Human B-lymphocytes Express α2-6-Sialylated 6-Sulfo-N-acetyllactosamine Serving as a Preferred Ligand for CD22/Siglec-2*

CD22/Siglec-2, an important inhibitory co-receptor on B-lymphocytes, is known to recognize α2-6-sialylated glycan as a specific ligand. Here we propose that the α2-6-sialylated and 6-GlcNAc-sulfated determinant serves as a preferred ligand for CD22 because the binding of a human B-cell line to CD22 was almost completely abrogated after incubating the cells with NaClO3, an inhibitor of cellular sulfate metabolism, and was also significantly inhibited by a newly generated monoclonal antibody specific to the α2-6-sialylated 6-sulfo-N-acetyllactosamine (LacNAc) determinant (KN343, murine IgM). The α2-6-sialylated 6-sulfo-LacNAc determinant defined by the antibody was significantly expressed on a majority of normal human peripheral B-lymphocytes as well as follicular B-lymphocytes in peripheral lymph nodes. The determinant was also expressed in endothelial cells of high endothelial venules of secondary lymphoid tissues, including lymph nodes, tonsils, and intestine-associated lymphoid tissues, more strongly than on B-lymphocytes, suggesting a role for CD22 in B-cell interaction with blood vessels and trafficking. These results indicate that the α2-6-sialylated 6-sulfo-LacNAc determinant serves as an endogenous ligand for human CD22 and suggest the possibility that 6-GlcNAc sulfation as well as α2-6-sialylation may regulate CD22/Siglec-2 functions in humans.

CD22/Siglec-2 (sialic acid-binding immunoglobulin-like lectins) is an important inhibitory receptor on B-lymphocytes. It controls the signaling threshold of B-cell receptors, preventing their overactivation (1)(2)(3)(4). It has inhibitory immunoreceptor tyrosine-based inhibitory motifs in its cytoplasmic domain and regulates B-cell receptor signaling by recruiting the tyrosine phosphatase SHP-1 (Src homology 2 domain-containing protein-tyrosine phosphatase-1) (5). It is known to also affect distribution and trafficking of B-lymphocytes, such as in the homing of IgD ϩ B-lymphocytes to the bone marrow (6). Disruption of CD22 is known to result in production of high affinity autoantibodies, an increase in follicular mature B-lymphocytes, and a reduction in the number of marginal zone B-lymphocytes (7,8).
CD22/Siglec-2 is known to specifically bind to its ligand, ␣2-6-sialylated glycan (9 -12). B-lymphocytes themselves significantly express ␣2-6-sialylated glycan on their surface, which can serve as a cis-ligand for endogenous CD22, thus masking the ligand binding activity of CD22 on the majority of B-lymphocytes. CD22/Siglec-2 is proposed to exert trans-interaction with target cells expressing ␣2-6-sialylated glycans only when expression of the endogenous cis-ligand is suppressed or when the trans-ligand on target cells has a significantly higher binding activity than the cis-ligand (13,14). For better understanding of CD22 function, it is important to know the diversity of ␣2-6-sialylated glycans and to find candidate glycans that preferentially bind to CD22.
In mice, it has long been known that CD22 prefers the ␣2-6-N-glycolylsialic acid terminus over the ␣2-6-N-acetylsialic acid terminus (15). Quite recently, it was proposed that the activity of CMP-NeuAc hydroxylase, which converts N-acetylsialic acid into N-glycolylsialic acid, plays a role in regulating CD22 activity (16). Another proposed candidate had been 9-Oacetylation of terminal sialic acid (17). Little is known in humans, however, about the heterogeneity of ␣2-6-sialylated glycans in terms of CD22 binding activity.
