Peptidoglycan (PGN) is the major cell wall component (90%, w/w) of Gram-positive bacteria and consists of N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) disaccharide repeating arrays that are cross-linked by short peptides. We hypothesized that PGN is a ligand for pathogen-associated pattern-recognition proteins. Mannose-binding lectin (MBL) and serum amyloid component P are two carbohydrate-binding innate immune proteins present in the blood. In this study we show that human MBL, but not serum amyloid component P, binds significantly to PGN via its C-type lectin domains, and that the interaction can be more effectively competed by GlcNAc than by MurNAc. Surface plasmon resonance analyses show that native MBL binds immobilized PGN with high avidity. Competition experiments also show that both native MBL and MBL(n/CRD), a 48-kDa recombinant trimeric fragment of MBL containing neck and carbohydrate recognition domains, have higher affinity for GlcNAc than for MurNAc. Protein arrays and ELISA show that PGN increases the secretion of TNF-α, IL-8, IL-10, MCP-2, and RANTES from PMA-stimulated human monocytic U937 cells. Interestingly, the presence of MBL together with PGN increases the production of IL-8 and RANTES, but reduces that of TNF-α. Our results indicate that Gram-positive bacterial is a biologically relevant ligand for MBL, and that the collectin preferentially binds to the GlcNAc moiety of the PGN via its C-type lectin domains. MBL inhibits PGN-induced production of proinflammatory cytokines while enhancing the production of chemokines by macrophages, which suggests that MBL may down-regulate macrophage-mediated inflammation while enhancing phagocyte recruitment.
Staphylococcus aureus infects humans (1) and often causes septic shock and multiorgan failure (2). The cell wall of Gram-positive bacteria, including that of S. aureus, has a multilayered three-dimensional matrix of glycan shells. The major cell wall component (∼90%, w/w) of these bacteria is peptidoglycan (PGN), 3 which consists of repeating arrays of disaccharides with β-1,4-linked N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc). An approximately four- to six-amino acid-long peptide chain, containing primarily Glu, Ala, and Lys residues, is covalently attached to the MurNAc moiety at carbon 3, whereas the GlcNAc end of the disaccharide is either free or forms a β-1,4 glycosidic bond with the MurNAc of the adjacent disaccharide unit. The basic muramyl peptide units are cross-linked by short stretches of oligo-glycine peptides (3). Host enzymes, such as lysozyme, hydrolyze the glycosidic bonds of this insoluble PGN (iPGN) and release low m.w. soluble PGN (sPGN). Lipoteichoic acid (LTA) is the second most abundant molecule (∼10%, w/w) present on the surface of Gram-positive bacteria. This molecule is a single-chain polymer made of phosphate-linked repeating units of alcohols such as glycerol or ribitol and anchored to the plasma membrane by its acyl chains (3). Frequently, carbohydrate moieties such as glucose, GlcNAc, small amino acids such as Ala, or short peptides are linked to the alcohol backbone. Although LTA can induce certain disease symptoms, PGN reproduces most of the clinical manifestations of Gram-positive bacterial infections, including fever, acute phase response, inflammation, septic shock, leukocytosis, sleepiness, malaise, abscess formation, and arthritis (3). Most of these effects are due to the PGN-induced release of cytokines and other inflammatory mediators from macrophages and other immune cells. The pathogen-associated molecular pattern of PGN is composed of an array of carbohydrates, which could act as an ideal ligand for pattern recognition proteins.
Several studies have shown that innate immune collectins avidly bind Gram-positive bacteria (3, 4, 5); however, the precise molecular targets on the bacterial surface have not been clearly defined. Whether pentraxins recognize Gram-positive bacteria, however, has not been thoroughly studied. We hypothesized that PGN could be one of the important Gram-positive bacterial surface ligands for these innate immune proteins. The major innate immune proteins present in the blood are the collectin, mannose-binding lectin (MBL), and the pentraxins, C-reactive protein and serum amyloid component P (SAP).
MBL belongs to the collectin family of proteins and is composed of subunits made of three identical polypeptide chains. In each trimer, polypeptide chains are joined at the N-terminal by interchain disulfide bonds (6) and also by the triple helical collagen-like regions (7, 8) that contain Gly-X-Y repeats (where X is any amino acid, and Y is often hydroxyproline or hydroxylysine). Three amphipathic α-helical neck domains (9) assemble as coiled-coil bundles and hold together three globular carbohydrate recognition domains (CRDs) (9, 10, 11). The trimeric subunits of the MBL further associate into high order oligomers consisting of two to six units (12) resembling a “bouquet of tulips,” similar to that of the lung-restricted collectin, surfactant-associated protein A (SP-A) (13), and a noncollectin, complement protein C1q (14). Although the trimeric subunits of the collectins have limited affinity (micromolar) for lipid (15, 16) and several carbohydrate targets (17, 18), their oligomeric assembly provides a high avidity (7, 19, 20) so that these proteins bind to ligands selectively and with high affinity (nanomolar to picomolar).
Pentraxins are also oligomeric lectins, but they have neither collagen-like regions nor the typical C-type lectin domains (21, 22). Binding of pentraxins to carbohydrates such as mannose and other ligands, however, is dependent on the presence of divalent cations (21, 23). Pentraxins are composed of five identical subunits that assemble noncovalently to form stable oligomers (22). SAP pentamers may stack to form decamers (24, 25). This high order assembly enables pentraxins to bind several microbial ligands with high avidity (nanomolar). However, whether pentraxins bind to the PGN of the Gram-positive bacteria is unknown.
