Endo180, an endocytic recycling glycoprotein related to the macrophage mannose receptor is expressed on fibroblasts, endothelial cells and macrophages and functions as a lectin receptor

Endo180 was originally identified as an antigen recognized by four different monoclonal antibodies (mAbs) which were raised as part of a panel of reagents designed to identify novel human fibroblast cell surface receptors (Isacke et al., 1990). Interest in this 180 kDa transmembrane glycoprotein stemmed from three key observations. First, N-terminal and tryptic peptide sequences obtained from the purified protein suggested that this was a novel receptor. Second, this protein was found to have a restricted cell type expression. Finally, and most interestingly, was the subcellular distribution and trafficking of the receptor. Immunofluorescence and immunoelectron microscopy studies revealed that in cultured fibroblasts, the Endo180 was concentrated on the plasma membrane into clathrin-coated pits. Moreover this cell surface protein only represented 10-30% of the total Endo180 and that the remaining 70-90% was localized to intracellular vesicles, identified by co-localization with the transferrin receptor as being endosomes. Using mAb Fab′ fragments it was then demonstrated that cell surface receptor could be internalized. This process was very rapid with greater than 60% of plasma membrane protein being taken up into the cell within 2 minutes of warming to 37°C. Importantly, using these Fab′ fragments the fate of the internalized receptor could be monitored. It was found that at least a significant fraction of the protein taken up from the cell surface into the endosomes recycled back to the plasma membrane within 60 minutes. Given its molecular size and recycling endocytic properties, it is proposed that this receptor is termed Endo180.

Receptor-mediated internalization of ligands is a vital physiological process carried out by all eukaryotic cells. It provides a mechanism by which a vast array of components such as nutrients, chemokines, hormones, toxins and pathogens are taken up by cells. Receptors that are internalized via clathrin coated pits can be divided into two groups (Trowbridge et al., 1993). The first are receptors, such as the epidermal growth factor receptor, for which internalization is driven by ligand binding and results in both the ligand and receptor being targeted for degradation in the lysosome. The second are receptors such as the transferrin, low density lipoprotein and asialoglycoprotein receptors whose internalization is constitutive and independent of ligand binding. In general, ligands internalized by these receptors dissociate in the low pH environment of the endosome leaving the receptors to recycle back to the plasma membrane. The biochemical, morphological and kinetic analysis of Endo180 (Isacke et al., 1990) strongly suggested that this glycoprotein was a novel, constitutively recycling receptor. In this manuscript we have made use of the specific anti-Endo180 mAbs and polyclonal antisera to facilitate cloning of the full length cDNA and undertake a detailed mechanistic and functional characterization of this endocytic receptor.

A human placental λgt11 library (Millán, 1986) was screened using an anti-Endo180 polyclonal antiserum, inserts were excised from positive colonies with EcoRI, ligated into pBluescript and partially sequenced. A human stromal λZAP cDNA library (J. Boulter, UCLA, USA) was screened with a human EST (I.M.A.G.E. clone number 1252230; Lennon et al., 1996) which had sequence identity to one of the positive clones isolated from the λgt11 library. Positives λZAP clones were rescued by in vivo excision of the pBluescript phagemid and characterized by PCR, restriction endonuclease digestion and sequencing. This resulted in the identification of λ10.2 as a putative full length Endo180 clone. The insert from λ10.2 was excised from pBluescript using NotI and XhoI and subcloned into the complementary sites in the pcDNA3 expression vector (Invitrogen). MDCK cells were transfected with pcDNA3-Endo180 as previously described (Sheikh and Isacke, 1996) and cultured for 24 hours before analysis.

AG1523, Flow2000 and MRC-5 fibroblasts, RPM-MC melanoma cells (Byers et al., 1991) and HepG2 cells were maintained in DME supplemented with 10% FCS. Human umbilical vein endothelial cells (HUVEC) were isolated from umbilical cords by digestion with collagenase type II as described previously (Wellicome et al., 1990) and used at passage 3. Human dermal microvascular endothelial cells (DMEC) were isolated from human foreskins and cultured on fibronectin-coated flasks as previously described (Mason et al., 1996) and used at passage 4 to 6. The human dermal microvascular endothelial cell line (HMEC-1; Ades et al., 1992), a gift from Dr E. Ades (CDC, Atlanta, USA) was cultured on 1% gelatin-coated tissue culture flasks in MCDB-131 growth medium (Gibco) supplemented with 10% FCS, 10 ng/ml epidermal growth factor (Becton Dickinson), 100 IU/ml penicillin, 0.1 mg/ml streptomycin and 2 mM L-glutamine. Human monocyte-derived macrophages were prepared by isolating peripheral blood mononuclear cells on a Percoll gradient, enriching for monocytes by selective attachment to plasma-coated dishes and differentiating in vitro by culture in autologous serum for 6 days (McCutcheon et al., 1998).

