Cell-specific glycoforms of sialoadhesin and CD45 are counter-receptors for the cysteine-rich domain of the mannose receptor.

We previously reported that CR-Fc, an Fc chimeric protein containing the cysteine-rich (CR) domain of the mannose receptor, binds to marginal zone metallophilic macrophages (Mo) and B cell areas in the spleen and to subcapsular sinus Mo in lymph nodes of naive mice (CR-Fc(+) cells). Several CR-Fc ligands were found in spleen and lymph node tissue lysates using ligand blots. In this paper we report the identification of two of these ligands as sialoadhesin (Sn), an Mo-specific membrane molecule, and the leukocyte common antigen, CD45. CR-Fc bound selectively to Sn purified from spleen and lymph nodes and to two low molecular weight isoforms of CD45 in a sugar-dependent manner. CR-Fc binding and non-binding forms of Sn, probably derived from CR-Fc(+) and CR-Fc(-) cells respectively, were selected from spleen lysates. Analysis of the glycan pool associated with the CR-Fc-binding form revealed the presence of charged structures resistant to sialidase, absent in the non-binding form, that could correspond to sulfated structures. These results confirm the identification of the CR region of the mannose receptor as a lectin. We also demonstrate that the same glycoprotein expressed in different cells of the same organ can display distinct sugar epitopes that determine its binding properties.

The mannose receptor (MR) 1 is a type I membrane glycoprotein that mediates the uptake of mannosylated glycoconjugates by macrophages (Mø), dendritic cells (DCs), and hepatic endothelium (1)(2)(3)(4)(5). It is defined as a pattern recognition receptor due to its ability to bind sugar structures not normally found in the extracellular milieu but abundant in microorganisms (1). Carbohydrate recognition takes place through eight C-type lectin domains located in the extracellular region. These car-bohydrate recognition domains are preceded by two poorly characterized domains, a cysteine-rich (CR) domain located at the amino terminus of the MR and a domain containing fibronectin type II repeats (6 -9). A search for ligands for these two domains in mouse tissues led to the identification of several cell subpopulations able to bind a chimeric protein containing the CR domain of the murine form of the receptor fused to the Fc region of human IgG1 (CR-Fc) (10). These CR-Fc ϩ cells included marginal zone metallophilic Mø (MZMMø) and subcapsular sinus Mø in spleen and lymph nodes (LN) of naive animals and, probably, follicular dendritic cells in immunized animals. During the course of an immune response CR-Fc ϩ cells with dendritic morphology appeared to migrate in draining LN through the follicles and interfollicular areas. Based on these data, an antigen transport pathway (highlighted by CR-Fc labeling) and an alternative role for the MR in antigen uptake and presentation were proposed (5,10,11).
A first step toward the interpretation of those results is the identification of the molecules that interact with CR-Fc. Preliminary data indicated the presence of several ligands in spleen and LN lysates with apparent molecular mass ranging from 360 to 100 kDa (10). This study presents the identification of two of these ligands as sialoadhesin (Sn), an Mø-restricted membrane molecule (12,13), and CD45 (14,15). Biochemical characterization of these interactions showed that CR-Fc binding to both molecules is sugar-dependent. These results indicate that the CR domain of the MR itself has lectin-like activity and that the different ligands recognized in tissue lysates could be different proteins expressed by CR-Fc ϩ cells sharing the same sugar structure. These findings provide additional evidence for the important role that tissue-specific post-translational modifications play in determining the function of a molecule and raise the possibility that Sn or CD45 may display different binding properties in different cells.

