CD22 (Siglec-2) is a well-known regulator of B cell signaling

Transcription

1 Masking of CD22 by cis ligands does not prevent redistribution of CD22 to sites of cell contact Brian E. Collins*, Ola Blixt*, Alexis R. DeSieno*, Nicolai Bovin, Jamey D. Marth, and James C. Paulson* *Department of Molecular Biology, The Scripps Research Institute, La Jolla, CA 92037; Shemyakin Institute of Bioorganic Chemistry, Ul. Miklukho-Maklaya 16 10, GSP-7, V-437 Moscow, Russia; and The Howard Hughes Medical Institute, University of California at San Diego, La Jolla, CA Edited by Robert L. Hill, Duke University Medical Center, Durham, NC, and approved March 8, 2004 (received for review February 5, 2004) CD22, a negative regulator of B cell signaling, is a member of the siglec family that binds to 2-6-linked sialic acids on glycoproteins. Previous reports demonstrated that binding of multivalent sialoside probes to CD22 is blocked, or masked, by endogenous (cis) ligands, unless they are first destroyed by sialidase treatment. These results suggest that cis ligands on B cells make CD22 functionally unavailable for binding to ligands in trans. Through immunofluorescence microscopy, however, we observed that CD22 on resting B cells redistributes to the site of contact with other B or T lymphocytes. Redistribution is mediated by interaction with trans ligands on the opposing cell because it does not occur with ligand-deficient lymphocytes from ST6GalI-null mice. Surprisingly, CD45, proposed as both a cis and trans ligand of CD22, was not required for redistribution to sites of cell contact, given that redistribution of CD22 was independent of CD45 and was observed with lymphocytes from CD45-deficient mice. Furthermore, CD45 is not required for CD22 masking as similar levels of masking were observed in the WT and null mice. Comparison of the widely used sialoside polyacrylamide probe with a sialoside streptavidin probe revealed that the latter bound a subset of B cells without sialidase treatment, suggesting that cis ligands differentially impacted the binding of these two probes in trans. The combined results suggest that equilibrium binding to cis ligands does not preclude binding of CD22 to ligands in trans, and allows for its redistribution to sites of contact between lymphocytes. CD22 (Siglec-2) is a well-known regulator of B cell signaling (1 3), an activity mediated through its recruitment of SH2 domain-containing phosphatase 1 (also known as SHP-1) to the B cell receptor by immunoreceptor tyrosine-based inhibition motifs in its cytoplasmic domain (4, 5). The siglecs are a subgroup of the Ig superfamily that have in common a NH 2 - terminal Ig domain that binds sialic acid containing carbohydrates of glycoproteins as a ligand. CD22 is unique among the siglecs for its high specificity for sialoside ligands containing the sequence Sia 2-6Gal that often terminates N-linked carbohydrate groups of glycoproteins (6, 7). The Sia 2-6Gal sequence recognized by CD22 is differentially expressed in a cell typespecific manner (8 10) and is abundant on B and T lymphocytes (11, 12). Several reports have suggested that the ligand-binding domain of CD22 can modulate its activity as a regulator of B cell signaling (5, 13 15). Although the precise ligand interactions that modulate CD22 function are not well understood, both cis (B cell) and trans (opposing cell) glycoprotein ligands appear to be important. CD22 is well documented to bind to endogenous B cell glycoproteins in cis. As elegantly demonstrated by Razi et al. (13), cis ligands mask human CD22 binding to synthetic multivalent sialoside probes unless the cells are first pretreated with sialidase or periodate. Indeed, masking by endogenous cis ligands has subsequently been demonstrated to be a general property of the entire siglec family (16 23). Initially, sialoadhesin (Siglec-1) was thought to be one of the few siglecs to escape masking by cis ligands because of its 17 Ig domains, which extend the N-terminal ligand-binding domain away from the cell surface (17, 24). More recently, however, sialoadhesin was found to be masked on splenic macrophages and largely unmasked on lymph node macrophages, suggesting that the degree of masking depends on the differential expression of cis ligands on the cell (18). These findings have generally been interpreted to mean that masking of siglecs by endogenous cis glycoprotein ligands can regulate the extent to which they are functionally available to interact with ligands on other cells in trans (17, 18, 20, 25 27). Cis ligands appear to be important