Versican Interacts with Chemokines and Modulates Cellular Responses*

We previously reported that versican, a large chondroitin sulfate proteoglycan, isolated from a renal adenocarcinoma cell line, ACHN, binds L-selectin. Here we report that versican also binds certain chemokines and regulates chemokine function. This binding was strongly inhibited by the chondroitinase digestion of versican or by the addition of soluble chondroitin sulfate (CS) B, CS E, or heparan sulfate. Furthermore, these glycosaminoglycans (GAGs) could bind directly to the chemokines that bind versican. Thus, versican appears to interact with chemokines via its GAGs. We next examined if versican or GAGs affect secondary lymphoid tissue chemokine (SLC)-induced integrin activation and Ca2+ mobilization in lymphoid cells expressing a receptor for SLC, CC chemokine receptor 7. Interestingly, whereas heparan sulfate supported both α4β7 integrin-dependent binding to mucosal addressin cell adhesion molecule-1 (MAdCAM-1)-IgG and Ca2+ mobilization induced by SLC, versican or CS B inhibited these cellular responses, and the extent of inhibition was dependent on the dose of versican or CS B added. These findings suggest that different proteoglycans have different functions in the regulation of chemokine activities and that versican may negatively regulate the function of SLC via its GAG chains.

Proteoglycans are proteins that carry glycosaminoglycans (GAGs). 1 The common GAGs are heparin, heparan sulfate (HS), chondroitin sulfate (CS), dermatan sulfate, and hyaluronic acid. These GAGs have a large number of sulfate or carboxyl groups and, hence, have a strong negative charge, which makes it possible for proteoglycans to bind through their GAG chains various positively charged molecules such as certain growth factors, cytokines, and chemokines (1,2). This interaction in the extracellular matrix or on the cell surface has been suggested to play important roles in vivo in the formation of immobilized gradients of these humoral factors, in their protection from proteolytic degradation, and in their presentation to specific cell-surface receptors (3)(4)(5). There is an increasing body of evidence indicating that HS proteoglycans are particularly important in promoting the oligomerization of chemokines and in facilitating their presentation to specific receptors (6 -8). Sporadic reports indicate that CS proteoglycans may also have this function (9,10). For instance, CS proteoglycans on human neutrophils specifically bind platelet factor 4 and are involved in the activation of neutrophils (9). An artificial proteoglycan modified with CS binds RANTES via its GAGs, which also induces the activation of T cells expressing this molecule on their surface (10). Thus, not only HS but also CS proteoglycans appear to bind chemokines and to play important roles in the regulation of chemokine functions.
Versican is a large CS proteoglycan (11,12) expressed in various tissues (13). We have recently reported that versican derived from a renal adenocarcinoma cell line, ACHN, interacts through its CS chains with leukocyte adhesion molecules such as L-selectin and CD44 (14,15), both of which have been implicated in leukocyte trafficking (16,17). Since versican has abundant CS side chains and is secreted into the extracellular matrix, we speculated that it might also interact with certain chemokines to help form haptotactic gradients in the tissues where chemokines and versican are colocalized. In the present study we examined the ability of versican to bind a large panel of chemokines and the biological consequences of such binding. Our results indicate that versican can bind specific chemokines through its CS chains and that the binding tends to downregulate the chemokine function. This raises the possibility that different proteoglycans have different functions in the regulation of chemokine activities and that certain proteoglycans, such as versican, may negatively regulate chemokine functions.