The generation and characterization of mAb E1/183 and the polyclonal anti-Endo180 antiserum have been described previously (Isacke et al., 1990). mAb A5/158 was generated by immunizing mice with human AG1523 fibroblasts and screening for mAbs directed against plasma membrane proteins (Isacke et al., 1986). Immunoprecipitation of ¹²⁵I-labelled cell surface proteins followed by limited proteolytic digestion with 0.25-10 μg Staphylococcus aureusV8 protease demonstrated that E1/183 and A5/158 recognized the same 180 kDa antigen.

Semi-confluent cultures of cells were lysed in sample buffer, sonicated, resolved on 10% SDS polyacrylamide gels and transferred to nitrocellulose membranes as previously described (Neame and Isacke, 1993). Detection of Endo180 was performed by incubation of the membrane with 4 μg/ml A5/158 anti-Endo180 mAb followed by 150 ng/ml HRP-conjugated anti-mouse Ig (Jackson Immunoresearch). Blots were developed using the enhanced chemiluminescence (ECL) kit (Amersham) and exposed to X-ray film (Fuji XR) at room temperature.

Confluent monolayers of endothelial cells were harvested by exposure to trypsin/EDTA (ICN) for 1 minute at 37°C. After repeated pipetting to ensure a single cell suspension, cells were incubated with 50 μg/ml mAb A5/158 or isotype matched irrelevant mAb for 30 minutes at 4°C, washed in HBSS plus 2.5% FCS, incubated with 20 μg/ml FITC-labelled rabbit anti-mouse Ig (Dako) for 30 minutes at 4°C, washed twice and fixed with 1% paraformaldehyde. Samples were analysed on a Becton Dickinson FACScan flow cytometer by counting 10⁴cells per sample. Flow cytometry of macrophages was performed as described previously (McCutcheon et al., 1998).

Cryosections of human term placenta were permeabilized with 0.5% Triton X-100 and stained with mAb E1/183 and counterstained with biotinylated Ulex europaeusagglutinin-I (UEA) lectin followed by Cy3-goat anti-mouse Ig and avidin-FITC. Cryosections of human skin were treated as described above and stained with mAb E1/183 followed by Cy3-goat anti-mouse Ig. Sections were then blocked with mouse serum and then counterstained with biotinylated anti-CD14 mAb (Serotec) followed by avidin-FITC. Confocal images were collected on the Leica TCS NT system. In all cases, parallel sections were stained for each marker separately to establish that the pattern of expression observed was not due to cross reactivity of second layer reagents.

HMEC-1 cells (cultured on gelatin-coated glass coverslips) and Flow2000 cells were fixed with 3% paraformaldehyde and processed as previously described (Neame and Isacke, 1993) using 100 μg/ml mAb E1/183 followed by 10 μg/ml rhodamine-conjugated anti-mouse Ig and confocal images were collected on the Leica TCS NT system.

35 mm dishes of AG1523 human fibroblasts were washed in phosphate-free DME and incubated for 16 hours in 2 ml of phosphate-free DME containing 2% FCS, 20 mM HEPES, pH 7.5, and 2 mCi [³²P]orthophosphate (Amersham). Parallel dishes were then left untreated or treated with 50 ng/ml TPA (12-O-tetradecanoyl-phorbol-13-acetate) for 10 minutes prior to lysis and preclearing as described previously (Neame and Isacke, 1992). Human Endo180 was immunoprecipitated using mAb E1/183, followed by goat anti-mouse Ig prebound to Protein A-agarose (Bio-Rad), resolved by SDS-PAGE and gels were exposed overnight at −70°C. Endo180 bands were excised from the dried gel and subjected to two-dimensional phosphoamino acid analysis as previously described (Cooper et al., 1983).

