Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
CD83 Is a Sialic Acid-Binding Ig-Like Lectin (Siglec) Adhesion Receptor that Binds Monocytes and a Subset of Activated CD8ⴙ T Cells1 Nathalie Scholler,2* Martha Hayden-Ledbetter,† Karl-Erik Hellström,* Ingegerd Hellström,* and Jeffrey A. Ledbetter† To help determine CD83 function, a cDNA encoding a soluble protein containing the CD83 extracellular domain was fused with a mutated human IgG1 constant region (CD83Ig) and expressed by stable transfection of Chinese hamster ovary cells. Purified CD83Ig bound to peripheral blood monocytes and a subset of activated CD3ⴙCD8ⴙ lymphocytes but did not bind to FcR. Monocytes that had adhered to plastic lost their ability to bind to CD83Ig after 90 min of in vitro incubation. CD83Ig bound to two of five T cell lines tested, HPB-ALL and Jurkat. The binding to HPB-ALL cells significantly increased when they were grown at a low pH (pH 6.5), whereas binding to Jurkat cells increased after apoptosis was induced with anti-Fas mAb. B cell and monocytic lines did not bind CD83Ig and neither did CD56ⴙ NK cells or granulocytes. Full-length CD83 expressed by a transfected carcinoma line mediated CD83-dependent adhesion to HPB-ALL cells. CD83Ig immunoprecipitated and immunoblotted a 72-kDa protein from HPB-ALL cells. Binding of CD83Ig to HPB-ALL cells was eliminated by neuraminidase treatment of the cells. We conclude that CD83 is an adhesion receptor with a counterreceptor expressed on monocytes and a subset of activated or stressed T lymphocytes, and that interaction between CD83 and its counterreceptor is dependent upon the state of glycosylation of a 72-kDa counterreceptor by sialic acid residues. In view of the selectivity of the expression of CD83 and its ligand, we postulate that the interaction between the two plays an important role in the induction and regulation of immune responses. The Journal of Immunology, 2001, 166: 3865–3872. T he marker CD83 is highly restricted to mature dendritic cells (DC),3 including Langerhans cells and interdigitating reticulum cells in the T cells zones of lymphoid organs (1, 2). It was independently discovered by two different teams, in 1992 by Zhou et al. (2) and in 1993 by Kozlow et al. (3). Mouse CD83 was recently cloned and is up-regulated during DC maturation (4). CD83 transcripts are also detectable in mouse and human brain mRNA by Northern hybridization, although it is not known what cells within the brain express CD83 (3, 5). Isolation of cDNA encoding CD83 revealed that it is a 45-kDa type 1 membrane glycoprotein member of the Ig superfamily (2). It is composed of a single extracellular V-type Ig-like domain, a transmembrane region, and a 40-aa short cytoplasmic domain. The CD83 structure is similar to that of several other members of the Ig superfamily. CD83 shows highly restricted cellular expression, it shares 23% overall identity with myelin protein Po, the most abundant glycoprotein in the peripheral myelin of mammals (6, 7), and has significant homologies with the B7 ancestral gene family Laboratories of *Tumor Immunology and †Immunobiology, Pacific Northwest Research Institute, Seattle, WA 98122 Received for publication September 26, 2000. Accepted for publication January 8, 2001. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by Pacific Northwest Research Institute and by National Institutes of Health Grants CA90143 (to J.A.L.), CA79490 (to K.E.H.) and CA85780 (to I.H.). 2 Address correspondence and reprint requests to Dr. Nathalie Scholler, Laboratory of Tumor Immunology, Pacific Northwest Research Institute, 720 Broadway, Seattle, WA 98122. E-mail address: [email protected] 3 Abbreviations used in this paper: DC, dendritic cell; Siglec, sialic acid-binding Ig-like lectin; PVDF, polyvinylidene difluoride. Copyright © 2001 by The American Association of Immunologists that includes B-G, butyrophilin, MOG, BT, BT2, B7c, B7-1, and B7-2 (8 –11). However, its function is not known. We now present data suggesting that CD83 mediates adhesion of DC to circulating monocytes and to a fraction of activated T cells or stressed T cells by a specific binding of CD83 to a 72-kDa counterreceptor (ligand). We further show that CD83Ig binding to its ligand is eliminated by neuraminidase, an enzyme specific for the most common sialic acid, N-acetylneuraminic acid. Thus, CD83Ig binds to a carbohydrate epitope that depends on sialic acid residues. This classifies CD83 as a sialic acid-binding Ig-like lectin (Siglec; Ref. 12). Our data further suggest that the formation of the carbohydrate epitope recognized by CD83 is influenced by cell growth conditions and can be rapidly altered by cellular stress and early transition to apoptosis. Materials and Methods CD83Ig fusion protein construction A population highly enriched for DC was isolated from 200 ml of human peripheral blood by discontinuous Nycodenz gradient centrifugation, as described elsewhere (13). Nycodenz was purchased as Nycoprep (13% (w/v) Nycodenz, 0.58% (w/v) NaCl, 5 mM Tris-HCl, pH 7.2, density ⫽ 1.068 ⫾ 0.001, 335 ⫾ 5 mOsm/kg) from Nycomed Pharma (Oslo, Norway). At the end of the purification procedure, RNA was directly extracted from DC by TRIzol (Life Technologies, Grand Island, NY) and reverse transcripted (Superscript II; Life Technologies). cDNA from DC was amplified with PCR primers containing a 5⬘ HindIII site: gaataagctt atg tcg cgc ggc ctc cag ctt ctg ctc c and a 3⬘ BglII site in the antisense primer: gag cca gca gca gga gaagatctt ccg ctc tgt att tc. The PCR product (457 bp) was cloned into pCDNA1 human IgG1 (a gift from Robert Peach, Bristol Myers Squibb Pharmaceutical Institute, Princeton, NJ). DNA from recombinant colonies was amplified by Qiagen plasmid maxi kit (Qiagen, Valencia, CA), sequenced, and transfected into COS7 cells. After 3 days, the presence of soluble protein