It has been reported that an organochemically synthesized ␣2-6-sialylated and 6-sulfated glycan serves as a good ligand, eventually better than simply ␣2-6-sialylated determinants, for recombinant human CD22-Fc in an in vitro assay system (18). Our recent preliminary experiments indicated that suppression of carbohydrate sulfation abrogates binding of cul-tured human B-lymphoid cells to immobilized CD22 in a cellbased assay system (see "Results"). These findings collectively raised the possibility that some sulfated determinants also serve as preferred ligands for human CD22 at the cellular level. To address this issue, we tried to clarify the role of sulfation in the carbohydrate ligands for CD22 using a newly generated monoclonal antibody specific to the ␣2-6-sialylated 6-sulfo-Nacetyllactosamine (LacNAc) 2 determinant.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Reagents-A cultured human colon cancer cell line (SW480) and B-lymphoid cell line (Daudi) were obtained from the Tohoku University Cell Resource Center for Biomedical Research (Sendai, Japan). A human B-lymphoid cell clone derived from Namalwa cells, selected for high expression of 6-sulfated carbohydrate determinants and transfected with a gene for fucosyltransferase VII, was prepared as described previously (19).
CD22-mediated Cell Binding Assays-Recombinant human CD22-Fc was obtained from R&D Systems (Minneapolis, MN) and coated on the bottom of 24-well plates at a concentration of 20 g/ml overnight at 4°C. 2Ј,7Ј-bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM)-labeled Namalwa cells (1 ϫ 10 6 /0.5 ml/well) were added, and the plate was placed on a rotating platform for incubation under shear (90 rpm) for 20 min at room temperature. Where indicated, the cells were preincubated with the inhibitor antibody (25 g/ml) before addition to 24-well plates. After non-binding cells were washed out three times with phosphate-buffered saline, they were lysed with 0.5% Nonidet P-40, and the attached cells were counted by measuring fluorescence intensity using an Arvo 1420 multilabel counter (Wallac, Gaithersburg, MD). Recombinant human E-selectin-Fc (R&D Systems) served as a control for sulfate-independent binding.
Binding of recombinant CD22-Fc to human peripheral B-lymphocytes was ascertained by flow cytometric analyses as described previously (19,34). Recombinant CD22-Fc was preincubated with affinity-purified biotinylated rabbit anti-human IgG (Dako, Glostrup, Denmark), followed by incubation with phycoerythrin-streptavidin (Dako) before application to the staining of peripheral lymphocytes. BD FACS TM lysing solution was used for lysing red blood cells in the flow cytometric analyses. The cells were preincubated with blocking antibodies (KN343 and/or GL7; 20 g/ml) for 30 min at 37°C where indicated. Control murine and rat IgM reagents were obtained from BD Biosciences.
Flow Cytometric Analysis and Immunohistochemical Examination of ␣2-6-Sialylated 6-Sulfated and Related Determinants-SW480 and transfectant cells were maintained in Dulbecco's modified Eagle's medium (high glucose), and the Namalwa clone was maintained in RPMI 1640 medium supplemented with 10% fetal calf serum. In some experiments, the cells were cultured in the presence of 40 mM NaClO 3 for 7 days. Peripheral blood samples were obtained from healthy donors. The whole blood samples were mixed and incubated with anti-carbohydrate monoclonal antibody (purified antibody at 1 g/ml or culture supernatant at a dilution of 1:10) at 4°C for 30 min. The cells were then washed three times with phosphate-buffered saline containing 0.5% bovine serum albumin and stained with a 1:200 dilution of fluorescein isothiocyanate-conjugated goat anti-mouse IgM (-chain specific; Chemicon International, Temecula, CA) at 4°C for 30 min. After lysing red blood cells using BD FACS TM lysing solution, the stained cells were analyzed on a FACSCalibur (BD Biosciences). Leu4 (CD3) for T-cells, Leu12 (CD19) for B-cells, MY31 (Leu19, CD56) for natural killer cells, anti-CD45RA antibody HI100 (IgG 2b ), anti-CD45RO antibody UCHL1 (IgG 2a ), and anti-CD22 antibody SHCL-1 (IgG 2b ) were obtained from BD Biosciences for cellsurface marker analysis.
