
Statistics -Data are presented as the mean + standard error of the mean (sem) of the indicated number of experiments (n) performed in duplicate. The significance of the differences was determined by the paired or unpaired Student't test.

Inhibitory Activity of Surface-Exposed Material from M. smegmatis on Mycobacterial
Phagocytosis -We first checked that the surface-exposed material (SEM) of M. smegmatis readily affected the ability of macrophages to internalize bacilli. The crude SEM was arbitrary dialyzed using a 100-Da molecular weight cut-off membrane, to eliminate small molecules from the culture medium. The amount of SEM to be used, corresponding to bacilli in contact with MDMs (50 particles/cell), was rationalized by using a final concentration of 150 ng protein/ml SEM. Under these conditions, SEM isolated from early log-phase grown M. smegmatis (day 3) decreased by 43 ± 2 % (n=2) the rate of phagocytosis of 3-day old bacteria.
Similarly, SEM isolated from late log-phase grown M. smegmatis (day 6) decreased by 41 ± 10 % (n=4) internalization of 6-day old bacteria. It was thus concluded that SEM exhibited an inhibitory effect on the phagocytosis of M. smegmatis that was independent from growth conditions.

Chemical Nature of the Inhibitory Compounds from the Surface-Exposed Material -
We estimated the approximate size of these molecules by dialysis. While the dialysis of SEM using a cut-off of 1000-Da did not affect its inhibitory effect (52 + 4 %, n=9) ( Fig. 2), no inhibition was observed with SEM dialyzed at a 6-8 kDa cut-off (4 + 1 % inhibition, n=3).
These data indicated that the inhibitory molecules possess a low molecular weight (between 1000 and 8000 Da).
Accordingly, the subsequent work was performed with SEM dialyzed with a 1000-Da Furthermore, SEM did not either modify phagocytosis of various inert particles such as latex beads and opsonized or non-opsonized zymosan (Fig. 2). It was thus concluded that SEM contained inhibitory molecules that specifically decrease the non-opsonic phagocytosis of mycobacteria but do not affect internalization of other particles.
An overnight incubation of SEM with non-specific/broad spectrum proteases, pronase and proteinase K, did not influence its inhibitory effect on M. smegmatis internalization (Fig.   3a), suggesting that the observed inhibition was not due to proteins. In contrast, when SEM was treated with alkali, a treatment that hydrolyses ester bounds, a pronounced decrease of its inhibitory activity was observed (Fig. 3a), implying that a significant portion of inhibitory molecules was alkali-labile. Based on the occurrence of ester linkages in most of the mycobacterial lipids, these molecules were isolated by partitioning SEM between chloroform, methanol and water. TLC analysis of the organic phase showed the presence of GPLs, 6monomycoloyltrehalose, PIMs, phosphatidylethanolamine (PE), phosphatidylinositol (PI) and phosphatidylglycerol (PG) (data not shown) as previously reported (21). The interphase and the aqueous phase contained mostly carbohydrates and proteins. As depicted on Fig. 3b, the organic phase exhibited the highest inhibitory effect on phagocytosis of M. smegmatis.
Although some inhibition was observed with the aqueous phase (Fig. 3b), an extensive extraction with organic solvents to eliminate residual lipids abolished this effect (data not shown). The inhibitory effect of the interphase was negligible (Fig. 3b). Altogether, these data indicated that most of the inhibitory activity of SEM was attributable to its lipid constituents.
Purification and Identification of the Inhibitory Substances of M. smegmatis -Lipids from the SEM were fractionated by adsorption chromatography on a Florisil column. They were eluted with chloroform, increasing concentrations of methanol in chloroform and finally with a mixture of chloroform, methanol and water. Five lipid fractions were obtained and tested at 100 µg lipid/ml for their ability to inhibit M. smegmatis phagocytosis. Only two of them exerted an inhibitory activity. The most polar fraction showed the highest inhibitory activity comparable to that of the organic phase (50 ± 6 and 47 ± 1 %, n=3, respectively).
TLC analysis (data not shown) demonstrated that this fraction contained glycolipids that shared with GPLs the same "string-of-beads" aspect and green-blue coloration with anthrone, and phospholipids such as PIMs, PI, PG and PE, suggesting that both surface-exposed GPLs and phospholipids possessed inhibitory activities. The second fraction that inhibited phagocytosis of M. smegmatis by 26 ± 4 %, (n=3) contained GPL-like molecules but not phospholipids (data not shown).
As GPL-like substances were present in the two inhibitory fractions, we next focused on these compounds. Because GPLs are present in both the outermost layer of mycobacteria cell envelope and the whole bacteria (26,32), they were isolated from whole bacterial lipids in order to obtain more material. The extract was first enriched in GPL-like molecules by methanol precipitation, a procedure known to concentrate phospholipids and mycoloylated glycolipids in the methanol-insoluble fraction. As expected, the extract enriched in GPL-like was able to inhibit the internalization of M. smegmatis (Fig. 4). At concentrations ranging from 1 pg to 100 µg lipid/ml, GPLs induced a biphasic effect on phagocytosis of M.
smegmatis. The inhibitory effect increased from 1 to 100 pg lipid/ml, with a maximal effect at 100 pg lipid/ml that remained stable up to 10 ng lipid/ml and then progressively decreased were also present in the SEM and showed by MALDI-TOF mass spectrometry that they were composed of the same species (Fig. 6).
