Molecular Cloning of a Dendritic Cell-associated Transmembrane Protein, DC-HIL, That Promotes RGD-dependent Adhesion of Endothelial Cells through Recognition of Heparan Sulfate Proteoglycans*

We isolated a novel molecule (DC-HIL) expressed abundantly by the XS52 dendritic cell (DC) line and epidermal Langerhans cells, but minimally by other cell lines. DC-HIL is a type I transmembrane protein that contains a heparin-binding motif and an integrin-recognition motif, RGD, in its extracellular domain (ECD). A soluble fusion protein (DC-HIL-Fc) of the ECD and an immunoglobulin Fc bound to the surface of an endothelial cell line (SVEC). This binding induced adhesion of SVEC to its immobilized form. Sulfated polysaccharides (e.g. heparin and fucoidan) inhibited binding of soluble DC-HIL-Fc and adhesion of SVEC. By contrast, an integrin inhibitor (RGDS tetramer) had no effect on binding to SVEC, but prevented adhesion of SVEC. This differential RGD requirement was confirmed by the finding that DC-HIL-Fc mutant lacking the RGD motif can bind to SVEC but is unable to induce adhesion of SVEC. Furthermore, DC-HIL appears to recognize directly these sulfated polysaccharides. These results suggest that DC-HIL binds to SVEC by recognizing heparan sulfate proteoglycans on endothelial cells, thereby inducing adhesion of SVEC in an RGD-dependent manner. We propose that DC-HIL serves as a DC-associated, heparan sulfate proteoglycan-dependent integrin ligand, which may be involved in transendothelial migration of DC.

unsurpassed potency in presenting antigens to naive T cells, thereby most efficiently initiating primary T cell-mediated immune responses (1). DC are widely distributed but comprise only a minuscule fraction (typically Ͻ5%) of the total cell population of a given peripheral tissue. After capturing antigens, tissue DC undergo maturation, losing endocytic capacity but acquiring increased immunostimulatory capacity (2). During maturation, DC migrate from peripheral tissues to T cell areas of secondary lymphoid organs where they activate T cells (2)(3)(4). Thus, DC migration is a critical step in initiating antigenspecific immune responses.
Several adhesion molecules (e.g. cutaneous lymphocyte-associated antigen (5), lymphocyte function-associated antigen-1/ CD11a (6), intercellular adhesion molecule-1 (ICAM-1)/CD54 (6), CD44 (7), E-cadherin (8), and ␣ 6 integrins (9)) regulate the migration of DC through the entire life cycle. For example, blood-circulating precursors of Langerhans cells (LC), which are DC that reside in epidermis, attach to the blood vessel wall prior to initiating transendothelial migration into the epidermis where they develop into APC with phenotypes distinct from other members of the DC family. The LC homing and anchoring in the epidermis are probably controlled by lymphocyte-associated antigen (5) and E-cadherin (8), respectively. Conversely, after antigenic stimulation, LC dissociate from surrounding keratinocytes by down-regulating E-cadherin expression (8,10), then adhere via ␣ 6 integrins to the basement membrane (9) with subsequent passage into the dermis, where they re-enter the afferent lymphatics or blood vessels. This endothelial migration is essential for completion of the DC life cycle. Despite this importance, very little is known about the molecular mechanisms underlying the adhesion of DC to endothelial cells.
Study on the unique properties of DC at molecular levels has been hampered by the paucity of DC within tissues and the relative difficulty in maintaining their viability ex vivo. To overcome this problem, we have previously established longlived DC lines, designated XS series, derived from the skin of newborn BALB/c mice (11). These XS cells retain many important features of LC, including their morphology, surface phenotype (B7-2, major histocompatibility class II, CD11b, ICAM-1, CD49d/e, and E-cadherin), responsiveness to cytokines, expression of cytokines and their receptors, and potent APC function (11,12). Having these cell lines, our goal is to illuminate the molecular properties that distinguish DC/LC from other APC, and our principal strategy has been to identify and define genes and gene products expressed selectively by DC and LC. We have previously isolated five novel molecules using a subtractive cDNA cloning method in which the cDNA library prepared from the XS52 DC line was subtracted with mRNA isolated from the J774 macrophage line. Two of these molecules, termed dectin-1 and dectin-2, are expressed selectively by XS52 DC (13). They are type II transmembrane glycoproteins that belong to the Ca 2ϩ -dependent (C-type) lectin superfamily and may serve as costimulatory molecules that are required for efficient activation of T cells by DC. We now focus on the third gene, originally designated as 2B4, which encodes a type I transmembrane protein that can function as a ligand for integrins most likely through recognition of heparan sulfate proteoglycans (HSPG) abundantly expressed on endothelial cell surfaces (14). Based on its expression pattern and function, we have renamed this molecule DC-associated, HSPG-dependent integrin ligand (DC-HIL). Here we report its structural, biochemical, and functional properties and discuss its potential roles in DC-endothelial cell adhesion.

