Collagen XVIII, a Basement Membrane Heparan Sulfate Proteoglycan, Interacts with L-selectin and Monocyte Chemoattractant Protein-1*

Leukocyte infiltration during inflammation is mediated by the sequential actions of adhesion molecules and chemokines. By using a rat ureteral obstruction model, we showed previously that L-selectin plays an important role in leukocyte infiltration into the kidney. Here we report the purification, identification, and characterization of an L-selectin-binding heparan sulfate proteoglycan (HSPG) expressed in the rat kidney. Partial amino acid sequencing and Western blotting analyses showed that the L-selectin-binding HSPG is collagen XVIII, a basement membrane HSPG. The binding of L-selectin to isolated collagen XVIII was specifically inhibited by an anti-L-selectin monoclonal antibody, EDTA, treatment of the collagen

L-selectin was initially identified as a lymphocyte homing receptor that plays a crucial role in lymphocyte migration into peripheral lymph nodes (1). Subsequently, this molecule was found to be expressed on other types of leukocytes, such as neutrophils and monocytes, and to be involved in leukocyte migration into inflamed tissues (2)(3)(4). In agreement with these reports, L-selectin-deficient mice show severely impaired lymphocyte homing and leukocyte infiltration to sites of inflammation (5,6).
L-selectin-binding molecules are found not only in the high endothelial venules of peripheral lymph nodes but also in nonvascular tissues, such as the white matter and choroid plexus of the central nervous system and the renal tubules (7). We have shown previously that leukocyte infiltration into the kidney interstitium induced by ureteral obstruction was significantly reduced by the administration of an anti-rat L-selectin blocking mAb 1 or a sulfated glycolipid, sulfatide, which binds to L-selectin in vitro (8,9). These observations prompted us to examine L-selectin-reactive molecules that are expressed in the kidney. We first identified sulfatide, a sulfated glycolipid expressed in the renal distal tubules, as an L-selectin-binding molecule in the kidney (9). Subsequent analyses revealed that, in addition to sulfatide, L-selectin-binding chondroitin sulfate proteoglycans and heparan sulfate proteoglycans (HSPGs) are both expressed in the kidney (10). Versican was identified as one of the L-selectin-binding chondroitin sulfate proteoglycans (11,12), whereas L-selectin-binding HSPGs with core protein sizes of 160 and 180 kDa (13) have remained undefined.
In the present study, we isolated these L-selectin-binding HSPGs from the rat kidney and identified them as collagen XVIII, a basement membrane HSPG. We also found that collagen XVIII interacts with monocyte chemoattractant protein-1 (MCP-1) and induces the integrin activation of monocytes, raising the possibility that collagen XVIII plays a role in renal inflammation by bridging selectin-and integrin-mediated cell adhesion.

EXPERIMENTAL PROCEDURES
Reagents-A recombinant soluble form of rat L-selectin (LEC-IgG) has been described previously (7). Human L-selectin-Ig and a recombinant soluble form of human vascular cell adhesion molecule-1 (VCAM-1) were purchased from R & D Systems Inc. (Minneapolis, MN). * This work was supported in part by grants-in-aid for the Center of Excellence Research and for Scientific Research on Priority Areas, Sugar Remodeling and Cellular Communications from the Ministry of Education, Science and Culture, Japan, and Taisho Pharmaceutical Co. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Preparation of the 1 M NaCl Fraction from the Renal Tubular Lysates-Renal tubules were collected from rat kidneys (male Wistar rats, 9-weeks old) according to the method of Krisko et al. (15). The fraction containing L-selectin-binding molecules (1 M NaCl fraction) was prepared as described previously (13).