Leukocytes recognize bacterial components such as PGN and amplify the immune response by expressing and secreting inflammatory cytokines at the site of infection (26, 27). The proinflammatory cytokines, including TNF-α, IL-1, and IL-6, help to increase the tissue permeability and recruit more phagocytes to the site of infection to effectively clear the bacterial infection. Chemokines such as MCP-1, MCP-2, IL-8, and RANTES help to defend the host against the infection by recruiting more phagocytes. Overproduction of proinflammatory cytokines, however, can result in septic shock and multiple organ failure (28, 29). MBL (30, 31) could either enhance or suppress (32) the amplification of inflammatory signals mediated by a variety of microbes (33). The effects of MBL on PGN-mediated induction of inflammatory cytokines by leukocytes, however, were unknown.
We have studied the interactions of two blood innate immune proteins, MBL and SAP, with PGN. In this study we show that MBL, but not SAP, binds significantly to PGN with high avidity. We found that MBL binds to PGN via its CRDs and typical C-type lectin-carbohydrate interactions, and that the GlcNAc moiety of the PGN is the preferred target for MBL compared with MurNAc. We also show that MBL inhibits PGN-mediated inflammatory cytokine secretion, but increases chemokine production by cultured U937 human monocytic cells. These findings establish a biologically relevant microbial ligand for MBL and suggest a potential mechanism for regulation of the immune response against Gram-positive bacterial infections by macrophages.
Materials and Methods
All chemicals and buffer salts were purchased from Sigma-Aldrich/FLUKA unless otherwise stated. Both biotinylated and nonbiotinylated mAbs to human MBL were purchased from Cambridge Biosciences. N-acetyl-d-glucosaminyl-(β1,4)-N-acetylmuramyl-l-alanyl-d-isoglu-tamine (GMDP) was purchased from Calbiochem.
Isolation and purification of sPGN from S. aureus
The soluble form of the PGN was isolated from liquid bacterial cultures as described previously (34). Briefly, S. aureus (RN4220) was grown in Todd-Hewitt broth medium at 37°C for 18 h, and 25 ml of this culture (A660, ∼0.1) was inoculated to 2.5 l of the same medium. The bacterial culture was further grown for ∼2 h at 37°C with vigorous shaking. At the end of this growth period, the A660 was ∼0.8–0.9. Bacterial cells were sedimented from the culture medium by centrifugation (1500 × g, 37°C, 15 min) and washed with warm PBS containing 28.5 mM glucose. These cells were grown in 1 l minimal medium supplemented with vitamins, 1 mM MgCl2, 0.1 mM MnCl2, benzyl-penicillin (75 μg/ml), and 10 μCi d,l-[14C]alanine (Amersham Biosciences) at 37°C for 1 h with vigorous shaking. The culture was boiled (95–100°C) for 30 min with frequent mixing and chilled to 4°C on ice water, and the sPGN-rich supernatant was separated from the debris by centrifugation at 6500 × g at 4°C for 20 min and filtered through a 0.2-μm membrane filter (Millipore).
The supernatant was concentrated to ∼80–100 ml by a heated (75°C) rotary evaporator (Büchi-461; Rotavapor). The concentrated sample was dialyzed (14,000 Da Mr cut-off; Spectrum) at 4°C, three times against 2 l of ddH2O and once against 2 l of PBS (pH 7.2). The dialyzed supernatant was sequentially passed through two columns: first through a 40-ml Sepharose 4B and then through a 40-ml vancomycin-Sepharose column. After disconnecting the Sepharose column, the vancomycin-Sepharose column was washed extensively with PBS. The sPGN that bound to the vancomycin affinity matrix was eluted by a 200-ml NH4OH pH gradient (7.5–10). The sPGN-containing fractions (2 ml each) were determined by monitoring the presence of 14C (Fig. 1,A), pooled, and concentrated by a 10,000-Da Mr cut-off centrifugation filter (Millipore) at 1500 × g for ∼20 min. The concentrated sPGN was additionally purified on a 1.6 × 30-cm TSK-HW 55 (F) (Fractogel; Merck) gel filtration column in 5 mM HEPES-NaOH (pH 7.4), 150 mM NaCl, and 5 mM EDTA buffer. The sPGN-containing fractions were determined by monitoring the presence of 14C as described above (Fig. 1,B), pooled, and stored at 4°C. The purity of the sPGN was determined by acid hydrolysis and HPLC. Amino acid analysis showed the presence of the four amino acids typically found in this polymer (Fig. 1 C). The concentration of sPGN was determined from the lysine content obtained from the amino acid analysis.
Removal of endotoxin
Endotoxin was removed from sPGN as described previously (34). Briefly, sPGN was mixed with cleaned polymyxin B-agarose, the mixture was incubated at 23°C for 30 min in a rotator and centrifuged at 1800 × g for 30 min, and the supernatant was filtered through 0.22-μm pore size filter to remove residual beads. The recovered endotoxin-free sPGN was stored at 4°C. Endotoxin levels in sPGN, iPGN, and MBL were determined using a Limulus amebocyte lysate assay kit (BioWhittaker). Because high Mr PGN would also be Limulus amebocyte lysate assay positive, endotoxin levels of iPGN were measured after digesting it with lysozyme as previously described (35). The endotoxin content of sPGN and iPGN preparations was <1 pg/mg (0.01 endotoxin unit (EU)/mg) and 1.8 ng/mg (18 EU/mg) of PGN, respectively. Endotoxin from native MBL preparations was also removed as described above (<200 pg/mg or <2 EU/mg protein).