in cell supernatant was checked by Western blot analysis and the fusion protein was purified by protein A-Sepharose 4B 0022-1767/01/$02.00 3866 CD83 IS A SIALIC ACID-BINDING Ig-LIKE LECTIN (SIGLEC) ADHESION RECEPTOR affinity chromatography (Zymed, South San Francisco, CA). Stable transfectants were generated in Chinese hamster ovaries cells by using CD83Ig cDNA cloned into pD18 (14). DMEM medium. In some experiments, incubations were carried in DMEM medium that had been supplemented with 2-fold serial dilutions of sucrose (from 0.6 to 0.1 M). CD83 retrovirus construction and generation of transfected cell line Labeling of beads CD83 cDNA was cloned into pLNCX vector (15). DNA from recombinant colonies was amplified by Qiagen plasmid maxi kit and transfected into ecotropic packaging cells (PE501) by using a calcium phosphate method (16). PE501 viral supernatant was used to infect PG13 cells, a primatespecific packaging line. PG13 supernatant was harvested, filtered, and used to infect 1C, a colon carcinoma line derived in our laboratory. Recombinant colonies were selected by G418 (Life Technologies). Media for cell culture and flow cytometry Cells were cultured with a standard medium (referred to as RPMI medium), which consisted of RPMI 1640 (Life Technologies) supplemented with glutamine (1%; Life Technologies), penicillin/streptomycin (1%; Life Technologies), and 10% FCS (Atlanta Biological, Norcross, GA). All labeling for flow cytometry was conducted at 4°C in a medium that consisted of DMEM (Life Technologies) supplemented with 5% FCS without azide (referred to as DMEM medium). In some experiments, this medium was supplemented with 0.6 M sucrose (Sigma-Aldrich, St. Louis, MO). Purification of PBL and of monocytes PBMCs (5–10 ⫻ 107) were isolated from 50 –100 ml fresh blood from healthy donors by sedimentation in Ficoll-Paque Plus (Amersham Pharmacia Biotech, Uppsala, Sweden) and washed twice in RPMI medium. For experiments involving T cell activation, the PBMCs were resuspended in RPMI medium and stimulated with Abs or with Ab-conjugated beads as described below. Induction of cellular stress and apoptosis Induction of cellular stress of HPB-ALL cells was conducted in four different ways: 1) by incubating the cells in a T75 culture flask with an airtight lid for 2– 4 days in RPMI medium without HEPES buffer; 2) by suspending 2–5 ⫻106 cells in 1 ml of RPMI medium, seeding them into six-well plates, and exposing them for 4 min to UV irradiation by an antimicrobial UV lamp placed 50 cm above the cells inside a biosafety cabinet; 3) by incubating the cells in culture medium at different pH (6 –7.4) or in culture medium supplemented with 25 mM HEPES buffer (Life Technologies); or 4) by exposing the cells to oxidative stress. This was accomplished by adding to the medium 2-fold serial dilutions of hydrogen peroxide (10 mM, 5 mM, 2.5 mM, 1.25 mM, and 0.625 mM) and incubating 2–5 ⫻106 cells for 10 min at 37°C. After two washes, the cells were incubated in RPMI medium at 37°C until labeling 1 h to 7 days later. Jurkat cell apoptosis was induced by anti-CD95 (Fas) mAb from Beckman Coulter (Palantine, IL). Petri dishes were coated with 1 g/ml antimurine IgM mAb from Beckman Coulter in bicarbonate buffer (Sigma) for 2 h at 37°C. Jurkat cells were washed three times with PBS and incubated for 1 h with 4-fold serial dilutions anti-Fas mAb. After two washes with culture medium, anti-Fas-coated Jurkat cells were incubated in the antimurine IgM-coated petri dish in culture medium at 37°C overnight. Flow cytometry analysis Monoclonal Abs recognizing the following Ags were used: CD83 (HB15A), IgG1 and IgG2a isotype controls from Immunotech, and CD11b (17), CD4, CD8, and CD3 from BD PharMingen (Lexington, KY). To detect apoptosis, we used an annexin V kit (Beckman Coulter) according to the manufacturer’s instructions. CD83Ig, CD80Ig (14), and CD40Ig (18) fusion proteins were biotinylated with EZ-Link N-hydroxy-succinimi-biotin (Pierce, Rockford, IL) kit according to the manufacturer’s procedure. In some experiments cells, the HPB-ALL line were incubated with neuraminidase (Sigma) for 15 min at 37°C (1 U/5 ⫻ 106cells) or with neuraminidase together with 2.5 mg/ml of a sialidase inhibitor (2,3-dehydro-2-deoxy-Nacetylneuraminic acid, Sigma). Cells were washed twice in DMEM medium and labeled at 4°C for 45 min with 1 g/ml biotinylated CD83Ig followed by two washes in DMEM medium and labeled for 15 min at 4°C with 3 l/100 l of PE streptavidin (BD PharMingen). Cold competition experiments Cells or beads were washed twice and incubated for 15–30 min at 4°C with 50 g/ml unlabeled CD83Ig or with 20 g/ml anti-CD83 mAb in DMEM medium. The cells then were labeled with PE-conjugated anti-CD83 mAb or with 1 g/ml biotin-conjugated CD83Ig as described previously in A total of 50 g of material to be labeled (CD83Ig, anti-CD83 mAb, and a mixture of anti-CD28 mAb and anti-CD3 mAb) were conjugated to magnetic beads (Tosylativated Dynabeads M-450; Dynal, Lake Success, NY) according to a published protocol (19). Beads conjugated with CD83Ig or anti-CD83 mAb were used to test the specificity of CD83Ig fusion protein. Anti-CD28/anti-CD3 mAb-conjugated beads were used to stimulate T cells. T cell stimulation with anti-CD28/anti-CD3 mAb-coated beads PBMC were incubated 5 days with conjugated or nonconjugated beads (control) in RPMI medium at 37°C. After that time, the beads were magnetically removed and the cells resuspended in RPMI medium supplemented with 10 IU/ml of rIL-2 (Roche Molecular Biochemicals, Indianapolis, IN) and cultivated for 2 wk. Cell lysate preparation HPB-ALL (5 ⫻ 107–10 8) cells were washed three times in ice-cold PBS, resuspended in 4 ml of lysis buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and protease inhibitor cocktail tablets, complete, mini; Roche Molecular Biochemicals), and