Amino Acid Sequencing Analysis of the 160-and 180-kDa Molecules-The 1 M NaCl fraction was dialyzed against deionized water and lyophilized. The sample was then solubilized in buffer A (50 mM Tris-HCl, 7 M urea, 0.1% octyl glucoside, 10 mM EDTA, 10 mM N-ethylmaleimide, 1 mM PMSF, pH 7.4) and dialyzed against buffer B (20 mM Tris-HCl, 0.15 M NaCl, pH 7.4). After dialysis, the buffer composition was adjusted to buffer B containing 0.046% Tween 20, 3 mM CaCl 2 , 0.93 mM PMSF, 4.6 g/ml pepstatin, 4.0 g/ml leupeptin. To 800 l of this solution, 63.45 l of CNBr-activated Sepharose 4B beads (Amersham Biosciences) coupled with 5 milliunits of heparitinase (0.0788 milliunits/l beads) was added, and the mixture was incubated at 37°C for 15 h with agitation. After brief centrifugation to remove the beads, the supernatant was concentrated, subjected to SDS-PAGE, and blotted onto a Hybond-C nitrocellulose membrane (Amersham Biosciences). Bands were visualized by staining with Ponceau S, and the 160-and 180-kDa bands were excised separately. In the next step, each membrane strip was blocked with 0.5% polyvinylpyrrolidone K30 (Wako Pure Chemicals Co., Osaka, Japan), 0.1 M acetic acid, washed with buffer C (50 mM Tris-HCl, pH 8.0), and then incubated with 0.5 g of trypsin (sequencing grade from bovine pancreas; Roche Molecular Biochemicals) in buffer C at 37°C for 15 h. After the reaction, peptides were fractionated on a reversed-phase C-18 column (3.9 ϫ 150 mm; Millipore Co., Bedford, MA) at 1 ml/min under a 40-min gradient from 0.1% (v/v) trifluoroacetic acid to 0.1% (v/v) trifluoroacetic acid, 12% (v/v) acetonitrile, 28% (v/v) 2-propanol. The peak fractions detected by absorbance at 220 nm were collected and dried. The fractions (peaks 1 and 2 of the 160-kDa protein and peak 2 of the 180-kDa protein) were subjected to gas-phase microsequencing using a model 492 protein sequencer (PerkinElmer Life Sciences).
Generation of Anti-rat Collagen XVIII Polyclonal Antibody RES.16 -A C-terminal fragment of rat collagen XVIII that corresponds to the mouse endostatin sequence (16) was amplified by PCR, inserted into the vector pQE-31 (Qiagen, Santa Clarita, CA), and expressed and purified as described recently (17). A polyclonal antibody RES.16 was raised against the recombinant rat endostatin using conventional methods. Briefly, rat endostatin in complete Freund's adjuvant (Sigma) was injected into a rabbit subcutaneously, followed by booster injections in incomplete Freund's adjuvant at 2-week intervals. The polyclonal antiserum was affinity-purified on a CNBr-activated Sepharose 4B (Amersham Biosciences) column coupled with the recombinant rat endostatin as described elsewhere (18). The specificity of the affinity-purified RES.16 antibody was verified by Western blotting against the recombinant rat and human endostatins and the hepatoblastoma-derived full-length human collagen XVIII.
Immunoprecipitation and Western Blotting Analyses-Immunoprecipitation and Western blotting were performed as described previously (13).
Purification of Collagen XVIII from the 1 M NaCl Fraction-Ten milliliters of the 1 M NaCl fraction was mixed with 40 ml of buffer D (50 mM Tris-HCl, 8 M urea, 1 mM PMSF, pH 8.0) and 2 ml of Ni-NTAagarose beads (Qiagen) at 4°C for 12 h. After incubation, the beads were packed in a column (Poly-Prep chromatography column; Bio-Rad) and washed with 10 ml of buffer D and then with 10 ml of buffer D containing 0.3 M NaCl. Collagen XVIII was eluted with 250 mM imidazole in buffer D, dialyzed against PBS, and stored at Ϫ80°C until use. A portion of the purified collagen XVIII was biotinylated as described previously (12).
Chemokine-induced cell adhesion to VCAM-1 was determined as follows. Purified collagen XVIII (1 g/ml) and/or VCAM-1 (0.5 g/ml) were immobilized in the wells of 4-mm-diameter 96-well EIA/RIA plates (Corning Glass) overnight at 4°C. After washing, the plates were blocked with 3% bovine serum albumin in PBS. Various concentrations of MCP-1 were added to the wells, which were incubated for 2 h, and then washed with PBS containing 1 mM CaCl 2 and 1 mM MgCl 2 . THP-1 cells (provided by Dr. Katsuhiko Ishihara, Osaka University Graduate school of Medicine) were labeled with 5 M 2Ј,7Ј-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (Molecular Probes Inc., Eugene, OR), and resuspended in RPMI 1640 medium containing 2% fetal calf serum (medium A). The cells were then added to the wells (5 ϫ 10 4 cells/well) in triplicate. After a 10-min incubation at 37°C, the wells were filled with medium A and sealed with parafilm. The plates were inverted for 20 min at 37°C to allow non-adherent cells to detach from the bottom surface of the wells. The remaining bound cells were lysed, and the fluorescence intensity was measured as described previously (12).

