Recognition of Bacterial Capsular Polysaccharides and Lipopolysaccharides by the Macrophage Mannose Receptor*

The in vitro binding of the macrophage mannose receptor to a range of different bacterial polysaccharides was investigated. The receptor was shown to bind to purified capsular polysaccharides from Streptococcus pneumoniae and to the lipopolysaccharides, but not capsular polysaccharides, from Klebsiella pneumoniae. Binding was Ca2+-dependent and inhibitable withd-mannose. A fusion protein of the mannose receptor containing carbohydrate recognition domains 4–7 and a full-length soluble form of the mannose receptor containing all domains external to the transmembrane region both displayed very similar binding specificities toward bacterial polysaccharides, suggesting that domains 4–7 are sufficient for recognition of these structures. Surprisingly, no direct correlation could be made between polysaccharide structure and binding to the mannose receptor, suggesting that polysaccharide conformation may play an important role in recognition. The full-length soluble form of the mannose receptor was able to bind simultaneously both polysaccharide via the carbohydrate recognition domains and sulfated oligosaccharide via the cysteine-rich domain. The possible involvement of the mannose receptor, either cell surface or soluble, in the innate and adaptive immune responses to bacterial polysaccharides is discussed.

The macrophage mannose receptor (MR) 1 is considered an important molecule of innate immunity mediating the nonopsonic phagocytic uptake by macrophages (MØs) of a wide variety of microbes including yeast, fungi, protozoa, and bacteria (1). It is a type I integral membrane glycoprotein expressed on most tissue MØs, certain endothelial cells, and in vitro derived dendritic cells (DCs) (2)(3)(4). In addition to phagocytosis, the MR mediates the endocytosis of soluble glycoconjugates leading to enhanced uptake of ligands in both MØs and DCs (5).
The MR belongs to the Ca 2ϩ -dependent lectin family that bind sugars through their carbohydrate recognition domains (CRDs) (6). The MR itself has eight CRDs located on a single polypeptide chain. CRDs 4 and 5 of the MR are central to sugar binding, and at least one further CRD proximal to the membrane is necessary for the binding and uptake of multivalent ligands (7,8).
The MR displays a mannose-type binding specificity recognizing monosaccharides in the following order of affinity: D-Man ϭ L-Fuc Ͼ D-Glc ϭ D-GlcNAc Ͼ Ͼ D-Gal (6). The binding of a monosaccharide by the CRD requires the presence of free equatorial hydroxyl groups at both the C 3 and C 4 positions of hexoses and at the C 2 and C 3 positions of the 6-deoxyhexose, L-Fuc. With respect to microbial ligands, a high affinity of the MR for polysaccharides displaying numerous terminal D-Man residues, as exemplified by yeast mannan, is well documented. The recognition of other complex microbial polysaccharides, particularly those expressed by bacteria, however, has not been investigated in any depth.
In this study, we investigate the binding of the MR to the capsular polysaccharides (CPSs) from different capsular serotype strains of Streptococcus pneumoniae and to the CPSs and lipopolysaccharides (LPSs) from different capsular and O-serotype strains of Klebsiella pneumoniae. Both bacterial species are major opportunistic pathogens, and in both cases antibodies directed against their CPSs confer protection. Vaccines currently in use against S. pneumoniae are either composed of CPS alone or are glycoconjugates based on the conjugation of CPS or oligosaccharides derived from CPS to a protein carrier.
In addition to binding through CRDs, the MR also has a second distinct lectin activity located in the N-terminal cysteine-rich (CR) domain of the molecule, this time with specificity for host sulfated oligosaccharides (9,10). Sulfated ligands for the CR domain are present in secondary lymphoid tissues, where they are expressed on metallophilic marginal zone MØs in the spleen, on subcapsular sinus MØs in lymph nodes, and also on a subpopulation of DCs (11,12).
A soluble form of the MR (sMR), released by constitutive proteolytic cleavage of the membrane-bound form by an endogenous cell metalloprotease, is also present in serum (13).
The CPSs of S. pneumoniae and K. pneumoniae are high molecular weight polysaccharides, typically 10 6 daltons and greater, and are composed of linear or branched repeating units containing from two to seven monosaccharides. S. pneumoniae CPSs variously contain D-GlcA and D-GalA acid residues and may be substituted with O-acetyl, pyruvate acetal, and glycerol phosphate groups (14). Several resemble teichoic acids in con-taining ribitol phosphate-linked repeat units. They frequently contain aminosugars, which include different aminohexoses and also N-acetyl-D-mannuronic acid. D-Gal, D-Glc, and L-Rha are common components, whereas D-Man is not found in any of these structures. All CPS preparations from S. pneumoniae contain 1-10%, by weight, common cell wall polysaccharide (CW-PS), which is the teichoic acid of this organism.
CPSs from K. pneumoniae commonly contain D-Glc, D-Man, D-Gal, and L-Rha. L-Fuc is found in only a very few serotype strains (15). They all contain one hexuronic acid residue in each repeat unit, which is nearly always D-GlcA acid, and are frequently substituted with pyruvate acetal or O-acetyl groups. Unlike CPSs from S. pneumoniae, phosphate groups, and aminosugars are absent from these polysaccharides. The uptake of K. pneumoniae by alveolar MØs mediated by recognition of CPS via the MR has been reported (16).
In contrast to the CPS of K. pneumoniae, the LPS O-antigens of this species display very limited structural variation with only nine recognized O-serotypes. Three main serotypes consist of linear polymers of ␣-linked D-mannose residues (serotype 03) and linear polymers of D-galacto-pyranose residues (galactan II) or both D-galacto-pyranose and D-galacto-furanose residues (galactan I) (serotypes O1 and O2) (Table III) (15). Other K. pneumoniae LPS types lack O-antigen (rough LPS).