incubated on ice for 30 min. Lysates were centrifuged for 30 min at 12,000 ⫻ g at 4°C, and the supernatants were harvested. In some experiments, HPB-ALL cells were treated by neuraminidase before lysis, as described in the flow cytometry analysis section. Immunoprecipitation Supernatants of cell lysates were incubated for 1 h with 50 l of streptavidin-Sepharose 4B conjugate (Zymed) at 4°C. Streptavidin-Sepharose was removed by centrifugation, and 1 ml of lysate was incubated overnight at 4°C with 50 g of biotinylated CD83Ig. Immunoprecipitated proteins were separated by 3 h of incubation with streptavidin-Sepharose followed by six washes with 1 ml of lysis buffer and one wash with PBS. Streptavidin-Sepharose was then resuspended in 2⫻ SDS sample buffer (Novex, San Diego, CA) with 1% 2-ME and boiled for 10 min, after which 20 l of the supernatant was loaded on a 4 –12% gradient gel (Novex). Subsequently, the migration gel was blotted and probed as described in the following section. Western blotting analysis Supernatants of cell lysates, streptavidin-Sepharose-purified samples, and CD83Ig were eluted in SDS-PAGE sample buffer containing 2-ME and then boiled. CD83Ig (10 g) was also reduced with 10 mM DL-DTT (Sigma) for 30 min at 37°C and free sulfhydryl residues were alkylated with 25 mM of iodoacetamide, pH 8 (Sigma), for 1 h at 37°C. Samples were run on a tris-glycine 4 –12% gradient gel (Novex). The amount of proteins in cell lysates was quantified with micro bicinchoninic acid protein reagent kit (Pierce) according to the manufacturer’s instructions. After migration proteins were transferred to polyvinylidene difluoride (PVDF) membranes (Novex), they first were probed with biotinylated CD83Ig 0.5 g/ml, washed four times (WesternBreeze; Invitrogen, Carlsbad, CA), and then probed with 1:5000 streptavidin-HRP (BD PharMingen). The signal was detected by ECL (Amersham Pharmacia Biotech) according to the manufacturer’s protocol and quantified by using OptiQuant version 03.00 (Packard Instruments, Meriden, CT). Adhesion assays Wild-type 1C cultured human carcinoma cells and CD83-transfected cells obtained from these (1C/CD83) were seeded at 5 ⫻ 105 cells/well into a 48-well plate (Costar, Cambridge, MA) and incubated in culture medium at 37°C for 24 h to allow their adhesion. After 24 h, the plates were washed one time with fresh medium to remove nonadherent cells. Nonstressed or HPB-ALL cells stressed by growth at low oxygen level were labeled with 50 Ci of 51Cr (Amersham Pharmacia Biotech) for 45 min at 37°C then washed twice and resuspended in RPMI medium at 2.5 ⫻ 106 cells/ml in the presence of 10 g/ml anti-2 integrin mAb 60.3 (20) or with 10 g/ml anti-CD83 mAb. Next, 2-fold serial dilutions of 51Cr-labeled HPB-ALL cells were distributed and incubated for 1– 4 h at 37°C with 1C or 1C/CD83 cells. Finally, the cells were washed once with PBS and once according to a procedure derived from gravity flow wash (21). According to this procedure, the plates were immersed in PBS in a large container and suspended upside down above the bottom of the container for 15 min to allow The Journal of Immunology 3867 nonadherent cells to detach. They then were turned slowly right side up, were removed from the container, and the PBS in each well was removed by aspiration. The adherent cells remaining after this procedure were lysed by 100 l of PBS plus 0.2% Triton X-100 (Fisher Scientific, Fairlawn, NJ). Lysates (40 l) were transferred into LumaPlate-96 plates (Packard Instruments) and counted with a Top-Count NXT (Packard Instruments). ever, after reduction with DTT and alkylation with iodoacetamide, CD83Ig migrated as a single band of an ⬃98-kDa monomer (Fig. 1D, lane 2). Results According to flow cytometry analysis of fresh PBMC, biotinylated CD83Ig was found to bind to ⬍1% of CD3⫹ cells and to ⬃4% of CD3⫺ cells in the lymphocyte-scatter gate (gate 1), whereas biotinylated CD40Ig used as control did not bind (Fig. 2, B and C). In the larger cell-scatter gate (gate 2), biotinylated CD83Ig bound to ⬎75% of cells that expressed CD11b (Fig. 2D), CD4 low⫹ (data not shown), and CD14 (Fig. 2E), i.e., cells with the distinctive characteristics of circulating monocytes. When CD83Ig labeling was performed in the presence of anti-CD83 mAb (HB15A), the binding of CD83Ig to the CD14⫹ cells consistently increased up to 90% (Fig. 2F), whereas an isotype control mAb did not increase the binding of CD83Ig to these cells. The binding of CD83Ig to monocytes was specific, because biotinylated CD40Ig did not bind at all (data not shown). This suggests that the HB15A mAb binding epitope is distinct from the active binding site of CD83 ligand, consistent with the lack of function described for this anti-CD83 mAb. The CD83Ig binding to monocytes decreased after 90 min of culture (Fig. 3, A and B). This decrease was less in the presence of the anti-CD18 (2 integrin) Ab 60.3 (Fig. 3, C and D), which blocks the adhesion of monocytes to plastic (22–24); because adhesion induces monocyte activation, this suggests that expression of the CD83 ligand on monocyte correlates with a resting stage. In contrast, the binding of CD83Ig to CD3⫹ T lymphocytes increased from ⬍1% for resting cells (Fig. 2C) to 6% for cells activated by 2 wk of culture with anti-CD3/anti-CD28-conjugated beads (Fig. 4B). CD3⫹CD8⫹ T lymphocytes bound to CD83Ig (Fig. 4C), whereas CD3⫹CD4⫹ T lymphocytes did not (Fig. 4D). A total of 90% of the cells binding to CD83Ig were costained with annexin V (Fig. 4E) as compared with ⬍6% of the cells binding to CD80Ig, used as control (Fig. 4F). Altogether, these data indicate that CD83Ig binds to a ligand the expression of which is regulated by cell activation and apoptosis. No binding of CD83Ig was found on Construction of CD83Ig and verification of its activity We constructed a CD83Ig fusion protein as described