Determination of sPGN concentration
The amino acid composition of an aliquot of endotoxin-free sPGN (50 μl) was determined by acid hydrolysis and HPLC. The sPGN contained a 1:3.7:4.3:1.1 molar ratio of Lys:Ala:Gly:Glu (Fig. 1). This ratio is approximately consistent with the composition of the purified PGN of S. aureus with (l-Ala)-Glu-Lys-(d-Ala)-(d-Ala) with Gly4 linkers (34). The concentration of the sPGN preparation was determined based on the molarity of Lys. Approximately 2.5–3.0 mg of sPGN/l medium was purified by this procedure. To obtain a 10 mg/ml stock solution of iPGN, the powdered form of the ligand (10 mg; Sigma-Aldrich) was dissolved in 1 ml of ddH2O, vortexed, and sonicated for 5 min.
Purification of human MBL and SAP
MBL and SAP were purified from human blood plasma based on a previously described procedure (36). Briefly, proteins in 500 ml of pooled plasma (HD Supplies) were selectively precipitated by polyethylene glycol (P3350; Sigma-Aldrich), dissolved in high salt, calcium-containing buffer (50 mM Tris (pH 7.8), 1 M NaCl, 20 mM CaCl2, and 0.05% (v/v) Tween 20). The redissolved proteins were sequentially passed through 25 ml of mannose-agarose, 5 ml of protein G (Amersham Biosciences), and 1 ml of MonoQ (Amersham Biosciences). In this purification procedure, MBL and SAP separated away from each other during ion exchange chromatography on a MonoQ column and were additionally purified by gel filtration chromatography on a Superose 6 column (1.6 × 30 cm). The identities of the purified MBL and SAP were confirmed by N-terminal amino acid sequencing (data not shown), and their concentrations were determined by amino acid analyses. These proteins migrated with their expected molecular mass under reducing and nonreducing conditions in denaturing SDS-PAGE (Fig. 2 A).
Cloning, expression, and purification of recombinant MBL(n/CRD)
A DNA fragment encoding residues Ser100-Ile248 of human MBL was amplified by PCR and ligated into a unique BamHI-EcoRI site of a modified pPIC9K vector (Invitrogen Life Technologies). Correct insertion was confirmed by automated DNA sequencing. The human MBL containing the neck and CRD, MBL(n/CRD), was expressed in the methylotropic yeast Pichia pastoris strain GS115 (Invitrogen Life Technologies) following the manufacturer’s protocol. Typically, four 1-l yeast cultures were grown in buffered glycerol-complex medium (Pichia medium) for 3 days at 30°C, and cells were harvested by centrifugation at 1000 × g for 5 min and resuspended in the same medium without glycerol, but containing 0.05% (v/v) methanol. The culture was incubated for an additional 3 days in 400 ml of medium to induce recombinant protein expression. The recombinant MBL(n/CRD)s of ∼16 kDa spontaneously formed ∼48-kDa noncovalent trimers. The culture supernatant was obtained by centrifugation (1000 × g for 5 min), adjusted to contain 10 mM EDTA/50 mM sodium citrate, and adjusted to pH 3.5 with 50 mM citric acid. This was mixed with 20 ml of SP-Sepharose (Amersham Biosciences) for 1 h at 23°C, packed into a column, and washed with 50 mM sodium citrate/citric acid (pH 3.5), and the recombinant protein was eluted with a 20-ml gradient from 0 to 1 M NaCl in the same buffer. The eluate containing MBL(n/CRD) was dialyzed in 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 10 mM CaCl2; loaded into a 20-ml mannan-Sepharose column; and extensively washed with Tris-saline-calcium buffer. Proteins that bound to the matrix were eluted with 20 mM Tris-HCl, 150 mM NaCl, and 5 mM EDTA. The eluted recombinant protein was purified using Superdex 75 (Amersham Biosciences) gel filtration in Tris-saline-EDTA buffer. On gel filtration, the protein eluted with an apparent Mr of ∼50 kDa, suggesting that the protein formed trimers. The samples were stored at 4°C until further analyses. The protein concentration was determined by amino acid composition analysis.
SDS-PAGE and Western blotting
Purified proteins were separated by SDS-PAGE (4–20% (w/v) gradient gel; Bio-Rad) under denaturing conditions. Samples were boiled for 5 min in sample loading buffer (50 mM Tris-HCl (pH 6.8), 2% (w/v) SDS, 0.1% (w/v) bromophenol blue, and 10% (v/v) glycerol) for nonreduced gels or in sample loading buffer with 5% (v/v) β-ME for reduced gels. The gels were stained with BIO-safe Coomassie G250 stain, or the proteins from the gels were transferred to an Immobilon-P (polyvinylidene difluoride), 0.45-μm pore size membrane using the semidry transfer method for Western blot analysis. Briefly, for Western blotting, the gels were equilibrated in semidry blotting buffer (49 mM Tris, 39 mM glycine, (pH 8.3), and 0.0375% (w/v) SDS) for 30 min. The gel and the Immobilon-P membrane, which was presoaked in methanol, were stacked in between 12 and 16 sheets of 3M chromatography papers (Whatman), which were presoaked in semidry blotting buffer. The stacked gel-membrane sandwich was carefully placed in a flat-bottom electrode of a semidry apparatus and secured with a lid. The proteins were transferred to the membrane using 0.8 mA/cm2 for 1.5 h with unlimited voltage, then the membrane with the protein was blocked with 3% (w/v) BSA in PBS with 0.02% (v/v) Tween 20 (PBST) overnight at 4°C. The membrane was washed three times for 15 min each time with PBST, then the membrane was incubated with biotinylated primary mAb (Hycult Biotechnology) against human MBL in PBST for 1.5 h at room temperature. The membrane was washed three times for 15 min each time with PBST, and the membranes were incubated with HRP-linked streptavidin in PBST for 25 min at room temperature and washed with PBST three or four times as described above. The protein-Ab complexes were detected by an ECL-Western blotting analysis system (Amersham Biosciences).