In this study, we investigate the interactions between bacterial polysaccharides and the MR in vitro using purified CPS and LPS and the following proteins: (a) CRDs 4 -7 of the MR fused to human IgG 1 Fc (CRD (4 -7) Fc), (b) the CR domain of the MR fused to human IgG 1 Fc (CR-Fc), and (c) the full-length sMR.

Preparation of Monoclonal Antibodies against the Mouse Mannose Receptor
Monoclonal antibody MR5D3 was generated by subcutaneous immunization of Fischer rats with 100 g of CRD (4 -7) Fc in complete Freund's adjuvant containing 10 mM Ca 2ϩ , followed after 14 days by a similar booster with incomplete Freund's adjuvant. A final boost of 50 g of CRD (4 -7) Fc in PBS with 10 mM Ca 2ϩ was given intraperitoneally 4 days prior to fusion. The splenic B cells were fused with the Y3 myeloma cell line as described in Ref. 23.
Hybridoma supernatants were screened for the presence of anti-MR antibodies in ELISA using CRD (4 -7) Fc-coated plates. Of several monoclonal cell lines that were obtained, MR5D3 was chosen for further application due to its performance in immunohistochemistry, fluorescence-activated cell sorting, enzyme-linked immunosorbent assays (ELISA), and immunoprecipitation. 2 MR5D3 was shown to be noninhibitory for the binding of the MR to carbohydrate ligands (data not shown).
The sMR was purified from culture supernatants of NIH 3T3-derived MR transfectants 2 by affinity chromatography using MR5D3 as follows. Transfected cells were grown in Dulbecco's modified Eagle's medium containing 10% (v/v) FCS, penicillin, streptomycin, and glutamine under the selection of G418 (400 g/ml). Once cells were confluent, this medium was replaced with the same medium containing 3% (v/v) FCS. After 13 days, the supernatants were harvested, centrifuged twice at 3,000 rpm for 15 min, filtered, and precleared using Gammabind PLUS Sepharose (Amersham Biosciences). The sMR was purified from the medium on a MR5D3 affinity column, prepared by cross-linking of MR5D3 to Gammabind PLUS Sepharose with dimethyl pimelimidate (Sigma). After extensive washing in 0.01 M phosphate buffer, pH 7.4, 0.154 M NaCl (PBS), bound proteins were eluted with 0.5% (v/v) diethylamine and neutralized using 1 M Tris-HCl, pH 7.0.

Antisera
Rabbit polyclonal antisera specific for the different CPS serotypes of K. pneumoniae and S. pneumoniae and for the S. pneumoniae CW-PS were obtained from the Statens Serum Institute.

Authentic Polysaccharide Ligands
Soluble multivalent monosaccharide or short oligosaccharide ligands attached to a soluble polyacrylamide support (sugar-PAAs) were obtained from Syntesome (Munich, Germany). Yeast mannan from Saccharomyces cerevisiae was purchased from Sigma.

Analytical Methods
CPS from K. pneumoniae was quantified by measuring the total uronic acid content (one D-GlcA present/polysaccharide repeating unit) (25). Protein content was measured by the Micro BCA™ assay (Pierce), and nucleic acid was measured by UV absorbance. Total carbohydrate was measured by the phenol-sulfuric acid reaction (26). LPS content was measured by a kinetic turbidometric Limulus amoebocyte lysate assay, using reagents from Charles River Endosafe and Escherichia coli 055 LPS as standard according to the manufacturer's instructions.

SDS-PAGE
SDS-PAGE was carried out according to Ref. 27. LPS was resolved on 15% (w/v) gels and visualized by silver staining according to Ref. 28. CPS was resolved on 10% (w/v) gels and visualized by combined Alcian blue and silver staining (29).

Bacterial Capsular Polysaccharides
Purified CPSs from S. pneumoniae were obtained from Dr. Chris Jones (National Institute for Biological Standards and Controls, Hertfordshire, UK). These polysaccharides contain less than 1% by weight protein and nucleic acid contaminants. S. pneumoniae CW-PS was obtained from the Statens Serum Institute.
K. pneumoniae CPSs were isolated and purified as follows. Bacteria were grown overnight at 37°C on lactose agar plates to maximize CPS production. The bacterial growth from 20 -30 plates was suspended in 25-50 ml of distilled water and heated for 10 min at 100°C to release capsular material. The CPS was precipitated with 80% (v/v) acetone at 4°C and recovered by spooling around a glass Pasteur pipette. The precipitate was dried overnight at room temperature and resuspended in PBS, and insoluble material was removed by centrifugation twice at 30,000 ϫ g for 30 min at 4°C.
The supernatant containing CPS was then centrifuged for a further 8 h at 100,000 ϫ g and 4°C, and the pellet containing LPS and protein was discarded. CPS was dialyzed against distilled water and freeze-dried.
In some cases, the CPS was also digested with ribonuclease type II-B (30 g/ml) and deoxyribonuclease EC 3.1.2.1 (70 g/ml) for 24 h, followed by the addition of subtilisin (70 g/ml) and incubation for a further 24 h. Incubations were carried out in 0.01 M Tris-HCl, pH 7.4, containing 4 mM Mg 2ϩ and 0.05% (w/v) thiomersal at an approximate concentration of 500 g/ml CPS at 37°C with shaking. Following digestion, samples were dialyzed extensively against distilled water, 25-kDa cut-off membrane, and freeze-dried.