35 mm dishes of Flow2000 human fibroblasts were cultured for 1 hour in methionine-free DME plus 4% dialysed FCS and then 300 μCi of [³⁵S]methionine (Amersham) per ml was added for a further 15 minutes. The dishes were washed and incubated with DME containing 5% FCS and 3 mM methionine for 0 to 24 hours and Endo180 immunoprecipitated with mAb E1/183 as described above. For endo H (endo-β-N-acteylglycosaminidase H; Boehringer Mannheim) treatment, immunoprecipitates were washed three times with PBS, boiled for 2 minutes in 25 μl of 50 mM Tris-HCl, pH 5.5, 0.02% SDS, and then incubated at 37°C for 3 hours with 2.5 mU of enzyme and 1 mM PMSF (phenylmethylsulphonyl fluoride). For endo F (endo-β-N-acetylglucosaminidase F) treatment, the immunoprecipitates were washed 3 times with PBS, boiled for 2 minutes in 25 μl of 100 mM sodium phosphate buffer, pH 6.1, 50 mM EDTA, 1% 2-mercaptoethanol and 0.1% SDS. Nonidet P-40 (NP-40) was added to 1% and samples incubated with 0.5 units of endo F in the presence of 1 mM PMSF for 3 hours at 37°C. For neuraminidase treatment, immunoprecipitates bound to Protein A-agarose were washed 3 times with PBS, resuspended in 25 μl of 50 mM sodium acetate, pH 5.5, 1 mM CaCl₂, 1 mM PMSF, 5 units/ml Clostridum perfringensneuraminidase type X (Sigma) and incubated at 37°C for 4 hours. To stop the reactions an equal volume of 2× concentrated SDS sample buffer was added, the samples boiled for 2 minutes and resolved by SDS-PAGE. Gels were exposed to X-ray film for 6 days.

Flow2000 human fibroblasts and HMEC-1 cells were grown to confluence on 9 cm dishes, rinsed with Tris saline (15 mM Tris-HCl, pH 7.5, 150 mM NaCl), lysed in 900 μl of detergent extraction solution (150 mM Tris-HCl, pH 8.0, 10 mM NaCl, 10 mM CaCl₂, 1 mM MgCl₂and 0.15% Triton X-100), incubated on ice for 15 minutes before centrifugation at 14,000 gin an Eppendorf microfuge and the supernatants collected. 1 ml mannose-, mannan-, fucose-, galactose-or -N-acetylglucosamine-agarose (Sigma) were loaded into PolyPrep chromatography columns (Bio-Rad) and pre-equilibrated by rinsing with 5 ml elution buffer (150 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM MgCl₂, 5 mM EDTA and 0.15% Triton X-100) followed by 5 ml of loading buffer (150 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM MgCl₂, 25 mM CaCl₂and 0.15% Triton X-100). 300 μl of cell lysate was loaded with an additional 700 μl of loading buffer and this fraction was collected as fraction 1. The column was rinsed with 4× 1 ml of loading buffer and washes (fractions 2-5) were collected. The column was then eluted with 5× 0.5 ml elution buffer (fractions 6-10). All fractions were precipitated by the addition of 20 μg of BSA and 0.5 volumes of 30% TCA. The samples were placed on ice for 10 minutes, centrifuged at 14,000 gin an Eppendorf microfuge, the pellets washed twice with ethanol/ether (1:1), vacuum dried, resuspended in 40 μl of sample buffer, resolved on 10% SDS-polyacrylamide gels and analysed by western blotting.

A polyclonal anti-Endo180 antiserum was used to screen a human placental λgt11 library and one clone (λ23.2) showed sequence identity to a number of human ESTs. One of these ESTs (I.M.A.G.E. clone number 1252230) was used to screen a human stromal λZAP cDNA library and PCR and restriction digest analysis suggested that clone λ10.2 contained a full length insert. The insert from λ10.2 (accession number AF134838) was fully sequenced and subcloned into the pcDNA3 eukaryotic expression vector.

Three lines of evidence demonstrate that clone λ10.2 encodes a full length Endo180 cDNA: (a) the purified Endo180 protein was subjected to N-terminal sequencing and shows a 22/26 amino acid match with the sequence of λ10.2 after cleavage of the signal sequence (Fig. 1A). In addition, protein sequence was obtained from nine Endo180 tryptic peptides. Sequence matches for 8 of these peptides are found within the human Endo180 cDNA (data not shown). (b) Clone λ23.2 isolated by polyclonal antibody screening is identical to the 3′ end of λ10.2 (bases 3,837 to 5,726: Fig. 1A), (c) clone λ10.2 was inserted into the pCDNA3 vector, expressed in MDCK epithelial cells and western blot analysis of transfected cells with an anti-Endo180 mAb revealed a single 180 kDa immunoreactive band (Fig. 1C). In these transfected cells, the distribution of Endo180 (data not shown) is identical to that observed for endogenous Endo180 (see below).