FIG. 1. Identification of the L-selectin-binding HSPGs in the rat kidney as collagen XVIII.
A, comparison of partial amino acid sequences of the L-selectin-binding HSPGs (peaks 160-1, 160-2, and 180-2) with that of human collagen XVIII (hCol XVIII). Identical residues are marked with asterisks. B, the 1 M NaCl eluate prepared from rat tubular lysates by DEAE-Sepharose anion exchange column chromatography (13) was incubated with human IgG 1 or LEC-IgG in the presence or absence of EDTA (20 mM), HRL2 (50 g/ml), or HRL3 (50 g/ml) and then with protein A-Sepharose beads. The precipitates were treated with heparitinase (250 milliunits/ml) at 37°C for 5 h. The samples were then subjected to SDS-PAGE and blotted onto a polyvinylidene fluoride membrane (Millipore Co.). The membrane was probed with 0.5 g/ml normal rabbit IgG or RES. 16.
Immunofluorescence Staining-Immunofluorescence staining of frozen sections of the rat kidney (male Wistar rat, 8-week-old) was performed as described previously (11). (13) with core protein sizes of 160 and 180 kDa were purified by preparative SDS-PAGE, and their partial amino acid sequences were analyzed. The sequences of two fragments from the 160-kDa protein and one fragment from the 180-kDa protein were determined (Fig. 1A). All the sequences obtained showed significant homology to that of human collagen XVIII (21)(22)(23)(24), suggesting that both the 160-and 180-kDa proteins are rat counterparts of human collagen XVIII.

Identification of L-selectin-binding HSPGs in the Rat Kidney as Collagen XVIII-L-selectin binding HSPGs
We next performed Western blotting analysis using a polyclonal antibody, RES.16, which had been raised against rat collagen XVIII. As shown in Fig. 1B, 160-and 180-kDa proteins that were precipitated with LEC-IgG were specifically reactive with RES. 16. The binding of these proteins to LEC-IgG was apparently mediated by the C-type lectin domain of L-selectin, given that the binding was blocked by EDTA and the anti-rat L-selectin-blocking mAb HRL3 but not by the non-blocking anti-rat L-selectin mAb HRL2. Collectively, these results show that the L-selectin-binding HSPGs with core protein sizes of 160 and 180 kDa were both rat collagen XVIII.
Purification of Collagen XVIII from the Rat Kidney-A GenBank TM data base search revealed that a partial sequence of rat collagen XVIII (GenBank TM accession number AF 189709) contains an unusual His 11 segment. Because His repeating sequences have been widely used as an affinity tag for recombinant proteins to facilitate their binding to matrices containing immobilized Ni 2ϩ ions, we next tried to purify rat collagen XVIII using nitrilotriacetic acid (NTA)-agarose coupled with Ni 2ϩ ions (Ni-NTA-agarose). The 1 M NaCl fraction from DEAE-Sepharose chromatography (13) containing collagen XVIII was applied to the Ni-NTA-agarose column, and the bound material was then eluted with 250 mM imidazole. As shown in Fig. 2, whereas the 1 M NaCl fraction contained many proteins, the imidazole eluate contained almost exclusively proteins of 160 and 180 kDa. These bands were reactive with the RES.16 antibody in Western blotting analysis. 2 These results indicate that collagen XVIII was purified to an apparent homogeneity by Ni-NTA-agarose column chromatography.