EXPERIMENTAL PROCEDURES
Animals-Female BALB/c mice (6 -10 weeks old) were purchased from Harlan (Indianapolis, IN) and housed in the pathogen-free facility of the Animal Resource Center at The University of Texas Southwestern Medical Center. To study tissue distributions and Langerhans cell (LC)-specific expression of DC-HIL mRNA, skin and other tissues were excised from mice sacrificed by overdose of methoxyflurane inhalation.
Epidermal Cell Isolation-Epidermal cells were isolated from abdominal skin of BALB/c mice using two sequential trypsin treatments and enriched for LC by centrifugation over Histopaque (1.083, Sigma Chemical Co., St. Louis, MO) as described previously (15)(16)(17). In some experiments, LC were depleted from this cell population using anti-Ia monoclonal antibody (mAb) plus complement treatment as described before (16).
Isolation of DC-HIL cDNA Clone-A DC-specific cDNA library was constructed by subtracting the cDNA library prepared from the XS52 DC with excess amounts of mRNAs isolated from the J774 macrophage. Twelve thousand independent clones from this library were screened by colony hybridization, slot blotting, and Northern blotting for the genes expressed by the XS52 DC but only minimally or not at all by J774 macrophages (15,18). The original clone 2B4 was one of the 50 clones selected in the above manner. The DNA sequences of both sense and antisense strands were determined by Taq dye deoxy termination cycle sequencing on a DNA sequencer (model 373A, Applied Biosystems, Foster City, CA).
Northern Blotting and RT-PCR Analyses-Northern blotting was performed as described previously (15). Briefly, total RNAs (10 g/lane) isolated from cell lines or mRNA (5 g/lane) from mouse organs were run on a vertical agarose gel, transferred onto a nylon membrane, and hybridized with the 32 P-labeled cDNA probe for DC-HIL, glyceraldehyde-3-phosphate dehydrogenase, or ␤-actin.
Immunoblotting of DC-HIL Protein-Anti-DC-HIL peptide Ab was generated by immunizing rabbits with a synthetic 20-aa peptide of C-GHEQYPNHMREHNQLRGWS (Alpha Diagnostics Intl., San Anto-nio, TX) (the amino-terminal cysteine was attached for thiol coupling) corresponding to aa 30 -48 in the ECD of DC-HIL. After three rounds of immunization, serum was collected and subjected to affinity purification of peptide-specific Ab using the same 20-mer peptide as described before (15).
Whole cell extracts were prepared from several different cell lines by lysis with 0.3% Triton X-100 in Dulbecco's PBS(Ϫ) for 15 min, followed by centrifugation for 20 min at 10,000 ϫ g. The full-length cDNA for DC-HIL was enzymatically excised from clone 2B4 and inserted into a mammalian expression vector pCMV5 (a gift from Dr. D. Russel, UT Southwestern Medical Center). COS-1 cells were transfected with the resulting vector (pCMV5-DC-HIL) or an empty vector using FuGene 6 (Roche Diagnostics, Indianapolis, IN), cultured for 3 days, and the whole cell lysate was prepared as described above. The samples were separated by 4 -15% SDS-PAGE, transferred onto polyvinylidene fluoride membrane (Millipore, Bedford, MA), and then blotted with 1 g/ml purified rabbit anti-DC-HIL or control rabbit IgG. After washing, the membrane was further blotted with horseradish peroxidase (HRP)conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratory, West Grove, PA) and developed with ECL system (Amersham Pharmacia Biotech, Piscataway, NY).
Sublocalization of DC-HIL-XS52 DC were incubated in 0.25 M sucrose for 10 min at 4°C. The pellet after centrifugation was resuspended and homogenized in 10 mM HEPES (pH 7.3) by 5-10 strokes with a 27-gauge needle on a 1-ml syringe, and then centrifuged for 10 min at 1000 ϫ g. The resulting supernatant was separated into cytosolic and membrane fractions by further centrifugation for 45 min at 50,000 ϫ g at 4°C. These samples were subjected to immunoblotting for determining localization of DC-HIL proteins.
For experiments in which cell surface expression of DC-HIL was assessed, XS52 DC and COS-1 cells transfected either with pCMV5-DC-HIL or a luciferase-expressing vector (pGL3-Control, Promega, Madison, WI) were surface-biotinylated by incubation with 0.5 mg/ml Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL) for 60 min at 4°C. After washing, cell lysates were prepared using 0.3% Triton X-100 and incubated with streptavidin-conjugated agarose beads (Sigma) by gently rocking overnight at 4°C. Following centrifugation, the supernatant was collected as the streptavidin-unbound fraction, and the beads were washed extensively with PBS and then eluted by boiling in Laemmli's sample buffer (streptavidin-bound fraction). DC-HIL or luciferase in each fraction was detected by immunoblotting with the respective Ab.
N-Glycosidase Treatment-The whole XS52 DC lysate was boiled at 95°C for 5 min with 1% SDS, and then incubated with or without 5 units/ml N-glycosidase (Roche Diagnostics) in the presence of 1% Nonidet P-40 (SDS at final concentration of 0.2%) overnight at 37°C. Samples were analyzed for molecular weights of DC-HIL by SDS-PAGE followed by immunoblotting.