in Materials and Methods (Fig. 1A). It was engineered without an Ig hinge region between the coding sequence for CD83 extracytoplasmic domain (432 bp) and the CH2 and CH3 domains and contains two mutations, one at 231 bp, transforming valine to proline, and the other at 531 bp, transforming a proline to a serine. These structural modifications eliminated the binding to FcR. CD83Ig did not bind to cells expressing Fc␥RI (U937), Fc␥RII (normal B cells and B cell leukemia lines, Raji, Ramos, Bjab), or Fc␥RIII (blood CD16 plus NK cells; data not shown). To check the specificity and proper folding of the CD83Ig fusion protein, experiments were performed that showed that PE-labeled anti-CD83 mAb bound to CD83Igconjugated beads and that a PE-labeled isotype control mAb did not (Fig. 1B). The binding of PE-labeled anti-CD83 mAb to the CD83Ig-conjugated beads was partially blocked by preincubation with an unlabeled anti-CD83 mAb (20 g/ml) for 15 min at 4°C (Fig. 1B). Conversely, CD83Ig bound to anti-CD83 mAb-conjugated beads whereas CD40Ig did not bind (Fig. 1C). The binding of CD83Ig to beads conjugated with anti-CD83 mAb was completely blocked by preincubation with an unlabeled anti-CD83 mAb (20 g/ml) for 15 min at 4°C (Fig. 1C, picture 2). 2-ME incompletely reduced CD83Ig, which migrated as a mixture of a 60-kDa monomer and a 120-kDa dimer (Fig. 1D, lane 1). How- FIGURE 1. Construction and verification of CD83Ig. A, CD83Ig was constructed by fusing the CD83 extracytoplasmic domain with a hingetruncated human IgG1. B, CD83Ig-conjugated beads were labeled with a PE-labeled isotype control mAb (picture 1), a PE-labeled anti-CD83 mAb after a preincubation with an unlabeled anti-CD83 mAb (picture 2), or a PE-conjugated anti-CD83 mAb in DMEM medium (picture 3). C, AntiCD83 mAb-conjugated beads were labeled with biotinylated CD40Ig plus PE streptavidin (picture 1), or with biotinylated CD83Ig plus PE streptavidin after a preincubation with an unlabeled anti-CD83 mAb (picture 2), or with biotinylated CD83Ig plus PE streptavidin in DMEM medium (picture 3). D, After treatment by 2-ME, CD83Ig migrated as a 60-kDa monomer and a 120-kDa homodimerized band. E, After treatment by DTT and iodoacetamide, CD83Ig migrated as a 98-kDa monomer band. CD83Ig binds to circulating monocytes and to a subset of activated T lymphocytes FIGURE 2. CD83Ig binds to circulating monocytes. PBMC from healthy donors were purified by Ficoll and stained immediately after purification. Cells were gated according to their forward and side angle light scatter proprieties (A). In B and C, lymphocytes (gate 1) were labeled with FITC-conjugated anti-CD3 mAb and (B) with biotinylated CD83Ig plus PE streptavidin; or (C) with biotinylated CD40Ig plus PE streptavidin. In D–F, monocytes from a different donor (gate 2) were labeled with biotinylated CD83Ig plus PE streptavidin and (D) FITC-conjugated CD11b. In E and F, monocytes were labeled with FITC-conjugated anti-CD14 mAb immediately after purification (E) or after 30 min of incubation at 4°C with 20 g/ml anti-CD83 mAb (F). 3868 CD83 IS A SIALIC ACID-BINDING Ig-LIKE LECTIN (SIGLEC) ADHESION RECEPTOR FIGURE 3. Adhesion down-regulates CD83 ligand expression on monocytes. Monocytes were labeled with CD83Ig plus PE streptavidin and FITC anti-CD11b (A) immediately after Ficoll preparation; (B) after 90 min in culture at 37°C; (C) after 24 h in culture at 37°C in medium; or (D) after 24 h in culture at 37°C in medium supplemented with the anti-CD18 mAb 60.3. CD56-positive NK cells, nor to granulocytes or erythrocytes (data not shown). The CD83 ligand is expressed on two of five tested T lymphocyte lines and its expression is influenced by cellular stress The binding of CD83Ig to B, T, myeloid, and monocyte cell lines was examined. CD83Ig did not bind to any of four B cell lines (Nalm6, Reh, Bjab, DHL10), or to two myeloid cell lines (Thp1, HL60) or to the monocyte cell line U937 (data not shown). Similarly, no binding was observed to Molt4, CEM, Hut78, or Jurkat cells in exponential growth. However, CD83Ig did bind to HPBALL cells (25) in light-scatter gates 1 and 2 (Fig. 5, A and B). Interestingly, we found that CD83Ig binding increased after 24 h of incubation in a flask with a tightly closed lid, and this increase was greater for cells in gate 2 than for cells in gate 1 (Fig. 5, C–F). This suggested that expression of the CD83 ligand might be regulated by cell growth conditions and/or by cellular stress and led us to study the effect of cellular stress, including apoptotic events, on CD83Ig binding. The increase of CD83Ig binding on HPB-ALL cells in gate 2 was not induced by UV irradiation, by hydrogen peroxide, or by incubation in a tight-close lid flask in the presence of HEPES (data not shown). This suggested that an additional condition was required to increase the binding of CD83Ig. Additional experiments demonstrated that the CD83Ig binding increase in gate 2 was correlated with the pH of the culture medium. When cells were incubated in a tight-close lid flask at pH 7, 10% of HPB-ALL cells bound to CD83Ig in gate 2 (Fig. 5E), whereas at pH 6.5 the binding increased up to 44% (Fig. 5F). In contrast, in gate 1 the binding of CD83Ig to HPB-ALL cells did not significantly vary with the pH (Fig. 5, C and D). Finally, 5– 6% of CD83Ig-bearing cells in gate 1 (Fig. 5, C and D) and none of the CD83Ig-bearing cells in gate 2 (Fig. 5, E and F) were labeled by annexin V. Thus, CD83Ig binding to HPB-ALL cells depends on the pH of the cell medium and is not dependent on apoptosis, because the cells that bound CD83Ig did not bind annexin V. Because HPB-ALL cells are not sensitive to apoptosis mediated by anti-Fas mAb (26), we chose Jurkat cells for studying CD83Ig FIGURE 4. CD83 ligand expression is up-regulated on CD8⫹-activated lymphocytes. PBL were activated 5 days with