Western and Ligand Blot
Analysis of CR-Fc binding and Sn and CD45 detection were performed following standard Western blot protocols. Briefly, samples (total tissue lysates or immunoprecipitated material) were resuspended in SDS-PAGE sample buffer (10 mM Tris-HCl, pH 6.8, 2% (w/v) SDS, 20% (v/v) glycerol, 0.001% (w/v) bromphenol blue), boiled, electrophoresed in a 5 or 6% SDS-PAGE, as indicated, and transferred to nitrocellulose (Hybond C plus; Amersham Pharmacia Biotech). Filters were blocked in 5% (w/v) non-fat milk, 0.1% (v/v) Tween 20 in phosphatebuffered saline and incubated with CR-Fc (10 g/ml) or rat IgG (5 g/ml) in blocking buffer. Binding was detected using peroxidase-conjugated F(abЈ) 2 mouse anti-human IgG (Jackson Immunoresearch Laboratories, West Grove, PA) in the case of Fc chimeric proteins or, in the case of rat IgG, peroxidase-conjugated goat anti-rat IgG (Chemicon, Harrow, UK) or F(abЈ) 2 mouse anti-rat IgG (Jackson Laboratories) and a chemiluminescence kit (ECL, Amersham Pharmacia Biotech). To avoid recognition by the secondary Abs of CR-Fc or rat mAb released from the Sepharose, primary reagent conjugated to digoxigenin or a rabbit polyclonal raised against domain 1 of Sn were used. Peroxidaseconjugated anti-digoxigenin F(abЈ) 2 (Roche Molecular Biochemicals) or F(abЈ) 2 donkey anti-rabbit IgG (Jackson Laboratories) was then used as secondary reagents.

N-Glycosidase Treatment
Digestion with peptide N-glycosidase F (Oxford GlycoSystems Ltd., Abingdon, UK) was performed following the manufacturer's recommendations.

Sugar Analysis of the CR-Fc-binding and Non-binding Isoform of Sialoadhesin
Sugar Extraction-N-Glycans were released from purified Sn by in situ digestion with peptide N-glycosidase F (Oxford GlycoSystems Ltd.) as described (26). Briefly, purified Sn was electrophoresed under reducing conditions, visualized by Coomassie staining, excised from the gel, alkylated and treated with peptide N-glycosidase F to release the Nlinked glycans.
Weak Anion Exchange (WAX) Chromatography-WAX was performed on a GlycoSep-C column (Oxford GlycoSystems Ltd) (7 ϫ 50 mm) attached to a Waters HPLC system as described previously (28).
Analysis of the Glycan Pool and Subsequent Digestion by Normal Phase Chromatography (NP-HPLC)-NP-HPLC was performed using a Glycosep-N column as described (29) using 50 mM ammonium formate, pH 4.4, as solvent A and acetonitrile as solvent B. The column was calibrated in glucose units using 2-AB-labeled dextran hydrolysate. Structures were assigned from the NP-HPLC elution positions expressed in glucose units, previously determined incremental values for the addition of monosaccharide residues to glycan cores (29) and the results of digestions of each glycan pool using arrays of exoglycosidases (30).

Sn Is the Major Component of the CR-Fc Ligands Preparation Obtained by Affinity Chromatography
By using the chimeric protein CR-Fc, several putative ligands for the CR domain of the MR were detected in spleen and peripheral (p) LN lysates by ligand blot analysis (10). To characterize these molecules, affinity chromatography was performed using CR-Fc coupled to protein A-Sepharose and spleen lysate as the ligand(s) source (see "Experimental Procedures"). CR-Fc ligands were eluted at high pH and collected in 2-ml fractions. Most CR-Fc binding activity was found in fractions 4 and 5, but activity was also present in fractions 3 and 6. A major band of 180 kDa apparent mass on non-reducing SDS-PAGE was found in fractions 4 and 5 after Coomassie Blue staining (Fig. 1A, ϪDTT). Additional proteins could be detected by silver stain (data not shown). Ligand blot analysis of these fractions confirmed that the 180-kDa protein bound CR-Fc and revealed the presence of other ligands not detectable by Coomassie Blue staining (Fig. 1A, CR-Fc binding). Under reducing conditions the apparent mass of the 180-kDa protein was shifted to Ͼ220 kDa, suggesting that the 180-kDa protein contained several internal disulfide bonds (Fig. 1A, ϩDTT). This pattern resembled that described for Sn, an Mø-restricted membrane molecule highly expressed by metallophilic and subcapsular sinus Mø (CR-Fc ϩ cells) but not restricted to these cell populations (12,13,17). Western blot analysis using the anti-Sn Ab 3D6, and amino-terminal sequencing (NH 2 -TWGVSSPKNVQG-) confirmed that Sn was the major component of the CR-Fc ligand preparation isolated by affinity chromatography. The specificity of the purification procedure was tested by probing with an Ab against F4/80, an Mø-specific membrane molecule absent in CR-Fc ϩ cells (10). No F4/80 antigen was detected in the preparation (data not shown).