modulators of the activity of CD22 as a regulator of B cell signaling. Indeed, inhibition of CD22 binding to cis ligands by using small molecule ligand mimics (14) or abrogation of its binding by site directed mutagenesis results in decreased SH2 domain-containing phosphatase 1 (SHP-1) recruitment and increased calcium flux after B cell receptor (BCR) crosslinking in vitro (15). CD22 can bind to a variety of B cell glycoproteins in vitro, including CD45 and IgM, which are often cited as potentially important functional cis ligands of CD22 (28). Significantly, however, there is no clear evidence that any one glycoprotein or multiple glycoproteins are the effective cis ligands on the B cell in situ. After its initial cloning, CD22 was described as a cell adhesion molecule that may mediate cell-to-cell communication by binding to ligands in trans on other immune cells (29, 30). Indeed, COS cells expressing recombinant CD22 bound lymphocytes and monocytes in a sialic acid-dependent manner (30). Such trans interactions were observed to influence T cell activation in vitro, given that the presence of a soluble CD22 constant region fragment chimera during CD3 ligation altered T cell signaling (31). Moreover, in mixed lymphocyte costimulatory assays, antibodies that inhibit CD22 binding to its ligand also dampened T cell activation (32). The presence of CD22 ligands on sinusoidal endothelial cells of bone marrow have also been suggested to play a role in recruitment of B cells in vivo (33). Recently, Lanuoe et al. (34) demonstrated that B cell activation by ligation of BCR with antigens expressed on a cancer cell is dampened if CD22 ligands are coexpressed on the cancer cell, suggesting an impact on B cell signaling by trans ligands on the antigenpresenting cell. The importance of both cis and trans ligands in CD22 function suggests dynamic interactions with multiple ligands on B cells and other cells it contacts (23). Clearly, interactions with trans ligands requires that it not be functionally masked by cis ligands. Activation of B cells has been demonstrated to result in unmasking of CD22, which has been proposed to expose the ligand binding site for interactions with trans ligands on other cells (13). However, to date, there has been no direct examination of the degree to which binding of cis ligands restricts the ability of CD22 to interact with trans glycoprotein ligands on an opposing This paper was submitted directly (Track II) to the PNAS office. Abbreviations: BCR, B cell receptor; SNA, Sambuccus nigra agglutinin; SAAP, streptavidinconjugated alkaline phosphatase; PAA, polyacrylamide; NeuGc 2-6Gal, N-glycolylneuraminic acid 2-6 galactose 1-4 N-acetylglucosamine. To whom correspondence should be addressed at: Department of Molecular Biology, The Scripps Research Institute,10550 North Torrey Pines Road, MEM L-71, La Jolla, CA jpaulson@scripps.edu by The National Academy of Sciences of the USA PNAS April 20, 2004 vol. 101 no cgi doi pnas

2 cell. In this report we demonstrate that, although cis ligands mask binding of multivalent sialoside probes, CD22 is still able to interact with glycoprotein ligands in trans on adjacent B or T cells causing it to redistribute to the site of cell contact. Moreover, CD45, a putative ligand of CD22, is neither required as a cis ligand for masking the binding of multivalent sialoside probes, nor as a trans ligand for recruitment of CD22 to sites of cell contact. The results suggest that interaction of CD22 with trans ligands on opposing cells is favored over the binding of ligands in cis. Methods Mice and Antibodies. Unless otherwise noted, antibodies used were obtained from Pharmingen, and the reagents were obtained from Sigma. C57BL 6 mice were obtained from The Scripps Research Institute and ST6GalI-null and CD45-null mice were generated as described (35, 36). Mice were male between the ages of 6 and 10 weeks. Preparation of N-glycolylneuraminic Acid 2-6 Galactose 1-4 N-acetylglucosamine (NeuGc 2-6Gal) Probes. Sialosides with amino linkers were generated as described (37) and were directionally biotinylated by using sulfosuccinimidyl-6 -(biotinamido)-6- hexanamido hexanoate (Sulfo-NHS-LC-LC-Biotin, Pierce) and bound to the alkaline phosphatase-conjugated streptavidin complex as described elsewhere (38). Briefly, streptavidinconjugated alkaline phosphatase (SAAP; 10 lof1mg ml in 10 mm cacodylate, ph 7.5) was mixed with 40 l of 55 M biotinylated NeuGc 2-6Gal, representing a 16-fold molar excess based on the number of biotin-binding sites on streptavidin. The resulting sialoside SAAP