anti-CD3/antiCD28conjugated beads and, after removal of beads, were incubated for 2 wk in medium supplement with 10 IU/ml of IL-2. Lymphocytes then were labeled with PE streptavidin and FITC isotype control mAb (A). In B–E, lymphocytes were labeled with biotinylated CD83Ig plus PE streptavidin and (B) FITC anti-CD3 mAb; (C) FITC anti-CD8 mAb; (D) FITC antiCD4 mAb; or (E) FITC annexin V. As control, lymphocytes were labeled with biotinylated CD80Ig plus PE streptavidin and FITC annexin V (F). binding to cells undergoing apoptosis. Jurkat cells were incubated with serial dilutions of anti-Fas Ab, as described in Materials and Methods. Although the binding of biotinylated CD86Ig, which was used as a control, did not vary (data not shown), apoptotic Jurkat cells expressed the CD83 ligand after 12 h of treatment with antiFas Ab coincidentally with the binding of annexin V. The binding of CD83Ig to Jurkat cells was proportional to the concentration of anti-Fas mAb (Fig. 5, G–I). Unlabeled CD83Ig blocks biotinylated CD83Ig binding in hypertonic medium but not in DMEM medium We were unable to block the binding of biotin-conjugated CD83Ig to HPB-ALL cells with unconjugated CD83Ig and therefore hypothesized that there could be a rapid, receptor-mediated internalization of the fusion protein, as described in some other systems (27). Thus, we tested CD83 ligand endocytosis in the presence of a high sucrose hypertonic medium known to block internalization by preventing clathrin-coated pit formation (28). Fig. 6 shows that it was possible to block the binding of biotin-conjugated CD83Ig with unlabeled CD83Ig in the presence of 0.6 M sucrose and that the blocking was proportional to the sucrose concentration. CD83Ig immunoprecipitates and immunoblots a 72-kDa protein from HPB-ALL cell lysates Lysates of HPB-ALL cells were immunoprecipitated with biotinconjugated CD83Ig and collected on streptavidin-Sepharose beads. The Journal of Immunology 3869 FIGURE 5. Cellular stress up-regulated CD83 ligand expression on T lymphocyte cell lines. A, HPB-ALL cells were analyzed in two light-scatter gates. B, HPBALL cells from gate 2 were stained with PE streptavidin as a negative control (black area) or with biotinylated CD83Ig plus PE streptavidin (lines). The signal was brighter when cells were cultivated with a tightly closed lid (straight line) than in standard culture conditions (dotted line). In C–F, HPB-ALL cells were labeled with biotinylated CD83Ig plus PE streptavidin and FITC annexin V. HPB-ALL cells were cultivated at pH 7 and gated on gate 1 (C) or gate 2 (D). HPB-ALL cells were cultivated at pH 6.5 and gated on gate 1 (E) or gate 2 (F). In G–I, cells from the Jurkat line were stained with biotinylated CD83Ig plus PE streptavidin after being cultured in the presence of (G) 10 g/ml isotype control mAb; (H) 1 g/ml anti-Fas mAb; or (I) 5 g/ml anti-Fas mAb. After washing, the beads were eluted with SDS sample buffer and separated on tris-glycine 4 –12% gradient gels (Fig. 7). After transfer, filters were blotted with biotinylated CD83Ig. Fig. 7 shows that a 72-kDa molecule binds to CD83Ig in HPB-ALL cell lysates, which were immunoprecipitated with biotinylated CD83Ig (lane 1) or directly blotted without immunoprecipitation (lanes 3– 6, FIGURE 6. Unlabeled CD83Ig blocks biotinylated CD83Ig binding in hypertonic medium but not in DMEM medium. HPB-ALL cells were labeled with PE streptavidin as a negative control (black) and with biotinylated CD83Ig plus PE streptavidin as a positive control (white). Binding was carried at 4°C in DMEM medium without azide (controls and dark gray bar) or in the presence of unlabeled CD83Ig (20 g/ml) and serial dilutions of sucrose. 2-fold serial dilutions). Biotinylated CD83Ig was detected in the control lane (lane 2). Thus, the ligand epitope detected by CD83Ig is not destroyed by boiling in SDS. The CD83 ligand binding site contains sialic acid Experiments were performed to test whether the epitope detected by CD83Ig on the 72-kDa protein was part of the protein itself or was a carbohydrate attached to the protein. We found that a treatment of HPB-ALL cells with neuraminidase diminished the CD83Ig binding (Fig. 8B), although treatment with sialidase usually enhances cell-cell interactions (29, 30) by removing negatively charged sialic acids. Inhibition of neuraminidase by a sialidase inhibitor, 2,3-dehydro-2-deoxy-N-acetylneuraminic acid, prevented the reduction of CD83Ig binding to HPB-ALL (Fig. 8C) as well as to fresh monocytes (data not shown). Lysates were prepared from HPB-ALL cells before or after treatment by neuraminidase, and their protein concentrations were quantified. Even FIGURE 7. CD83Ig immunoprecipitates HPB-ALL lysates. A SDS trisglycine 4 –12% gradient gel was loaded on lane 1 with HPB-ALL lysates immunoprecipitated with 10 g/ml of biotinylated CD83Ig and purified with streptavidin-Sepharose, on lane 2 with biotinylated CD83Ig alone (200 ng), and on lanes 3– 6 with 2-fold serial dilutions of HPB-ALL lysates. After migration, the gel was blotted on PVDF membrane, probed with biotinylated CD83Ig, and the signal was detected by streptavidin-HRP chemoluminescence. 