A final purification step was carried out by gel filtration chromatography under dissociating conditions on a TSK G5000 PW xl column (300 ϫ 7.5 mm). The buffer used was PBS containing 0.25% (w/v) sodium deoxycholate. Chromatography was carried out at 60°C at a flow rate of 0.25 ml/min. Absorbance at 205 and 280 nm and refractive index of the eluant were monitored. Fractions were collected at 2-min

Lipopolysaccharides
LPSs from Haemophilus influenzae type b strain Eagen, Pseudomonas aeruginosa strain PAC1, and K. pneumoniae were obtained by extraction with hot aqueous 45% (w/v) phenol (30). LPSs were dialyzed extensively against distilled water followed by centrifugation at 100,000 ϫ g for 18 h at 4°C. The LPS pellet was resuspended in water and quantified by gravimetric measurement. Salmonella typhimurium rough LPS types, Rb 2 , Rb 3 , Rc, Rd 1 , Rd 2 , and Re were purchased from Accurate Chemical and Scientific Corp. (New York, NY). LPSs from Neisseria meningitidis serogroup B, immunotype L3 and mutant 4 type, were gifts from Dr. S. R. Andersen. The structures of the LPSs from H. influenzae, S. typhimurium, and N. meningitidis are shown in Table IV. The structure of P. aeruginosa PAC1 LPS is not known.

Mannose Receptor Binding Assays
The binding of the mannose receptor to different polysaccharides was determined by ELISA, either directly by measuring binding to plates coated with different ligands or indirectly by inhibition assays as follows.
Polysaccharides were coated onto the wells of ELISA plates (Nunc; Maxisorb) by incubation in 0.154 M NaCl overnight at 37°C (50 l/well). Plates were sealed with parafilm and incubated in a sealed damp box. Bacterial CPSs and LPSs were coated at 50 g/ml, authentic sugar polyacrylamide substrates at 5 g/ml, and yeast mannan at 10 g/ml. After coating, plates were washed five times. The coating efficiency of the different CPSs was confirmed by titration with anti-CPS-specific antisera.
CRD (4 -7) Fc, CR-Fc, and sMR were incubated in the wells of coated plates at 2 g/ml (50 l/well) for 2 h at room temperature. Plates were washed five times. The MR fusion proteins were detected by incubation with anti-human IgG Fc-specific, alkaline phosphatase conjugate, species-absorbed (Jackson Laboratories). Bound sMR was detected by incubation with MR5D3 at 10 g/ml for 2 h at room temperature followed by detection with anti-rat IgG Fc-specific, alkaline phosphatase conjugate, species-absorbed (Jackson). Plates were washed five times and developed with p-nitrophenyl phosphate substrate (Sigma). Absorbance was measured at 405 nm after 30 min. Readings were measured against a blank of uncoated wells incubated with MR protein. Background readings were no more than 0.05 absorbance units higher than the reagent blank. All assays were carried out in duplicate or triplicate. Yeast mannan-coated wells were used as a positive control in assays measuring CRD (4 -7) Fc and sMR binding.
Inhibition Assays-The inhibition of binding of CRD (4 -7) Fc and sMR to ELISA plates coated with different ligands was measured. Inhibition was carried out in a high salt buffer consisting of 10 mM Tris-HCl, pH 7.5, containing 10 mM Ca 2ϩ , 1 M NaCl, and 0.05% (w/v) Tween 20. All incubations were at room temperature.
Either CRD (4 -7) Fc or sMR were preincubated at 2 g/ml with different concentrations of inhibitor, either monosaccharides or polysaccharides, for 30 min followed by incubation for 2 h in the wells of coated ELISA plates. The plates were washed five times, and the detection of bound CRD (4 -7) Fc and sMR was carried out as described above.

RESULTS
Purification of the sMR-The sMR obtained by affinity chromatography was quantified by the BCA assay and analyzed by SDS-PAGE and gel filtration chromatography. The sMR migrated as a single band on SDS-PAGE ( Fig. 1a) with an apparent molecular mass of ϳ180 kDa. The protein similarly eluted as one main peak on gel filtration chromatography (Fig 1b, peak 2), with an apparent molecular mass just greater than the 158-kDa protein standard consistent with the presence of monomers of the full-length soluble form of the molecule. The presence of a minor higher molecular weight component (peak 1) may represent a very small amount of sMR dimer or aggregate. Purification of CPS from K. pneumoniae-CPS extracts prior to gel filtration contained on average by weight 10% protein, 5% LPS, and 1% or less nucleic acid. Extensive enzymic digestions with nucleases and protease followed by dialysis reduced the levels of protein and nucleic acid by ϳ50%, suggesting a strong association of CPS with protein or peptide fragments. In order to further remove protein and also LPS, further purification was carried out by a gel filtration step under dissociating conditions. All CPSs gave very similar profiles (Fig. 2a). One main carbohydrate peak (peak a) eluted close to the void volume of the column, molecular mass 2,000 kDa and greater, and a second minor peak of carbohydrate (peak b) eluted, together with the majority of the protein, close to the total volume of the column.
Recovery of total carbohydrate was 80% and greater as measured by the phenol-sulfuric acid reaction. Carbohydrate was recovered from both pools as described under "Experimental Procedures," and aliquots were analyzed by SDS-PAGE (Fig. 2,  b and c). Carbohydrate in pool 1 migrated as a broad high molecular weight band that could be visualized only by using Alcian blue prior to silver staining, confirming the presence and identity of CPS. The carbohydrate in pool 2 gave the characteristic banding pattern of LPS, which was readily visualized by silver staining alone.
The CPS recovered in pool 1 was quantified by measuring the total uronic acid content. Protein content was no more than 1%, and LPS content, estimated by the Limulus amoebocyte lysate assay, was no more than 0.1% by weight. No nucleic acid was detectable. It was also confirmed that there was no reduction in the reaction of CPS with anti-CPS-specific antisera following gel filtration, confirming that no degradation of CPS had taken place during purification.