probe was FITC-labeled without further purification. Briefly, 50 l of the sialoside SAAP solution was diluted with 180 l of 0.1 M Na bicarbonate, ph 9.0, mixed with 20 l of5mg ml FITC (Pierce) in 0.1 M Na bicarbonate, ph 9.0, and labeled in the dark overnight. To stop the reaction, 300 l of1methanolamine, ph 8.0, was added and incubated for 4 h at room temperature. Free sialoside-biotin and FITC were removed by extensive dialysis against PBS. NeuGc 2-6Gal 1-4GlcNAc polyacrylamide (PAA) was synthesized as described (37). Lymphocyte Preparations. Single-cell suspensions from spleen, brachial lymph node, and bone marrow were obtained by grinding tissue between two frosted glass slides in RPMI medium % FBS 50 M 2- and passing over a cotton-plugged pipette. Erythrocytes were lysed with a 5 min incubation at room temperature in 150 mm ammonium chloride, 10 mm potassium carbonate, 0.1 mm EDTA, ph 7.2, washed and resuspended in buffer. Pure B cell populations were obtained by negative selection with anti-thy1.2, anti-ly6g, and guinea pig complement. Populations were checked for purity by flow cytometry and determined to be 95% B220-positive and 1% CD3 positive. Flow Cytometry. Cells ( ; resuspended in 10 mg ml BSA in PBS at cells per ml) were incubated with 1 g of the PAA probe for 2honiceandthen1hwith1 g of streptavidinphycoerythrin (Molecular Probes) and anti-b220-cychrome at 2 g ml. For sialoside SAAP probes, cells were resuspended in 100 l of buffer at cells per ml and incubated with 1 l of probe (see above) for 30 min with or without additional antibodies. For sialidase pretreatment, cells were resuspended at cells per ml in PBS containing 2 mg ml BSA and 400 milliunits ml Arthrobacter ureafaciens sialidase (Kyoto Research Laboratories, Marukin Chuyu, Kyoto). After 2hat37 C with end-over-end rotation, the cells were washed and stained as above. Cells were stained with FITC-labeled Sambuccus nigra agglutinin (SNA; Vector Labs) by incubating cells in 10 mg ml BSA in PBS with 2 g ml SNA for 30 min on ice. Flow Fig. 1. CD22 redistributes to the site of cell-to-cell contacts in murine splenocytes. Representative immunofluorescent images of murine splenocytes are shown. Freshly isolated splenocytes were stained with Hoechst for nuclei (Left), Texas red-labeled anti-igm (Center), and FITC labeled anti-cd22 (Right). cytometry data were acquired on a FACSCaliber flow cytometer and analyzed with CELLQUEST (BD Biosciences). Immunofluorescence Microscopy. Splenocytes (10 6 ) were incubated on ice with the indicated antibody [1 g of Texas red-labeled anti-igm (Jackson ImmunoResearch), FITC-labeled anti-cd22, biotinylated anti-thy1.2, or 2 g of SNA-FITC] in PBS BSA for 30 min. Bound Thy1.2 was detected with 0.5 g of Texas red-labeled streptavidin. Stained cells were allowed to settle onto a glass slide on ice and fixed with 4% paraformaldehyde in PBS. Fixed cells were permeabilized with 0.05% Triton X-100 in PBS, and nuclei were detected with 0.5 g ml Hoechst before mounting with Prolong Antifade (Molecular Probes). Results Ligand-Dependent Recruitment of CD22 to the Sites of Cell Contact. During immunofluorescence microscopy experiments with freshly isolated resting splenocytes, we noted differential localization of CD22 on single cells as compared with those in lymphocyte clusters. Although both CD22 and IgM were uniformly localized on the cell surface of isolated cells (not shown), CD22 was preferentially localized at sites of cell contact in lymphocyte clusters, as exemplified in Fig. 1. In this cluster of four cells, one resting B cell is intensely stained with anti-igm and distributed evenly over the entire B cell surface. In contrast, staining with anti-cd22 is found primarily at the sites of cell contact. As illustrated by the representative clusters in Fig. 2, redistribution is seen at sites of cell contact between adjacent B cells (Fig. 2A Upper), and B cells with T cells (Fig. 2B Upper). In general, CD22 staining observed with B to T cell contacts are much smaller than those observed with B to B cell contacts. The small sites of contact are similar in morphology to those observed by Zal et al. (39) with anti-cd3 as a marker of a T cell synapse. The pattern of CD22 redistribution to contact sites was observed for virtually all B cells present in lymphocyte clusters, which typically ranged from 50 70% of all cells. To investigate whether redistribution of CD22 to sites of cell contact is mediated by its ligand-binding domain, we used ST6GalI-null mice, which lack the CD22 ligand NeuGc 2-6Gal (35). In contrast to the results