Immunocytochemistry-Visualization of DC-HIL proteins in XS52 DC was performed using anti-DC-HIL Ab and a peroxidase-based immunostaining kit (LSAB, Dako, Carpenteria, CA) according to manufacturer recommendations. Briefly, XS52 DC cultured on a coverslip were air-dried and subjected to the following sequential treatment: 15 min fixation in 4% paraformaldehyde, 20 min permeabilization in 0.5% saponin, 10 min blocking in goat IgG, and 60 min staining with anti-DC-HIL Ab (10 g/ml) in the presence of 0.5% saponin, and followed by color development using HRP and 3-amino-9-ethylcarbazole substrate. Subsequently, the specimens were counter-stained by hematoxylin and mounted onto slides. Images were taken by slide films using an Olympus BH-2 microscope; these were digitized and image contrast was adjusted using Photoshop (Adobe, San Jose, CA).
Preparation of DC-HIL-Fc Fusion Protein-DC-HIL-Fc protein, consisting of, from the amino-terminus, the extracellular domain (ECD) of DC-HIL and a Fc region of human IgG 1 (hIgG 1 Fc), was produced in COS-1 cells. The DNA fragment encoding the ECD (aa 34 -511) was PCR-amplified with primers containing HindIII and XbaI sites at the 5Ј and 3Ј ends, respectively. Using these sites, the PCR product was ligated in-frame to the coding sequence of the Fc region sequence (789 bp), which had been cloned into pSecTagB vector (Invitrogen, Carlsbad, CA) at XbaI and ApaI sites in the 5Ј and 3Ј ends (pSTB-DC-HIL-Fc). An expression vector for Dectin-1-Fc was constructed by replacement of the ECD in pSTB-DC-HIL-Fc with the ECD of Dectin-1 (aa 71-213) (15). We have reported previously that the Dectin-1 recombinant protein tagged with the amino-terminal histidine binds selectively to T cells (15), whereas the C-terminal Fc fusion proteins lack the binding activity. The Fc-encoding expression vector was constructed by insertion of the DNA fragment encoding hIgG 1 Fc into the pSecTagA vector using XbaI and ApaI sites. These expression vectors were introduced into COS-1 cells by FuGene 6 (Roche Diagnostics). At 96 h after transfection, the culture supernatant was harvested and immunoglobulin fusion proteins were purified by protein A-agarose beads (Life Technologies, Inc.) according to manufacturer recommendations.
Site-directed Mutagenesis-The RGD sequence of the pSTB-DC-HIL-Fc was mutated to RAA using the QuikChange site-directed mutagenesis kit according to the manufacturer recommendations (Stratagene, La Jolla, CA). Sense and antisense oligonucleotides (nucleotides 160 -193) with the mutation (RGD to RAA) were synthesized and high pressure liquid chromatography-purified. The sense strand, 5Ј-CCAGT-GTGGAGGAGGGCAGCAGGCAGGTGGAAGG-3Ј (underlined sequence encodes RAA); and the antisense strand, 5Ј-CCTTCCACCTGC-CTGCTGCCCTCCTCCAACACTGG-3Ј. The nucleotide fragment coding entire ECD of DC-HIL was ligated into pBluescript II KS(Ϫ) vector (Stratagene) and used as a template DNA for the following PCR. Using the above oligonucleotides, the entire plasmid vector with the designed mutation was PCR-amplified with the cycling protocol; 18 cycles of 1-min annealing at 52°C and 18-min extension at 68°C. The resulting PCR products were treated with DpnI restriction enzyme to digest selectively the template DNA and then transformed into Escherichia coli. The nucleotide sequence of full-length mutated ECD was confirmed by DNA sequencing. The DNA fragment for wild-type ECD in pSTB-DC-HIL-Fc was replaced with the mutated ECD and produced in COS-1 cells, followed by affinity chromatography with protein A-agarose.
Cell Binding Assays of Soluble DC-HIL-Fc-Binding properties of soluble DC-HIL-Fc were examined by flow cytometric analyses: Different cell lines were nonenzymatically harvested and incubated with typically 10 g/ml DC-HIL-Fc or control Fc fusion proteins, Dectin-1-Fc or Fc, in binding buffer (Hanks' balanced salt solution containing 1.3 mM CaCl 2 , 0.8 mM MgSO 4 , 3% fetal calf serum, and 10 mM HEPES) for 30 min on ice. After washing, the cells were labeled with 5 g/ml biotin-conjugated Fab fragment of goat anti-human Fc␥ (Jackson Im-munoResearch Laboratory) for 10 min, followed by fluorescein isothiocyanate-conjugated streptavidin at 1:100 dilution (Jackson ImmunoResearch Laboratory) for 10 min. Binding of soluble DC-HIL-Fc to cells was examined by FACSCalibur (Becton Dickinson, San Jose, CA). In some experiments, the binding was performed in the presence of inhibitors (heparin, fucoidan, dextran sulfate (500 kDa), chondroitin sulfate A, mannan, RGDS and RPKP tetramer peptides, and EDTA; all purchased from Sigma).