Specificity of Binding of CRD (4 -7) Fc, CR-Fc, and sMR for Authentic Standards-The binding properties and specificities of the proteins, CRD (4 -7) Fc, CR-Fc, and sMR, toward authentic ligands were analyzed prior to investigating binding to bacterial polysaccharides.
The measurements obtained in direct binding assays were directly proportional to the concentration of protein used in the assay (Fig. 3), and all subsequent assays were standardized using 2 g/ml protein.
The specificity and affinity of binding were further investigated by monosaccharide inhibition assays (Fig. 4). The sMR and CRD (4 -7) Fc both showed the highest affinity toward D-Man and L-Fuc with lower affinity for D-Glc, D-GlcNAc, and L-Rha. The K i values were all in the millimolar range, consistent with the known low affinity of binding of the MR to monovalent ligands. As in earlier studies (7,8), the inhibition assays reported here were carried out in buffer containing 1 M NaCl, but very little difference was found when assays were performed using physiological saline during the present study (data not shown).
In conclusion, CRD (4 -7) Fc and sMR both showed similar specificities toward different monosaccharides, which are consistent with the known sugar binding properties of the MR (7). In addition, it was shown that the MR recognizes the nonhost but common bacterial monosaccharide, L-Rha, with similar affinity to that shown for D-Glc and D-GlcNAc. Binding to L-Rha most probably occurs through C 3 and C 4 equatorial hydroxyl groups.
Direct Binding of the MR to CPSs from S. pneumoniae-The direct binding of CRD (4 -7) Fc and sMR to pneumococcal CPSs is shown in Fig. 5. The MR, both CRD (4 -7) Fc and sMR, recognized all except three of the CPS structures, serotypes 1, 4, and 18C, where binding was not detectable. The CW-PS, a minor component of most or all of the CPS samples, also did not act as a ligand for the MR and therefore is unlikely to contribute to the observed binding toward CPS.
The relative binding of CRD (4 -7) Fc and sMR to the different CPSs was very similar, although overall, a higher percentage binding relative to yeast mannan was obtained with the fusion protein. The affinity of binding to the different CPSs also appeared to be very similar. However, three of the CPS types (2, 14, and 19F) consistently gave higher readings in direct binding. No binding was detected when Ca 2ϩ was omitted from the buffer and CR-Fc did not recognize any CPS structure.
To further investigate the specificity of the interaction with the MR, inhibition assays were carried out using different concentrations of D-Man and D-Glc (Fig. 6). In all cases, binding of CRD (4 -7) Fc to CPS could be completely inhibited by incubation FIG. 1. Purification of the sMR. sMR was affinity-purified from culture supernatants of MR-transfected NIH 3T3 cells using MR5D3 as described under "Experimental Procedures." The recovered protein was quantified by the BCA assay and analyzed by SDS-PAGE (a) and gel filtration chromatography (b). a, SDS-PAGE profile of affinity-purified sMR. SDS-PAGE was carried out on a 10% (w/v) slab gel under reducing conditions, and proteins were stained with Coomassie Blue. Lane 1, molecular weight markers; lanes 2 and 3, sMR. b, gel filtration profile of affinity-purified sMR. Gel filtration chromatography was carried out on a TSK G3000 SW (7.5 ϫ 600 mm) column. The buffer used was 10 mM Tris-HCl, pH 7.4, 0.154 M NaCl, and 10 mM Ca 2ϩ . The flow rate was 0.25 ml/min, and the absorbance of the eluant was monitored at 214 and 280 nm. The arrows indicate the elution positions of gel filtration molecular mass protein standards (Bio-Rad) of 158, 44, 17, and 1.3 kDa as indicated. v and t indicate the void and total volumes of the column, respectively. with 5 mM D-Man. A concentration of 0.5 mM D-Man resulted in ϳ50% inhibition. Incubation with 5 mM D-Glc resulted in inhibition from 30 to 55% and from 50 to 80% inhibition at 10 mM.
As expected, higher concentrations of both sugars were required to inhibit binding of the MR to yeast mannan. Inhibition was approximately half that obtained for binding to CPScoated plates at the same sugar concentrations. A concentration of 10 mM D-Man was required to obtain complete inhibition of binding to yeast mannan. D-Gal tested at a 10 mM concentration had no effect on binding.
Inhibition of Binding by S. pneumoniae CPSs in Solution-To determine whether S. pneumoniae CPSs could act as ligands for the MR when in solution, as well as when surfacebound, CPSs were tested for their ability to inhibit the binding of the MR to coated ELISA plates. In all cases where direct binding to CPS was observed, the same CPSs inhibited binding in solution.
