Identification and Characterization of a Novel Siglec, Siglec-7, Expressed by Human Natural Killer Cells and Monocytes*

We describe the characterization of sialic acid-binding Ig-like lectin-7 (siglec-7), a novel member of the siglec subgroup of the immunoglobulin superfamily. A full-length cDNA encoding siglec-7 was isolated from a human primary dendritic cell cDNA library. Siglec-7 is predicted to contain three extracellular immunoglobulin-like domains that comprise an N-terminal V-set domain and two C2-set domains, a transmembrane region and a cytoplasmic tail containing two tyrosine residues embodied in immunoreceptor tyrosine-based inhibition motif-like motifs. Overall, siglec-7 exhibited a high degree of sequence similarity to genes encoding CD33 (siglec-3), siglec-5, OBBP1/siglec-6, and OBBP-like protein and mapped to the same region on chromosome 19q13.3. When siglec-7 was expressed on COS or Chinese hamster ovary cells, it was able to mediate high levels of sialic acid-dependent binding to human erythrocytes and soluble sialoglycoconjugates, suggesting that it may be involved in cell-cell interactions. Among human peripheral blood leukocytes, siglec-7 was found to be present at low levels on granulocytes, intermediate levels on monocytes, and relatively high levels on a major subset of natural killer cells and a minor subset of CD8+ T cells. Immunoprecipitation experiments indicated that siglec-7 is expressed as a monomer of ∼65 kDa.

Sialic acid-binding immunoglobulin-like lectins (siglecs) 1 are a recently defined subset of the immunoglobulin superfamily of cell surface proteins. To date, six members have been characterized in mammals, namely sialoadhesin (siglec-1) (1), CD22 (siglec-2) (2), CD33 (siglec-3) (3), myelin-associated glycoprotein (siglec-4A) (4), siglec-5 (5), and OBBP1/siglec-6 (6); the latter is also described as CD33L1 (7). In addition, the gene encoding another siglec-like sequence, OBBP-like protein, has been reported (8), but there is no information on the potential sialic acid binding properties of this protein. The extracellular regions of siglecs are made up of an N-terminal V-set Ig-like domain followed by varying numbers of C2-set Ig-like domains, ranging from 1 in CD33 to 16 in sialoadhesin. Apart from myelin-associated glycoprotein, which is found exclusively in the nervous system, siglecs are generally expressed in a cell type-restricted manner on cells of the hemopoietic and immune systems. Thus, sialoadhesin is a macrophage-restricted adhesion molecule (9), CD22 is found on B cells (10), siglec-5 is found on neutrophils and monocytes (5), and CD33 is a marker of immature myeloid cells (11). These restricted expression patterns imply highly specific functions, as illustrated by extensive studies on CD22, a negative regulator of B cell activation (reviewed in Ref. 12), and myelin-associated glycoprotein, which is involved in myelin-axon interactions (reviewed in Ref. 13).
The common theme of sialic acid recognition suggests that carbohydrate binding is likely to be important in the biological functions of these molecules. Where studied, each siglec has a particular preference for both the nature of the sialic acid (14 -16) and its linkage to subterminal sugars (3)(4)(5)17). Mutagenesis and structural studies have established that the sialic acid binding site of sialoadhesin is on the V-set domain. A conserved arginine on the F-strand has been shown by x-ray crystallography to form a salt bridge with the carboxylate of sialic acid (18), and site-directed mutagenesis studies with sialoadhesin (19), CD22 (20), myelin-associated glycoprotein (21), and CD33 (22) have shown that the corresponding arginine is essential for sialic acid-dependent binding by all siglecs. Other important interactions revealed by x-ray crystallography (18) and NMR (23) include hydrophobic contacts between conserved aromatic amino acids on the A and G ␤-strands of the V-set domain with the N-acetyl and glycerol side-groups of N-acetyl neuraminic acid. All siglecs so far characterized also have an unusual arrangement of conserved cysteine residues in the V-set and adjacent C2-set domains. These are predicted to result in the formation of an intrasheet disulfide bond in the V-set domain and an interdomain disulfide (18).