Binding Assays of DC-HIL-Fc to Heparin and Fucoidan-The 96-well ELISA plates were coated with heparin-BSA or fucoidan overnight at 4°C. After extensive washing with PBS, 1 g/ml DC-HIL-Fc or control Fc fusion proteins were incubated for 60 min at room temperature. After removing unbound fusion proteins, the wells were incubated with biotin-conjugated Fab fragment of goat anti-human Fc␥ and then incubated with HRP-conjugated streptavidin at 1:100 dilution for 10 min. The 3,3Ј,5,5Ј-tetramethylbenzidine (TMB) substrate reagent (PharMingen) was added to the wells, and the enzymatic reaction was allowed to progress at room temperature for 60 min. After the reaction was stopped by addition of 1 M phosphoric acid, the absorbance at 450 nm was measured by an ELISA plate reader. The background binding to BSA-coated wells or noncoated wells was negligible (A 450 nm Ͻ 0.03). Each value represents the mean of three separate reactions.
Adhesion Assays of SVEC to Immobilized DC-HIL-Fc-SVEC were metabolically labeled with 3 Ci/ml [ 3 H]thymidine (ICN, Costa Mesa, CA) in 10% fetal calf serum-Dulbecco's modified Eagle's medium overnight. After harvest with 0.5 mM EDTA and subsequent washing with binding buffer, SVEC were counted for radioactivity by a liquid scintillation counter (usually 2-4 cpm/cell). During labeling of SVEC, the 96-well ELISA plates were coated with 100 g/ml goat anti-human Fc␥ IgG overnight at 4°C. After washing, DC-HIL-Fc or control Fc fusion proteins (10 g/ml) were added to the wells and incubated for 1 h at 4°C for immobilization, and then washed with binding buffer. As an additional control, vitronectin (5 g/ml) was also immobilized. The labeled SVEC (1 ϫ 10 4 cells) were incubated for 60 min, in most experiments, at 37°C in the Fc fusion protein-coated wells. After extensive washing, SVEC adhered to the wells were lysed with 1 N NaOH and counted for the radioactivity of 3 H (cpm). Adhesion activity was shown by the cpm/total input cpm; each value represents the mean of three independent wells. Like SVEC, other cell lines that were used for the DC-HIL binding assays were also tested for the adhesion.

RESULTS
Cloning of DC-HIL cDNA and Its Deduced Amino Acid Structure-We employed a subtractive cDNA cloning strategy to isolate genes expressed by XS52 DC but not by J774 macro-phage. Fifty clones were selected and were searched for their identities in the GenBank data base (updated in 1997). Five novel genes were identified, one of which was designated 2B4 (the original clone name).
The clone 2B4 contained a cDNA insert of 2279 bp, including the largest open reading frame (nucleotides 91-1815) with sequences that matched Kozak consensus sequences (20). The open reading frame encoded a putative polypeptide, termed DC-HIL, consisting of 574 amino acids (aa) with features typical of type I integral membrane proteins (Fig. 1). DC-HIL consists of a leader sequence (aa 1-19), a long extracellular domain (ECD, aa 20 -499), a transmembrane domain (aa 500 -523), and a cytoplasmic domain (aa 524 -574). The ECD contains 11 potential N-glycosylation sites (NX(S/T)) and several putative O-glycosylation sites based on the stretch of proline-, serine-, and threonine-rich region (21), and a proline-rich region (aa 320 -352) that presumably forms a hinge, as seen in proteins like IgA (22), which can mediate protein-protein interactions (23). Other functional motifs identified were an RGD sequence (aa 64 -66), an integrin-binding sequence, and a KRFR sequence (aa 23-26) that matches a heparin-binding motif composed of a stretch of basic residues (BBXB, where B represents a basic residue) (24). The cytoplasmic tail contained an immunoreceptor tyrosine-based activation motif (ITAM, YXXI, aa 529 -532) and two lysosomal targeting di-leucine motifs (LL, aa 548 -549 and 566 -567). In a homology search, the aa sequence of DC-HIL had identity to mouse nmb (GenBank accession number: AJ251685) with 99.4% (by the Clustal method analyzed with the Lasergene Program, DNA Star, Madison, WI) containing four substitutions of aa residues. The human nmb (GenBank accession number: NM002510), a polypeptide that has been reported to be expressed preferentially by lowly metastatic melanoma cells (25), had homology of 71.1% to DC-HIL (Fig. 2). In addition, DC-HIL showed homology with rat osteoactivin, for which the GenBank data annotates the abundant expression in osteopetrotic bones (88.3%); QNR-71, which is responsible for melanin production in quail neuroretina cells (26) (48%); and pMEL17s melanoma markers (27) (24%). Less homology was identified with lysosome-associated membrane protein (LAMP) family members (11-13%), such as hLAMP-1 and -2 (28, 29), mCD68 (30), and hDC-LAMP (31). These structural characteristics led us to postulate that DC-HIL is a type I transmembrane protein that is heavily glycosylated, is involved in heparin binding, promotes integrin-mediated cell adhesion, and is capable of transducing tyrosine-based signaling.