Each CPS was tested at 50 g/ml for its ability to inhibit binding of CRD (4 -7) Fc (2 g/ml) to ELISA plates coated with the homologous CPS. In each case, binding was inhibited between 70 and 90%. To test whether the CPS could also inhibit binding to heterologous CPS structures, type 14 CPS (50 g/ml) was incubated with CRD (4 -7) Fc (2 g/ml) and binding to differ-FIG. 2. Purification of K. pneumoniae CPS. a, gel filtration profile of K. pneumoniae CPS. K. pneumoniae CPSs were extracted and purified by gel filtration chromatography as described under "Experimental Procedures." Chromatography was carried out on a TSK G5000 xl column (filtration range for dextrans, 7,000 to 50 kDa) in PBS containing 0.25% (w/v) sodium deoxycholate at 60°C. The flow rate was 0.25 ml/min. The refractive index (panel c) and the absorbance of the eluant at 205 nm (panel a) and 280 nm (panel b) were monitored. Peaks a and b were pooled, dialyzed against distilled water, and freeze-dried, and residual detergent was removed by precipitation with 80% (v/v) ethanol. Total carbohydrate in pools a and b was quantified by the phenolsulfuric acid reaction. Recovery was 80% or greater. The figure shows a representative profile of CPS from strain K31. All CPS extracts gave very similar results. v and t indicate the void and total volumes of the column, respectively. b and c, SDS-PAGE analysis of pools a and b. The CPS and LPS content of pools a and b was analyzed by SDS-PAGE on either 10 or 15% (w/v) gels. LPS was visualized by silver staining, and CPS was visualized by staining with Alcian blue followed by silver staining. K. pneumoniae CPS was not detectable without the use of Alcian blue. b, lanes 1 and 2, LPS extracted from capsular serotype strain K55, 0.4 and 0.2 g; lanes 3 and 5, K55 CPS extract, pool a; lanes 4 and 5, K55 CPS extract, pool b. Separation was on a 15% (w/v) gel, and detection was by silver staining. LPS is clearly visible in pool 2, showing the characteristic O-antigen banding pattern, identical to that shown by LPS extracted with 45% (w/v) aqueous phenol. LPS is not detectable in pool a. c, lanes 1 and 2, K3 CPS extract, pool b; lanes 3 and 4, K3 CPS extract, pool a. Separation was on a 10% (w/v) gel and detection with Alcian blue followed by silver staining. Lanes 1 and 2 show the Oantigen banding pattern of K3 LPS (the low molecular weight core-LPS component has been eluted from the gel); lanes 4 and 5 show the high molecular weight CPS migrating as a broad smear at the top of the gel but no lower molecular weight components or banding pattern indicative of LPS.

TABLE V
Direct binding of the MR to authentic standards Soluble sugar-polyacrylamide substrates were coated onto the wells of ELISA plates, and direct binding of MR fusion proteins or sMR assayed by incubation with 2 g/ml protein for 2 h at room temperature followed by detection of bound proteins as described under "Experimental Procedures." Results are the average of three assays Ϯ 10% and are shown as absorbance readings at 405 nm taken after 30-min incubation with substrate and in parentheses as percentage binding relative to yeast mannan.  (4 -7) Fc and sMR were tested to establish the concentrations required for detection of binding to polysaccharide substrates. Proteins were incubated for 2 h at room temperature in the wells of coated plates, and bound CRD (4 -7) Fc and sMR were detected as described under "Experimental Procedures." ࡗ and f, binding of CRD (4 -7) Fc to yeast mannan and ␣-D-Man-PAA, respectively. ‚ and Ⅺ, binding of the sMR to yeast mannan and ␣-D-Man-PAA, respectively. ent CPSs measured. Again inhibition was from 60 to 90%, with inhibition of binding to yeast mannan lower at 40%.
The capacity of CPS to inhibit was further investigated by measuring the inhibition of binding of CRD (4 -7) Fc to ␣-D-Man-PAA-coated plates by different concentrations of CPS from 10 -200 g/ml (Fig. 7). Little or no inhibition occurred at 10 g/ml. At 50 g/ml, inhibition was less than that observed for binding to CPS, ranging from 10 to 40%, reflecting the higher affinity of the coating substrate for CRD (4 -7) Fc. Inhibition at 100 g/ml ranged from 30 to 60%, and inhibition at 200 g/ml ranged from 50 to 80%.
Of the three CPS serotypes (1, 4, and 18C) that were not recognized by the MR in direct binding assays, two (1 and 4) also failed to inhibit in solution, whereas 18C showed low inhibition at the higher concentrations. The CW-PS, which was not recognized in direct binding assays, also did not inhibit.
In contrast to CPS, incubation of CRD (4 -7) Fc with yeast mannan resulted in 50% inhibition of binding to ␣-D-Man-PAA at a very low concentration of 50 ng/ml and 100% inhibition at 500 ng/ml.
Binding to K. pneumoniae CPSs-When tested in direct CPSs were coated onto the wells of ELISA plates, and the direct binding of CRD (4 -7) Fc and sMR was assayed by incubation with 2 g/ml protein and detection of bound protein as described under "Experimental Procedures." All CPSs coated the plates as shown by titration with anti-CPS specific antisera. Coating concentrations were optimal from 10 g/ml upwards for detection with antisera. For detection with the MR, a concentration of 50 g/ml was used for coating, which resulted in maximal readings for each CPS. The results are expressed as percentage binding relative to yeast mannan (100%). Results are Ϯ10% taken as the average of 4 -6 experiments. Open bars, CRD (4 -7) Fc; shaded bars, sMR.  (4 -7) Fc was incubated at 2 g/ml with different concentrations of monosaccharide for 2 h in the wells of ELISA plates coated with CPS. Detection of CRD (4 -7) Fc was as described under "Experimental Procedures." Inhibition of binding to yeast mannan is shown for comparison. binding assays, crude extracts of K. pneumoniae CPS prior to the gel filtration purification step all showed very low levels of binding to both CRD (4 -7) Fc and sMR. This binding was much lower, ϳ10% that obtained with pneumococcal CPSs, and was inhibitable with D-Man (data not shown). These extracts, however, contained up to 5% by weight LPS, which is also a potential ligand for the MR.
After further purification, all except three of the 15 CPSs tested showed no binding to either CRD (4 -7) Fc or sMR. All binding activity was instead contained in the fractions from gel filtration containing protein and LPS. The three CPSs that still showed some binding to the MR were K3, K46, and K64. Binding was still weak compared with the pneumococcal CPSs, however, at ϳ10% relative to yeast mannan. Binding was again inhibitable with D-Man and Ca 2ϩ -dependent. None of the CPSs were bound by CR-Fc.