Immunohistochemical staining of human lymphoid tissues was performed using frozen sections of 10-m thickness prepared at autopsy at the Aichi Cancer Center Hospital. The avidin-biotin complex technique for immunohistochemical study was performed as described in the instructions for the respective kits for murine and rat IgM (VEC-TASTAIN) provided by Vector Laboratories (Burlingame, CA). The micrographs were recorded using a Microphot-FXA microscope system equipped with a DS-5M-L1 digital camera (Nikon, Tokyo, Japan), and images were processed using Adobe Photoshop Version 5.0J.

RESULTS
Generation of Anti-␣2-6-Sialylated 6-Sulfo-LacNAc Monoclonal Antibody KN343-As we did not have purified or synthetic ␣2-6-sialylated 6-sulfo-LacNAc for use as an immunogen for generation of monoclonal antibody at the initial stage of this study, we used human colon cancer cells (SW480) transfected with the gene for the GlcNAc␤:6-O-sulfotransferase HEC-GlcNAc6ST (SW480/HEC-GlcNAc6ST cells) as an immunogen. The obtained hybridoma clones were initially screened for their reactivity with SW480/HEC-GlcNAc6ST cells, but not with parental SW480 cells, to ascertain their 6-sulfation dependence (Fig. 1A). Antibodies G152 and G72, which we previously raised against 6-sulfated determinants carrying ␣2-3-sialylated termini, served as positive controls. The 6sulfation-dependent clones were further subjected to a second screening using SW480/HEC-GlcNAc6ST cells treated with sialidase S (specific to the NeuAc␣2-3 terminus) or sialidase A (cleaves NeuAc␣2-3/6/8 termini). A KN343 (murine IgM) clone was selected because it reacted with sialidase S-treated cells but not with sialidase A-treated cells, which is compatible with its specificity for the NeuAc␣2-6-linked terminus (Fig.  1B). Antibodies G152 and G72, which are specific to the 6-sulfated determinants carrying NeuAc␣2-3-linked termini, which should not react with either sialidase S-or sialidase A-treated cells, served as controls for the experimental conditions.
These findings established that KN343 is a specific antibody for the ␣2-6-sialylated 6-sulfo LacNAc determinant at both the ELISA and cellular levels. Sulfated determinants often give false-positive results in ELISAs because of their strong negative electric charges, and it should be borne in mind that this antibody shows the same specificity also at the cellular level.

␣2-6-Sialylated 6-Sulfo-LacNAc Determinant as CD22 Ligand
Roles of Sulfate Residues in Carbohydrate Recognition by Recombinant CD22-A clone of Namalwa cells was found to most strongly express the ␣2-6-sialylated 6-sulfo-LacNAc determinant among the various cultured human B-lymphocyte lines we tested. The Namalwa clone did not express CD22 and was thought to be suitable for the cell binding assays using immobilized recombinant CD22. The Namalwa cells strongly bound to immobilized recombinant CD22, and this was almost completely abrogated by preincubating the cells with NaClO 3 , an inhibitor of cellular sulfate metabolism ( Fig. 2A). This find-ing suggests a role for ligand sulfation in the binding of CD22. In line with this, the treatment abrogated expression of the ␣2-6-sialylated 6-sulfo-LacNAc determinant defined by antibody KN343, whereas expression of non-sulfated ␣2-6sialylated determinants defined by antibody GL7 rather slightly increased by 36% in median fluorescence intensity (Fig. 2B). These findings strongly suggest the importance of sulfated ␣2-6-sialylated determinants in the binding of CD22 to Namalwa cells. Binding of the cells to recombinant CD22 was significantly inhibited by the addition of antibody KN343 to the assay system (Fig. 2C), confirming the involvement of the sulfated ␣2-6sialylated determinants in CD22 binding.