Characterization of the Interaction between L-selectin and Collagen XVIII-By having obtained purified collagen XVIII, we next examined if collagen XVIII could directly interact with L-selectin. As shown in Fig. 3A, biotinylated collagen XVIII bound to immobilized LEC-IgG but not to control human IgG, and the binding was inhibited by HRL3 or EDTA but not by HRL2. Treatment of the biotinylated collagen XVIII with heparitinase but not chondroitinase ABC or keratanase inhibited its binding to LEC-IgG (Fig. 3B) 3. Binding of collagen XVIII to L-selectin is inhibited by HRL3, EDTA, and heparitinase treatment. A, biotinylated collagen XVIII was added to wells coated with human IgG or LEC-IgG (5 g/ml) that had been incubated with 30 g/ml HRL2 or HRL3 or 20 mM EDTA. B, biotinylated collagen XVIII that had been treated with 0.1 unit/ml chondroitinase ABC (CHase ABC), heparitinase, or keratanase, or left untreated was added to wells coated with human IgG or LEC-IgG (5 g/ml). The extent of binding was determined by ELISA as described previously (12). Each bar represents the mean Ϯ S.D. of triplicate determinations.
for the interaction of collagen XVIII with L-selectin, we next performed an inhibition assay using various chemically modified glycosaminoglycans. As shown in Fig. 4A, the binding of biotinylated collagen XVIII to LEC-IgG was inhibited in a dose-dependent manner by intact heparin containing both Nand O-sulfation. The blocking ability of intact heparin was comparable with that of chondroitin sulfate (CS) E, which strongly interacts with L-selectin (12,19). The binding was also inhibited by CS B, which also interacts with L-selectin (12). On the other hand, the binding was not inhibited by chemically modified heparins, such as CDSNAc-heparin, CDSNS-heparin, and NDSNAc-heparin, indicating that both N-and O-sulfation of HS/heparin chains are required for the interaction of collagen XVIII with L-selectin. Consistently, disaccharide composition analysis showed that HS chains of collagen XVIII contained abundant N-and O-sulfated disaccharide units (Fig.  4B). Peak areas of ⌬DiHS-0S, ⌬DiHS-6S, ⌬DiHS-NS, ⌬DiHSdi(6,N)S, ⌬DiHS-di(U,N)S, and ⌬DiHS-tri(U,6,N)S in Fig. 4B  were 8.1, 28.2, 19.9, 15.7, 20.4, and 7.7%, respectively.
The L-selectin-Collagen XVIII Interaction Mediates Cell Adhesion-To determine whether the L-selectin-collagen XVIII interaction could mediate cell adhesion, we next performed a cell-binding assay using L-selectin transfectants. As shown in Fig. 6, rat L-selectin transfectants (RM-1 cells) bound to collagen XVIII-coated wells at a higher level than did untransfected EL-4 cells, and the binding was inhibited by HRL3 or EDTA but not by HRL2. These results indicate that the L-selectincollagen XVIII interaction can mediate cell adhesion.
Localization of Collagen XVIII in the Rat Kidney-We next The solid bars on the right show the binding of biotinylated collagen XVIII to the wells coated with human IgG 1 or LEC-IgG (5 g/ml) in the absence of GAGs. B, disaccharide composition analysis of HS chains of collagen XVIII. Collagen XVIII was treated with a mixture of heparitinase, heparitinase I and heparitinase II, and the resultant disaccharides were derivatized with 2-AB and analyzed by high performance liquid chromatography on an amine-bound silica column as described previously (19). The elution positions of the 2-AB-derivatized standard disaccharides, ⌬DiHS-0S (peak 1), ⌬DiHS-6S (peak 2), ⌬DiHS-NS (peak 3), ⌬DiHS-di(6,N)S (peak 4), ⌬DiHS-di-(U,N)S (peak 5), and ⌬DiHS-tri(U,6,N)S (peak 6) are shown. examined the tissue distribution of L-selectin ligands and collagen XVIII in the kidney (Fig. 7). Consistent with our previous observations (7, 9 -11), LEC-IgG bound abundantly to tubules but only slightly, if at all, to the vascular bundles in the outer medulla of the rat kidney (Fig. 7A), whereas the anti-collagen XVIII antibody RES.16 reacted with the vascular bundles and the basement membranes of the tubules of the outer medulla (Fig. 7B). Thus, L-selectin ligands and collagen XVIII appeared to colocalize in the basement membranes of the renal tubules of the outer medulla. In contrast, collagen XVIII was also detected in the glomeruli and tubules in the renal cortex, where no LEC-IgG binding was observed (Fig. 7, C and D), suggesting that the collagen XVIII expressed in these regions is not reactive with L-selectin.
