Lipoarabinomannans Activate the Protein Tyrosine Kinase Hck in Human Neutrophils

Abstract

The mycobacterial lipoarabinomannans (LAMs) are glycosylphosphatidyl-myo-inositol-anchored lipoglycans with diverse biological activities. It has been shown that purified LAMs interact directly, or indirectly, through receptors with the plasma membrane receptors of target cells located in domains rich in glycosylphosphatidylinositol-anchored proteins that contain Src family protein tyrosine kinases. To examine whether LAMs could activate Src-related kinases, human neutrophils were exposed to mannosylated LAMs (ManLAMs) purified from the vaccinal strain Mycobacterium bovis BCG and to phosphoinositol-capped LAMs (AraLAM or PILAM) obtained from the nonpathogenic species Mycobacterium smegmatis. We report first that both ManLAMs and PILAMs activate Hck in a rapid and transient manner and second that complete deacylation of ManLAM abolished its effect on Hck activity, thereby demonstrating that acylation of LAM but not mannosylation is critical for Hck activation. These data indicate that Hck is involved in the signaling pathway of LAMs, molecules known for their ability to trigger several responses in eukaryotic cells.

Mannosylated lipoarabinomannans (ManLAMs) are immunogenic lipoglycans of Mycobacterium tuberculosis, Mycobacterium leprae, and Mycobacterium bovis BCG, the vaccinal strain (4, 28). Whatever their mycobacterial origins, ManLAMs have the same tripartite amphipathic structural model composed of two highly branched homopolysaccharides, namely, d-mannan and d-arabinan, a phosphatidyl-myo-inositol (PI) anchor, and mannooligosaccharide caps (Fig. 1). The PI anchor, located at the reducing end of the mannan, can be mono- or polyacylated (Fig. 1). Phosphoinositol-capped LAMs (PILAMs), previously called AraLAMs from the absence of mannooligosaccharide caps, have been isolated from nonpathogenic species, Mycobacterium sp. and Mycobacterium smegmatis, and are characterized by phosphoinositol caps (10, 28).

FIG. 1.

Schematic presentation of M. bovis BCG ManLAMs. The PI anchor acyl forms of parietal (a) and cellular (b) ManLAMs are detailed. Parietal ManLAMs contain a single fatty acid (FA1), attributed to the unusual 12-O-methoxypropionyl-12-hydroxystearic acid, on C-1 of the glycerol unit (16). Cellular ManLAMs are characterized by four acyl forms; the two major forms, present in equal amounts, contain two fatty acids (FA1 and FA2) and three fatty acids (FA1, FA2, and FA3) (17). P, phosphate; Manp, mannopyranose.

ManLAMs, which can be regurgitated by infected macrophages (30), have been shown to exert several biological effects on phagocytic cells and T lymphocytes (4, 28). Some of them are abolished when the mannooligosaccharide caps or the acyl chains are removed (4, 28). These effects could be initiated by the binding of ManLAMs to membrane receptors or by direct interaction of the molecule with the membrane lipid bilayer. On one hand, LAMs can bind to receptors: ManLAMs have been shown to bind to the mannose receptor through mannooligosaccharide caps (2, 22, 27), and AraLAMs bind to the glycosylphosphatidylinositol (GPI)-anchored protein CD14 (2) through the lipid core (21). On the other hand, purified LAMs, ManLAMs, and AraLAMs, can be directly integrated into specialized plasma membrane domains of target cells enriched in endogenous GPI-anchored molecules; the acyl chains are critical in this process (9).