Because GPLs are species-specific (33,34), the purified GPL-like compounds from + 9 % and from 31 ± 9 to 31 ± 8 %, n = 4, respectively). However, when lipid concentration was increased from 10 ng/ml to 10 µg/ml, phagocytosis of M. tuberculosis progressively decreased (data not shown). At 10 µg/ml the inhibitory effect was similar to that obtained with 10 ng lipid/ml on M. smegmatis phagocytosis (Fig. 7). As a control, we checked that 10 µg lipid/ml from M. smegmatis decreased to the same extent the phagocytosis of M. kansasii (from 27 + 1 to 16 + 2 %, n = 2), while they did not interfere with the internalization of zymosan (Fig. 7). This led us to propose that the GPL-like molecules that composed S21 specifically inhibited the non-opsonic phagocytosis of mycobacteria. As these compounds exhibited TLC mobilities similar to those of GPL-like substances found in fractions S13-18 ( Fig. 5b), we postulated that subtle structural features might exist between the isolated compounds.
Structural features of GPL-like substances -The structure of the various constituents of S8-21 was determined by NMR, MALDI-TOF mass spectrometry and analysis of chemical degradation products according to the procedures shown in Fig. 1. Preliminary MALDI-TOF mass spectrometry analysis has led us to choose fractions S8, S13 and S20 as representative of three types of GPL-like compounds. Their one- (Fig. 8) and two-dimensional 1 H-NMR spectra were very similar to those of the previously characterized C-type GPLs (33) and confirmed the presence of i) phenylalaninyl, threoninyl, alaninyl and alaninol residues, ii) deoxysugar units, iii) β-hydroxylated long-chain, iv) double bonds, and v) methoxyl groups in these compounds (35). Molecules in fractions S8, S13 and S20 were thereafter called GPLs I, II and III, respectively (Fig. 9). In addition, signals at 2.62 and 2. showed that the hydroxyl and methoxyl groups were located on the phenylalanyl diacid moiety of the unsaturated molecules. It followed then that fractions S8-20 contained all the structural features found in C-type GPLs (Fig. 9).
Structures of the various GPLs of M. smegmatis -The MALDI-TOF mass spectrum of the native S8 (Fig. 10a)  The most active fractions S17-21 contained GPL-like molecules (called GPLs III) exhibiting mobilities on TLC similar to those of GPLs II (Fig. 5b). This observation was unexpected since S17-21 were eluted from the gel only when water was added to the mixture of chloroform and methanol. When these fractions were analyzed by MALDI-TOF mass spectrometry, however, the observed [M + Na] + peaks (Fig. 10b)  Na] + peaks due to the remaining molecules were observed for the major series at 1008.9, 1036.9 and 1064.9 m/z (Fig. 10c). Perdeuteriomethylation of GPLs III followed by MALDI-TOF mass spectrometry analysis (Fig. 1), showed that hydroxylated compounds incorporated 9 CD 3 (including 4 on the peptide core) whereas only 8 CD 3 were incorporated in methoxylated molecules (data not shown). Acid hydrolysis of the perdeuteriomethylated products ( Fig. 1)  have been previously shown to use common receptors, such as mannose receptors and CR3, to infect macrophages (unpublished data and (9,10,19,41). Therefore, one can assume that mycobacteria express distinct ligands that can be recognized by a same receptor but with various ranges of affinity, a matter that remains to be addressed.
C-type GPLs share a common lipopeptidyl core composed of a long-chain fatty acyl linked to a tripeptide and terminated by alaninol. In all known C-type GPLs the alaninol residue is substituted by a O-CH 3 -rhamnosyl or dirhamnosyl residue (33). In M. smegmatis, a di-O-acylated-6-deoxy-talosyl residue is linked to the allo-threoninyl residue. These simplest forms are found in all GPL-containing mycobacterial species and, accordingly, are called nsGPLs or apolar GPLs. The allo-threoninyl-linked sugar unit of nsGPLs may be further glycosylated in some GPL-containing mycobacteria, such as M. avium (33,34). These discrete modifications confer to GPLs variable TLC patterns and antigenic properties (20,33,34). The mechanisms by which surface-exposed GPLs participate in the binding and phagocytosis of mycobacteria deserve consideration. GPLs from M. avium, essentially through its lipopeptide fragment, have been reported to disturb cell membrane ultrastructure and to change the expression of surface receptors of murine macrophages (43). In addition, mycobacterial GPLs are able to get inserted into phospholipid monolayers (44) and to disturb its properties (45). Such an insertion of molecules may alter interactions of mycobacteria with their host cell. However, while all GPLs share the same lipid core, only GPLs II and III were active. In addition, GPLs have no effect on the internalization of control particles such as zymosan, suggesting a specific recognition of structurally defined molecules. Taken together, these observations ruled out a non-specific effect of GPLs under our experimental conditions.
Further studies are clearly needed to elucidate the precise mechanisms by which GPLs affect the internalization of mycobacteria, either directly as ligands of receptor and/or indirectly as modulators of a specific receptor function.
In conclusion, the outermost layer of M. smegmatis contained new classes of C-type GPLs and phospholipids that efficiently inhibited the non-opsonic phagocytosis of mycobacteria. As such, these molecules may help to design pharmacological drugs with a new therapeutic strategy, consisting in the inhibition of mycobacterial multiplication by preventing their cyclic internalization into macrophages. Mycobacteria are internalized by several receptors into human macrophages. Some of them are pattern recognition receptors, which can elicit different intracellular signals depending on the ligands used (13). Therefore, identification of surface exposed mycobacteria ligands should also help to dissect the signalling pathways of receptors already known to internalize mycobacteria, as well as to discover new receptors involved in the infection of macrophages by mycobacteria.            S8 S11 S13 S15 S18 S20