Purified Sn Interacts with CR-Fc
To confirm that Sn was a ligand for CR-Fc, ligand blot analysis was performed on Sn purified independently from spleen using an Sn-specific affinity column (13) and Sn isolated from the CR-Fc-ligand preparation using 3D6-Sepharose (see "Experimental Procedures"). As shown in Fig. 1B, Sn from both sources bound CR-Fc in an Fc-independent fashion. Binding of CR-Fc to Sn was dose-dependent and less than 25 ng of Sn could be detected in this assay (data not shown).

CR-Fc Binds Selectively to Sn Purified from Spleen and LN
By immunohistochemistry, CR-Fc did not label splenic red pulp Mø, medullary Mø in LN, or stromal Mø in bone marrow, all Sn ϩ cells. It neither recognized thioglycollate-elicited Mø treated with mouse serum (thio-Mø ϩ NMS), which express large amounts of Sn (32), nor the cell line 9B12, a CHO-derived stable transfectant expressing full-length Sn (data not shown). Therefore, CR-Fc did not label all Sn ϩ cells. To test if the absence of correlation between Sn expression and CR-Fc binding was due to lack of accessibility or reduced Sn expression in the CR-Fc Ϫ ,Sn ϩ cell populations, Sn was immunoprecipitated from spleen, pLN, 9B12, and thio-Mø ϩ NMS protein lysates using 3D6-Sepharose (see "Experimental Procedures") and tested for its ability to interact with CR-Fc. Only Sn purified from spleen and pLN bound CR-Fc even when Sn from thio-Mø ϩ NMS or 9B12 were in excess ( Fig. 2A). It should be noted that the CR-Fc binding specific activity of Sn from pLN is higher than that from spleen. This result correlates with the data obtained by immunohistochemistry, since the percentage of Sn ϩ , CR-Fc ϩ cells is higher in LN.

N-Glycosidase Treatment of Sn Abrogates CR-Fc Binding
Sn migrated as a broad band on SDS-PAGE indicating that it consisted of a range of glycoforms (data not shown). To determine the role that glycosylation might play in the interaction of Sn with CR-Fc, Sn immunoprecipitated from spleen, pLN, and thio-Mø ϩ NMS protein lysates was treated with peptide N-glycanase F, to remove N-linked glycans and tested for the ability of the deglycosylated protein to bind CR-Fc. As shown in Fig. 2B, the removal of N-linked sugars completely abolished the ability of Sn to bind CR-Fc compared with control samples incubated in the same buffer but without the enzyme. Under these conditions the binding of 3D6 (an antibody that recognizes a conformation-dependent epitope) was reduced in the deglycosylated sample but was still detectable, indicating that N-glycans play a role in maintaining the structure of Sn. To test how CR-Fc binding was affected by protein conformation, Sn from pLN was reduced and alkylated and used in the binding assay. Under these conditions, the epitope recognized by 3D6 was eliminated, but CR-Fc binding, although reduced, could still be observed indicating that this interaction was maintained after the disruption of disulfide bonds (Fig. 2C).

Identification of CD45 as a Putative Ligand for CR-Fc
In addition to Sn, other CR-Fc ligands were present in the preparation obtained by affinity chromatography (Figs. 1 and  3). In an attempt to identify these molecules, the preparation of CR-Fc ligands was probed by Western blot with different mAbs. One of these Ab was YBM42.2.2, a rat IgG that recognizes all isoforms of CD45 (18). As shown in Fig. 3, YBM42.2.2 recognized a 100-kDa band that is strongly labeled by CR-Fc and two minor bands of about 180 kDa (see Fig. 4A for a better resolution). This result indicated that CD45 was present in this preparation but in a mostly degraded form that could have originated from two of the low molecular weight isoforms. Confocal analysis of splenic CR-Fcϩ cells showed that they express CD45 (data not shown). To assess the contribution of CD45 to CR-Fc binding, spleen and LN from CD45Ϫ/Ϫ mice (16) were tested by immunohistochemistry as described (10). CR-Fc la-  (20) (data not shown). Protein lysates from spleens of CD45Ϫ/Ϫ and wt mice were tested for CR-Fc binding as described under "Experimental Procedures" (Fig. 3B). In general there was a reduction in CR-Fc binding, but the only qualitative difference between the strains was found in the 180-kDa area where two bands were absent in the tissue from the knock-out mice. These results are in agreement with the data presented in Fig. 3A.