with the WT mice (Fig. 2A Upper), CD22 in B cells of ST6GalI-null mice was uniformly localized in both B to B cell contacts (Fig. 2A Lower) and B to T cell contacts. Experiments were also performed mixing B cells purified from WT and ST6GalI-null splenocytes (Fig. 2C). In these experiments, WT cells were differentiated from the ligand-deficient, ST6GalI-null cells by their ability to bind to the SNA lectin as described (35). CD22 on both WT and ST6GalI-null B cells was found to redistribute to sites of cell contact with WT lymphocytes. In contrast, contact between a SNA-positive WT B cell with a SNA negative ST6GalI-null B cell resulted in redistribution of CD22 on the ST6GalI-null B cell only (Fig. 2C). The results show that CD22 on either WT or ST6GalI-null cells is capable of redistributing to sites of cell contact when the IMMUNOLOGY Collins et al. PNAS April 20, 2004 vol. 101 no

3 (Fig. 3A). Indeed, there was essentially no probe binding to native B cells from either strain, and sialidase treatment to destroy cis ligands allowed binding of the probe at equivalent levels to both WT and CD45-null B cells. If CD45 were required for CD22 masking, the levels of probe binding to nontreated CD45-deficient cells would be similar to the sialidase treated (unmasked) WT cells. Thus, cis ligands other than CD45 are sufficient to produce complete masking of probe binding. Comparison of the distribution of CD45 and CD22 in B cells from WT and CD45 mice is also shown in Fig. 3. In B cells of WT mice, although CD22 is recruited to the sites of cell adhesion, CD45 immunostaining was evenly distributed across the cell surface (see Fig. 3B). With B cells from CD45-null mice, redistribution of CD22 to the sites of cell contact was observed indistinguishable from that observed with WT mice (Fig. 3C). Taken together, these data suggest that CD45 clearly is also not required as a trans ligand for redistribution of CD22 to the sites of cell contact. Fig. 2. Redistribution of CD22 to the sites of lymphocyte contacts requires Sia 2-6Gal terminated glycans. CD22 redistributes to the sites of cell contact between B lymphocytes as well as T and B lymphocytes. (A) Representative immunofluorescent images of B-to-B lymphocyte contacts with splenocytes from WT (Upper) or ST6GalI-null (Lower) mice stained with Hoechst for nuclei (Left), Texas red-labeled anti-igm (Center), and FITC-labeled anti-cd22 (Right). (B) B-to-T lymphocyte contacts with splenocytes isolated from WT (Upper) or ST6GalI-null (Lower) mice stained with Hoechst for nuclei (Left), Texas red-labeled anti-thy 1.2 to mark T cells (Center), and FITC-labeled anti-cd22 (Right). (C) B-to-B lymphocyte contact with purified B cells isolated from WT and ST6GalI-null mice stained with Hoechst for nuclei (Left), FITClabeled SNA to detect the product of ST6GalI (Sia 2-6Gal) on the WT B cell (Center), and Texas red-labeled anti-cd22 (Right). opposing cell expresses glycoprotein ligands, indicating that redistribution is mediated by the interaction of CD22 in trans with glycoprotein ligands on the opposing cell. CD45 Is Not Required for CD22 Masking by Cis Ligands or Redistribution to Sites of Cell Contact. CD45 is the most abundant B cell surface glycoprotein and has been proposed as both a cis ligand on B cells (28) and as a trans ligand on T cells (31, 40, 41). To determine whether CD45 influenced the masking of CD22 to multivalent sialoside probes as a cis ligand or was required for redistribution of CD22 as a trans ligand, we compared the masking and redistribution of CD22 on B cells from WT and CD45-null mice. CD22 on human and murine B cells has been demonstrated to be masked from binding a multivalent sialoside PAA probe unless cis ligands are destroyed by sialidase (13, 37). It is well documented that human CD22 recognizes ligands with both N-acetylneuraminic acid and N-glycolylneuraminic acid, whereas murine CD22 prefers NeuGc, expressed abundantly on murine but not human cells (42, 43). We used murine CD22- specific probe (NeuGc 2-6Gal-PAA) (37) but saw no difference in the masking of CD22 on B cells of WT and CD45-null mice Differential Masking of Two Multivalent Sialoside Probes by Endogenous Cis Ligands. We recently reported a multivalent sialoside probe for CD22 prepared by adsorption of biotinylated sialosides to streptavidin-alkaline phosphatase (NeuGc 2-6Gal-SAAP) (38). Initial use of this probe in cell binding experiments suggested that CD22 was partially unmasked on a subpopulation of B cells (44). To determine whether this result reflected a true difference in masking of