Cell and Tissue Distributions of DC-HIL mRNA-Northern blot analysis showed DC-HIL mRNA (2.9 kb) to be expressed at relatively high levels by XS52 DC, but only minimally by macrophages J774 and Raw (Fig. 3A). Importantly, DC-HIL mRNA was undetectable in other tested cell lines, including ␥␦ T cells (7-17 DETC), two ␣␤ T cells (HDK-1 and D10), B cell hybridoma clones (5C5), keratinocytes (Pam 212), and dermal fibroblasts (NS01). As noted in Fig. 3B, DC-HIL mRNA was detected in greatest quantities in bone marrow and adipose tissues, in modest levels in thymus, and at low levels in skin. Although mRNA was not detectable by Northern analysis in lymph node and spleen, RT-PCR demonstrated its expression in lymph node (data not shown). Tissue distribution of DC-HIL mRNA does not well correlate with that of DC, suggesting that the expression may be restricted to certain subpopulations of DC. The expression of DC-HIL mRNA in skin was consistent with its establishment from skin-derived XS52 DC (12), which resemble skin resident DC, i.e. LC in many features (11,12). We then assessed by RT-PCR the source of DC-HIL mRNA expression in epidermis (Fig. 3C). DC-HIL mRNA was detected in epidermal cell suspensions freshly prepared from BALB/c mice. Importantly, depletion of LC (Iaϩ) by treatment with anti-Ia mAb plus complement abrogated most of the signal for DC-HIL. The absence of PCR signals for IL-1␤, which is known to be specific for LC in mouse epidermis (32,33), shows complete depletion of LC. This corroborates our observations that DC-HIL mRNA was detected by Northern blot in the LC-like XS52 cells, but was totally absent from cells derived from other epidermal populations (i.e. Pam 212 keratinocytes and the 7-17 epidermal ␥␦ T cells) (Fig. 3A). On the other hand, the incomplete abrogation of DC-HIL mRNA by the depletion suggests a different minor source for this signal in epidermis, possibly melanocytes, because DC-HIL mRNA was detected in several murine melanoma lines by RT-PCR (data not shown). Collectively, our data suggest that DC-HIL mRNA is expressed constitutively and preferentially by XS52 DC and epidermal LC.

Identification of DC-HIL Protein-
The DC-HIL protein was characterized biochemically by immunoblotting with affinitypurified rabbit Ab raised against a synthetic peptide corresponding to aa 30 -48 of DC-HIL. This Ab detected under reducing conditions a broad band migrating between 90 and 110 kDa in COS-1 cells that was transfected with full-length cDNA for DC-HIL but not with the empty vector (Fig. 4A). The same Ab recognized in XS52 DC lysates two bands of 95 and 125 kDa. Consistent with Northern blot data (Fig. 3A), DC-HIL protein was expressed preferentially by XS52 DC, detectable in J774 macrophage and B16 melanoma cells, but undetectable in J558 myeloma and BW5147 thymoma cells (Fig. 4B). The lower two bands visible in COS-1 transfectants, XS52 DC, and other cell lines were also detected by control rabbit IgG (data not shown). The estimated molecular size of the DC-HIL protein was considerably larger than that predicted from the fulllength amino acid sequence (67 kDa), a discrepancy probably due to glycosylation, because DC-HIL contains 11 putative N-glycosylation sites (Fig. 1). Indeed, N-glycosidase treatment of XS52 DC lysate reduced significantly the molecular size of two bands (95 and 125 kDa) (Fig. 4C): A major band (76 kDa) was still larger than the predicted molecular size, whereas a minor band (66 kDa) was almost identical to the size. This heterogeneity in DC-HIL proteins may be produced as a result of differential N-and/or O-glycosylation, which may also explain the disparity in molecular size between native DC-HIL in XS52 DC and its recombinant form in COS-1 cells. DC-HIL is thus expressed preferentially by XS52 DC as a heavily glycosylated protein.
Subcellular Localization of DC-HIL-The subcellular localization was determined by a biochemical method. As shown in Fig. 5A, the 95-and 125-kDa immunoreactivities were detected exclusively in the membrane fraction, documenting that DC-HIL is a membrane-integrated polypeptide. To further separate the membrane fractions, whole cell lysates of surfacebiotinylated XS52 DC were incubated with streptavidinconjugated agarose beads and separated into unbound (cytoplasm) and bound (plasma membrane) fractions. Immunoblotting with anti-DC-HIL Ab revealed that a majority of DC-HIL proteins was recovered in the streptavidin-unbound fraction (intracellular fraction), with less amounts in the plasma membrane fraction. Similar results were obtained using COS-1 cells transfected with full-length cDNA for DC-HIL (Fig. 5C). On the other hand, recombinant luciferase, used as an intracellular marker when expressed transiently in COS-1 cells, was unrecoverable in the streptavidin-bound fraction (Fig. 5C), validating the efficacy of our system.