3870 CD83 IS A SIALIC ACID-BINDING Ig-LIKE LECTIN (SIGLEC) ADHESION RECEPTOR FIGURE 8. CD83 ligand binding site contains a sialic acid. A, HPBALL cells grow at pH 6.5 were stained with PE streptavidin as a negative control (dotted line) or with biotinylated CD83Ig plus PE streptavidin (black area). B, HPB-ALL cells were stained with biotinylated CD83Ig plus PE streptavidin in flow cytometry medium (dotted line) or after a treatment with neuraminidase (1 U/ml) for 15 min at 37°C (black area). C, HPB-ALL cells were stained with biotinylated CD83Ig plus PE streptavidin after a neuraminidase treatment (dotted line) or after neuraminidase treatment in presence of 2.5 mg/ml of a neuraminidase inhibitor for 15 min at 37°C (black area). D, A SDS tris-glycine 4 –12% gradient gel was loaded with 2-fold serial dilutions of HPB-ALL cell lysates, without (lanes 1 and 2) and with neuraminidase treatment (lanes 3 and 4). Protein concentrations of cell lysates were determined by micro bicinchoninic acid to be 25 g (lanes 1 and 3) and 50 g (lanes 2 and 4). After migration, the gel was blotted on PVDF membrane and probed by biotinylated CD83Ig plus HRPstreptavidin. The signal was detected by chemoluminescence, scanned, and quantified by OptiQuant program. The difference of digital light units after background subtraction was 37% between lanes 1 and 3 and 31% between lanes 2 and 4. though the protein concentration of the cell lysates was unchanged, the CD83Ig reactivity with the 72-kDa protein was reduced by 31–37% (Fig. 8D) with neuraminidase treatment of HPB-ALL cells, as measured by OptiQuant program (Packard, Meriden, CT). This indicates that the ability of the CD83 ligand to bind to CD83 is dependent on glycosylation by sialic acid residues. The binding between CD83 and the CD83 ligand mediates adhesion We used cells from 1C colon cancer line transfected with CD83 retrovirus (1C/CD83; Fig. 9A) to test their adhesion to HPB-ALL (which express the CD83Ig ligand); wild-type 1C cells were used as a control. HPB-ALL cells were incubated for 3 days at pH 7.4 or at pH 6.5, harvested, and labeled with chromium 51. After two washes to remove unincorporated radioactivity, 51Cr-labeled cells were incubated for 1– 4 h with adherent 1C or 1C/CD83 in RPMI medium (pH 7.4) supplemented with either 10 g/ml of the antiintegrin mAb 60.3 to avoid nonspecific adherence or with 10 g/ml of an anti-CD83 mAb. The strongest binding was observed after 3 h of incubation (Fig. 9, B and C). After three washes with PBS, adherent cells were lysed with 100 l of lysate buffer, and 40 l was harvested and counted. HPB-ALL cells showed a higher level of adhesion to 1C/CD83 than to 1C, both in medium (data not shown) and in the presence of 60.3 (Fig. 9B). The anti-CD83 mAb blocked the adhesion of HPB-ALL cells incubated at pH 6.5 to 1C/CD83 (Fig. 9C). This adhesion pattern exactly followed the pattern of the CD83 ligand, indicating that the CD83-CD83 ligand interaction mediates adhesion. FIGURE 9. The binding between CD83 and CD83 ligand mediates adhesion. A, 1C wild-type cells (dotted line) or 1C/CD83 cells (black area) were labeled with PE anti-CD83 mAb. B, 51Cr-labeled HPB-ALL cells either in exponential growth or stressed by growth at a low pH (pH 6.5), were incubated 3 h with adherent 1C (wild type; white) or 1C/CD83 (black) in RPMI medium supplemented with 10 g/ml of anti-2 integrin mAb 60.3. C, Serial dilutions of 51Cr-labeled stressed HPB-ALL CD83 (5 ⫻105; 2.5 ⫻ 105; 1.25 ⫻ 105; 0.625 ⫻ 105/well) were incubated for 3 h with adherent wild-type 1C cells (squares) or transfected 1C/CD83 cells (circles) in RPMI medium (open) or supplemented with 10 g/ml antiCD83 mAb (filled). Discussion CD83 is selectively expressed by mature DC and has been used to define the maturation stage and purity of DC populations (1, 31). Its restricted expression suggests that CD83 may have a specialized function during Ag presentation by DC, perhaps contributing to their ability to efficiently activate T cells. However, little data has been presented to support this hypothesis, and anti-CD83 mAbs have not been reported to alter the ability of DC to activate T cells or to synthesize cytokines. We present here evidence that CD83 is an adhesion receptor, because a soluble CD83Ig fusion protein binds to blood monocytes, two of five leukemia T cell lines, and a subset of activated CD8⫹ lymphocytes and that this binding increased as a result of cellular stress. In addition, transfection of CD83 into a carcinoma line allowed them to bind to HPB-ALL T cells. These results suggest that CD83 contributes to cell adhesion by facilitating DC interactions with monocytes and subpopulations of activated and/or stressed CD8⫹ T cells. DC can be derived from monocytes by stimulation with IL4 and GM-CSF and/or Ag (32–34). Because the CD83 ligand disappears quickly during monocyte culture, adhesion between CD83 on DC and its ligand on monocytes may represent a regulatory mechanism to control APC maturation. Conversely, mature DC may stimulate the monocytes to release chemokines and/or cytokines to amplify immune responses. The Journal of Immunology In cultures of lymphocytes, binding of CD83Ig was seen on 6% of CD8⫹ T cells after 2 wk of activation through their CD3 and CD28 receptors. This binding increased 2- to 3-fold in the absence of APC, and addition of mature DC but not CD83Ig to the T cell cultures down-regulated the expression of CD83 ligand (data not shown). This suggests that CD83-specific interactions between subpopulations of T cells and DC may be important when the T cells are at a specific stage of maturation. It is not yet known whether or not the activated T cells that expresses the CD83 ligand represent subpopulations of stressed cells; however, activated cells that bound CD83Ig were also annexin V positive. Neither is it known whether interaction of the ligand-positive stressed cells with DC is a mechanism for removal or rescue of such cells. A recent publication by Cramer et al. (35) suggests that mouse CD83 ligand is expressed by B cells labeled with an anti-CD45R mAb (B220), because mouse CD83Ig bound to B220⫹ splenocytes from normal BALB/c mice but not to splenocytes from MT (B cell knockout) mice. In contrast, in human peripheral blood we did not detect any CD83Ig binding to B cells labeled with an antiCD19 mAb, while in mouse blood we could detect human CD83Ig binding to circulating monocytes (data not shown). In our study, we demonstrated that cellular stress including cell growth at low pH or progression into apoptosis after anti-Fas