Collagen XVIII Interacts with Chemokines to Induce Integrin Activation in Leukocytes-HSPGs are proposed to function as "reservoirs" for chemokines and to present them to leukocytes efficiently (29). Among various chemokines, MCP-1 has been shown to play a critical role in renal inflammation (30 -32). We thus examined the possibility that collagen XVIII interacts not only with L-selectin but also with MCP-1 and presents it to leukocytes. As shown in Fig. 8A, MCP-1 bound to immobilized collagen XVIII in a dose-dependent manner. The binding was inhibited by heparitinase but not by chondroitinase ABC treatment, indicating that MCP-1 interacts with HS chains of collagen XVIII. Binding of MCP-1 to immobilized collagen XVIII was inhibited by LEC-IgG in a dose-dependent manner (Fig. 8B). Similarly, binding of LEC-IgG to immobilized collagen XVIII was inhibited by MCP-1 (Fig. 8C), suggesting that MCP-1 and L-selectin interact with the same or overlapping carbohydrate structure in the HS chains of collagen XVIII.
We then investigated the possible functional consequences of the collagen XVIII-MCP-1 interaction. To this end, MCP-1 was applied to wells that had been coated with collagen XVIII alone or with a mixture of collagen XVIII and VCAM-1. We then applied to each well a monocytic cell line, THP-1, that expresses a receptor for MCP-1, CCR2B, and ␣ 4 ␤ 1 integrin. The THP-1 cells did not bind collagen XVIII in an L-selectin-dependent manner under the conditions used in this experiment, and thus were useful to determine the effect of chemokine presentation by collagen XVIII on integrin-mediated cell adhesion. As shown in Fig. 9A, THP-1 cells bound to the wells that had been coated with a mixture of collagen XVIII and VCAM-1 but not to the wells coated with collagen XVIII or VCAM-1 alone in a manner dependent on the dose of MCP-1, indicating  6. Binding of rat L-selectin transfectants to collagen XVIII. 2Ј,7Ј-Bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein-labeled rat L-selectin transfectants (RM-1) or parental cells (EL-4) (1 ϫ 10 5 cells/well) were allowed to bind to wells coated with (black bars) or without (striped bars) collagen XVIII (1 g/ml) in the presence or absence of 50 g/ml anti-L-selectin mAbs HRL2 or HRL3 or 25 mM EDTA as described previously (12). Each bar represents the mean Ϯ S.D. of triplicate determinations. that MCP-1 that had been immobilized on collagen XVIII activated the binding of THP-1 cells to VCAM-1. This up-regulated adhesion was apparently mediated by ␣ 4 ␤ 1 integrin, since the adhesion was blocked by either the anti-␣ 4 integrin mAb HP2/1 or the anti-␤ 1 integrin mAb 4B4 (Fig. 9B). DISCUSSION In this study, we have purified L-selectin-binding HSPGs in the kidney and identified them as collagen XVIII. Treatment with heparitinase, but not with any other GAG-degrading enzymes, abolished the L-selectin binding (Fig. 3B), indicating that HS chains of collagen XVIII bind to L-selectin. Unlike sulfatide, collagen XVIII bound to L-selectin in a divalent cationdependent manner. This binding was inhibited by the mAb HRL3, which inhibits the binding of L-selectin to lymph node high endothelial venules (14). These results indicate that HS chains on collagen XVIII are recognized by the lectin domain of L-selectin, which is known to interact with sLe X -like sugar chains on L-selectin ligands such as CD34 and glycosylation- After washing, MCP-1 binding was determined as described previously (41), except that streptavidin-conjugated alkaline phosphatase (1:500, Promega, Madison, WI) and Blue Phos TM (Kirkegaard Perry Laboratories, Gaithersburg, MD) were used to detect the binding. B, binding of MCP-1 (0.2 g/ml) to wells coated with (black bars) or lacking (striped bars) collagen XVIII (0.2 g/ml) in the presence or absence of LEC-IgG (3, 30, and 300 g/ml). The binding was detected as described in A. C, binding of LEC-IgG (3 g/ml) to wells coated with (black bars) or lacking (striped bars) collagen XVIII (0.2 g/ml) in the presence or absence of MCP-1 (0.3, 3, and 30 g/ml). The binding of LEC-IgG to collagen XVIII was detected by ELISA (method 2) as described previously (12).