The intracellular localization was further examined by immunocytochemistry. When XS52 DC were membrane-permeabilized with saponin, most DC-HIL accumulated in a single, large perinuclear vesicle and also in small vesicles scattered toward the periphery (Fig. 6). This reactivity was significantly reduced when cells were stained in the absence of saponin (data not shown). Together, these results suggest that DC-HIL mostly localize in the cytoplasm, but is present at lower levels on the surface of XS52 DC.
Binding of Soluble DC-HIL-Fc to Cell Lines-To study the function of DC-HIL on DC, we prepared a soluble Fc fusion protein, DC-HIL-Fc. DC-HIL-Fc was produced in COS-1 cells as a disulfide-linked homodimer (data not shown), which was highly purified by affinity chromatography with protein Aagarose, and detected as a single band by SDS-PAGE and subsequent CBB staining (Fig. 7A). By using this recombinant protein, together with control proteins consisting of Dectin-1 ECD and hIgG 1 Fc (Dectin-1-Fc) or hIgG 1 Fc (Fc) alone, we were able to test binding of DC-HIL-Fc to the cell surface by flow  ␤-actin (B). C, epidermal cells freshly isolated from skin tissues of adult BALB/c mice were examined for DC-HIL mRNA expression by RT-PCR using the primer (see "Experimental Procedures") designed to amplify nucleotides 1029 -1404. Some epidermal cells were treated with anti-Ia mAb plus complement to deplete LC. The extent of depletion was assessed by measuring IL-1␤ mRNA, which is known to be expressed exclusively by LC in the mouse epidermis. cytometric analysis. Among the cell lines tested, only mouse vascular endothelial cells (SVEC), mouse melanoma cells (Queens), and dermal fibroblasts (NS47) showed significant binding (Fig. 7B). No binding was detected with the control Fc fusion proteins Dectin-1-Fc or Fc. We focused on characterizing the interaction of DC-HIL and SVEC, because this system may provide a useful model for understanding transendothelial migration of DC.
Because the ECD of DC-HIL contained both RGD and heparin-binding motifs (Fig. 1), we determined which motif is responsible for the binding of soluble DC-HIL-Fc. Inhibition studies showed that synthetic RGD-containing peptide (1 mM RGDS) fails to inhibit binding on SVEC (Fig. 7C), whereas binding was reduced significantly by 1 g/ml heparin (Fig. 7C). Further experiments demonstrated that other highly sulfated polysaccharides such as dextran sulfate (500 kDa) and fucoidan were also inhibitory in a dose-dependent manner, whereas the less sulfated polysaccharide, chondroitin sulfate A, and a nonsulfated polysaccharide mannan were much less or not inhibitory at all (Fig. 7D). These results suggest that HSPG, but not RGD, is involved in the binding of DC-HIL ECD to SVEC. We then tested whether DC-HIL binds directly to heparin or fucoidan. DC-HIL-Fc showed dose-dependent binding to wells coated with either heparin-BSA or fucoidan (Fig. 8A). Binding to heparin was inhibited by heparin, fucoidan, and dextran sulfate, but less so by chondroitin sulfate A or mannan (Fig.  8B). On the other hand, binding to fucoidan was not inhibited by heparin (Fig. 8C). These data suggest that a class of binding sites on DC-HIL is shared by heparin, fucoidan, and dextran sulfate, and a second class not recognized by heparin. Thus, DC-HIL binds to HSPG probably via multiple sites, including a heparin-binding motif. Recognition of HSPG on EC by DC-HIL was supported by the findings that just preincubation of SVEC with these inhibitors did not affect binding of DC-HIL-Fc (Fig.  8D), and that pretreatment with chlorate, a reversible inhibitor of glycosaminoglycan sulfation, significantly reduced binding (Fig. 8E). It is therefore highly likely that DC-HIL binds to SVEC through recognition of HSPG on the surface of SVEC.
Adhesion of SVEC to Immobilized DC-HIL-Fc-The presence of the RGD motif in the ECD of DC-HIL led us to test the possibility that binding of DC-HIL-Fc to SVEC leads to cell adhesion. We employed an in vitro adhesion assay in which cells were allowed to adhere to the recombinant proteins-immobilized plates. Typically 50 -60% of input SVEC adhered to plates coated with DC-HIL-Fc in a dose-dependent (Fig. 9A) and time-dependent manner (Fig. 9B), but they adhered to Dectin-1-Fc or Fc-coated plates only negligibly (data not shown). Maximal adhesion of SVEC was observed at 45-to 60-min incubation and at 37°C temperature. Importantly, an RGDS peptide but not control peptide RPKP inhibited adhesion of SVEC in a dose-dependent manner, with maximum inhibition (90%) at 0.5 mM (Fig. 9C). Moreover, chelation of divalent cations by EDTA, which interferes with the integrin subunit assembly (34,35), also inhibited adhesion dose-dependently. These data strongly suggest that SVEC adhere to immobilized DC-HIL-Fc through recognition of its RGD sequence by some members of integrin family, which are expressed on EC and play pivotal roles in the adhesion to leukocytes (36,37). Of interest was that the adhesion of SVEC was also inhibited by most of the sulfated polysaccharides that could inhibit soluble DC-HIL-Fc binding on SVEC (Fig. 7D). On the other hand, adhesion of SVEC to vitronectin, a well known extracellular matrix component containing RGD sequence as well as heparin-binding motifs (38,39), was blocked completely by RGDS peptide but not affected by any of these sulfated polysaccharides (data not shown). Thus DC-HIL-mediated adhesion is more dependent on HSPG recognition than vitronectin.