probe binding by cis ligands, we directly compared the binding of the NeuGc 2-6Gal-PAA and NeuGc 2-6Gal-SAAP probes for binding to native murine B cells as shown in Fig. 4. Although there was no binding of the PAA-based probe to untreated B cells, there was significant and reproducible binding of the SAAP-based probe (Fig. 4A). Both probes gave substantial increases in binding following sialidase pretreatment to remove sialic acids on the cell surface. Similar binding of the NeuGc 2-6Gal-SAAP probe to native B cells was observed with cells isolated from bone marrow, spleen, and lymph nodes as shown in Fig. 4B. In bone marrow, only the B220hi cells were observed to bind the probe, consistent with the expression of CD22 (data not shown). Analysis of probe binding to B cells from each of these tissues from eight different mice gave an average of 11% ranging from 6% to 20% of the B220hi cells. Probe binding was sialoside specific given that no binding is observed with SAAP alone (Fig. 4B), and periodate pretreatment of the probe, which removes the glycerol side chain of sialic acid required for CD22-mediated recognition, eliminated probe binding (data not shown) (45). We used the lectin SNA, specific for the Sia 2-6Gal sequence recognized as a ligand of CD22, but could not detect a difference in the expression of CD22 ligands in the cell populations that bound the SAAP probe compared with those that did not (Fig. 4C). In similar experiments, a soluble CD22-Ig chimera also showed no difference in binding to these two populations of cells (44). These data illustrate that the increased binding of the NeuGc 2-6Gal-SAAP probe does not appear to result from a decrease in the concentrations of cis CD22 ligands. Thus, relative to the NeuGc 2-6Gal-PAA probe, the NeuGc 2-6Gal-SAAP probe appears to more effectively compete in trans with endogenous cis ligands and bind to a significant fraction of B cells. Taken together, these data suggest that even synthetic multivalent ligands can differentially compete for endogenous cis ligands. Discussion The activity of CD22 as a regulator of B cell signaling is modulated by the interaction of its extracellular ligand-binding domain with sialoside ligands on the same cell (cis) or adjacent cells (trans) (14, 15, 26, 34, 35, 46). To date, few studies have investigated the dynamic interactions of cis and trans ligands and their relative roles in CD22 function. Clearly, the ligand-binding sites of CD22 are cgi doi pnas Collins et al.

4 IMMUNOLOGY Fig. 4. Cis ligands differentially mask binding of two multivalent sialoside probes. Binding of NeuGc 2-6Gal-SAAP to native B cells does not require unmasking, as measured by a decrease in SNA receptors. (A) Splenocytes mock-pretreated or A. ureafaciens sialidase-pretreated were incubated with the NeuGc 2-6Gal-SAAP probe or the NeuGc 2-6Gal-PAA probe and anti- B220, as described in Materials and Methods. (B) Native (untreated) lymphocytes from bone marrow, spleen, and lymph node were incubated with anti-b220 and either NeuGc 2-6Gal-SAAP probe (Upper) or the SAAP carrier alone (Lower) without any pretreatment. The percentage of the B220hi cells that bound either the sialoside SAAP probe or SAAP carrier is indicated in the upper right quadrant of each panel. (C) Native splenocytes were stained with SNA, anti-b220, and NeuGc 2-6Gal-SAAP. (Left) Binding of NeuGc 2-6Gal- SAAP to native B220 cells. Gates were set for cells that exhibited background (R1) or high (R2) binding of the sialoside probe. (Right) SNA binding to gated B cells with background (R1, dotted line) or high (R2; solid line) levels of NeuGc 2-6Gal-SAAP probe. Fig. 3. CD22 masking and recruitment to the sites of lymphocyte contact in the absence of CD45. (A) Masking status of CD22 as measured by binding of NeuGc 2-6Gal-PAA probe (Upper) to WT (black traces) and CD45-null (red traces) splenocytes, pretreated with (thick trace) or without (thin trace) sialidase to remove cis ligands. Splenocytes from WT (Left Lower) or CD45 (Right Lower) mice were stained with anti-cd45 (B220) and anti-cd22. (B) Representative immunofluorescent images of splenocytes from WT mice stained with anti-cd45 (Left) and anti-cd22 (Right). (C) Representative immunofluorescent images of splenocytes from CD45-deficient mice stained with Hoechst for nuclei (Left) or anti-cd22 (Right). predominately occupied by cis ligands on B cells, given that they mask the binding of synthetic multivalent sialoside ligands unless first destroyed by sialidase or periodate (13, 37). Several reports have suggested that CD22 is unmasked