Although siglecs are able to mediate sialic acid-dependent binding as isolated proteins, there are major differences in their binding activities when expressed on plasma membranes (3,5,24,25). This could be due, at least in part, to the occupation of their ligand binding sites by sialic acids present in the glycocalyx. With CD33, no binding is detectable in transfected COS cells, but following sialidase treatment of the cells, high levels of binding can be demonstrated (3). In contrast, siaload-hesin can mediate strong sialic acid-dependent binding on both transfected cells (1) and on macrophages that naturally express the receptor (9). Because sialoadhesin contains 17 Ig-like domains compared with only two in CD33, one possibility is that the sialic acid binding site in sialoadhesin extends away from the plasma membrane, thereby promoting cell-cell interactions.
In this paper, we describe the properties of a new member of the siglec family, siglec-7. Siglec-7 is the first example of a siglec to be found predominantly on natural killer (NK) cells. It is also unusual compared with other siglecs in being able to mediate sialic acid-dependent binding on non-sialidase-treated transfectants, strongly suggestive of a role in cell-cell interactions. Finally, the presence of immune receptor tyrosine-based inhibition motif (ITIM)-like motifs in the cytoplasmic tail suggests that it may be involved in regulating NK cell activation events.
Identification and Characterization of Siglec-7 cDNA-Using the amino acid sequence of CD33, a specific homology search was performed against a data base containing more than one million expression sequence tags obtained from over 700 different cDNA libraries. Several clones corresponding to the same CD33-like cDNA were identified in cDNA libraries from the following sources: placenta, breast cancer, spleen, human primary breast cancer, human primary dendritic cell, human gall bladder, apoptotic T-cell, spleen, and chronic lymphocytic leukemia. One of these, HDPUW68, in the pSPORT mammalian expression vector (Life Technologies, Inc.), was isolated from a human primary dendritic cell cDNA library and found to contain a full-length cDNA. The predicted amino acid sequence encoded by HDPUW68 is referred to as siglec-7 from here on. A computer search of nucleotide and protein sequence was carried out on June 24, 1999, using the Blast GeneSearch (NCBI, National Institutes of Health, Bethesda, MD). Manipulations of sequences and alignments were performed using Baylor College of Medicine molecular biology software (available on the Internet) (Human Genome Center, Baylor College of Medicine, Houston, TX).
Chromosomal Localization-Metaphase spreads prepared from phytohemagglutinin-stimulated human lymphocytes were hybridized with a biotinylated insert from HDPUW68 as described (26). A total of 50 metaphase cells were analyzed.
Northern Blot Analysis-Two human multiple tissue Northern blots containing approximately 2 g of poly(A) ϩ RNA per lane from various human tissues were purchased from CLONTECH (Palo Alto, CA) and hybridized with 32 P-labeled insert from HDPUW68 as described previously (5).
Cells-The following cell lines were provided by the ICRF Cell Production Service: COS-1, Balb/c 3T3 A31, Chinese hamster ovary K1 (CHO), KG1b, HL-60, U937, and Daudi. The NK-like cell lines, YT and NKL, were obtained from Dr. Gillian Griffiths (Oxford University) and Dr. Jerome Ritz (Harvard Medical School) respectively. CHO cells stably expressing full-length sialoadhesin (Sn-CHO) were obtained as described (27). COS-1 cells were cultured in Dulbecco's modified Eagle's medium with 5% heat-inactivated fetal calf serum, CHO cells were cultured in Ham's F-10 medium with 5% fetal calf serum, and all other cell lines were cultured in RPMI 1640 medium with 5 or 10% fetal calf serum. Human red blood cells (RBCs) were obtained from healthy donors and stored at 4°C in Alsever's solution for up to 2 weeks. Human blood leukocytes were obtained from whole blood by dextran sedimentation followed by lysis of contaminating RBCs. Mononuclear fractions for flow cytometry were obtained by density gradient centrifugation using Ficoll-Paque (Amersham Pharmacia Biotech).