Recognition of LPS by the MR-LPS is the second main potential ligand for the MR on the surface of Gram-negative bacteria in addition to CPS. To test whether LPS could act as a ligand for the MR, LPSs were extracted from representative O-serotype strains of K. pneumoniae. The results for direct binding to K. pneumoniae LPSs as well as to LPSs from a number of other bacteria are shown in Table VI Tables III and IV. As expected, the polymannose O-serotype LPS was an excellent ligand. More unexpectedly, however, both the rough K. pneumoniae LPS and polygalactose O-serotypes of LPS were good ligands, with binding being almost equivalent to that observed with yeast mannan for the rough LPS and O2 serotype. LPSs from other bacterial species (Pseudomonas, Haemophilus, and Neisseria) did not bind. Also, the LPS from K. pneumoniae strain K36 with a ribose/Gal O-antigen failed to bind. A series of rough mutant LPSs from Salmonella all showed a very low level of binding to the MR. As with CPS, the binding specificities of CRD (4 -7) Fc and sMR were similar, and binding could be inhibited by incubation with 10 mM D-Man and D-Glc but not D-Gal.
The percentages of inhibition of binding of CRD (4 -7) Fc to K. pneumoniae LPS by different concentrations of D-Man and D-Glc are shown in Fig. 8. Except for serotype O1, higher concentrations of monosaccharide were required for inhibition than seen with S. pneumoniae CPS, consistent with a higher affinity for LPS recognition. In particular, the O2 serotype showed inhibition values similar to yeast mannan.
Again, as with CPS, CR-Fc did not bind to any of the LPSs tested, and no binding occurred in the absence of Ca 2ϩ .
Dual Lectin Activity of the MR-To test whether the sMR could bind both to polysaccharide through its CRD domains and to ligand for the CR domain simultaneously, sMR was preincubated with either yeast mannan, pneumococcal CPS or LPS, and then incubated in ELISA plate wells coated with SO 4 -3-␤-D-Gal-PAA. Bound sMR was detected with monoclonal antibody MR5D3, and bound CPS was detected with rabbit anti-CPS-specific antiserum. The secondary antibodies used in both cases to detect bound MR5D3, and bound rabbit antibodies were species-absorbed.
In no instance, did binding of the sMR to polysaccharide prevent subsequent recognition via the CR domain (Fig. 9). In contrast, incubation of the sMR with polysaccharides known to be ligands for the CRD domains increased the amount of sMR detected bound to SO 4 -3-␤-D-Gal-PAA-coated plates. Large polysaccharides, such as CPS, presumably display several binding sites for the sMR per molecule. Multiple attachment of sMR to polysaccharide may therefore increase the effective valency of the sMR and hence the affinity for subsequent binding to SO 4 -3-␤-D-Gal. LPS is similarly multivalent, existing as micelles in solution. DISCUSSION In this paper, we report the binding of the MR to CPSs from S. pneumoniae and to the LPS but not to the CPS from K. pneumoniae. The finding that the MR was able to recognize pneumococcal CPS was surprising, since these polysaccharides have none of the structural features associated with known MR specificities. Polysaccharides binding the MR may be expected   K17 (15, 35). The presence or absence of O-antigen and the banding patterns indicative of polygalactose or polymannose O-antigen were confirmed by SDS-PAGE and silver staining (data not shown). It was noted that LPSs from capsular strains K3 and K55 contained a relatively high proportion of LPS molecules with O-antigen, whereas LPS from K1 and K36 contained less O-antigenic material. The structures of the K. pneumoniae LPS O-antigens and of the other LPS types used in this assay are shown in Tables III and IV. LPSs were coated onto the wells of ELISA plates at 50 g/ml, and the direct binding of CRD (4 -7) Fc and sMR was assayed as described under "Experimental Procedures." The results shown are percentage binding relative to yeast mannan. to display, for example, multiple terminal D-Man residues or possibly terminal residues with lower affinity such as D-Glc or D-GlcNAc or, in the case of bacterial polysaccharides, L-Rha. Other potential recognition sites lie within the polysaccharide chain where the appropriate hydroxyl groups of the monosaccharides are not substituted. To date, however, there is no information as to whether the MR can recognize nonterminal residues. CPSs from certain capsular serotype strains of K. pneumoniae are potential ligands for the MR according to these criteria, but none were able to bind the MR in in vitro assays other than possibly in three cases with very low affinity. The specificity of the binding to CPS and LPS was demonstrated by complete inhibition with D-Man, partial inhibition with D-Glc, no inhibition with D-Gal, and Ca 2ϩ -dependence. Since CRD 4 is the only CRD known to bind D-Man by itself (36), this strongly suggests that CRD 4 is also involved in binding to bacterial polysaccharides. The fusion protein containing CRDs 4 -7 behaved in a very similar manner to the full-length sMR, indicating that CRDs 1-3 and CRD 8 are not involved directly in CPS binding. At least three CRDs (4, 5, and 7) are known to be required for high affinity and binding to multivalent ligands (7). Each CRD shows weak affinity for a single monosaccharide, with high affinity being achieved through multiple weak interactions.