stimulation can increases the binding of CD83Ig. It is possible that the CD83Ig binding to mouse splenocytes described by Cramer et al. may be related to the stage of B cell activation or cell stress. It is also possible that there are differences between mouse and human leukocytes and that the mutation in human CMP-sialic acid hydroxylase (36) may alter the sialic acid recognition by CD83Ig. CD83 is structurally related to the B7 ancestral gene family, and its closest homology is 23% of identity with the myelin protein Po, which is an I-type lectin that recognizes a sulfated carbohydrate. Therefore, it is highly interesting that the CD83 ligand contains a sialic acid, classifying CD83 as a siglec i.e., it belongs to a subfamily of I-type lectins that can bind sialic acids and presently includes nine members. All members of the siglec family share a restricted cell expression, to the hemopoietic and immune systems for most of them (37), and to the nervous system for myelin-associated glycoprotein (siglec 4A; Ref. 38). Although four of eight siglecs that have been characterized today share ⬎60% protein sequence homology and are closely linked to 19q13.333– 41, the binding of four other siglecs, siglec 3 (CD33; Ref. 39), siglec 5 (40), siglec 8 (41), and siglec 9 (42, 43), varies with respect to both the nature of the sialic acid and its linkage to subterminal sugars (37). The extracellular regions of these siglecs are made up of an N-terminal V-set Ig-like domain followed by a variable number of C2-set domains. Importantly, sialic acid binding depends on the N-terminal V-set domain (44). We conclude that CD83 is an adhesion receptor that belongs to the siglec family and that its binding site depends on glycosylation on sialic acids of a protein of 72 kDa. We further conclude that formation of the carbohydrate epitope recognized by CD83 is highly sensitive to cell-growth conditions and can be rapidly altered by cell stress and/or early transition to apoptosis. Our data suggest that interaction between CD83 and its ligand plays a role in intracellular communications involving DC, circulating monocytes, and certain populations of activated and/or stressed T lymphocytes. Acknowledgments We thank Dr. Sen-Itiroh Hakomori for helpful discussions and valuable advice. We also thank Dr. Peter S. Linsley, who introduced N.S to studies aimed at defining the CD83 ligand. 3871 References 1. Zhou, L. J., and T. F. Tedder. 1995. Human blood dendritic cells selectively express CD83, a member of the immunoglobulin superfamily. J. Immunol. 154: 3821. 2. Zhou, L. J., R. Schwarting, H. M. Smith, and T. F. Tedder. 1992. A novel cellsurface molecule expressed by human interdigitating reticulum cells, Langerhans cells, and activated lymphocytes is a new member of the Ig superfamily. J. Immunol. 149:735. 3. Kozlow, E. J., G. L. Wilson, C. H. Fox, and J. H. Kehrl. 1993. Subtractive cDNA cloning of a novel member of the Ig gene superfamily expressed at high levels in activated B lymphocytes. Blood 81:454. 4. Berchtold, S., T. Jones, P. Muhl-Zurbes, D. Sheer, G. Schuler, and A. Steinkasserer. 1999. The human dendritic cell marker CD83 maps to chromosome 6p23. Ann. Hum. Genet. 63:181. 5. Twist, C. J., D. R. Beier, C. M. Disteche, S. Edelhoff, and T. F. Tedder. 1998. The mouse CD83 gene: structure, domain organization, and chromosome localization. Immunogenetics 48:383. 6. Filbin, M. T., F. S. Walsh, B. D. Trapp, J. A. Pizzey, and G. I. Tennekoon. 1990. Role of myelin Po protein as a homophilic adhesion molecule. Nature 344:871. 7. Filbin, M. T., and G. I. Tennekoon. 1991. The role of complex carbohydrates in adhesion of the myelin protein, Po. Neuron 7:845. 8. Pham-Dinh, D., M. G. Mattei, J. L. Nussbaum, G. Roussel, P. Pontarotti, N. Roeckel, I. H. Mather, K. Artzt, K. F. Lindahl, and A. Dautigny. 1993. Myelin/ oligodendrocyte glycoprotein is a member of a subset of the immunoglobulin superfamily encoded within the major histocompatibility complex. Proc. Natl. Acad. Sci. USA 90:7990. 9. Vernet, C., J. Boretto, M. G. Mattei, M. Takahashi, L. J. Jack, I. H. Mather, S. Rouquier, and P. Pontarotti. 1993. Evolutionary study of multigenic families mapping close to the human MHC class I region. J. Mol. Evol. 37:600. 10. Gruen, J. R., S. R. Nalabolu, T. W. Chu, C. Bowlus, W. F. Fan, V. L. Goei, H. Wei, R. Sivakamasundari, Y. Liu, H. X. Xu, et al. 1996. A transcription map of the major histocompatibility complex (MHC) class I region. Genomics 36:70. 11. Henry, J., M. T. Ribouchon, C. Offer, and P. Pontarotti. 1997. B30.2-like domain proteins: a growing family. Biochem. Biophys. Acta 235:162. 12. Crocker, P. R., S. Mucklow, V. Bouckson, A. McWilliam, A. C. Willis, S. Gordon, G. Milon, S. Kelm, and P. Bradfield. 1994. Sialoadhesin, a macrophage sialic acid binding receptor for haemopoietic cells with 17 immunoglobulin-like domains. EMBO J. 13:4490. 13. McLellan, A. D., G. C. Starling, and D. N. Hart. 1995. Isolation of human blood dendritic cells by discontinuous Nycodenz gradient centrifugation. J. Immunol. Methods 184:81. 14. Hayden, M. S., L. S. Grosmaire, N. A. Norris, L. K. Gilliland, G. Winberg, D. Tritschler, T. T. Tsu, P. S. Linsley, R. S. Mittler, P. D. Senter, H. P. Fell, and J. A. Ledbetter. 1996. Costimulation by CD28 sFv expressed on the tumor cell surface or as a soluble bispecific molecule targeted to the L6 carcinoma antigen. Tissue Antigens 48:242. 15. Miller, A., and G. Rosman. 1989. Improved vectors for gene transfer and expression. BioTechniques 7:980. 16. Hillova, J., M. Hill, G. Goubin, and D. Dantchev. 1975. Infectivity of Rous sarcoma cell DNA: comparison of two techniques of transfection assay. Intervirology 5:367. 17. Prieto, J., P. G. Beatty, E. A. Clark, and M. Patarroyo. 1988. Molecules mediating adhesion of T and B cells, monocytes and granulocytes to vascular endothelial cells. Immunology 63:631. 