The presence of MBL in the eluates and supernatants was detected by sandwich ELISA. Anti-human MBL Ab raised in rabbits (100 μl, 1 μg/ml) was immobilized overnight at 4°C in the wells of 96-well microtiter plates (Nunc Maxi-Sorp Immuno Plates) in 0.1 M sodium carbonate buffer (pH 9.6). After washing with 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM CaCl2, and 0.02% (v/v) Tween 20 (TBSCT), wells were blocked with 5% (w/v) BSA for 1 h at 37°C and washed again with TBSCT. Wells were then incubated with eluate or supernatants, obtained from MBL-PGN binding experiments, for 1 h at 37°C. After washing, wells were incubated with biotinylated anti-human MBL Ab raised in rabbits (100 μl, 0.2 μg/ml) for 1 h at 37°C and washed three times in TBSCT. The wells were incubated with streptavidin-HRP conjugate (1/10,000) for 20 min at 37°C and washed three times with TBSCT. Finally, the MBL-biotinylated Ab complexes were detected by an H2O2-tetramethylbenzidine-based chromogenic substance (Bio-Rad), according to the manufacturer’s instructions. Absorbance was measured at 450 nm in a spectrophotometer (Multiscan Ascent; Labsystems).
To determine the TNF-α present in the U937 cell culture supernatant, a Quantikine ELISA kit (R&D Systems) was used. Monoclonal anti-human TNF-α (100 μl, 4 μg/ml) was immobilized overnight at 4°C in the wells of 96-well microtiter plates (Nunc Maxi-Sorp Immuno Plates) in PBS buffer (pH 7.2). After washing with PBST buffer, wells were blocked with 5% (w/v) BSA for 1 h at 37°C and washed again with PBST. Wells were then incubated with cell culture sample or standard in PBST for 2 h at room temperature. After washing, wells were incubated with biotinylated goat anti-human TNF-α Ab (100 μl, 300 ng/ml) for 2 h at room temperature and washed three times in PBST. The wells were incubated with streptavidin-HRP conjugate (1/200) for 20 min at room temperature and washed three times with PBST. Finally, the TNF-α-biotinylated Ab complexes were detected by H2O2-tetramethylbenzidine-based chromogenic substance, according to the manufacturer’s instructions. Absorbance was measured at 450 nm in a spectrophotometer (Multiscan Ascent; Labsystems). A standard curve was fitted, and the TNF-α concentrations in the cell culture supernatants were calculated from the standard curve.
Surface plasmon resonance (SPR) analyses
The sPGN (250 μg/ml) was dialyzed in 100 mM NaOAc (pH 4.0) buffer and injected over a National Health Service/N-hydroxysuccinimide/N,N-(3-dimethylaminoproply)-N′-ethyl-carbodimide hydrochloride-activated CM5 BIAcore chip at 5 μl/min for 10 min. Nonreacted sites present on the chip were neutralized by 1 M ethanolamine. The blank flow cell was treated in the same way, except that 100 mM NaOAc (pH 4.0) buffer was used instead of sPGN. All flow cells were normalized with 50% (v/v) glycerol before the first use.
Yeast mannan was biotinylated and immobilized on a streptavidin chip (BIAcore 2000) as described previously (37). Briefly, mannan (1 mg/ml) was reacted with 1.5 mM NaIO4 in 100 mM sodium acetate buffer (pH 5.5) for 30 min on ice, and the reaction was stopped with 15 mM glycerol. The mixture was dialyzed in 100 mM sodium acetate buffer (pH 5.5) for 18 h at 4°C and reacted with 5 mM biotin-LC-hydrazide (Pierce) for 2 h at 23°C in the same buffer. The mixture was dialyzed in 100 mM sodium acetate buffer (pH 5.5) for 18 h at 4°C to remove nonreacted biotin and was stored at the same temperature. The biotinylated mannan in 100 mM sodium acetate (pH 5.5) buffer was immobilized on individual streptavidin-coated chips by injecting it at 5 μl/min for 10 min, leaving flow cell 1 as a blank. Free streptavidin in all flow cells was blocked with biotin, and the chip was normalized with 50% (v/v) glycerol solution.
The binding of MBL (0–50 μg/ml) and MBL(n/CRD) (0–600 μg/ml) to the biotinylated mannan was analyzed in 10 mM HEPES-NaOH (pH 7.4), 5 mM CaCl2, 150 mM NaCl, 0.02% (w/v) NaN3, and 0.005% (v/v) surfactant P20 buffer at a flow rate of 10 μl/min. In competition assays, MBL (1.25 μg/ml) or MBL(n/CRD) (100 μg/ml) was allowed to bind (10 μl/min) to biotinylated mannan that was immobilized on a streptavidin chip in the presence of different concentrations of competitors (0–40 mM). The SPR response obtained in the absence of any competitor was considered to be 100%, and the relative binding values were calculated for each concentration of competitor.