Selected Isoforms of CD45 Bind CR-Fc in a Sugar-dependent Manner
To test direct binding of CR-Fc to CD45, spleen and pLN lysates were immunoprecipitated using YBM42.2.2-Sepharose, as described under "Experimental Procedures," and bound proteins were eluted using diethylamine or by boiling in 0.05% SDS. Early experiments showed that Sn co-precipitated with CD45 (data not shown) and that, for the satisfactory resolution of the different CR-Fc ϩ bands, samples had to be electrophoresed in a 5% SDS-PAGE gel. To establish the contribution that the presence of Sn could have on CR-Fc binding, an additional preclearance step was introduced in the immunoprecipitation protocol; after incubation with protein A-Sepharose, lysates were precleared with 3D6-Sepharose. CR-Fc binding was tested as described under "Experimental Procedures," and the results are shown in Fig. 4A. Two low molecular weight isoforms of CD45 (ϳ180 kDa) bound CR-Fc, and this interaction was independent of the presence of Sn. Peptide N-glycanase F treatment confirmed the sugar requirement for CR-Fc binding to CD45 (Fig. 4B).

CR-Fc Binding and Non-binding Forms of Sn and CD45 Can Be Selected from Spleen Lysates
To confirm the data obtained by immunoprecipitation, Sn and CD45 were purified from spleen lysates as described under "Experimental Procedures," and CR-Fc binding and non-binding forms were separated using a CR-Fc-protein A-Sepharose column. Briefly, 0.5-ml fractions containing purified Sn (4 -11) or CD45 (7-23) were pooled, diluted in 0.5% Triton X-100, 140 mM NaCl, 10 mM Tris-HCl, pH 8, containing protease inhibitors to 25 and 50 ml, respectively, and applied to an 800-l CR-Fcprotein A-Sepharose column (2 mg CR-Fc/ml). After extensive washing in lysis buffer containing 0.5% (v/v) Triton X-100, bound proteins were eluted using 0.5% (v/v) diethylamine in 0.5% (v/v) Triton X-100. Collected fractions (binding forms, Sn-b and CD45-b) and flow-through (non-binding forms, Sn-nb and CD45-nb) were analyzed by Coomassie Blue staining (not shown) and tested for CR-Fc and antibody binding (Fig. 5). Approximately 50% of Sn present in spleen bound to CR-Fc in agreement with the high expression of Sn detected in CR-Fcϩ MZMMø. Two forms of Sn can be detected in the unbound fraction, a CR-Fc non-binding glycoform (upper band) and probably an unglycosylated precursor (lower band). As expected from the fact that most splenic cells are CD45ϩ and a minor proportion is CD45ϩ CR-Fcϩ, only a small fraction of the total CD45 present in spleen-bound CR-Fc (data not shown). Fractions containing the CR-Fc binding forms of CD45 were enriched in the lower isoforms (ϳ180 kDa). These results confirm the specificity of the selection procedure and the results obtained by immunoprecipitation.

Comparison of N-Glycan Pools Associated with Binding and Non-binding Forms of Sn
The preparations of Sn used in this study include the initial preparation obtained by CR-Fc affinity chromatography shown

(␣-CD45) as described under "Experimental
Procedures." Most CR-Fc binding activity was detected in pooled fractions 4 and 5 which presented a complex binding pattern. In fractions 3 and 6 most binding activity was restricted to a 180-kDa band (probably Sn). Several forms of Sn (3D6ϩ bands) were observed in sample 4/5. A major 100-kDa band and two minor 180-kDa bands were recognized by the anti-pan CD45 antibody in the same sample. The 100-kDa band could have originated by proteolytic cleavage of the two 180-kDa bands that correspond to two low molecular weight isoforms of CD45. B, pattern of CR-Fc binding to spleen lysates from CD45 Ϫ/Ϫ and wild type (wt) mice. Total protein lysates from spleens of CD45 Ϫ/Ϫ (2 mice, 132 g, 168 g) and wild type mice (2 mice, 120 g, 108 g) were tested for CR-Fc binding and for the presence of Sn and CD45 as described under "Experimental Procedures." Two CR-Fcϩ bands that co-migrated with low molecular weight isoforms of CD45 were absent in lysates of CD45 Ϫ/Ϫ mice (indicated by arrows). Higher Sn expression was detected in the spleens from CD45 Ϫ/Ϫ mice.