The PI termini of LAMs are related to the GPI moiety that anchors proteins to membranes (4, 28). In eukaryotic cells, proteins with a GPI anchor are often localized in particular areas of the plasma membrane which are enriched in cholesterol and glycosphingolipids. Following membrane solubilization at low temperature by most types of mild detergent, such as Triton X-100, GPI-anchored proteins are found in detergent-insoluble complexes called rafts or glycosphingolipid-enriched membrane domains (GEMs) (8). Protein tyrosine kinases of the Src family are associated with the internal sides of GEMs and are implicated in transducing signals of several GPI-anchored proteins (25). Among the Src family tyrosine kinases, Hck, which is specifically expressed in phagocytes (25), has been shown to increase the expression of tumor necrosis factor in macrophages (7), a function also served by LAMs (4). Therefore, we decided to examine whether Hck could participate in the transduction signals elicited by ManLAM. To avoid the potential interaction of ManLAMs with the mannose receptor, a receptor not located in GEMs, this study was performed with neutrophils, phagocytes which do not express the mannose receptor (19). Neutrophils were isolated from the blood of healthy donors and separated by the dextran Ficoll method, as previously described (13). The final cell preparation, containing more than 95% neutrophils (15), was adjusted to 2.5 × 10⁶ cells/ml in minimal essential medium (Life Technologies, Cergy Pontoise, France) buffered with 10 mM HEPES, pH 7.4, and stimulated for various times with ManLAMs. Hck was then solubilized and immunoprecipitated, and its kinase activity was assayed as previously described (13, 29). Briefly, neutrophil proteins were solubilized in a buffer containing 1% Nonidet P-40, and after immunoprecipitation, Hck tyrosine kinase activity was assayed using acid-denatured rabbit muscle enolase as a nonspecific substrate and 10 μCi of [γ-³²P]ATP (6,000 mCi/mmol) (13). The proteins were then separated on a sodium dodecyl sulfate–8% polyacrylamide gel, and Hck-dependent phosphorylation of enolase was quantified by using the Image QuaNT program on a Molecular Dynamics Storm840 imager. Using a new extraction procedure, two separate pools of ManLAMs, called parietal and cellular, were obtained from BCG (16). Cellular and parietal ManLAMs differ mainly in the structure of the PI anchor (Fig. 1). Briefly, the anchor of the parietal ManLAM contains a single fatty acyl residue, 12-O-(methyoxypropanoyl)-12-hydroxystearic acid. In contrast, four acyl forms were characterized for the cellular ManLAMs (17) (Fig. 1), among which the triacylated and diacylated acyl forms are the most abundant (17). In the di- and triacylated forms, the glycerol moiety was found to be mainly esterified by tuberculostearic acid and palmitic acid, while the myo-inositol was replaced by palmitic acid in the triacylated ManLAM. The cellular ManLAM GPI anchor structure is shared by all the ManLAMs investigated to date, while the parietal form appears unique with respect to the fatty acid structure.

Neutrophils were exposed for different periods of time to 20 μg of cellular ManLAM/ml. At 2.5 min, Hck was activated 1.9-fold over the basal activity. The level of Hck activation was of the same magnitude as the response elicited by 3 mg of zymosan/ml opsonized in human serum (15), used in these experiments as a positive control for Hck activation (29) (see Fig. 3). The activation was transient and was back to basal value at 10 min (Fig. 2A). Lipopolysaccharide (LPS) has been shown to activate Hck through CD14 binding (23). LPS contamination of our LAM preparations did not exceed 10 pg per μg of LAM. Thus, the LPS concentration in experiments using 20 μg of LAM/ml reached 0.2 ng/ml. This is far below the concentration needed to prime neutrophil responses in the absence of LPS-binding protein (6) and, as depicted in Fig. 3, cellular ManLAM elicited Hck activation at concentrations as low as 0.02 μg/ml, when the LPS concentration is negligible. Furthermore, ManLAM treatment with polymyxin, a component interacting with the lipid A moiety of LPS, did not affect Hck activation ([1.6 ± 0.1]-fold over basal value) when compared to the response induced by 2 μg of untreated ManLAM/ml ([1.4 ± 0.1]-fold over basal value; n = 2). Taken together, these data ruled out a potential involvement of LPS in the ManLAM-dependent response.

FIG. 3.

Effects of various concentrations of cellular ManLAMs. Neutrophils were incubated with various concentrations of cellular ManLAMs for the period of time shown in Fig. 2 (n = 6) or with 3 mg of serum-opsonized zymosan (OZ)/ml for 5 min as a positive control for Hck activation (n = 3). The data were analyzed at the peak of Hck activation and are presented as the means ± standard errors of n individual experiments.

FIG. 2.

Time course of Hck activation in neutrophils stimulated with cellular ManLAMs. Neutrophils were incubated with 20 μg of cellular ManLAMs/ml for the indicated periods of time. Hck was then solubilized and immunoprecipitated, and the in vitro phosphorylation of the exogenous substrate enolase by Hck was measured as detailed in the text. The data are the mean ± standard error of four independent experiments. (Inset) Representative kinase assay showing phosphorylation of enolase.

We then examined whether the degree of acylation of ManLAM could modulate its effects on Hck activation. As shown in Fig. 4, parietal ManLAM containing a single fatty acyl residue induced activation of Hck with kinetics and magnitude similar to those of cellular ManLAM (Fig. 2). Therefore, whether ManLAMs were mono- or polyacylated did not make a difference in Hck activation. To further investigate the role of acyl chains, deacylated ManLAMs were prepared by incubating 300 μg of ManLAMs in 200 μl of 0.1 N NaOH for 2 h at 37°C (16). After neutralization with HCl, the reaction products were dialyzed against deionized water. No significant amounts of fatty acids, as detected by gas chromatography, were released from deacylated ManLAMs when they were exposed to strong alkaline hydrolysis (1 N NaOH). Deacylated ManLAMs failed to activate the kinase (Fig. 5 and 6), indicating that acyl chains of ManLAMs play a critical role in the activation of Hck.