on minor subsets of resting B cells (26, 44) or becomes unmasked on a portion of activated B cells following BCR ligation (13), which would allow binding of CD22 to trans ligands on opposing cells. Although reduced levels of cis ligands will no doubt facilitate binding of CD22 to ligands in trans, it is not a prerequisite. Indeed, as shown in this report, despite the presence of cis ligands that prevent the binding synthetic multivalent sialoside probes to CD22 on resting B cells, CD22 binds to trans ligands on adjacent B or T lymphocytes, causing its redistribution to sites of cell contact. All these observations are consistent with a simple model involving competition of cis and trans ligands of CD22. Collins et al. PNAS April 20, 2004 vol. 101 no

5 Masking of CD22 to multivalent sialoside probes is due to occupancy of its ligand binding site by cis ligands (13). CD22 exhibits relatively weak binding to its preferred sialoside ligand Sia 2-6Gal 1-4GlcNAc. The K d of both murine and human CD22 for this sequence ranges from 0.1 to 0.3 mm, regardless of whether it is presented on synthetic sialosides or attached to an N-linked oligosaccharide of a glycoprotein, such as CD45 or CD4 (7, 38, 47). Thus, CD22 is in rapid equilibrium with its carbohydrate ligands, and the degree of occupancy of its binding site depends on their local concentration. We estimate that the sialic acid concentration at the surface of the B cell is about 110 mm based on a lymphocyte volume of 210 m 3 (48), a glycocalyx thickness of 44 nm (49), and a cell surface sialic acid content of 2.5 g per 10 7 lymphocytes (50). Because the abundant N-linked oligosaccharides of B cell glycoproteins contain mainly the Sia 2-6Gal linkage, we can assume that the preferred ligand of CD22 is present at least 25% or mm. This represents a concentration 100-fold higher than the K d of CD22, resulting in the high occupancy that can occlude binding of a soluble multivalent sialoside probe presented in trans. Of the two sialoside probes examined in Fig. 4, the SAAPbased probe more effectively competed as a trans ligand and established sufficient multivalent interactions to remain bound to a portion (10 20%) of the native B cells. The difference in the binding properties of the SAAP- and PAA-based probes is likely due to a difference in their unique geometry and physical presentation of sialosides. At present, however, the molecular features that distinguish these two probes are not clear. Commercial preparations of SAAP exhibiting high molecular weight conjugates by SDS gels were the most effective, indicating that high valency is key (38). Establishing the precise molecular features required for probes to compete with endogenous ligands will require evaluation of more chemically defined multivalent sialoside ligands (51). The dynamics of CD22 interacting with its sialoside ligand are completely different at sites of cell contact. For example, consider the case of adjacent B cells. Because the cis and trans ligands are both sialosides of B cell glycoproteins, the candidate ligands are present at the same concentration. However, in one case they are presented to CD22 from the same cell in cis and in the other case they are presented from the opposing cell in trans. The fact that CD22 redistributes to the site of cell contact strongly suggests that trans ligands favorably compete with cis ligands for occupancy of the CD22 ligand binding. In view of the three-dimensional constraints for docking with the ligand binding site of CD22, the potential for individual sialoside sequences to serve as a functional ligand of CD22 will differ when they are presented in cis or trans. In principle, the same glycoprotein can be a more favorable ligand when presented in trans than when presented in cis. Thus, it is conceivable that the same or different B cell glycoproteins participate in masking of CD22 in cis and redistribution of CD22 in trans. The B and T cell glycoproteins that serve as functional ligands of CD22 have not been identified. Numerous glycoproteins are immunoprecipitated from cell lysates of B cells and T cells with CD22-Ig chimeras, demonstrating that they carry the sialoside sequence recognized by CD22 (28, 31). Although the functional ligand(s) of CD22 is likely to be among them, solubilization of the cell membrane disrupts the spatial constraints imposed by the membrane anchors of CD22 and its cis and trans ligands. Thus, glycoproteins brought down in immunoprecipitation experiments are not necessarily those that interact with CD22 in situ. Because CD45 has been