FIG. 4. CR-Fc binds selected isoforms of CD45 in a sugar-dependent fashion.
A, CR-Fc binding to CD45 immunoprecipitated from pLN lysates. pLN lysates were immunoprecipitated with 3D6-Sepharose (a) or YBM42.2.2-Sepharose (b and c), with (b) or without (c) a preclearing step using 3D6-Sepharose. Immunoprecipitated material was electrophoresed under non-reducing conditions, transferred to nitrocellulose, and probed with digoxigenin-labeled 3D6 (␣-Sn) (10 g/ ml), CR-Fc (10 g/ml), or YBM42.2.2 (5 g/ml) (␣-CD45). Binding was detected by incubation with peroxidase-conjugated anti-digoxigenin F(abЈ) 2 (␣-Sn), anti-human Fc F(abЈ) 2 (CR-Fc), or anti-rat IgG (␣-CD45). In the sample immunoprecipitated with 3D6-Sepharose (a), CR-Fc recognized a single band corresponding to Sn, and no CD45 could be detected. In samples b and c, which contained undetectable (b) or trace (c) levels of Sn, CR-Fc recognized two bands (ϳ180 kDa) that correspond to the low molecular weight isoforms of CD45 (indicated by arrowheads). Note that rat IgG is recognized in samples a-c by the anti-rat IgG secondary antibody (3rd panel). B, CR-Fc binding to CD45 is sugar-dependent. CD45 immunoprecipitated from pLN tissue lysates was incubated overnight in the absence (Ϫ) or in the presence (ϩ) of PNGase F and probed for CR-Fc or YBM42.2.2 binding (␣-CD45), as described under "Experimental Procedures." Treatment of CD45 with N-glycosidase increased its mobility in SDS-PAGE and abrogated its ability to interact with CR-Fc.
in Fig. 1 (a), and the CR-Fc-binding (b) and non-binding forms (c) of Sn described in the previous section. Similar results were obtained with samples a and b (data not shown).

Analysis of the N-Linked Glycan Pools Released from Sn-b by NP-HPLC-The 2-AB-labeled N-glycan pool released from
Sn-b was resolved by NP-HPLC (Fig. 6). 14 major peaks were assigned on the basis of their elution positions, measured in glucose units, obtained by comparison with the elution profile of a series of oligomers from a standard dextran hydrolysate and from previously determined incremental values for the addition of monosaccharides residues to glycan cores (29). The preliminary assignments were confirmed by exoglycosidase digestion of aliquots of the glycan pool (30). Each assigned peak eluted at its predicted position following digestion of the entire pool with the specific exoglycosidases shown in each panel. The effect of the enzyme arrays on 4 specific sugars (A2G2FS2, A4G4, A3G3FS, A3G3FS2) is indicated by the arrows. Further digestions with arrays containing almond meal ␣1-3/4-fucosidase, jack bean ␣-mannosidase, or Streptococcus pneumoniae ␤-N-acetylhexosaminidase indicated that there are no significant amounts of outer arm fucose residues nor are there polylactosamine or oligomannose structures (data not shown). Thus Sn-b contained a range of bi-, tri-, and tetra-antennary complex type sugars most of which were core fucosylated and sialylated. The major structure was the disialylated bi-antennary core fucosylated glycan A2G0F. The enzyme arrays in panel iv of Fig. 6 digests all non-substituted unextended complex sugars to the core glycans A2G0, A3G0, and A4G0. Glycans that are not digested to these core structures, including unassigned peaks in panels i-iii, could contain sulfated sugars. WAX analysis of Sn-b glycan pool ϩ Abs (Fig. 7, WAX, Sn-b, pool BЈ) suggests that these structures might be monosulfated since their elution profile corresponds to those with charge equivalent to 1 sialic acid residue.