FIG. 4.

Time course of Hck activation in neutrophils stimulated with parietal ManLAMs. Neutrophils were incubated with 20 μg of parietal ManLAMs/ml for the indicated periods of time, and then samples were processed as described in the legend to Fig. 2. The data are expressed as the mean ± standard error of four independent experiments.

FIG. 5.

Time course of Hck activation in neutrophils stimulated with deacylated LAMs. Neutrophils were incubated with 20 μg of ManLAMs (■) or deacylated LAMs (▴) per ml for the indicated periods of time, and Hck kinase activity was assayed as described in the legend to Fig. 2. One experiment representative of three independent experiments is shown.

FIG. 6.

Acylation but not mannosylation of LAMs is required for Hck activation. Neutrophils were incubated with 20 μg of cellular (c) ManLAMs, deacylated ManLAM, parietal (p) ManLAM, or PILAM/ml for the periods of time shown in Fig. 2 and 5. The data were analyzed at the peak of Hck activation and are presented as means ± standard errors of three to seven individual experiments.

Experiments were then conducted with PILAMs isolated from an avirulent species, M. smegmatis (28). As depicted in Fig. 6, PILAMs also induced activation of Hck in a manner which was not significantly different from the response evoked by cellular ManLAM. Therefore, mannose moieties of LAMs, and thus mannose-recognizing receptors, are not critical for Hck activation. This also indicates that the capacity of LAMs to activate Hck is not related to their mycobacterial origin.

This is the first demonstration that LAMs can activate a kinase of the Src family, Hck. Hck is rapidly and transiently activated in human neutrophils exposed to very low concentrations of ManLAMs and PILAMs, indicating that these molecules are very efficient stimuli. Very few reports are available on LAM-dependent intracellular signals in phagocytes. ManLAMs inhibit protein kinase C (3) and activate the tyrosine phosphatase SHP-1 in human mononuclear cells (11). We demonstrate here that Hck is probably part of the very early transducing signals elicited by LAMs in human neutrophils. Interestingly, we have previously reported that phagocytosis of mycobacteria by neutrophils is not associated with activation of Hck (15), further supporting previous data showing that mycobacteria and LAMs do not necessarily trigger the same responses (24).

Hck is found in distinct subcellular compartments in human neutrophils (13, 29). It is likely that the kinase gains access to substrates present in these two compartments and mediates distinct functions. It has been shown to be associated with the membranes of lysosomal granules and, to a lesser extent, with the plasma membrane (13, 29), especially in GEMs, where it is regulated by several GPI-anchored proteins (5, 14).

How ManLAMs can activate Hck is not yet clear. We demonstrate here that the mannose moieties of ManLAM are not critical for Hck activation, thus ruling out involvement of mannose-recognizing receptors. In addition, AraLAMs have been described as ligands of CD14 (20), a GPI-anchored protein initially described as a receptor for LPS (26), and activation of Hck through GPI proteins, such as CD14, CD59, and CD87, has been reported by several groups (5, 14, 23). It is rapid and transient, and the magnitude of activation is similar to that obtained here with LAMs (5, 14, 23). Therefore, the profiles and magnitudes of activation of Hck by GPI proteins and LAMs look similar. The mechanism by which GPI-anchored proteins on the outer face of the plasma membrane can communicate with Hck on the inner leaflet has not been elucidated. Using antibodies directed against several of these GPI proteins often induced coimmunoprecipitation of Src family kinases, but the precise nature of this association remains unresolved. Furthermore, it has been shown that ManLAMs and AraLAMs can be directly integrated into GEMs through their GPI anchors, independently of interaction with surface receptors (9). Prior deacylation of LAMs abrogated this membrane insertion (9) and abolished activation of Hck (see above). It is possible that insertion of molecules into GEMs may modify the organization of these membrane domains and thereby affect their signaling pathways. However, we cannot rule out the possibility that LAMs could also directly interact with the plasma membranes outside GEMs. Regardless of how LAMs interact with host cells, we show that ManLAMs and PILAMs are able to activate Hck and that acyl chains play a critical role.

ManLAMs have been described as stimulating agents for expression and secretion of tumor necrosis factor alpha in human neutrophils (18), functions also served by Hck (7). ManLAMs regurgitated by M. tuberculosis-infected macrophages induce T-cell chemotaxis, which can be blocked by tyrosine kinase inhibitors (1). In addition, chemotaxis is also under the control of Src family tyrosine kinases, since it has been shown to be deficient in Hck^−/− and Fgr^−/− double-knockout mice (12). Therefore, it is possible that these cell responses initiated by ManLAMs require activation of Hck as part of the signal transduction pathway.