U937 monocytic cell culture
U937 cells were maintained in endotoxin-free RPMI 1640 with 10% (v/v) heat-inactivated FCS, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C in a 5% (v/v) CO2 environment. Actively growing cells from the flasks (Nunc) were harvested by centrifugation at 450 × g (1400 rpm) for 5 min and washed with RPMI 1640 before the experiments. Cells were seeded in 96-well plates (5 × 105 cells/well) in the presence of 10 nM PMA and incubated overnight at 37°C in a 5% (v/v) CO2 environment. The plates were centrifuged at 450 × g for 5 min, and unattached cells were removed by aspiration. Cells were additionally incubated for 24 h in the absence of PMA in RPMI 1640 with 10% (v/v) heat-inactivated FCS. The plates were centrifuged at 450 × g for 5 min, and unattached cells were removed by aspiration. The iPGN (1 mg/ml) was preincubated with MBL (0, 5, or 10 μg/ml) in 250 μl of RPMI 1640 at 37°C for 1 h and washed twice with the same medium. Cells were incubated with medium containing iPGN (125 μg/ml) and a final concentration of 0–1.25 μg/ml MBL in the presence or the absence of polymyxin B (25 μg/ml) for 5 h at 37°C in a 5% (v/v) CO2 environment. Supernatant from the wells was harvested after centrifuging the cells at 450 × g for 5 min and was stored at −80°C.
Tissue culture medium supernatant (100 μl) was incubated with protein array nitrocellulose membranes that contained 23 protein-specific Abs in duplicate (Ray Biotech). Bound proteins were detected by a mixture of biotinylated Abs and ECL reagents as recommended by the manufacturers (Ray Biotech). The signal intensities of the proteins were determined using National Institutes of Health image analyzer software (〈http://rsb.info.nih.gov/nih-image/〉).
The mean and SD or SE were calculated by Excel software (Microsoft). Student’s t test was used to calculate significance, and p < 0.05 was considered to represent a statistically significant difference between two sample means. ANOVA and Tukey-Kramer multiple mean comparisons (p < 0.05) were performed using JMP Statistical Discovery software (〈www.jmpdiscovery.com〉).
MBL, but not SAP, binds to iPGN
To determine whether MBL and SAP interact with PGN, we purified these two proteins from human serum (Fig. 2,A) and allowed them to bind to S. aureus iPGN. After incubating 500 μg of iPGN with 1 μg of MBL in the presence of 5 mM CaCl2 or 2 mM EDTA for 30 min at 37°C, the mixtures were washed in the binding buffer, and bound proteins were released by 5 mM EDTA elution. MBL was detected in eluate and supernatant by Western blotting. Polyclonal Ab against MBL specifically recognized an ∼30-kDa protein, which is consistent with the Mr of MBL under reducing and denaturing conditions. This Ab did not cross-react with any other component present in the blot. This blotting result showed that MBL specifically bound to iPGN in the presence of calcium ions, but not in EDTA (Fig. 2 B).
Interaction between iPGN and SAP was also tested as described above for MBL in the presence of calcium ions and EDTA. The presence of SAP was detected in the pellet and supernatant by Coomassie Blue staining of SDS-PAGE (4–20% (w/v) gradient gel). A small amount of SAP (<10%) was detected in the iPGN pellet in the presence of calcium (Fig. 2,C, lanes 2 and 3), but not EDTA (Fig. 2,C, lanes 4 and 5). Therefore, we consider that SAP does not bind to iPGN to any significant degree (Fig. 2,C). These results (Fig. 2, B and C) established that MBL, but not SAP, binds to iPGN efficiently.
MBL binds to iPGN in a calcium-dependent manner
MBL is a C-type lectin and can bind to carbohydrates only in the presence of calcium ions (18, 38). Therefore, to determine the cation requirement of the MBL-iPGN interactions, we conducted binding experiments in the presence of 5 mM concentrations of MgCl2, MnCl2, ZnCl2, CaCl2, or EDTA. The amounts of MBL present in the PGN pellets and supernatants were determined by sandwich ELISA. The results showed that MBL bound to iPGN only in the presence of CaCl2, and none of the other divalent cations tested could substitute for calcium ions (Fig. 3 A). These results suggest that the MBL probably binds PGN via its C-type lectin domains.
Binding of MBL to iPGN is inhibitable by the saccharide components of PGN
PGN contains repeating GlcNAc-MurNAc disaccharide moieties and cross-linking short peptides (3, 28). To determine which carbohydrate moiety of PGN interacts with MBL, we incubated MBL (1 μg) with iPGN (500 μg) in the presence of 20 mM mannose, GlcNAc, or MurNac. MBL present in the PGN pellets was released by 5 mM EDTA elution and compared with the MBL in the supernatants. The amount of MBL that bound to iPGN was determined by sandwich ELISA (Fig. 3,B). Only one of the two saccharide moieties present in PGN, the GlcNAc, but not MurNAc, competed the interaction between MBL and iPGN. Mannose was used as a positive control to identify the presence of C-type lectin-carbohydrate interactions (18), and it also inhibited the MBL-iPGN interaction. Therefore, our experimental set-up worked as expected (Fig. 3 B). These results are consistent with the interpretation that MBL preferentially binds the GlcNAc moiety of iPGN.