Roles of Sulfate Residues in Cell-Cell Adhesion Mediated by CD22-
The roles of sulfate residues in glycan recognition by CD22 were further ascertained by monolayer cell adhesion assays. For this, Daudi cells strongly expressing CD22 (Fig. 3A) were employed for the adhesion experiments with parental SW480 or SW480/HEC-GlcNAc6ST cells. The sialidasetreated Daudi cells exhibited significant adhesion to SW480/ HEC-GlcNAc6ST cells, which express the ␣2-6-sialylated 6-sulfo-LacNAc determinant, but did not adhere to parental SW480 cells, which do not express the determinant (Fig. 3B). The adhesion of Daudi cells was abrogated by sialidase or NaClO 3 treatment of SW480/HEC-GlcNAc6ST cells (Fig.  3B) or by the addition of antibody KN343 (Fig. 3C). Antibody GL7 exhibited only marginal inhibition of cell adhesion, suggesting the important role of sulfated residues in the adhesion.
It has been proposed that endogenous CD22 ligands mask the interaction of CD22 with trans-ligands (13,35). As Daudi cells strongly express non-sulfated ␣2-6-sialylated determinants defined by GL7 as well as CD22 but only weakly express the ␣2-6-sialylated 6-sulfated determinants defined by KN343 (Fig. 3A), the cells were thought to be suitable for testing the role of endogenous sulfated ligands of CD22 on its interaction with trans-ligands expressed in SW480/HEC-GlcNAc6ST cells. As expected, sialidase treatment of Daudi FIGURE 1. Specificity of antibody KN343 to the ␣2-6-sialylated 6-sulfated LacNAc determinant. A, results of flow cytometry indicating that antibody KN343 recognizes cell-surface 6-sulfated determinants. The antibody was reactive only with the cells transfected with the gene for a 6-O-sulfotransferase (HEC-GlcNAc6ST), but not with parental SW480 cells. Antibody G152, which is specific to the ␣2-3-sialylated 6-sulfated Lewis X determinant (20), served as a control. B, results of flow cytometry indicating that antibody KN343 recognizes cell-surface ␣2-6-sialylated determinants. The reactivity of antibody KN343 with a clone of Namalwa cells was lost after sialidase A treatment, but not after sialidase S treatment. Recovery of the reactivity was observed when the sialidase A-treated cells were incubated with ␣2-6-sialyltransferase, but not with ␣2-3-sialyltransferase. Antibody G152 again served as a control; its reactivity was lost after either sialidase A or S treatment and was recovered only when the sialidase-treated cells were incubated with ␣2-3-sialyltransferase. C, results of ELISA using pure carbohydrate determinants indicating glycan specificity of antibodies KN343 (upper panel) and GL7 (lower panel). Antibody KN343 specifically reacted with the ␣2-6-sialylated 6-sulfated LacNAc determinant, but did not react with other determinants. Antibody GL7, which is known to be reactive with nonsulfated ␣2-6-sialyl-LacNAc determinants (16), served as a control. NOVEMBER 2, 2007 • VOLUME 282 • NUMBER 44 cells markedly enhanced binding of the cells to SW480/HEC-GlcNAc6ST cells. The treatment of Daudi cells with 25 mM NaClO 3 for 5 days was found to confer a similar enhancement of binding to SW480/HEC-GlcNAc6ST cells (Fig. 3D). The effects of NaClO 3 and sialidase were not additive. These findings suggest that endogenous ␣2-6-sialylated 6-sulfated determinants in Daudi cells, although only weakly expressed, significantly mask the interaction of CD22 with trans-ligands.
Recombinant CD22-Fc was found to bind to CD19 ϩ B-lymphocytes. The binding was significantly inhibited by the addition of antibody KN343, but was hardly inhibited by antibody GL7 (Fig. 4C). The inhibitory effect of the simultaneous addition of antibodies KN343 and GL7 was not much different from that exerted by antibody KN343 alone (Fig. 4C). These findings again corroborate the importance of the ␣2-6-sialylated 6-sulfo-LacNAc determinant as the ligand for CD22.