Analysis of the human Endo180 cDNA sequence reveals a number of striking features:

• The cDNA clone encodes for a protein with a large extracellular domain containing an N-terminal signal sequence followed by a cysteine-rich domain, a fibronectin type II domain and 8 C-type carbohydrate recognition domains (CRDs). The single transmembrane domain is followed by a short 43 amino acid cytoplasmic domain (Fig. 1A). This arrangement is unusual due to the presence of multiple CRDs within a single polypeptide backbone and has only been found in three other related proteins; the macrophage mannose receptor (Ezekowitz et al., 1990; Taylor et al., 1990; Harris et al., 1992), the phospholipase A₂receptor (Higashino et al., 1994; Ancian et al., 1995) and the DEC-205/MR6 receptor (Jiang et al., 1995; McKay et al., 1998).

• Sequence analysis indicates that Endo180 is the probable human orthologue of the recently described mouse cDNA clone identified in a screen for novel proteins containing C-type lectin domains (Wu et al., 1996; accession number MMU56734). Overall, the human Endo180 has 87% nucleotide identity with its mouse counterpart and greater than 90% identity in the fibronectin type II repeat, CRD1, 2, 3 and 6 and the transmembrane domain.

• ESTs derived from the ENDO180gene map to the D17S791-D17S794 interval on human chromosome 17q (Deloukas et al., 1998). Database searching identified a 174,428 bp genomic sequence from human chromosome 17 (accession number AC005821) which contains 29 exons and 28 introns of Endo180 covering a 29 kb stretch (Fig. 1Aand 1B). From the similarity of this intron-exon boundary with the 30 exons of the human and mouse macrophage mannose receptors (Kim et al., 1992; Harris et al., 1994), it is predicted that the 29 exons in the AC005821 sequence represent exons 2-30 of human Endo180 and that exon 1 is separated from exon 2 by a large intron of greater than 32 kb.

• The sequence of human Endo180 shows a high degree of similarity to a cDNA clone recently isolated in a strategy to clone large cDNAs from human brain (Ishikawa et al., 1998; accession number AB014609). Comparison of the λ10.2 and AB014609 sequences show that the former has 8 bp and 72 bp 5′ and 3′ extensions, respectively, and the latter contains an open reading frame disrupted by a 5 bp deletion at position 3341-3345. The presence of these 5 bp in clone λ10.2 suggests that the deletion in the AB014609 cDNA results from a cloning artefact rather than representing an alternative splicing event. Additionally, clone λ10.2 shows two base changes and these changes are also found in the AC005821 genomic chromosome 17 sequence. The first is a G to A change at position 243 which would result in an amino acid change from Val to Ile at amino acid 43. Interestingly, when this region was subjected to protein sequence analysis, both Val and Ile were assigned at this position (Fig. 1A) suggesting that this base change represents a natural amino acid polymorphism. The second (position 2,717) is an A to G change which does not alter the amino acid sequence and may also be a polymorphism.

Together the cDNA and genomic sequence data strongly suggest that human Endo180 and the novel murine C-type lectin identified by Wu et al. (1996)represent the human and mouse counterparts of a fourth family member related to the macrophage mannose receptor, the phospholipase A₂receptor and the DEC-205/MR6 receptor (Taylor, 1997; Stahl and Ezekowitz, 1998).

We have previously demonstrated that Endo180 has a restricted tissue and cell type distribution with high levels found in fibroblastic cells both in vivo and in vitro and little or no expression on epithelial cells or haematopoietic cell lines (Isacke et al., 1990). Subsequently, by northern blot analysis, Wu et al. (1996)demonstrated a similar expression pattern of murine Endo180 transcripts with high levels found in lung and kidney but little or no expression in brain, thymus or adult liver. In order to gain clues as to the functional role of Endo180 it was important to further characterize its distribution. As described here, Endo180 is structurally and functionally (see below) related to the macrophage mannose receptor. These two receptors clearly have a distinct distribution. For example, unlike Endo180, the macrophage mannose receptor is expressed both in the liver and thymus (Magnusson and Berg, 1993; Fiete et al., 1997; Linehan et al., 1999) but it is not known whether there is any overlap in expression in other cell types such as macrophages. In addition, by in situ hybridization, Wu et al. (1996)detected the presence of Endo180 transcripts at highly endothelialized sites such as those in the choroid plexus and kidney glomerulai. However, the limitation in resolution of this type of analysis precludes a definitive localization of Endo180 protein to endothelial cells. To address these particular issues we have examined the expression of Endo180 on endothelial cells and macrophages both in vivo and in vitro using anti-Endo180 mAbs.