Heparin inhibited binding of soluble DC-HIL-Fc to SVEC, and blocked adhesion of SVEC to immobilized DC-HIL-Fc. However, RGDS had no effect on binding of soluble protein to SVEC, but prevented cell adhesion to immobilized DC-HIL-Fc. Therefore, we investigated whether DC-HIL-Fc interaction FIG. 5. Subcellular localization of DC-HIL. The subcellular localization was examined by biochemical method. A, crude homogenate prepared from XS52 DC was centrifuged at 50,000 ϫ g for 45 min, and the supernatant (cytosol fraction) and the pellet (membrane fraction) were separately recovered and tested for the amounts of DC-HIL by immunoblotting. An aliquot (equivalent to 1 ϫ 10 5 cells) in each fraction was loaded on SDS-PAGE. B, surface proteins on XS52 DC were labeled with biotin at 4°C, followed by detergent extraction of cells and precipitation by streptavidin-conjugated agarose beads. After centrifugation, the supernatant (streptavidin-unbound fraction, shown by "U") and the eluate from the beads (streptavidin-bound fraction, "B") were immunoblotted for DC-HIL. The sample proteins loaded for streptavidin-unbound and bound fractions were equivalent to 1 and 3 ϫ 10 5 cells, respectively. C, the efficacy of the surface biotinylation method was confirmed by immunoblotting for recombinant DC-HIL (left) and luciferase (Luc, used as an intracellular control protein, right) in COS-1 cells transfected with the corresponding cDNAs. The sample proteins loaded for streptavidin-unbound (U) and bound fractions (B) were the same as those in B. The data shown are representative of three independent experiments. with sulfated polysaccharides was needed prior to integrinmediated adhesion. Notably, BW5147 thymoma cells, which lacked the ability to capture DC-HIL-Fc (Fig. 7B) did not adhere to immobilized DC-HIL-Fc (Fig. 10). On the other hand, BW5147 cells showed constitutive adhesion to vitronectin, which was comparable to those of SVEC and Queens (Fig. 10), indicating a good correlation between cell binding and cell adhesive capacities of DC-HIL. Moreover, SVEC once adhered to immobilized DC-HIL-Fc were no longer removable by incubation with sulfated polysaccharides (data not shown). These data suggest that RGD-dependent cell adhesion is preceded by interaction of this molecule with HSPG.
Binding and Adhesion Profiles of RGD-deficient DC-HIL-Fc Mutant-To more definitively demonstrate involvement of RGD sequence in adhesion of SVEC to immobilized DC-HIL-Fc, we generated RGD-deficient DC-HIL-Fc, termed RAA mutant, by altering RGD to RAA using site-directed mutagenesis. This strategy has successfully documented the critical role of RGD in the interaction of several different molecules with integrins (40 -42). RAA mutant DC-HIL-Fc was produced in COS-1 cells and purified as highly as the wild-type was (Fig. 11A). When tested for its binding ability to SVEC, the RAA mutant showed an almost identical binding profile to the wild-type, including the inhibition by heparin (Fig. 11B). In sharp contrast, when tested for the adhesive activity, the mutant was unable to induce adhesion of SVEC (Fig. 11C). Thus, these results clearly demonstrate that the RGD-integrin interactions are indispensable for the adhesion of SVEC to immobilized DC-HIL-Fc but not for the binding of soluble DC-HIL-Fc to SVEC. DISCUSSION In searching for DC-specific genes that provide unique properties of DC, a number of DC-specific or -associated, novel molecules have been identified, e.g. DC-SIGN (43), Langerin (44), and DC-LAMP (31). In fact, functional studies on these molecules demonstrated that they are involved in establishing distinct properties of DC. In addition to dectin-1 and dectin-2, we added to the current list of DC-associated molecules a new member, termed DC-HIL, that may contribute to unique mechanisms for transendothelial migration of DC.
DC-HIL showed significant homologies with melanoma/ melanosome-associated membrane proteins and lower degree with members of the LAMP family (Fig. 2). In addition, its structural and biochemical features are also in accordance with these membrane protein families, which include: 1) a prolinerich hinge region containing numerous O-glycans, 2) highly N-glycosylated forms, 3) lysosomal targeting motifs, 4) a tyrosine-based signaling motif in a cytoplasmic tail, 5) high accumulation in a perinuclear region (Fig. 6) (28,29), and 6) relatively low expression on the cell surface (Fig. 5) (45)(46)(47). Because DC do not have melanosomes, we speculate that DC-HIL may be a newly identified member of the LAMP family and may play roles in the intracellular events.