Analysis of Sn-b and Sn-nb Glycans on the Basis of Charge Using WAX HPLC-2-AB-labeled N-glycan pools of Sn-b (A-C)
and Sn-nb (D-F) untreated (A-F) or treated with sialidase (AЈ-FЈ) were separated according to charge by WAX chromatography (Fig. 7, WAX). The column was calibrated with fetuin standard sugars (neutral, mono-(S1), di-(S2), and tri-(S3)sialylated). Three pools were collected from each sample: A, AЈ, D, and DЈ, 0 -5 min, contained neutral polysaccharide; B, BЈ, E and EЈ, 5-20 min, contained charged glycans eluting at positions equivalent to mono-to tetra-sialylated structures; and C, CЈ, F and FЈ, 20 -30 min, containing putative highly charged structures. Subsequent analysis of each pool by NP-HPLC chromatography revealed that almost all of the material in pools A and AЈ consisted of a contaminant polymer (data not shown).
NP-HPLC analysis of pool B confirmed that most N-linked glycans in Sn-b were charged. Most of the material in pools C and F was non-carbohydrate. Only Sn-b-derived glycans in pool C contained any glycans when the pools were analyzed by NP-HPLC, and these were a minor component (data not shown).
Whereas most of the peaks observed in pool E (5-20 min, from Sn-nb) disappeared after sialidase digestion (Fig. 7, NP-HPLC, Sn-nb, compare panels E and EЈ), two major components in the corresponding fraction from Sn-b were resistant to this enzyme (Fig. 7, NP-HPLC, Sn-b, compare B and BЈ). This was consistent with the possibility that terminally sulfated glycans were major components of the Sn-b glycan pool (compare Fig. 6, ii, abs treatment, and Fig. 7, NP-HPLC, Sn-b, panel  BЈ). In contrast, NP-HPLC analysis of pool EЈ confirmed that the majority of the charged sugars present in pool E from Sn-nb contained sialic acid. Accordingly, an increase in the amount of neutral sugars was observed after sialidase digestion of this sample (data not shown).