LPS
In this study, the monomeric form of the sMR was shown to be able to bind both ligands for the CRDs and CR domain. In contrast to the sMR, the CRD (4 -7) Fc fusion protein is dimeric due to dimerization via the Fc region. As such, CRD (4 -7) Fc might be expected to have a higher affinity than the sMR. A direct comparison of the affinity of binding between the two molecules was not carried out as part of this investigation. However, in a separate experiment, not reported here, preincubation and cross-linking of CRD (4 -7) Fc with MR5D3 strongly indicated that increasing the valency of the MR increased the affinity of binding for CRD ligands. Higher affinity acquired by multimerization of several MR molecules was also indicated for binding to CR domain ligands as shown by the increased binding observed following attachment of the sMR to a large polysaccharide backbone. Cross-linking of the MR by high molecular weight polysaccharides may also occur on the cell surface if the relevant CRD domains are accessible. Recognition of pneumococcal CPSs by the MR shows that criteria other than the primary polysaccharide structure must be taken into account. One possible explanation is the recognition of conformational epitopes. According to NMR data, pneumococcal CPSs have an extended, flexible, ribbon-like structure with no stabilization of secondary structure (37,38). It is possible that this flexibility allows the positioning of hydroxyl groups between different monosaccharide residues at suitable positions to allow for binding to CRD domains. Thus, rather than contact be-  (4 -7) Fc was incubated at 2 g/ml with different concentrations of monosaccharide for 2 h in the wells of ELISA plates coated with LPS. Detection of CRD (4 -7) Fc was as described under "Experimental Procedures." Inhibition of binding to yeast mannan is shown for comparison. YM, yeast mannan.
FIG. 9. Dual recognition of microbial polysaccharide and sulfated ligand by the mannose receptor. a, increase in detection of binding of sMR to SO4 -3-␤-D-Gal-PAA following incubation with yeast mannan, CPS, and LPS. sMR (2 g/ml) was preincubated with yeast mannan, CPS, and LPS for 30 min followed by incubation in the wells of SO 4 -3-␤-D-Gal-PAA-coated plates and detection of bound sMR as previously described. Yeast mannan co-incubation was at 10 and 1 g/ml and CPS and LPS at 50 g/ml. The percentage increase in detection of bound sMR relative to control binding with no added polysaccharide is shown. b, detection of CPS bound to SO 4 -3-␤-D-Gal-PAA-coated plates following incubation with sMR. CPS and sMR were co-incubated as previously described. CPS bound to the plate was detected with rabbit anti-CPS specific antisera followed by anti-rabbit IgG alkaline phosphatase conjugate, species-absorbed. Open bars, incubation of S0 4 -3-␤-D-Gal-PAA-coated plates with CPS alone; filled bars, co-incubation of CPS and sMR. tween one CRD and one monosaccharide residue, one CRD may make contact with two closely adjacent monosaccharides. It is known that the repeating units of pneumococcal CPSs have preferred secondary conformations and that conformational epitopes within these polysaccharides are recognized by antibodies (39,40). The polysaccharide structural moieties recognized by the MR are present on both surface-bound CPS and soluble CPS as shown by direct binding and inhibition assays. Compared with yeast mannan, however, the pneumococcal CPSs are relatively weak ligands. This was shown by (a) the amount of MR detected directly bound to CPS-coated plates, (b) by the concentrations of monosaccharide required to inhibit binding to CPS, and (c) by the concentrations of CPS needed to inhibit in solution.
In comparison with S. pneumoniae, CPSs from K. pneumoniae showed no binding or in three cases very poor binding to the MR. K. pneumoniae CPSs have well defined secondary helical conformations shown by x-ray diffraction analysis (41). If conformation is a main factor in CPS recognition, then a more rigid secondary polysaccharide conformation may result in a less varied display of sugar hydroxyl groups and preclude MR binding. Where side branches are present in the polysaccharide repeating unit, these are usually displayed outwards from the main ␣-helical chain. K17 and K64 CPSs both have single residue branches of L-Rha, K46 has a short side-branch terminating in D-Glc, K27 has a single D-Glc, and K60 has three single branch residues of D-Glc per repeating unit. If such branch sugars are suitably orientated, they may act as sites of recognition for the MR. In accordance, both K64 and K46 CPSs showed weak binding to the MR; however, no binding to K17, K27, or K60 could be detected, so this idea cannot be substantiated. The other CPS to show weak binding to the MR, K3, has a side branch of a single D-Man residue substituted with cyclic pyruvate. Whereas pyruvylation should prevent recognition, it is possible that substitution is only partial or that some depyruvylation has occurred during purification. It would be of interest to know whether the three CPSs that appear to support low affinity binding have any unique features in their secondary conformations.
In contrast to the very poor binding shown toward CPSs from K. pneumoniae, the LPS bound with relatively high affinity. As with the pneumococcal CPSs, LPS was recognized both when surface-bound and when in solution.
The O-antigenic side chain of the O3 K. pneumoniae LPS serotype consists of a linear polymer of D-Man residues. As expected, this LPS type bound well to the MR. In addition, however, both the polygalactan O-antigen serotypes (O1 and O2a) and the rough type LPS were recognized. The rough LPS and O2a (K3) LPS serotype in particular showed very strong binding to the MR comparable with yeast mannan.
The structures of the core oligosaccharide regions of K. pneumoniae LPS have only recently been determined (42, 43) and are generally not as well defined as those from Salmonella or Escherichia coli. Analysis of the core regions from eight serotypes revealed that they all had similar structures variously containing terminal D-Glc and D-Hep residues as well as, unusually for LPS, D-GalA. LPS from a rough strain of K. pneumoniae expressed an unusual small oligosaccharide composed of D-Hep residues (44). The ring structure of D-Hep has the same configuration as D-Man and is therefore potentially a high affinity ligand for the MR, although this has not been tested directly. At this stage, we have not determined which part of the LPS molecule is bound by the MR (i.e. whether the Oantigenic side chains are bound directly or whether all binding is directed toward the core oligosaccharide regions).