The sPGN and its molecular components inhibit the interaction between MBL and mannan
To determine whether MBL could bind to the free form of sPGN and its subcomponents, we conducted competition experiments by SPR. MBL is known to bind to mannan with high affinity (18, 38, 39); hence, we immobilized biotinylated mannan onto a streptavidin chip and allowed MBL to bind to this ligand. SPR analyses showed that MBL bound to mannan with high affinity (Fig. 4,A). Soluble PGN effectively competed the interaction between MBL and mannan in a concentration-dependent manner (Fig. 4,B). Similar experiments were conducted with components of PGN as competitors. These results were converted to percent inhibition, considering the binding of MBL to mannan in the absence of any competitor as 100%. In competition experiments, mannose effectively competed the interaction between MBL and mannan, whereas glucose and related saccharides poorly inhibited the binding (17, 38). Our results showed similar competitive abilities with mannose, glucose, and maltose as previously observed (Fig. 4,B), suggesting that our BIAcore system behaved similarly to that of the conventional solid phase ELISA. When comparing the saccharides present in PGN, GlcNAc inhibited the interaction between MBL and mannan better than did MurNAc. Peptide components of the PGN, however, did not inhibit the binding between MBL and mannan (Fig. 4 B). These results show that MBL binds to sPGN in solution, and that it preferentially binds to the GlcNAc moiety of the ligand.
The sPGN and its molecular components inhibit the interaction between recombinant MBL(n/CRD) and mannan
To confirm that the globular domain of MBL can bind to these ligands in a similar manner to that of the native protein, we conducted similar competition experiments. MBL(n/CRD) fragments bound to immobilized mannan in a concentration-dependent manner (Fig. 4,C). Almost all the bound recombinant protein molecules, however, disassociated from the ligand immediately after the end of injection, indicating that the affinity of the interaction was low (Fig. 4,C). We also studied the interaction between MBL(n/CRD) (100 μg/ml) and mannan in the presence of different concentrations of competitors. The sPGN effectively inhibited the interaction between MBL(n/CRD) and mannan, indicating that the recombinant protein fragment bound to the ligand in solution (Fig. 4,D). Individual carbohydrate components, mannose and maltose, inhibited the interactions to high and low degrees, respectively. These results confirmed that recombinant MBL(n/CRD) had the appropriate carbohydrate-binding characteristics expected for MBL (17, 38). A saccharide component of PGN, GlcNAc, inhibited the interaction between MBL(n/CRD) and mannan very effectively. MurNAc and MurNAc with two amino acids, Ala-Gln, were, however, poor competitors, and the inhibition occurred only after reaching specific concentrations of these compounds (5–10 mM). The relative inhibitory abilities of saccharides determined in this experiment are similar to that of the competition effect seen in native MBL-mannan interaction (Fig. 4, A and B). These results therefore indicate that MBL preferentially binds to the GlcNAc moiety of PGN via its CRDs.
MBL binds immobilized sPGN with high affinity
Bacterial cell wall PGN is digested by host lysozyme and released into the surrounding environment as sPGN fragments (27). Hence, to determine whether MBL could bind to soluble PGN, we studied the interaction between these molecules by SPR response analyses. We first grew S. aureus in the presence of penicillin, which prevents transpeptide bond formation, and isolated sPGN by vancomycin-agarose affinity and gel filtration chromatographies. We immobilized this PGN onto a BIAcore CM5 sensor chip and allowed MBL to bind to the ligand in the presence of 5 mM CaCl2. MBL avidly bound to PGN in a concentration-dependent manner (Fig. 5,A). The SPR responses also showed that MBL accumulated on the ligand-coated surface during protein injection in a time-dependent manner (100–340 s). Most of the MBL that bound to PGN, however, did not disassociate after stopping the protein injection (340–400 s). This type of SPR response indicates that MBL binds PGN with very high affinity. Because native MBL is an oligomeric protein with a different number of subunits (two to six trimers), calculation of its affinity for sPGN from the SPR data would not be accurate. SPR data, however, were analyzed based on a simple model of A + B = AB, where A was the fully assembled MBL (18 mer) and B was the PGN, to obtain an approximate affinity value. These calculations indicated that native MBL bound to PGN with nanomolar-picomolar affinity (Kd = 0.1–3.2 nM). Although MBL(n/CRD) bound to PGN, it quickly disassociated from the ligand (Fig. 5 B). Scatchard plot analysis of the SPR response of MBL(n/CRD) binding to sPGN was performed with the assumption that trimers behave as a single unit. The calculations showed that MBL(n/CRD) interacted with immobilized sPGN with an affinity of 0.71 μM (Kd). This type of interaction suggests that polyvalent binding is important for stable interaction between MBL and PGN.