Western blot analysis of cultured cells revealed a high level of Endo180 expression in fibroblasts, large vessel endothelial cells (HUVEC) and both primary (DMEC) and immortalized (HMEC-1) small vessel endothelial cells. By contrast no expression was detected in a hepatoma cell line (HepG2) in agreement with the lack of detection of Endo180 protein (Isacke et al., 1990) and transcripts (Wu et al., 1996) in the adult liver (Fig. 2A). Flow cytometric analysis demonstrated that Endo180 is expressed at the plasma membrane of both large and microvascular endothelial cells with a unimodal distribution (Fig. 2B). To determine whether this expression of Endo180 by cultured endothelial cells reflected expression in vivo, the distribution of Endo180 was examined in sections of human term placenta counterstained with an endothelial specific lectin, UEA (Fig. 3a-c). In these sections it was found that Endo180 was expressed on a subset of microvascular vessels. Strong co-localization with UEA was observed on the larger microvessels but Endo180 was not detected on the smaller microvessels. In addition, dispersed Endo180-positive/UEA-negative cells were identified in placenta mesenchyme which morphologically resembled Hofbauer cells, the macrophage-like placental scavenger cells.

The expression of Endo180 by macrophages in vivo was confirmed by staining human skin sections with mAbs directed against Endo180 mAb and CD14 (Fig. 3d-f). Dispersed throughout the dermis, double labelled cells are clearly visible, by contrast the epidermis is Endo180-negative. To determine whether the expression was restricted to dermal macrophages, human peripheral blood monocytes were differentiated in vitro and their maturation to macrophages confirmed by their expression preparations. In all of the distribution studies described here, identical results were obtained with the different anti-Endo180 mAbs (data not shown).

The ability of the macrophage mannose receptor, the phospholipase A₂receptor and the DEC-205/MR6 receptor to be endocytosed from the cell surface suggests that this family of receptors have a role in the internalization of extracellular components (Kruskal et al., 1992; Jiang et al., 1995; Zvaritch et al., 1996). Although cell surface Endo180 expression was detected on endothelial cells and macrophages (Figs 2and 4) it was suspected that this represented only a proportion of the total Endo180 as flow cytometric analysis of permeabilized cells demonstrated that the majority of Endo180 is intracellular (data not shown). To investigate the cell surface and intracellular localization of Endo180, microvascular endothelial cells and fibroblasts were examined by confocal microscopy (Fig. 4). Consistent with the flow cytometry data, permeabilized endothelial cells exhibited much stronger Endo180 staining with the protein being localized throughout the cytoplasm to intracellular vesicles (Fig. 4D). In the non-permeabilized cells, Endo180 is localized on the cell surface in a fine punctate distribution (Fig. 4C). This distribution of CD14, CD11c and HLA-DR (Fig. 2C). On these monocyte-derived macrophages, a unimodal cell surface distribution of Endo180 was detected and this expression was consistent in different macrophage pattern is identical to that observed in human fibroblasts (Fig. 4C,D) where it has previously been demonstrated that Endo180 co-localizes with clathrin to plasma membrane coated pits and with the transferrin receptor to intracellular endosomes (Isacke et al., 1990).

The ability of receptors to be recruited into clathrin coated pits and subsequently endocytosed, requires that their cytoplasmic domains can interact with intracellular adaptin complexes. An examination of the Endo180 cytoplasmic domain revealed that although this domain is highly conserved between human and mouse Endo180, there is only 24%, 30% and 12% identity to the cytoplasmic sequences of the macrophage mannose receptor, phospholipase A₂receptor and DEC-205/MR6 receptor, respectively (Fig. 5). However, two putative endocytosis motifs are present in the Endo180 sequence. The first of these is based on Tyr1452. Tyrosine-based endocytosis motifs have been extensively characterized and optimally have the sequence FxNxxY forming a tight turn structure (Trowbridge et al., 1993). Although Tyr1452 is conserved between family members, human Endo180 has a glycine rather than asparagine three residues upstream and murine Endo180 has a phenylalanine residue in the equivalent position. The second putative endocytosis motif centres on the di-hydrophobic Leu1468/Val1469 residues which have a glutamic acid four amino acids upstream. It is known from studies on other receptors that di-hydrophobic motifs can mediate intracellular trafficking (Sandoval and Bakke, 1994) and the presence of an upstream acidic residue has been shown to be important for targeting to endocytic vesicles (Pond et al., 1995). Comparison of the cytoplasmic domains between family members reveals a second interesting issue. Both of the putative endocytosis motifs in Endo180 are upstream from an adjacent serine residue, a feature that is not found in the macrophage mannose receptor. [³²P]orthophosphate labelling showed a low level of constitutive Endo180 phosphorylation but interestingly phosphorylation was dramatically and rapidly enhanced in response to activation of protein kinase C. Moreover, phosphoamino acid analysis demonstrated that this protein kinase C mediated phosphorylation is restricted to serine residues (Fig. 6).