mentioned both as a potential cis ligand for CD22 on B cells (28) and trans ligand on T cells (31, 40), we sought to determine whether CD45 was required as a ligand for CD22 by evaluating splenocytes from CD45-null mice. As shown in Fig. 4, there was no difference between B cells from WT and CD45-null mice with respect to masking of CD22 by cis ligands and redistribution of CD22 to sites of lymphocyte contact by trans ligands. Although this similarity does not preclude participation of CD45 as a ligand in WT cells, other glycoprotein ligands are sufficient to mediate these functions. The observation that CD22 on resting B cells can redistribute to sites of cell contact has important implications for its function as a regulator of B cell signaling and possibly its function as a cell adhesion receptor. It is generally accepted that CD22 must be in close proximity to the IgM receptor complex to effect negative regulation of signaling by recruitment of the SH2 domaincontaining phosphatase 1 (SHP-1) (5, 52). In a recent report, Lanoue et al. (34) compared activation of murine B cells by antigens expressed on a cancer cell line with activation by the same cell line transfected with the cdna for the ST6GalI sialyltransferase to generate trans ligands of CD22. ST6GalI expression in the antigen-presenting cells inhibited activation of WT B cells but had no effect on activation of B cells isolated from CD22-null mice. These results suggested that expression of ST6GalI in the antigen-presenting cells caused inhibition of B cell activation by generating trans ligands of CD22. In view of the fact that cis ligands mask CD22 on the resting B cell, it was suggested that CD22 might redistribute to the site of B cell receptor contact by unmasking following stimulation as proposed by Razi et al. (13, 34). However, as demonstrated in this report, unmasking would not be necessary for trans ligands to cause redistribution of CD22 to the site of contact with the cancer cell. Moreover, this would result in colocalization of CD22 with the B cell receptor at the site of antigen presentation before activation, enhancing its ability to dampen signal transduction. These principles could also play a role in preventing B cell activation in other physiologically relevant circumstances. For example, during B cell presentation of antigens through MHC to T cells, recruitment of CD22 to the site of cell contact (see Fig. 2) could help prevent inadvertent B cell activation (e.g., through CD40 CD40L ligation). This may be of critical importance as T cells sample B cell MHC before finding its cognate antigen. Similarly, CD22 may contribute to suppressing inadvertent activation of B cells to self antigens expressed on cells bearing CD22 ligands (53). CD22 binding in trans is likely to have critical functions other than affecting B cell signaling directly. Potential CD22 ligands exist on bone marrow sinusoidal epithelial cells and may be critical for B cell homing (33). Additionally, T cell status may be affected, as CD22 constant region fragment chimeras have been shown to attenuate T cell receptor-mediated signaling in vitro (31). Furthermore, potential CD22 ligands (i.e., Sia 2-6LacNAc) are expressed in other critical immune regulators, such as splenic dendritic cells. Thus, further investigation into the affect CD22 interacting with its trans ligands, which we have shown here can occur despite CD22 appearing masked, is critical for understanding the role of CD22 in regulating immune status. Finally, given that most members of the siglec family have been demonstrated to be subject to masking by cis ligands, we suggest that the same principles observed here for CD22 are likely to be relevant to the interactions of other siglecs with ligands presented in trans on opposing cells. Specifically, that interactions with biologically relevant trans glycoprotein ligands may effectively recruit the siglec to sites of cell contact despite the presence of cis ligands that functionally mask its binding to sialoside probes. We thank Dr. Yasuhiro Ohta at Kyoto Research Laboratories and Dr. William Raschke for the generous gifts of A. ureafaciens sialidase and CD45-deficient mice, respectively; Ms. Anna Tran-Crie for her expert assistance in manuscript preparation; and the members of the Paulson lab for their scientific input. This work was supported by National Institutes of Health Grants GM25042 (to B.E.C.), GM60938 and AI (to J.C.P.), and P01-HL57345 (to J.D.M.) cgi doi pnas Collins et al.