In this study we focused on characterizing the function of the ECD rather than the intracellular domain of DC-HIL. The ECD contains an RGD motif and a heparin-binding motif, both of which are frequently found in adhesion molecules, leading us to hypothesize a role as an adhesion molecule. Using the recom- After washing, the adherent cells were lysed and measured for the radioactivity, and shown as percent cpm relative to total input cpm. B, time course of SVEC adhesion. In the same adhesion assays, the adhesive activity was measured at the various time points after incubation of SVEC in the wells precoated with 10 g/ml DC-HIL-Fc. C, blocking of SVEC adhesion by polysaccharides. SVEC were incubated for 60 min in the wells precoated with 10 g/ml DC-HIL-Fc in the presence of sulfated or nonsulfated polysaccharides at the 10-fold increasing concentrations. Inhibitory activity is shown by percentage of 3 H cpm obtained in the absence of inhibitors. D, blocking with inhibitors for integrin-mediated adhesion. Increasing doses of inhibitors (an RGDS tetramer, an RPKP control tetramer, and EDTA) were added to the same adhesion assays. Note that adhesion of SVEC to control fusion proteins typically showed as little as 1%. Data shown are representative of five independent experiments. binant fusion protein DC-HIL-Fc, we found two independent functions. The soluble form of DC-HIL-Fc binds to limited cell types, including EC, and the immobilized form of DC-HIL-Fc promotes adhesion of SVEC. In further studies, we demonstrated that: 1) binding of DC-HIL to cells is inhibited by heparin and fucoidan; 2) DC-HIL recognizes heparin and fucoidan; 3) SVEC adhesion to immobilized forms absolutely requires RGD recognition; and 4) the adhesion is induced by the recognition of HSPG. Based on these findings, we propose dual roles of DC-HIL: first, recognition of HSPG, which functions to specify the target cells; second, a ligand for integrins, which mediates an RGD-dependent cell adhesion only when DC-HIL binds to cells. Although it is unclear at present how the interaction of DC-HIL and HSPG on EC induces an RGDdependent cell adhesion, two possible molecular models are considered: First, the interaction activates integrins expressed on SVEC, resulting in recognition of the RGD. Second, the interaction induces conformational changes of DC-HIL, thereby exposing the cryptic RGD sequence, as suggested for von Willebrand factor and thrombospondin (48,49). Further studies will be required for determining which model accounts for the DC-HIL-induced cell adhesion.
This working model shares some features with the two-step adhesion cascade for neutrophil/EC interactions, in which the primary adhesion involves three members of selectin family and their oligosaccharide ligands. Subsequently, secondary adhesion involves ␤2 (CD18) integrins on neutrophils and immunoglobulin gene superfamily counter-receptors on EC, primarily ICAM-1 (37). However, DC-HIL-mediated adhesion differs from this two-step model in that DC-HIL serves also as one of the counter-receptors for integrins. In this respect, DC-HIL is distinguishable from other adhesion molecules on the cell surface: e.g. Thy-1, Ly-5, and NCAM are all capable of interacting with heparin (50 -52), whereas they do not function as ligands for integrins. This feature of DC-HIL may exemplify that DC display not only common but also unique mechanisms for their transendothelial migration.
Heparan sulfate, as products of the enzymatic degradation of HSPG, is rapidly released from cell surface and extracellular matrices under conditions of inflammation and tissue damages (53,54). Recently, this heparan sulfate has been demonstrated to induce phenotypic as well as functional maturation of DC characterized by enhanced surface expression of major histocompatibility class II, CD80 (B7-1), and CD86 (B7-2), lowered rate of antigen uptake, and increased allostimulatory capacity (55). Although it is not defined how heparan sulfate induces maturation signals to DC, direct binding of heparan sulfate to its putative receptors on DC is thought to be one of the most possible mechanisms. Because a toll-like receptor, which is expressed on DC, binds to peptidoglycan, including lipopolysaccharide and heparan sulfate (56,57), this receptor is proposed to be a candidate responsible for inducing maturation signals. It is thus tempting to propose that DC-HIL is a second candidate for such a receptor, which may transduce maturation signals to DC via the residues ITAM on its cytoplasmic tail.
We were unable to block adhesion of XS52 DC to SVEC by pretreatment of SVEC with DC-HIL-Fc even at 75 g/ml (data not shown). Because our anti-DC-HIL antibody does not recognize proteins expressed on the cell surface of XS52 DC (probably the epitope for the Ab is cryptic in the molecule), we have not examined blocking of SVEC adhesion by Ab. Successful blocking may require antibodies reactive with the DC-HIL proteins on cell surface. It thus remains uncertain to which extent DC-HIL contributes to the DC/EC adhesion under physiological conditions. Nevertheless, DC-HIL is potentially involved in endothelial adhesion of DC, and its preferential expression of the protein in DC suggests that it plays critical roles in the unique mechanisms for DC migration.