18. Hollenbaugh, D., J. Douthwright, V. McDonald, and A. Aruffo. 1995. Cleavable CD40Ig fusion proteins and the binding to sgp39. J. Immunol. Methods 188:1. 19. Garlie, N. K., A. V. LeFever, R. E. Siebenlist, B. L. Levine, C. H. June, and L. G. Lum. 1999. T cells coactivated with immobilized anti-CD3 and anti-CD28 as potential immunotherapy for cancer. J. Immunother. 22:336. 20. Beatty, P. G., J. A. Ledbetter, P. J. Martin, T. H. Price, and J. A. Hansen. 1983. Definition of a common leukocyte cell-surface antigen (Lp95–150) associated with diverse cell-mediated immune functions. J. Immunol. 131:2913. 21. Song, Y., D. A. Withers, and S. Hakomori. 1998. Globoside-dependent adhesion of human embryonal carcinoma cells, based on carbohydrate-carbohydrate interaction, initiates signal transduction and induces enhanced activity of transcription factors AP1 and CREB. J. Biol. Chem. 273:2517. 22. Aicher, A., M. Hayden-Ledbetter, W. A. Brady, A. Pezzutto, G. Richter, D. Magaletti, S. Buckwalter, J. A. Ledbetter, and E. A. Clark. 2000. Characterization of human inducible costimulator ligand expression and function. J. Immunol. 164:4689. 23. Jonjic, N., P. Jilek, S. Bernasconi, G. Peri, I. Martin-Padura, S. Cenzuales, E. Dejana, and A. Mantovani. 1992. Molecules involved in the adhesion and cytotoxicity of activated monocytes on endothelial cells. J. Immunol. 148:2080. 24. Carlos, T., A. Dobrina, R. Ross, and J. Harlan. 1990. Multiple receptors on human monocytes are involved in adhesion to cultured human endothelial cells. J. Leukocyte Biol. 48:451. 25. Ledbetter, J. A., G. L. Schieven, V. M. Kuebelbeck, and F. M. Uckun. 1991. Accessory receptors regulate coupling of the T-cell receptor complex to tyrosine kinase activation and mobilization of cytoplasmic calcium in T-lineage acute lymphoblastic leukemia. Blood 77:1271. 26. Russo, M., R. Palumbo, I. Tedesco, G. Mazzarella, P. Russo, G. Iacomino, and G. L. Russo. 1999. Quercetin and anti-CD95(FAS/Apo1) enhance apoptosis in HPB-ALL cell line. FEBS Lett 462:322. 27. Francisco, J. A., P. A. Kiener, P. Moran-Davis, J. A. Ledbetter, and C. B. Siegall. 1996. Cytokine activation sensitizes human monocytic and endothelial cells to the cytotoxic effects of an anti-CD40 immunotoxin. J. Immunol. 157:1652. 3872 CD83 IS A SIALIC ACID-BINDING Ig-LIKE LECTIN (SIGLEC) ADHESION RECEPTOR 28. Heuser, J. E., and R. G. Anderson. 1989. Hypertonic media inhibit receptormediated endocytosis by blocking clathrin-coated pit formation. J. Cell Biol. 108:389. 29. Varki, A. 1994. Selectin ligands. Proc. Natl. Acad. Sci. USA 91:7390. 30. Varki, A. 1996. “Unusual” modifications and variations of vertebrate oligosaccharides: are we missing the flowers for the trees? Glycobiology 6:707. 31. Zhou, L. J., and T. F. Tedder. 1995. A distinct pattern of cytokine gene expression by human CD83⫹ blood dendritic cells. Blood 86:3295. 32. Morse, M. A., L. J. Zhou, T. F. Tedder, H. K. Lyerly, and C. Smith. 1997. Generation of dendritic cells in vitro from peripheral blood mononuclear cells with granulocyte-macrophage-colony-stimulating factor, interleukin-4, and tumor necrosis factor-␣ for use in cancer immunotherapy. Ann. Surg. 226:6. 33. Romani, N., D. Reider, M. Heuer, S. Ebner, E. Kampgen, B. Eibl, D. Niederwieser, and G. Schuler. 1996. Generation of mature dendritic cells from human blood: an improved method with special regard to clinical applicability. J. Immunol. Methods 196:137. 34. Zhou, L. J., and T. F. Tedder. 1996. CD14⫹ blood monocytes can differentiate into functionally mature CD83⫹ dendritic cells. Proc. Natl. Acad. Sci. USA 93: 2588. 35. Cramer, S. O., C. Trumpfheller, U. Mehlhoop, S. More, B. Fleischer, and A. von Bonin. 2000. Activation-induced expression of murine CD83 on T cells and identification of a specific CD83 ligand on murine B cells. Int. Immunol. 12:1347. 36. Brinkman-Van der Linden, E. C., E. R. Sjoberg, L. R. Juneja, P. R. Crocker, N. Varki, and A. Varki. 2000. Loss of N-glycolylneuraminic acid in human evolution: implications for sialic acid recognition by siglecs. J. Biol. Chem. 275: 8633. 37. Nicoll, G., J. Ni, D. Liu, P. Klenerman, J. Munday, S. Dubock, M. G. Mattei, and P. R. Crocker. 1999. Identification and characterization of a novel siglec, siglec-7, expressed by human natural killer cells and monocytes. J. Biol. Chem. 274: 34089. 38. Kelm, S., A. Pelz, R. Schauer, M. T. Filbin, S. Tang, M. E. de Bellard, R. L. Schnaar, J. A. Mahoney, A. Hartnell, P. Bradfield, et al. 1994. Sialoadhesin, myelin-associated glycoprotein and CD22 define a new family of sialic aciddependent adhesion molecules of the immunoglobulin superfamily. Curr. Biol. 4:965. 39. Freeman, S. D., S. Kelm, E. K. Barber, and P. R. Crocker. 1995. Characterization of CD33 as a new member of the sialoadhesin family of cellular interaction molecules. Blood 85:2005. 40. Cornish, A. L., S. Freeman, G. Forbes, J. Ni, M. Zhang, M. Cepeda, R. Gentz, M. Augustus, K. C. Carter, and P. R. Crocker. 1998. Characterization of siglec-5, a novel glycoprotein expressed on myeloid cells related to CD33. Blood 92:2123. 41. Floyd, H., J. Ni, A. L. Cornish, Z. Zeng, D. Liu, K. C. Carter, J. Steel, and P. R. Crocker. 2000. Siglec-8: a novel eosinophil-specific member of the immunoglobulin superfamily. J. Biol. Chem. 275:861. 42. Foussias, G., G. M. Yousef, and E. P. Diamandis. 2000. Identification and molecular characterization of a novel member of the siglec family (SIGLEC9). Genomics 67:171. 43. Zhang, J. Q., G. Nicoll, C. Jones, and P. R. Crocker. 2000. Siglec-9, a novel sialic acid binding member of the immunoglobulin superfamily expressed broadly on human blood leukocytes. J. Biol. Chem. 275:22121. 44. Crocker, P. R., M. Vinson, S. Kelm, and K. Drickamer. 1999. Molecular analysis of sialoside binding to sialoadhesin by NMR and site-directed mutagenesis. Biochem. J. 341:355.