MBL reduces PGN-induced TNF-α production
MBL could either increase or decrease the inflammatory cytokine production by leukocytes in response to microbial infections (30, 31, 40, 41). Studies in knockout mouse models suggested that MBL was necessary for down-regulating TNF-α and IL-6 in the blood and preventing septic shock (42), but the role of PGN in these processes was unknown. To determine the biological effect of MBL binding to PGN, we first studied the effect of MBL on TNF-α production by U937 human monocytic cells. We stimulated U937 cells with PMA, added different amounts of iPGN, and determined the amount of TNF-α released into the supernatant by ELISA. Because polymyxin B sulfate has been used to eliminate the effect of endotoxin in similar cell assays (33), we incubated different amounts of iPGN in the culture medium in the presence of 25 μg/ml polymyxin B sulfate. The iPGN induced TNF-α production by stimulated U937 cells in a concentration-dependent manner (Fig. 6,A). The presence of polymyxin B sulfate (25 μg/ml) reduced the amount of TNF-α produced by the U937 cells, particularly when high concentrations of the ligand were used (Fig. 6,A). We titrated the effect of polymyxin B sulfate (0–200 μg/ml) on TNF-α production by PMA-treated monocytes in the presence of 125 μg/ml iPGN. A concentration of 25 μg/ml polymyxin B sulfate was sufficient to eliminate TNF-α production induced by the traces of endotoxin present in this amount of PGN. Increasing the polymyxin B sulfate concentration beyond 25 μg/ml did not change TNF-α production by these cells under these conditions (Fig. 6 B). Therefore, we maintained this condition (125 μg/ml iPGN and 25 μg/ml polymyxin B sulfate) in subsequent experiments. We interpret this result to mean that polymyxin B sulfate removed contaminating endotoxin, as previously reported by others (43). These results therefore show that iPGN induces TNF-α production by PMA-stimulated U937 cells in a concentration-dependent manner.
To study the effect of MBL on TNF-α production by U937 cells, we preincubated iPGN (1 mg/ml) and MBL (0, 5, and 10 μg/ml) in a small volume of the culture medium, washed the complexes, diluted the iPGN-MBL complexes to a final concentrations of 125 μg/ml iPGN and 0, 0.625, or 1.25 μg/ml MBL, and incubated them with PMA-stimulated cells. TNF-α concentrations in the U937 cells culture supernatant were determined by ELISA, as described above. MBL suppressed PGN-mediated TNF-α production by these monocytic cells in a concentration-dependent manner (Fig. 6 C). The endotoxin-free MBL (5 μg/ml; <1 fg/ml endotoxin) did not induce TNF-α production by PMA-stimulated monocytic cells (data not shown). These results, therefore, show that MBL suppresses the inflammatory signal of the leukocytes in response to Gram-positive bacterial PGN.
MBL suppresses PGN-induced proinflammatory cytokine TNF-α and enhances chemokines IL-8 and RANTES
To understand the signaling pathways regulated by MBL, we incubated the PMA-treated U937 cells with iPGN (125 μg/ml) in the absence or the presence of MBL (0.5–1 μg/ml) and determined the changes in 23 protein markers, including proinflammatory and anti-inflammatory cytokines, and different classes of chemokines by protein arrays. Consistent with the data shown in Fig. 6, the presence of 25 μg/ml polymyxin B did not change the cytokine profile (data not shown). These macrophages produced only one proinflammatory cytokine, TNF-α, but not IL-1, IL-6, or IL-12, in response to iPGN (Fig. 7). This cytokine profile indicated that PGN did not enhance the production of multiple proinflammatory cytokines, but specifically enhanced TNF-α production. MBL suppressed PGN-induced TNF-α production, and it was consistent with our previous experiment (Fig. 6). Furthermore, in response to PGN, these macrophages secreted more IL-8, an important ELR+CXC class chemokines that acts as a chemoattractant for neutrophils, and MBL enhanced its production (Fig. 7). These macrophages also secreted more CC class chemokines, such as MCP-2 and RANTES, in response to PGN, and MBL further enhanced the levels of RANTES. This type of signaling suggests that MBL is a molecule that could enhance the recruitment of T cells, monocytes, macrophages, dendritic cells, and other phagocytes in response to PGN.
Macrophages produced only one anti-inflammatory cytokine, IL-10, in the presence of PGN. The IL-10 level was slightly, but not statistically significantly, increased in the presence of MBL (1 μg/ml). Taken together, these results indicate that MBL effectively suppresses the activation and inflammatory state of the macrophages and maintains their phagocytic abilities.
PGN, the major cell wall component of Gram-positive bacteria, forms a matrix that is made of repeating arrays of disaccharide GlcNAc-MurNAc and short peptides. This insoluble polymer is digested by host lysozyme and released as sPGN fragments at the site of infection. We determined that MBL, but not SAP, binds both the insoluble and soluble forms of PGN with high affinity. We also showed that MBL binds to both of these PGNs via its CRDs by typical C-type lectin-carbohydrate interactions. Our results also show that MBL efficiently and preferentially recognizes the GlcNAc moiety present in PGN rather than MurNAc. Furthermore, MBL suppresses PGN-mediated TNF-α production by PMA-treated human macrophage cells. MBL also increases the production of chemokines by these cells. These findings establish that PGN is a biologically relevant ligand for MBL, and that the innate immune collectin helps to recruit phagocytes to clear the ligand and bacteria.
Although the binding of MBL to bacteria has been known for several decades (5, 44), only the interaction between this collectin and the Gram-negative bacterial ligand LPS has been studied previously. In this study we considered the interaction of MBL and SAP with Gram-positive bacteria. MBL is known to bind to several clinically relevant Gram-positive bacteria (5, 45), and the relative degrees of its binding to these bacteria appear to be related to the structure of LTA (46). Although the constituents that make LTA vary considerably among bacterial strains, PGN structure is well conserved among many Gram-positive bacteria (3, 46). We, therefore, considered that PGN is a general ligand for pattern recognition lectins (Figs. 1 and 2).
Our SDS-PAGE and Western blot analyses show that MBL, but not SAP, binds iPGN effectively (Fig. 2). Hence, the collectin MBL is one of the important acute phase proteins that could recognize Gram-positive bacteria via the cell wall PGN. Solid phase determination of carbohydrate specificities suggest that MBL (17, 38, 39), but not SAP (23), binds GlcNAc with high affinity. SAP preferentially binds to mannose and galactose pyruvate acetyl moieties (23). Our finding suggests that neither GlcNAc (23) nor MurNAc is a good ligand for SAP.