6 1. Nitschke, L., Carsetti, R., Ocker, B., Kohler, G. & Lamers, M. C. (1997) Curr. Biol. 7, Sato, S., Miller, A. S., Inaoki, M., Bock, C. B., Jansen, P. J., Tang, M. L. & Tedder, T. F. (1996) Immunity 5, Otipoby, K. L., Andersson, K. B., Draves, K. E., Klaus, S. J., Farr, A. G., Kerner, J. D., Perlmutter, R. M., Law, C. L. & Clark, E. A. (1996) Nature 384, Doody, G. M., Justement, L. B., Delibrias, C. C., Matthews, R. J., Lin, J., Thomas, M. L. & Fearon, D. T. (1995) Science 269, Cyster, J. G. & Goodnow, C. C. (1997) Immunity 6, Hanasaki, K., Powell, L. D. & Varki, A. (1995) J. Biol. Chem. 270, Powell, L. D., Jain, R. K., Matta, K. L., Sabesan, S. & Varki, A. (1995) J. Biol. Chem. 270, Kitagawa, H. & Paulson, J. C. (1994) J. Biol. Chem. 269, O Hanlon, T. P., Lau, K. M., Wang, X. C. & Lau, J. T. (1989) J. Biol. Chem. 264, Wen, D. X., Svensson, E. C. & Paulson, J. C. (1992) J. Biol. Chem. 267, Powell, L. D., Sgroi, D., Sjoberg, E. R., Stamenkovic, I. & Varki, A. (1993) J. Biol. Chem. 268, Lo, N. W. & Lau, J. T. (1999) Glycobiology 9, Razi, N. & Varki, A. (1998) Proc. Natl. Acad. Sci. USA 95, Kelm, S., Gerlach, J., Brossmer, R., Danzer, C. P. & Nitschke, L. (2002) J. Exp. Med. 195, Jin, L., McLean, P. A., Neel, B. G. & Wortis, H. H. (2002) J. Exp. Med. 195, Yamaji, T., Teranishi, T., Alphey, M. S., Crocker, P. R. & Hashimoto, Y. (2002) J. Biol. Chem. 277, Angata, T. & Brinkman-Van der Linden, E. (2002) Biochim. Biophys. Acta 1572, Nakamura, K., Yamaji, T., Crocker, P. R., Suzuki, A. & Hashimoto, Y. (2002) Glycobiology 12, Barnes, Y. C., Skelton, T. P., Stamenkovic, I. & Sgroi, D. C. (1999) Blood 93, Crocker, P. R. & Varki, A. (2001) Immunology 103, Nicoll, G., Ni, J., Liu, D., Klenerman, P., Munday, J., Dubock, S., Mattei, M. G. & Crocker, P. R. (1999) J. Biol. Chem. 274, Razi, N. & Varki, A. (1999) Glycobiology 9, Crocker, P. R. (2002) Curr. Opin. Struct. Biol. 12, Crocker, P. R., Mucklow, S., Bouckson, V., McWilliam, A., Willis, A. C., Gordon, S., Milon, G., Kelm, S. & Bradfield, P. (1994) EMBO J. 13, Daniels, M. A., Hogquist, K. A. & Jameson, S. C. (2002) Nat. Immunol. 3, Floyd, H., Nitschke, L. & Crocker, P. R. (2000) Immunology 101, John, B., Herrin, B. R., Raman, C., Wang, Y. N., Bobbitt, K. R., Brody, B. A. & Justement, L. B. (2003) J. Immunol. 170, Law, C. L., Aruffo, A., Chandran, K. A., Doty, R. T. & Clark, E. A. (1995) J. Immunol. 155, Stamenkovic, I. & Seed, B. (1990) Nature 345, Engel, P., Nojima, Y., Rothstein, D., Zhou, L. J., Wilson, G. L., Kehrl, J. H. & Tedder, T. F. (1993) J. Immunol. 150, Sgroi, D., Koretzky, G. A. & Stamenkovic, I. (1995) Proc. Natl. Acad. Sci. USA 92, Tuscano, J., Engel, P., Tedder, T. F. & Kehrl, J. H. (1996) Blood 87, Nitschke, L., Floyd, H., Ferguson, D. J. & Crocker, P. R. (1999) J. Exp. Med. 189, Lanoue, A., Batista, F. D., Stewart, M. & Neuberger, M. S. (2002) Eur. J. Immunol. 32, Hennet, T., Chui, D., Paulson, J. C. & Marth, J. D. (1998) Proc. Natl. Acad. Sci. USA 95, Virts, E. L., Diago, O. & Raschke, W. C. (2003) Blood 101, Collins, B. E., Blixt, O., Bovin, N. V., Danzer, C. P., Chui, D., Marth, J. D., Nitschke, L. & Paulson, J. C. (2002) Glycobiology 12, Blixt, O., Collins, B. E., van den Nieuwenhof, I. M., Crocker, P. R. & Paulson, J. C. (2003) J. Biol. Chem. 278, Zal, T., Zal, M. A. & Gascoigne, N. R. (2002) Immunity 16, Stamenkovic, I., Sgroi, D., Aruffo, A., Sy, M. S. & Anderson, T. (1991) Cell 66, Aruffo, A., Kanner, S. B., Sgroi, D., Ledbetter, J. A. & Stamenkovic, I. (1992) Proc. Natl. Acad. Sci. USA 89, Angata, T., Varki, N. M. & Varki, A. (2001) J. Biol. Chem. 276, Kelm, S., Brossmer, R., Isecke, R., Gross, H. J., Strenge, K. & Schauer, R. (1998) Eur. J. Biochem. 255, Danzer, C. P., Collins, B. E., Blixt, O., Paulson, J. C. & Nitschke, L. (2003) Int. Immunol. 15, Brinkman-Van der Linden, E. C., Sjoberg, E. R., Juneja, L. R., Crocker, P. R., Varki, N. & Varki, A. (2000) J. Biol. Chem. 275, Smith, K. G. & Fearon, D. T. (2000) Curr. Top. Microbiol. Immunol. 245, Bakker, T. R., Piperi, C., Davies, E. A. & Merwe, P. A. (2002) Eur. J. Immunol. 32, Segel, G. B., Cokelet, G. R. & Lichtman, M. A. (1981) Blood 57, Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K. & Watson, J. D. (1994) Molecular Biology of the Cell (Garland, New York). 50. Kataoka, S., Kikuchi, T. & Toyota, T. (1985) Tohoku J. Exp. Med. 145, Kalovidouris, S. A., Blixt, O., Nelson, A., Vidal, S., Turnbull, W. B., Paulson, J. C. & Stoddart, J. F. (2003) J. Org. Chem. 68, Justement, L. B. (2000) Curr. Top. Microbiol. Immunol. 245, O Keefe, T. L., Williams, G. T., Batista, F. D. & Neuberger, M. S. (1999) J. Exp. Med. 189, IMMUNOLOGY Collins et al. PNAS April 20, 2004 vol. 101 no