Generation of Monoclonal Antibodies to Siglec-7-Balb/c 3T3 A31 cells were co-transfected, by electroporation, with a 10:1 ratio of HD-PUW68 and pcDNA3. G418-resistant clones expressing siglec-7 were identified by their ability to bind human RBCs and designated siglec-7-3T3. Balb/c mice were immunized twice, intraperitoneally, at an interval of 14 days, with 10 7 live siglec-7-3T3 cells. One of the mice was boosted with 10 7 live cells, 4 days prior to the fusion of spleen cells with the SP2 myeloma, and hybridomas were generated following standard methods (28). A positive well, reacting with siglec-7-3T3 cells, was cloned three times by limiting dilution, and the mAb was designated S7 (IgG1). S7 was used either as tissue culture supernatant or following purification with protein A-Sepharose.
Human RBC Binding Assays to Siglec-7-CHO Cells-CHO cells stably expressing siglec-7 were generated by cotransfection of CHO cells with a 10:1 ratio of HDPUW68 and pcDNA3. G418 resistant clones expressing siglec-7 were identified by their ability to bind anti-siglec-7 mAbs and designated siglec-7-CHO. For binding assays, siglec-7-CHO, Sn-CHO, and wild-type CHO cells were plated overnight at 5 ϫ 10 4 cells/well on 24 well plates. RBCs were washed three times in PBS containing 0.25% bovine serum albumin (PBA), resuspended at 0.25% (v/v) in Dulbecco's modified Eagle's medium ϩ 0.2% bovine serum albumin, and 1 ml of the cell suspension was added to the wells. After 60 min at 37°C, nonadherent cells were removed by three gentle washes in Dulbecco's modified Eagle's medium ϩ 0.2% PBA, and RBC rosetting was quantified by measuring the pseudoperoxidase activity of hemoglobin as described (4). Sialidase pretreatment of CHO cells or RBCs was carried out with 0.05 units/ml V. cholerae sialidase in Dulbecco's modified Eagle's medium for 2-3 h at 37°C, followed by three washes in PBA.
Binding Assays with Polyacrylamide Glycoconjugates-COS-1 cells were transfected by electroporation with full-length cDNAs encoding siglec-7, mouse sialoadhesin (1), mouse CD22 (4), siglec-5 (5), or CR1 as a negative control (29), and binding assays were carried out 48 -72 h later. The COS cells were detached with PBS ϩ 5 mM EDTA and either left untreated or treated with 0.05 units/ml sialidase for 2 h at 37°C. Cells were incubated with saturating concentrations (20 g/ml) of 2,3-PAA or 2,6-PAA for 1 h at room temperature, washed in PBA, and incubated with 125 I-streptavidin diluted in PBA to 0.5 Ci/ml for 1 h at 4°C. After washing three times in PBA, bound radioactivity was counted using a Beckman gamma counter. FACS analysis was carried out in parallel to determine the percentage of COS cells expressing each molecule.
FACS Analysis-Single, double, and triple labeling were performed following standard protocols (30). Following staining, cells were fixed in 2% formaldehyde and analyzed on a Becton-Dickinson FACS analyzer.
Immunoprecipitation-Wild-type CHO cells or CHO cells expressing siglec-7 at 2 ϫ 10 7 /ml were labeled with TRANS 35 S-LABEL TM at 0.2 mCi/ml, and lysates were prepared in 1% Triton X-100. Immunoprecipitations were carried out following standard procedures (28), and precipitated material was analyzed by SDS-PAGE on 4 -12% gradient polyacrylamide gels followed by autoradiography.