Because MBL bound to PGN (Fig. 2), we considered this cell wall component, a GlcNAc-containing macromolecule, as a biologically relevant ligand for MBL, and analyzed the MBL-PGN interactions in detail. Calcium ion requirement (Fig. 3,A), competition experiments with saccharide ligands in an ELISA set-up (Fig. 3,B), and SPR experiments (Figs. 4 and 5) suggest that MBL preferentially recognizes the GlcNAc moiety of the PGN via the typical C-type lectin-carbohydrate interactions by its CRDs. Our finding, therefore, indicates that the GlcNAc present on PGN is the biologically relevant ligand for MBL.
Although PGN, present on the bacterial cell wall surface, is a highly cross-linked insoluble compound, digested fragments of this ligand enter into the surrounding environment during an infection (27). Cosedimentation studies show that MBL effectively binds iPGN (Figs. 2 and 3), whereas SPR analyses show that MBL recognizes both free and immobilized forms of sPGN ligand effectively (Figs. 4 and 5). Furthermore, both iPGN and sPGN interact with MBL via similar mechanisms and are ligands for MBL. Although the trimeric MBL(n/CRD) domains do not bind to carbohydrate ligands such as mannan (Fig. 4, C and D) and PGN (Fig. 5,B) with sufficient avidity, native oligomeric MBL binds well to these ligands (Figs. 4, A and B, and 5F4 A). These results emphasize the necessity for multivalent CRD-ligand interactions for an efficient recognition of arrays of carbohydrate ligands by MBL. Although SP-A is similar to MBL in its structure and carbohydrate binding affinities, it does not bind PGN (33, 47). SP-D, however, has different carbohydrate binding affinities than MBL and does not recognize GlcNAc with high affinity (48). One report shows that SP-D can recognize iPGN (47); however, signaling mechanisms mediated by SP-D-PGN interactions and its biological roles are unclear.
To understand the biological role of MBL-PGN interactions, we examined the effect of MBL on PGN-mediated cytokine and chemokine production by stimulated U937 cells. We show that PGN induces the production of TNF-α by these cells (Figs. 6, 7); this is consistent with previous studies (26, 33). A recent study showed that MBL null mice die of septic shock in response to the S. aureus infection due to the elevated levels of TNF-α and IL-6 in their blood. Our results show that PGN up-regulates only TNF-α, but not IL-6 or any other proinflammatory cytokines, by macrophages. This result suggests that PGN alone contributes to the synthesis of only one type of proinflammatory cytokine, whereas other components may be responsible for IL-6 production. Interestingly, MBL inhibited the induction of this cytokine by PGN. Murakami et al. (33) showed that SP-A does not bind to PGN, but it inhibits PGN-mediated inflammatory signaling via directly interacting with TLR-2. This SP-A-mediated event appears to inhibit the signal transduction and subsequent TNF-α production by U937 cells (33). Hence, different collectins, MBL (Figs. 6 and 7) and SP-A (33), may inhibit TNF-α production and proinflammatory signals by macrophages via different mechanisms.
SP-A, SP-D, and MBL are known to bind to CD14, a receptor that recognizes LPS and PGN (49), via different mechanisms. SP-D (50) and MBL (51) bind to CD14 via their C-type lectin domains, but SP-A (50) binds to this receptor via protein-protein interactions, and hence, complex biological interactions may occur under different conditions. Furthermore, MBL and SP-D (47), but not SP-A (33), appear to bind both PGN and LTA. Modulation of bacterial clearance, cytokine response, and septic shock are dependent on the synergistic effects of PGN and LTA (28, 29). Hence, recognition and regulation of both PGN and LTA levels may be critical to maintain a nondetrimental cytokine profile.
In summary, we have identified that MBL binds to soluble and insoluble PGN effectively. The PGN binding of MBL is mediated by the CRDs and their C-type lectin interactions with carbohydrates. Furthermore, MBL preferentially binds to the GlcNAc moiety of PGN, and our data show that MBL inhibits PGN-mediated TNF-α production and increases the secretion of chemokines by macrophages. These findings suggest that MBL may reduce PGN-mediated inflammation.
We thank Antony C. Willis for amino acid analysis of PGN and protein preparations, and Jackie Shaw for maintaining U937 cell cultures.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by European Union Contract QLK2 CT 2000 00325 and Medical Research Council, U.K., research grants (to J.N. and K.B.M.R.). P.N. is a recipient of postdoctoral fellowships from the Wellcome Trust, U.K., and the Canadian Institutes of Health Research/Canadian Lung Association.
Abbreviations used in this paper: PGN, peptidoglycan; C1q, complement component 1q; CRD, carbohydrate-recognition domain; EU, endotoxin unit; GlcNAc, N-acetylglucosamine; GMDP, N-acetyl-d-glucosaminyl-(β1,4)-N-acetylmuramyl-l-alanyl-d-isoglutamine; iPGN, insoluble PGN; LTA, lipoteichoic acid; MBL, mannose-binding lectin; MurNAc, N-acetylmuramic acid; PBST, PBS with 0.02% (v/v) Tween 20; SAP, serum amyloid component P; SP, surfactant-associated protein; sPGN, soluble PGN; SPR, surface plasmon resonance; TBSCT, TBS with calcium and Tween 20.