DISCUSSION
Human CD22/Siglec-2 has been known to specifically recognize ␣2-6-sialylated glycans. The reactivity of CD22 with ␣2-6-sialylated and 6-sulfated glycans was first noted in the printed covalent glycan microarray technique using synthetic oligosaccharide and recombinant CD22-Fc protein by Blixt et al. (18). Our results indicate that ␣2-6-sialylated 6-sulfated LacNAc determinants are really expressed in situ on human B-lymphocytes in peripheral blood as well as in lymphoid tissues such as peripheral lymph nodes and can confer significant binding of CD22 at the cellular level.
Previous glycan array assays (18) indicated that additional 6-GlcNAc sulfation of ␣2-6-sialylated glycan enhances CD22 binding relative to the corresponding non-sulfated glycan and suggested that ␣2-6-sialylated and 6-sulfated LacNAc glycan could be a preferred high affinity ligand for human CD22 (10,18). In line with this, our results indicate the importance of 6-GlcNAc sulfation because NaClO 3 treatment of the cells, which resulted in loss of 6-GlcNAc sulfation but retained ␣2-6-sialylation, almost totally abrogated binding to CD22. The strong inhibition of CD22 binding with antibody specific to the ␣2-6-sialylated 6-sulfated LacNAc determinant also serves to corroborate that additional 6-GlcNAc sulfation significantly enhances CD22 binding.
Endogenous CD22/Siglec-2 ligands present on B-lymphocytes are known to be ␣2-6-sialylated glycans, which mask the CD22 interaction with trans-ligands. Our results suggest that the ␣2-6-sialylated 6-sulfated LacNAc determinant can also serve as a cis-ligand for endogenous CD22 and participate in the masking. It has long been assumed that ␣2-6-sialyltransferases such as ␤-galactoside ␣2-6-sialyltransferase-1 (ST6Gal-1) regulate expression of CD22 ligands and therefore regulate the masking and unmasking of CD22/Siglec-2. Recently, however, it was shown that N-glycolylation of sialic acid by CMP-NeuAc hydroxylase significantly affects CD22 ligand activity in murine system, as glycans carrying NeuGc␣2-6 termini are much preferred ligands for murine CD22 compared with glycans carrying NeuAc␣2-6 termini (16). Because humans have no NeuGc due to deletion of exons in the CMP-NeuAc hydroxylase gene (37,38), this regulatory mechanism is not applicable to humans. Instead, our results suggest a role for 6-O-sulfotransferases, which are known to be affected by various stimuli and depend on cellular activation status (34, 39 -41), in the regulation of CD22 ligand activity, as well as ␣2-6-sialyltransferases.
An unexpected finding of this study is that the ␣2-6-sialylated and 6-sulfated LacNAc determinant is strongly expressed in HEVs in lymphoid tissues, including peripheral lymph nodes, tonsils, and intestine-associated lymphoid tissues. Endothelial expression of the determinant seems to be HEV-specific because endothelial cells lining the blood vessels in the spleen and thymus did not express the determinant. Expression of the determinant in HEVs was much stronger than that in B-lymphocytes, judging from the results of immunohistochemical staining. This is not unusual given that 6-O-sulfotransferases HEC-GlcNAc6ST and GlcNAc6ST-1 are strongly expressed in HEVs (23,42) and synthesize the ␣2-3-sialylated 6-sulfo-Lewis X determinant, which serves as a specific ligand for L-selectin (19,20,33,43,44). Our results thus suggest that these two 6-O-sulfotransferases also synthesize the ␣2-6-sialylated 6-sulfo-LacNAc determinant in HEVs. The physiological functions for the endothelial CD22 ligands remain to be clarified. It has long been suggested that CD22 may be involved in B-lymphocyte trafficking and homing, including that to the bone marrow (6,35). The reduced number of B-lymphocytes in peripheral lymph nodes was noted in 6-O-sulfotransferase-deficient mice (44). The strong expression of the CD22 ligand ␣2-6-sialylated 6-sulfated LacNAc determinant in HEVs suggests its involvement in the homing and recruitment of peripheral B-lymphocytes.