As well as recognition of LPSs from K. pneumoniae, some binding to the rough LPSs from Salmonella was observed, but at a much lower affinity. These LPSs variously contain terminal D-Hep residues as well as terminal D-Glc or D-Gal residues. The roughest LPS mutant, containing only terminal 3-deoxy-D-manno-octulosonic acid, was also recognized. Similarly to CPS, therefore, the structural features of the LPS polysaccharides supporting MR binding are not clear, again indicating a conformational dependence. Not all LPSs tested were found to bind. Those from P. aeruginosa, H. influenzae, and N. meningitidis showed no binding at all. It may be of interest to note that H. influenzae and N. meningitidis are pathogens and that P. aeruginosa causes serious infection in cystic fibrosis patients, whereas K. pneumoniae is not pathogenic under normal circumstances.
Previous studies have reported that the binding and phagocytosis of whole cells of K. pneumoniae by alveolar MØs occurs through the recognition of CPS by the MR (16). In in vitro assays using purified CPS and the purified MR protein, however, we were unable to demonstrate any significant binding. K. pneumoniae CPS is released from cells as a complex of CPS, LPS, and protein as well as being associated with the cell surface (45). The crude CPS extract in all the strains analyzed, except K26, showed very weak binding to the MR in vitro (data not shown). On further purification, however, all binding, in nearly all cases, was shown to be due to the LPS component. It is possible that LPS on the cell surface may contribute to binding to MØs or that other MØ receptors such as class A scavenger receptors (46), complement receptor type 3 (47,48), or the recently identified ␤-glucan receptor, dectin (49), are involved.
The Ca 2ϩ -dependent lectins of the collectin family, mannosebinding protein (MBP) and the two surfactant proteins A and D, recognize a wide variety of different microorganisms (50 -52). MBP recognizes different monosaccharides with the same binding specificity as the MR, and the mechanism of sugar recognition by the CRD of MBP is analogous to that of CRD 4 of the MR. The prediction would be for both lectins to recognize similar microbial polysaccharides. In this respect, MBP has been reported to recognize rough LPS structures from Salmonella but not to bind to whole cells of S. pneumoniae (53,54). In the case of MBP, binding to the purified bacterial component did not always reflect actual binding to the bacterial surface experimentally. This is perhaps not surprising, since different surface components may interact in concert, or one component may prevent binding of another. In the case of S. pneumoniae, for example, the pneumococcal surface protein A extends from the cell wall and protrudes outside the capsule. The exposed part of this protein has a highly electronegative charge, which reduces complement activation on the bacterial surface (55). Surface protein A may also repel the binding of other molecules such as the MBP. With respect to K. pneumoniae, it is possible that the LPS is still accessible even in capsular strains (56).
Similar to the conclusions made during the present study for MR recognition, no direct correlation could be made between MBP binding to microbial polysaccharides and the primary polysaccharide structure (53,54), again indicating an important conformational effect regarding recognition of complex polysaccharides.
The sugar binding specificities of the surfactant proteins A and D are not as clearly defined as (and differ from) those of the MR and MBP. They have both been reported to bind bacterial CPS and LPS, and knockout mice have demonstrated an important role for surfactant protein A in pulmonary innate immunity (51,52).
Both purified bacterial polysaccharide and bacterial polysaccharide or oligosaccharide in the form of protein glycoconju-gates are used in vaccines. The initial interaction of the CPS with the host is mediated by components of innate immunity such as natural IgM anti-carbohydrate antibodies, complement, and lectins. The MR, either on MØs or as a soluble protein, forms one such constituent, and its overall importance is not known. Factors such as the binding, uptake, transport, localization, and rate of degradation and clearance of polysaccharide antigens, however, are poorly understood, and all may be modified by MR recognition. It should be noted that a physiological concentration of blood glucose of ϳ6 mM may be sufficient to reduce binding of weak ligands to the MR.
Some bacterial polysaccharides, including pneumococcal CPSs, activate the alternative complement pathway, resulting in the covalent attachment of complement components such as C3d (57). Covalent attachment of complement results in targeting of the polysaccharide to splenic marginal zone B cells and to follicular DCs, both of which express high levels of the complement receptors CD21 and CD35. The spleen and particularly marginal zone B cells are important in anti-carbohydrate immune responses. Complement activation and attachment to CPS may increase (or be critical to) the antibody response, similar to the effect observed with protein antigens (58). The MR is a heavily glycosylated molecule, most probably sialylated, which may serve to reduce CPS-mediated complement activation. If bound at a sufficient density, sMR may also affect recognition and cross-linking of B cell receptors by the CPS. Overall, the effect would be to potentially reduce the anti-CPS immune response.
Another possibility is that the sMR targets antigen to cells expressing ligands for the CR domain of the MR (i.e. metallophilic marginal zone MØs and DCs) (11,12,59). The structures of the sulfated oligosaccharide ligands expressed by these cells are not known. In agreement with Ref. 10, we show SO 4 -3-Gal to be a good ligand as well as SO 4 -3-GalNAc and SO 4 -4-Gal-NAc. Expression of SO 4 -4-GalNAc occurs on pituitary glycoprotein hormones and is responsible for the rapid clearance of these glycoproteins from the bloodstream by hepatic endothelial cells mediated by MR recognition (60). Sulfated GalNAc, however, is an unusual modification, probably specific to pituitary hormones, and it is more likely that terminal SO 4 -3-D-Gal is recognized on metallophilic marginal zone MØs and DCs.
We have shown that the sMR can bind at the same time CPS through its CRD domains and sulfated oligosaccharide through the CR domain. It has been speculated that DCs expressing ligands for the CR domain may serve to transport antigens via surface-bound sMR. Such DCs have been traced migrating from sites of immunization to B cell areas in secondary lymphoid tissues (12,59). CPS may therefore be targeted to specific cells via sMR attachment. The potential effects of MR recognition of CPS and LPS on immune responses are currently under investigation.