RESULTS
Characterization of Siglec-7 as a Novel Siglec-HDPUW68, a clone derived from a human primary dendritic cell cDNA library, was identified as an EST sharing a high degree of sequence similarity with human CD33 cDNA. Examination of its full-length sequence of 1748 base pairs revealed a single long open reading frame encoding a type 1 membrane protein of 467 amino acids belonging to the Ig superfamily. Based on its sequence similarity with other siglecs and its ability to bind sialic acid (see below), this protein has been designated siglec-7 (Fig. 1). The extracellular region of siglec-7 contains a hydrophobic signal peptide and three Ig-like domains that are made up of an N-terminal V-set domain and two C2-set domains. There are eight potential N-linked glycosylation sites. Following the transmembrane region, there is a cytoplasmic tail of 91 amino acids.

Structural Features Characteristic of Siglecs-Examination
of the first two domains at the N terminus of siglec-7 revealed the presence of the characteristic structural features of the siglec subgroup of Ig superfamily proteins (Fig. 1). There is precise conservation of the unusual pattern of cysteines found in these proteins (18) and the key amino acids involved in sialic acid binding, in particular the critical arginine at position 124 and the two conserved aromatic residues at positions 26 and 134 on the A and G strands of the V-set domain (18). These are tyrosine and tryptophan, respectively, in siglec-7 (Fig. 1).
Chromosomal Localization and Expression of the Siglec-7 Gene-The gene encoding siglec-7 was mapped by in situ hybridization to the long arm of chromosome 19, in the 19q13.3 band ( Fig. 2A), closely linked to the CD33, siglec-6, siglec-5, and OBBP-like genes (5, 7, 8, 35). Northern blot analysis (Fig. 2B) revealed the presence of a major siglec-7 mRNA transcript of 2.0 kilobases, with the highest levels in placenta, liver, lung, and spleen. High levels were also detected in extracts of peripheral blood leukocytes, but low or undetectable levels were present in other tissues examined (Fig. 2B).
Siglec-7 Mediates Sialic Acid-dependent Binding to Human RBCs and to Glycoconjugates-To investigate the sialic acid binding properties of siglec-7, we initially performed binding assays in which human RBCs were added to transiently transfected COS cells. High levels of binding were observed with native RBCs, and binding was abolished when the RBCs were pretreated with sialidase, demonstrating that the binding was sialic acid-dependent (data not shown).
To determine the sialic acid linkage preference of siglec-7, binding assays were carried out with polyacrylamide conjugates, carrying either 3Ј or 6Ј sialyllactose or lactose. In these experiments, COS cells were transiently transfected with siglec-7, sialoadhesin, CD22, siglec-5, or CR1 as a negative control. FACS analysis showed that 20 -30% of the cells expressed each molecule 3 days after the transfection (data not shown). Transfected cells were either untreated or treated with sialidase immediately before the binding assay to remove potentially inhibitory sialic acids in the COS cell glycocalyx (Fig. 3). With sialidase-treated COS cells, the properties of siglec-7 were more similar to those of siglec-5, binding glycoconjugates carrying sialic acid in either ␣2,3 or ␣2,6 linkages. In contrast, sialoadhesin and CD22 exhibited a clear preference for ␣2,3 or ␣2,6-linked sialoglycoconjugates, respectively (Fig. 3). Interestingly, siglec-7 also bound strongly to the glycoconjugates in the absence of COS cell pretreatment with sialidase, unlike sialoadhesin and siglec-5, which showed greatly increased binding following the treatment. The high levels of binding seen with CD22 on untreated COS cells is likely to be due to the absence from COS cells of ST6Gal1, a glycosyltransferase required to make CD22 ligands (36). No specific binding was observed when lactose-PAA was used as a negative control.
The high constitutive binding of siglec-7 to sialylated ligands was further investigated. CHO cells stably expressing siglec-7 bound RBCs at high levels either with or without pretreatment of the CHO cells with sialidase (data not shown). In contrast, RBC binding to sialoadhesin-expressing CHO cells was greatly increased by sialidase pretreatment of the CHO cells (data not shown).