Sulfation of Sn Correlates with CR-Fc Binding
Analysis of Sn-b and Sn-nb N-linked glycan pools suggested the presence of sulfated structures in Sn-b. To investigate this possibility, Sn was immunoprecipitated from 35 SO 4 -labeled spleen protein lysates, prepared as described under "Experimental Procedures," and fractionated into Sn-b and Sn-nb using CR-Fc-protein A-Sepharose. In agreement with the sugar analysis described in the previous section, 35 SO 4 was only incorporated into Sn-b (Fig. 8A, left panel). This labeling could be inhibited in the presence of chlorate, an inhibitor of ATPsulfurylase, the first enzyme in the synthesis of 3Ј-phosphoadenosine 5Ј-phosphosulfate, the high energy donor of sulfate (31) (Fig. 8B). These results suggest that sulfated structures in Sn-b mediate binding to the CR domain of the MR. DISCUSSION In this study we present the identification of two of the counter-receptors recognized by the CR domain of the MR in secondary lymphoid organs, Sn and CD45. Neither of these proteins is exclusively expressed by CR-Fc ϩ cells. Sn is an Mø-specific surface molecule highly expressed by MZM and subcapsular sinus Mø (CR-Fc ϩ cells) but also present in splenic red pulp Mø, medullary Mø in LN, and bone marrow stromal Mø (CR-Fc Ϫ ) (17,33). CD45 is present in all nucleated cells of hemopoietic origin (14,15). Characterization of CR-Fc binding to the isolated proteins by ligand blotting demonstrated that this chimeric protein is able to discriminate between proteins obtained from different sources (in the case of Sn) and between the different isoforms of CD45 present in spleen and LNs. These results suggested that CR-Fc recognized a post-translational modification and underlined the correlation between the ligand blot and in situ labeling studies. Based on this it can be predicted that only the CR-Fc ϩ cells detected in situ are able to modify Sn and CD45 to generate ligands for CR-Fc. The fact that only two low molecular weight isoforms of CD45 bind to CR-Fc is consistent with Mø binding of CR-Fc in spleen and LNs since these are the forms found in cultured Mø in vitro. 2 CR-Fc binding is lost after peptide N-glycosidase F digestion and, at least in the case of Sn, is structure-independent. These data confirm that the CR domain of the MR itself has lectin-like binding activity (34) and indicate that cell-specific glycosyltransferase(s), responsible for the transformation of Sn and CD45 (among others) into CR-Fc ligands, are present in CR-Fc ϩ cells. The influence of site of synthesis on the binding properties of a molecule has been previously described. Only the form of GlyCAM 1 expressed by endothelial cells of peripheral and mesenteric lymph nodes contains the sulfate-modified carbohydrate required for L-selectin binding. Different carbohydrate modifications have been found in GlyCAM 1 expressed by lactating mammary gland epithelial cells, a non-binding form (35).
Inhibition assays were performed in an attempt to define the sugar moiety recognized by the CR domain of the MR. CR-Fc 2 L. Martínez-Pomares, unpublished observations. The number of charges predicted on samples eluting at this region is 5ϩ. After sialidase digestion, all charge due to sialic acid is removed, and sugars with charge only due to sialic acid elute in the neutral region (data not shown). Digested products eluting after 5 min in chromatograms BЈ (Sn-b) and EЈ (Sn-nb) contain charges other than that due to sialic acid. After lyophilization, each fraction was dissolved in H 2 O/CH 3 CN and analyzed by NP-HPLC. NP chromatograms are shown for the region in which N-linked sugars elute (80 -135 min). After sialidase treatment, Sn-b peaks shift their elution positions (BЈ versus B) but not to neutral structures indicating that these structures contain charged moieties other than sialic acid. This is consistent with the presence of SO 4 (Fig. 8). In contrast, NP-HPLC analysis of WAX fractions E and EЈ shows that all charged structures in Sn-nb have become neutral after sialidase. NP chromatogram EЈ is included to explain that there are no structures contained in this fraction that elute within the N-linked range (see the corresponding insert (0 -180 min) showing the total gradients).