One issue that is not addressed by these studies is whether endocytosis of Endo180 from the plasma membrane accounts for all of the intracellular receptor detected by confocal microscopy. It is known that other lectins such as the mannose-binding ERGIC-53 protein function intracellularly to sort immature or incorrectly processed glycoproteins in the biosynthetic pathway (Arar et al., 1995). To examine the post-translational processing, Endo180 was immunoprecipitated from [³⁵S]methionine pulse-chased cells and then treated with different endoglycosidases. After 15 minutes of [³⁵S]methionine labelling a single 170 kDa immature form is detected, which matures into the 180 kDa form with increasing chase times (Fig. 7). Treatment of this immature 170 kDa form with endo H or endo F results in the removal of approximately 15 kDa and 10 kDa, respectively, of carbohydrate, but no change in mobility was detected after neuraminidase treatment. The mature 180 kDa form of Endo180 immunoprecipitated after a 4 or 24 hour chase is resistant to endo H treatment, but approximately 10 kDa of carbohydrate is removed by endo F and approximately 5 kDa by neuraminidase. Endo F hydrolyses the glycosidic bond of both complex and high mannose oligosaccharides N-linked to asparagine residues, while endo H only hydrolyses high mannose structures. These data demonstrate that maturation of Endo180 is accompanied by modification of N-linked carbohydrates to complex structures and the addition of neuraminidase-sensitive terminal sialic acid residues. As all of the Endo180 is fully processed, these studies indicate that intracellular Endo180 does not represent immature glycoprotein that has been retained in the biosynthetic pathway.

As described above, a key characteristic of the 4 macrophage mannose receptor-related proteins is the presence of multiple CRDs within their extracellular domains. However, it is important to note that these domains are defined on the basis of size and the presence of certain conserved amino acids resulting in an overall structural homology and that the presence of a CRD in a protein does not necessarily confer carbohydrate binding capacity. Moreover, despite the presence of 8 CRDs within the macrophage mannose receptor, detailed analysis has revealed that the binding of glycoconjugates terminating in mannose, N-acetylglucosamine or fucose is primarily mediated by CRD4 (Weis et al., 1998). Within CRD4, carbohydrate binding requires four cysteine residues to create nested disulphide bonds plus other key residues for the formation of two Ca²⁺binding sites and the packing and formation of the hydrophobic cores (Weis et al., 1991; Mullin et al., 1994). An examination of the Endo180 sequence reveals that the Endo180 CRD4 has retained the four cysteine residues at equivalent spacing but key residues required for Ca²⁺and sugar co-ordination are not conserved. However, as shown in Fig. 8, CRD2 in both mouse and human Endo180 has a strong consensus binding site including the critical Ca²⁺/sugar binding site residues, correctly spaced cysteine residues and 26/28 of the remaining conserved amino acids found in functional CRDs. This sequence analysis suggests that if Endo180 has lectin activity, ligand binding will be primarily mediated by this domain. Interestingly, many of these residues are also conserved in Endo180 CRD1 although there is a loss of conservation in Ca²⁺site 2.