Expression of Siglec-7 on Human Peripheral Blood Leukocytes-Using S7, a mAb prepared to siglec-7, detailed analysis of the expression of the molecule was carried out by FACS analysis, using human peripheral blood leukocytes. First, expression on granulocyte, monocyte, and lymphocyte subsets was compared (Fig. 4A), revealing weak expression on granulocytes, intermediate expression on monocytes, and high expression on a subpopulation of lymphocytes. To characterize the lymphocyte-reactive cells in more detail, double labeling was carried out in which staining for siglec-7 was combined with staining for CD3 (pan T cell), CD4 and CD8 (T cell subsets), CD19 (pan B cell), and CD56 (NK cells) (Fig. 4B). The dominant populations expressing siglec-7 were CD56 ϩ cells, with a small subset of CD3 ϩ CD8 ϩ T cells also binding antibody. The proportion of CD56 ϩ cells expressing siglec-7 varied between individuals and within the same individual over time. However, there was consistently a difference in expression between CD56-high cells (a small population of NK cells) of which a high proportion were positive for siglec-7 (mean, 78%; range, 60 -92%) and CD56-mid cells (a much larger population), of which an average of 65% were positive (range, 52-85%). The CD8 cells that were siglec-7 ϩ fell mainly into the CD8-mid range and were predominantly CD56 ϩ NK cells (not shown). Only a very small proportion of the CD8-high cells, representing cytotoxic T cells, were siglec-7 ϩ (mean 2.7%). These cells corresponded to the ϳ2% siglec-7 ϩ CD3 ϩ cells shown in Fig. 4B.
The phenotype of the CD56 ϩ NK subset that expressed siglec-7 was examined in more detail using triple staining. Interestingly, among CD56-mid cells, siglec-7 was not distributed uniformly; its expression was associated with increased expression of CD38, CD45RA, and CD16 and lower expression of CD45RO (Fig. 4C). The phenotype of the minor population of CD8-high, siglec-7 ϩ T cells was also examined for comparison. These cells were low in CD38, CD62L, MHC class II, and CD69 and high in CD44 and CD45RO and thus were distinct from the phenotype of the NK cell subset (data not shown).
Finally, FACS staining of various human cell lines was also Molecular Characterization of Siglec-7-To determine the molecular mass of siglec-7 on CHO cells, immunoprecipitations were performed using cell lysates prepared from metabolically labeled cells. A single, heterogeneous band of ϳ65 kDa was observed under both reducing and nonreducing conditions (Fig. 5). DISCUSSION In this study, we characterized the binding properties and cellular expression pattern of a novel siglec, siglec-7. Siglec-7 is the first example of a siglec that is expressed predominantly on NK cells. Siglec-7 also shows unusually high constitutive bind-ing to sialylated glycoconjugates when expressed on COS cells and CHO cells, strongly suggesting its involvement in cell-cell interactions of NK cell subsets and other cells that express the molecule naturally. In addition to NK cells, high levels of siglec-7 were present on a subset of CD8 T cells and intermediate levels were present on monocytes. Monocytes thus express several siglecs simultaneously, including CD33 (siglec-3) (11), siglec-5 (5), and siglec-7. Most circulating monocytes are thought to enter tissues and differentiate to tissue macrophages, but further studies are needed to determine whether siglec-7 is retained or lost following monocyte differentiation.