FIG. 8. Incorporation of 35 SO 4 into the CR-Fc-binding form of Sn.
A, sulfation of Sn correlates with CR-Fc binding. Sn was immunoprecipitated from 35 SO 4 -labeled spleen protein lysates (1200 l, 600 l/spleen) using 3D6-Sepharose, and fractionated into Sn-b (b) and Sn-nb (nb) using CR-Fc-protein A Sepharose. Sn-nb (flow-through) was subsequently concentrated using 3D6-Sepharose. Both samples were eluted using 100 l of 0.5% (v/v) diethylamine in 0.5% (v/v) Triton X-100 as described under "Experimental Procedures." 60 and 10 l of each sample were electrophoresed under non-reducing conditions in two 5% SDS-PAGE. The gel containing 60-l samples was used to visualize 35 SO 4 -labeled proteins by fluorography (left panel). The second gel was used for the detection of Sn by Western blot using 3D6 and peroxidaseconjugated anti-rat IgG (right panel). B, sulfation of Sn can be inhibited by chlorate. Splenic cells were metabolically labeled with 35 SO 4 , as described under "Experimental Procedures," in the absence (Ϫ) or in the presence (ϩ) of 20 mM NaClO 3 (Na Chlorate). Sn was immunoprecipitated, fractionated into Sn-b and Sn-nb, and analyzed by SDS-PAGE and fluorography (top panel) and Western blot (bottom panel) as above. Detection of Sn was performed using a rabbit polyclonal specific for the domain 1 of Sn and peroxidase-conjugated donkey anti-rabbit IgG. binding (2 g/ml) to tissues could not be competed for by Dmannose (10 mM), mannan (1 mg/ml), L-fucose (10 mM), Dgalactose (10 mM), N-acetylglucosamine (10 mM), 6Ј N-acetylneuramin-lactose (1 mM), 3Ј N-acetylneuramin-lactose (1 mM), or dextran sulfate (10 mg/ml) (data not shown). Only the presence of fucoidan (10 mg/ml) inhibited this interaction, although the large amounts of sugar required and the undefined components present in this reagent make interpretation difficult. Involvement of sialic acid in this system was ruled out by the ability of Sn treated with sialidase from Vibrio cholerae to bind CR-Fc (data not shown).
Fiete et al. (34) recently reported that the CR domain of the MR domain recognizes the sugar structure SO 4 -4-GalNAc␤1,4GlcNAc␤1,2Man␣-present in lutropin and other glycoprotein hormones (36). The two enzymatic activities required for this modification have been found in several bovine tissues including spleen (37). Analysis of the glycan pool associated with the CR-Fc binding and non-binding forms of Sn is compatible with these data since it revealed the presence of putative sulfated structures in the Sn-b isoform that could be responsible for MR CR domain recognition (Figs. 7 and 8). The ability to bind CR-Fc might not necessarily correlate with the presence of SO 4 -4-GalNAc␤1,4GlcNAc␤1,2Man␣-, additional specificities could be present in lymphoid tissues, since the motif required for recognition by the PX(K/R)-specific GalNActransferase (38) is not present in Sn or CD45. Sn and CD45 do not account for all the CR-Fc ligands detected by ligand blots on total tissue lysates (10) (Figs. 1 and 3A). Affinity purification on Sn-, CD45-depleted lysates should yield additional counter-receptors.
In agreement with the reported presence of CR domain ligands in glycoprotein hormones produced in anterior pituitary (34), CR-Fc labeled isolated cells in this organ (data not shown), and our CR-Fc preparation bound to bovine lutropin-Sepharose in a salt-dependent, Ca 2ϩ -independent fashion (data not shown). These results emphasize the close correlation between tissue labeling and biochemical analysis.
Sn in MZMMø and subcapsular sinus Mø is a major membrane molecule thought to be involved in cell-cell adhesion (12,17,39). The presence of Sn in CD45 preparations is suggestive of an interaction between these molecules (Fig. 4A, left panel). No evidence for signal transduction motifs is found in the cytoplasmic tail of Sn (12), but the intracellular region of CD45 contains a tyrosine phosphatase domain (14,15). These results open the possibility for signal transduction in CR-Fcϩ cells following binding of the CR domain of the MR.
Glycoprotein hormones secreted by CR-Fc ϩ cells in anterior pituitary could be endocytosed by MR ϩ cells in liver (sinusoidal endothelium or Kupffer cells) through the recognition of SO 4 -4-GalNAc␤1,4GlcNAc␤1,2Man␣-by the CR domain of the MR (40). In the case of the spleen and LN, the two ligands found for this domain (Sn and CD45) are integral membrane proteins for which no cleaved, soluble forms have been described. We proposed a soluble form of the MR (sMR), found in Mø-conditioned media and mouse serum (11), as the counter-ligand for the specialized forms of Sn and CD45 (among others) present in the CR-Fc ϩ cells of secondary lymphoid organs. This interaction could mediate delivery of native antigen to follicular areas (10). Assessment of the biological relevance of these findings will require the characterization of the CR-Fc ϩ cells in situ and in vitro, study of the effects that sMR or CR-Fc have on their endocytic and phagocytic activity, and analysis of their interaction with other cell populations (MR Ϫ or MR ϩ ).
In this study we provide further evidence for the important role that post-translational modifications play in modulating protein function. MR has been shown to undergo proteolytic cleavage to release a soluble form (11), and the binding prop-erties of its CR domain seem to be influenced by the site of synthesis (liver versus lung) (40) probably due to differential post-translational modification(s). Accordingly, a heterogeneous population of MR molecules with different binding activities was obtained in CHO cells transfected with MR-specific cDNA (41). These results highlight the versatility of the MR and the importance of in vivo studies to unveil its multiple roles in immunological and physiological processes.
Ligands for the CR domain of the MR are themselves produced by post-translational modification(s) that take place in specific cell populations located in anterior pituitary (secreted glycoprotein hormones), secondary lymphoid organs (membrane-bound glycoforms of Sn and CD45 among others), and thymus (5). In particular, the characterization of the glycan pools associated with the CR-Fc binding and non-binding forms of Sn demonstrate that the same glycoprotein expressed in different cells of the same organ can display distinct sugar epitopes that determine its binding properties.