To determine whether Endo180 exhibits Ca²⁺-dependent carbohydrate binding, cell lysates were passed over mannose, mannan, fucose, galactose or N-acetylglucosamine columns and wash fractions and EDTA eluted fractions were assayed for Endo180 content. By this type of analysis, Endo180 extracted from Flow2000 fibroblasts did not bind to mannose, fucose (Fig. 9A,B), galactose or mannan (data not shown). By contrast, Endo180 was retained on N-acetylglucosamine columns and eluted in the presence of EDTA (Fig. 9C). This was not a cell line specific phenomenon as identical binding was observed with Endo180 extracted from AG1523 fibroblasts (data not shown). To determine whether this carbohydrate binding was restricted to fibroblast Endo180, experiments were repeated with lysates from the microvascular endothelial cell line, HMEC-1. As shown in Fig. 9D, endothelial Endo180 did not bind mannose but was retained on N-acetylglucosamine columns and eluted after chelation of the Ca²⁺ions (Fig. 9E). The same pattern of binding was also observed with Endo180 extracted from the large vessel endothelial cells, HUVEC (data not shown). These experiments directly demonstrate that Endo180 has lectin activity and that its ligand specificity is distinct from that of the macrophage mannose receptor.

In this manuscript we describe the cloning and characterization of human Endo180. Sequence analysis demonstrates that this receptor and the novel murine C-type lectin identified by Wu et al. (1996)represent the human and mouse counterparts of a fourth family member related to the macrophage mannose receptor, the phospholipase A₂receptor and the DEC-205/MR6 receptor (Stahl and Ezekowitz, 1998).

Despite the overall structural similarity between these family members, there are important functional distinctions, most notably in their carbohydrate binding abilities. By mutational analysis it has been shown that CRD5 is the critical domain involved in the binding of phospholipase A₂to its receptor with CRDs 3, 4 and 6 being required but of less importance (Higashino et al., 1994; Nicolas et al., 1995). However, phospholipase A₂is a non-glycosylated protein and this binding is via a protein:protein interaction and is Ca²⁺independent. Similarly, the DEC-205/MR6 receptor has been suggested to play a role in antigen presentation of internalized glycoproteins in dendritic cells and thymic epithelium but like the phospholipase A₂receptor, none of its CRDs has conserved the key amino acids involved in carbohydrate and Ca²⁺binding. Consequently it is unlikely that the CRDs in either of these two receptors has functional lectin activity. By contrast, these key amino acids are conserved in CRD2 and CRD4 of Endo180 and the macrophage mannose receptor, respectively (Fig. 8), and both these receptors have been demonstrated to mediate Ca²⁺-dependent carbohydrate binding (Fig. 9; Weis et al., 1998). However, Endo180 and the macrophage mannose receptor display two important differences in their sugar binding. First, they have distinct ligand specificities. The macrophage mannose receptor binds mannose, fucose and N-acetylglucosamine (Stahl and Ezekowitz, 1998; Weis et al., 1998) whereas Endo180 does not bind mannose or fucose but does bind N-acetylglucosamine (Fig. 9). Second, although Ca²⁺site 2 is conserved within the macrophage mannose receptor CRD4, Ca²⁺site 1 is not and instead CRD4 has a distinct mechanism for binding the second cation (Mullin et al., 1994, 1997). CRD2 within Endo180 has conserved the Ca²⁺site 1 indicating that it is more similar in structure to other mannose binding proteins (Weis et al., 1991) than the macrophage mannose receptor. Interestingly, the N-terminal cysteine-rich domain of the macrophage mannose receptor can mediate Ca²⁺-independent binding of oligosaccharides with terminal SO₄-4-GalNAcβ1,4GlcNAcβ1,2Manα structures (Fiete et al., 1997, 1998). Moreover, Martínez-Pomares et al. (1996, 1999) have demonstrated that this domain can bind cell surface ligands such as sialoadhesin and CD45 independently of the CRDs. Although the cysteine-rich region is the least well conserved of all the different domains within the receptor family, it will be important to determine if this region in Endo180 can mediate Ca²⁺-independent carbohydrate binding. Despite the differences in mechanisms of ligand binding, a feature in common to the macrophage mannose receptor, the phospholipase A₂receptor and the DEC-205/MR6 receptor is their ability to function as endocytic receptors (Kruskal et al., 1992; Jiang et al., 1995; Zvaritch et al., 1996). In previous biochemical and morphological studies, we have demonstrated that in fibroblasts Endo180 is localized to clathrin coated pits on the plasma membrane from where it is rapidly and constitutively internalized into endosomal compartments and then recycled back to the cell surface (Isacke et al., 1990). In this manuscript we have extended these morphological studies to demonstrate that the endocytic capacity of Endo180 is not a cell type specific phenomenon as an identical subcellular distribution is found in endothelial cells. In addition, we have undertaken biochemical analysis to further characterize the biosynthesis and trafficking of Endo180. The majority of Endo180 is found in its mature 180 kDa form within 2 hours of synthesis and this maturation is accompanied by a resistance to endo H digestion and a sensitivity to neuraminidase digestion (Fig. 7). The formation of endo H resistant complex oligosaccharides and the addition of sialic acid takes place in the medial and trans-Golgi stacks. Together with the morphological data (Fig. 4) and results obtained using Fab′ fragments to monitor receptor trafficking (Isacke et al., 1990), this indicates that all of the intracellular Endo180 protein has been fully post-translationally modified and subsequently internalized from the plasma membrane.