Collagen XVIII is a triple helical endothelial/epithelial basement membrane protein characterized by multiple interrupted triple helical regions and a unique C-terminal endostatin domain (21)(22)(23)(24). Collagen XVIII became the focus of medical interest when an 18-kDa anti-angiogenic peptide with tumorsuppressing activity, named endostatin, turned out to be the C-terminal part of collagen XVIII (16). More recently, it was reported that collagen XVIII is modified with HS chains in the human kidney (18) and in various organs in the chick (35). Our present finding that the binding of rat collagen XVIII to Lselectin and MCP-1 is mediated by HS chains of collagen XVIII is consistent with these previous reports.
Amino acid sequencing and Western blotting analyses showed that the 160-and 180-kDa proteins are both rat collagen XVIII (Fig. 1). To date, three variants of mouse collagen XVIII (36,37) and two variants of human collagen XVIII (38) have been reported. In the mouse and human, collagen XVIII transcripts corresponding to the short variant are predominantly expressed in the kidney (36,38). The human collagen XVIII short variant is 180 kDa after heparin lyase II and III treatment (18). It is unclear at present whether the 160-and 180-kDa proteins detected in the rat kidney represent forms produced by alternative splicing or proteolytic cleavage. However, the presence of two forms does not appear to be due to the cleaving off of the C-terminal 18-kDa endostatin domain from the core protein, because both proteins were reactive with the RES.16 antibody, which recognizes the rat endostatin domain.
It appears that proteoglycans can selectively interact with particular adhesion molecules depending on differences in their GAG modifications. As reported previously, versican modified with CS B and CS C (12) as well as CS E (19) binds L-and P-selectin and CD44, whereas the rat chondrosarcoma proteoglycan aggrecan, which is modified exclusively with CS A (28), selectively binds CD44 (Fig. 5). Collagen XVIII modified with highly sulfated HS (Fig. 4B) bound both L-and P-selectin but not CD44 (Fig. 5). It should be noted, however, that only a subset of collagen XVIII, such as that expressed in the basement membrane of distal tubules, bound L-selectin, whereas collagen XVIII in other locations did not (Fig. 7). This result implies that only a subset of collagen XVIII species present in the rat kidney is modified with L-selectinreactive HS chains. Although the precise carbohydrate structure required for L-selectin binding has yet to be determined, our present results suggest that both N-and O-sulfation of heparin/HS chains are required for the interaction with Lselectin (Fig. 4A).
Although only a weak expression of MCP-1 in tubular epithelial cells has been reported in the normal kidney (39), a strong expression of MCP-1 has been detected in various renal diseases. For example, strong tubular epithelial expression of MCP-1 was found in kidney biopsies of patients with membranous nephropathy, IgA nephropathy, and glomerulosclerosis (39). MCP-1 expression was also found in the glomeruli from biopsies of patients with proliferative glomerulonephritis, Wegener's disease, and lupus nephritis (40). Our finding that collagen XVIII interacts with MCP-1 and presents it to a monocytic cell line to induce ␣ 4 ␤ 1 integrin activation in vitro (Figs. 8 and 9) indicates that collagen XVIII may bind locally produced MCP-1 and present it to infiltrating monocytes in the diseased kidney.
Under physiological conditions, L-selectin-reactive molecules are selectively localized to the epithelial cells of distal tubules, where leukocytes do not normally traffic. The paucity of leukocytes in the kidney parenchyma under physiological conditions can probably be explained by the absence of leukocyte accessibility to collagen XVIII and other L-selectin-binding molecules in the kidney. However, in the diseased kidney, where massive mononuclear cell infiltration can be observed, it is possible that collagen XVIII interacts with L-selectin on the infiltrating cells and/or presents chemokines to them, and may accelerate the progression of inflammatory processes. Experimental verification is now required to assess the actual involvement of collagen XVIII in leukocyte migration under pathological conditions.