NK cells are bone marrow-derived granular lymphocytes that play an important role in natural immunity to infectious diseases and have the capacity to kill certain virally infected cells and tumor cells that have down-regulated MHC class I antigen expression (reviewed in Ref. 37). The killing and proinflammatory activities of NK cells are regulated through a variety of cell surface receptors that can mediate either activatory or inhibitory signals. The best understood receptors are those that recognize MHC class I molecules at the cell surface and deliver a negative signal, thereby protecting normal host cells from cytotoxicity. These receptors can belong either to the C-type lectin superfamily (38) or the Ig superfamily, although in humans the majority are members of the Ig superfamily known as killer cell Ig-like receptors (reviewed in Ref. 39). The inhibitory killer cell Ig-like receptors contain at least one ITIMlike motif that mediates binding and activation of the cytoplasmic tyrosine phosphatase, SHP-1, resulting in inhibition of cytotoxicity and cytokine secretion (reviewed in Ref. 40). In addition to killer cell Ig-like receptors, a number of other regulatory Ig-related receptors have been characterized that are expressed more broadly among leukocytes. These include Iglike transcripts, NKP46, CD89 (Fc␣R), and leukocyte-associated Ig-like receptor 1. Like the killer cell Ig-like receptors, these Ig-related receptors comprise both activatory and inhibitory classes of receptors (reviewed in Ref. 40) and are tightly clustered on chromosome 19q13.4. This is now recognized as a region that is especially enriched in leukocyte-expressed Ig superfamily members described as the leukocyte receptor complex (41). CD33 and siglecs-5, -6, and -7 have features in common with several of these proteins. As well as being members of the Ig superfamily expressed on leukocytes, both sets of receptors have ITIM-like sequences in their cytoplasmic tails and are tightly linked to the leukocyte receptor complex on chromosome 19q13. 3-13.4.
The presence of the ITIM-like regions in the cytoplasmic tail of siglec-7 indicates a potential role in NK cell regulation. A recent study on human CD33, which has an arrangement of ITIM-like motifs very similar to that of siglec-7, demonstrated that pervanadate-treatment of myeloid cells expressing CD33, or cross-linking of CD33 itself, resulted in tyrosine phosphorylation of CD33 and recruitment of the tyrosine phosphatases SHP-1 and SHP-2 (22). However, the functional consequences of these events are not clear because CD33 can only mediate binding to ligands on other cells following sialidase treatment of cells bearing the receptor (3). In contrast, siglec-7, expressed on either CHO or COS cells, binds constitutively to sialylated ligands on other cells in the absence of sialidase pretreatment. This suggests that on NK cells and other cells, such as monocytes, that naturally express siglec-7, the lectin binding site is available for interacting with sialylated ligands on other cells. Further studies are required to investigate this possibility, as well as the potential functional consequences of siglec-7 engagement and cross-linking.
The phenotypic analysis of siglec-7 expression on CD56 ϩ NK cells revealed a striking correlation with expression of the CD38 antigen. CD38 is an ectoenzyme with ADP-ribosyl cyclase and hydrolase activities that is expressed on diverse cell types in the immune and hemopoietic systems (reviewed in Ref. 42). Following ligation, CD38 has been shown to mediate activatory signals (43), and CD31 has been identified as a specific ligand on endothelium and hemopoietic cells (44). Interestingly, siglec-7 ϩ NK cells were also enriched in cells expressing CD16. CD16 is a low affinity Fc receptor that is important for antibody-dependent and -independent (45) cellular cytotoxicity and for regulating NK cell proliferation, cytokine release, and apoptosis (46). The levels of CD16 expression on circulating NK cells have been shown to correlate with a more activated phenotype (47). However, siglec-7 ϩ NK cells completely lacked CD45RO, a marker that is normally associated with recent activation (48). This observation raises the interesting possibility that siglec-7 can become down-regulated following NK cell activation. In contrast to NK cells, CD45RO was expressed on the minor subset of siglec-7 ϩ CD8 ϩ T cells, but the CD69 activation marker was absent. This suggests that siglec-7 is expressed predominantly on a subset of memory CD8 ϩ T cells.
Carbohydrate recognition by NK cells has long been thought to be important in NK cell biology, particularly in target cell recognition. Several receptors have been identified on NK cells that belong to the C-type lectin family, but apart from L-selectin, which is implicated in NK cell-endothelial cell interactions (49), it is unclear whether these proteins have the capacity to mediate carbohydrate recognition (reviewed in Ref. 38). A number of studies have also implicated sialic acids in the recognition functions of NK cells (50 -52). Siglec-7 is the first sialic acid binding receptor to be identified that is expressed predominantly on NK cells. A challenge for the future will be to determine the functions of this protein in NK cell biology, carbohydrate recognition, and host defense.