The endocytosis of proteins via clathrin-coated pits requires that they have a ‘positive’ signal in their cytoplasmic domains for association with intracellular adaptin complexes (Trowbridge et al., 1993; Kirchhausen et al., 1997; Marsh and McMahon, 1999). As with the other family members, examination of the Endo180 cytoplasmic domain sequence reveals the presence of two putative endocytosis motifs (Fig. 5). However, a distinct feature in Endo180 is the presence of a C-terminal serine residue adjacent to both putative endocytosis motifs. We demonstrate here that phosphorylation of Endo180 is dramatically and rapidly upregulated after treatment of cells with phorbol esters and that this phosphorylation occurs on serine residues. Evidence that this phosphorylation is directly mediated by protein kinase C comes from studies showing that immunopurified Endo180 can be phosphorylated by this kinase in vitro (Isacke et al., 1990). Receptor phosphorylation can be an important modulator of intracellular trafficking (Trowbridge et al., 1993). For example, phosphorylation is required for efficient endocytosis of the polymeric immunoglobulin receptor (pIgR; Casanova et al., 1990) and the B₂adrenergic receptor (Pizard et al., 1999). In the case of CD4, protein kinase C mediated phosphorylation results in an increased rate of endocytosis from the cell surface and the diversion of CD4 from the recycling pathway to a pre-lysosomal compartment (Pelchen-Matthews et al., 1993).

An examination of the cell type distribution of Endo180 in vivo and in vitro has revealed high levels of expression on fibroblasts, microvascular endothelial cells and macrophages. The data presented here strongly suggests that in these cells Endo180 will function to internalized glycosylated ligands with terminal N-acetylglucosamine residues via clathrin-coated pit mediated endocytosis. Like mannose, N-acetylglucosamine is not a common terminal sugar on mammalian oligosaccharides (Drickamer and Taylor, 1998) suggesting that Endo180 and the macrophage mannose receptor have complementary roles in glycoconjugate clearance. An unusual aspect of the macrophage mannose receptor is its ability to mediate both clathrin-coated pit endocytosis of glycoconjugates and phagocytosis of microorganisms (Ezekowitz et al., 1990; Kruskal et al., 1992). Despite these internalization events utilizing distinct cellular machinery (Aderem and Underhill, 1999; Marsh and McMahon, 1999) both are dependent on the macrophage mannose receptor cytoplasmic domain and are similarly impaired by mutation of the conserved tyrosine residue (Kruskal et al., 1992). As demonstrated here, Endo180 is expressed on macrophages and therefore it also could potentially function in host defence in the recognition and clearance of non-opsonized microorganisms. If this is the case, then given the differences in ligand specificity between the two receptors it would be expected that they would mediate the uptake of different pathogens. Finally, if Endo180 can mediate phagocytosis it is possible that it may additionally be an important receptor in fibroblasts and endothelial cells which are known to function as non-professional phagocytes in vivo (Rabinovitch, 1995) and utilise lectin receptors to mediate the internalization of apoptotic cells (Hall et al., 1994; Oka et al., 1998).

We thank Dr Peter van der Geer (UCSD, California) for the Endo180 glycosylation analysis, Dr Kurt Drickamer (Oxford) for his help and advice with the sugar binding assays, Dr Ian Dransfield (Edinburgh) for his analysis of the human monocyte-derived macrophages, Dr Jim Boulter (UCLA, California) for generously providing the human stromal λZAP library, Dr Peter Clark (Imperial School of Medicine) for his help and advice with the immunohistochemistry and James Legg (Imperial College) for invaluable assistance with the image analysis. Dr Tony Hunter and Dr Ian Trowbridge (Salk Institute, California), Professor Colin Hopkins (University College London) and Dr Stephen Neame (Eisai London Research) provided essential support and contributions to the early stages of this project. EST clones were provided by the UK HGMP Resource Centre. This research was supported by the Medical Research Council, The Wellcome Trust and the Cancer Research Campaign.