EXPERIMENTAL PROCEDURES
Reagents-Versican was isolated from the culture supernatant of a human renal adenocarcinoma cell line, ACHN, as described previously (14). Epithelial-derived neutrophil attractant 78 (ENA-78), growth-related gene (Gro-␣), interleukin-8, ␥ interferon-inducible protein 10 (IP-10), platelet factor 4, lymphotactin, MCP-1, -2, -3, MIP-1␣, MIP-1␤, and RANTES were purchased from Pepro Tech, Inc. (Rocky Hill, NJ). Recombinant human SDF-1␤ (18) was purchased from Genetics Institute. Inc. (Cambrige, MA). Recombinant human liver and activation-regulated chemokine (LARC), pulmonary and activation-regulated chemokine (PARC), and thymus and activation-regulated chemokine (TARC) were produced by using a baculovirus expression system and purified as described previously (19 -21). EBI1-ligand chemokine was * This work was supported in part by a grant-in-aid for Center of Excellence Research and Scientific Research on Priority Areas, Sugar Remodeling and Cellular Communications from the Ministry of Education, Science, and Culture, Japan, a grant from the Science and Technology Agency, Japan, and a grant from Ono 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.
Dot Blot Analysis-Various concentrations of chemokines were spotted onto a Hybond C nitrocellulose membrane (Amersham Pharmacia Biotech) and air-dried for 5 h. The membrane was blocked overnight with 3% BSA, 0.1% NaN 3 in PBS at 4°C. After blocking, the membrane was washed with PBS containing 1 mM CaCl 2 , 1 mM MgCl 2 (PBSϩ) containing 0.05% Tween 20, and 0.1% BSA. Subsequently, the membrane was incubated with biotinylated versican (3 g/ml) in the presence or absence of various GAGs (10 g/ml) for 1 h at room temperature. After washing, ABC reagent (Vector Laboratories Inc., Burlingame, CA) was applied, and bound versican was detected using ECL (Amersham Pharmacia Biotech).
ELISA-Versican (20 g/ml) or GAGs (50 g/ml) were coated on 4-mm-diameter 96-well EIA/RIA plates (Costar, Corning Inc.) overnight at 4°C. The wells were washed with PBSϩ, and blocked with FCS for 2 h at room temperature. SLC was added to the versican-or GAGcoated wells and incubated for 2 h. After washing, the wells were incubated with biotinylated goat anti-human SLC (1 g/ml, DAKO) in PBSϩ containing 0.05% Tween 20 and 1% FCS for 1 h at room temperature. After washing, the wells were incubated with streptavidin-conjugated alkaline phosphatase (1:500, Promega, Madison, WI) for 1 h. To quantify the reaction, Blue Phos (Kirkegaard Perry Laboratories, Gaithersburg, MD) was added, and the optical density was read at 620 nm in a microtiter plate reader (InterMed, Tokyo, Japan). To determine the binding of MCP-1 to versican, versican (20 g/ml) was coated on 4-mm-diameter 96-well EIA/RIA plates (Costar, Corning Inc.) overnight at 4°C. The wells were washed with PBSϩ and blocked with 3% BSA in PBSϩ for 2 h at room temperature. MCP-1 (0 -3 g/ml) was added to the versican-coated wells and incubated for 2 h. After washing, the wells were incubated with biotinylated goat anti-human MCP-1 (1 g/ml, Santa Cruz, Biotechnology Inc. Santa Cruz, CA) and then with peroxidase-conjugated streptavidin in PBSϩ containing 0.1% BSA and 0.05% Tween 20 for 1 h at room temperature. To quantify the reaction, o-phenylenediamine (0.4 mg/ml) was used as the substrate, and the optical density at 490 nm was read in a microtiter plate reader. In some experiments, immobilized versican was treated with chondroitinase ABC (0.6 units/ml, pH 8.0), chondroitinase B (0.06 units/ml, pH 8.0), or keratanase (0.6 units/ml, pH 8.0) overnight at 37°C. After blocking, SLC (1 g/ml) was added to the wells and incubated for 1 h at room temperature. After washing, bound SLC was detected with streptavidin-conjugated alkaline phosphatase and Blue Phos as described above. To examine the efficacy of the enzyme digestion, either mAb 2B1 (3 g/ml) or CS56 (1:500 dilutions) was added to the wells after the enzyme treatment. Peroxidase-conjugated anti-mouse IgG or antimouse IgG ϩ IgM was then added to detect the binding of the primary antibodies. The binding was quantified by the o-phenylenediamine reaction as described above.
Cell Adhesion Assay-Cell adhesion to MAdCAM-1 induced by SLC was determined as follows. SLC (0 -3 g/ml) and MAdCAM-1-IgG (100 ng/well) or human IgG (100 ng/well) were coimmobilized on 4-mmdiameter 96-well EIA/RIA plates (Costar, Corning Inc.) overnight at 4°C. After washing twice with PBS, plates were blocked with FCS. TK-1 cells labeled with 2Ј,7Ј-bis-(2-carboxyethyl)-5-(and 6)carboxyfluorescein acetoxymethyl ester (BCECF-AM) (2 M) were then added to the wells (1.2 ϫ 10 5 cells/well) in triplicate. After a 20-min incubation at 37°C, the wells were filled with RPMI-FCS and sealed with Parafilm. The plates were inverted for 30 min at 37°C to allow nonadherent cells to detach from the bottom surface of the wells, after which the medium containing unbound cells was removed by suction. The remaining bound cells were lysed with 0.1% Nonidet P-40, and the fluorescence intensity was read in a Fluoroskan II microplate fluorometer (Labsystem, Tokyo, Japan). The results are expressed as the percentage of cells bound and represent triplicate determinations.
Cell adhesion to MAdCAM-1 induced by SLC that had been immobilized on versican or GAGs was determined as follows. Various combinations of versican (300 ng/well), CS A, CS B, or HS (50 g/ml) and MAdCAM-1-IgG (100 ng/well) or human IgG (100 ng/well) were coimmobilized in the wells overnight at 4°C. After blocking with FCS, SLC (0 -3 g/ml) was added to the wells for 2 h at room temperature. Unbound SLC was removed by washing with PBS. Cell adhesion to each well was determined as described above.
Ca 2ϩ Mobilization Assay-The Ca 2ϩ mobilization assay was carried out according to the method of Nagira et al. (28). In brief, L1.2, L1.2/ CCR7, and Jurkat cells were incubated at 37°C for 1 h with 2 M Fura-2/AM (Sigma) at 1.0 ϫ 10 6 cells/ml in RPMI without FCS in the dark. Cells were washed twice and resuspended at 1.0 ϫ 10 6 cells/ml in PBSϩ containing 1% FCS. A 1.5-ml sample of cell suspension was placed in a cuvette and set into a spectrofluorophotometer (RF1500, Shimadzu, Japan) with constant stirring. SLC (1 g/ml), SDF-1␤ (1 g/ml), or leukotriene D4 (Sigma, 0.2 g/ml) was preincubated with versican (0.1, 0.2, or 0.4 g/ml) or GAGs (50 g/ml) for 10 min at room temperature and added to the cell suspension. Emission fluorescence intensity at 510 nm was measured upon excitation at 340 nm (F340) and 380 nm (F380). Data were presented as the ratio of F340 divided by F380 (R340/380). After each measurement, R340/380 in the presence or absence of excess calcium was determined by the sequential addition of 100 l of 1% Triton X-100 with or without 100 l of 0.5 M EDTA.
Binding of Radiolabeled SLC to CCR7 Transfectant Cells-The recombinant SLC was radioiodinated using Bolton-Hunter reagent (Amersham Pharmacia Biotech) according to the manufacturer's instruction. L1.2 or L1.2/CCR7 cells (5 ϫ 10 5 cells) were incubated with 125 I-labeled SLC (10 ng/ml) alone or in the presence of unlabeled SLC (2 g/ml) or increasing concentrations of versican (0.001-10 g/ml), CS A, CS B, or HS (0.5, 5, or 50 g/ml) for 1 h. After incubation, cell suspensions were placed on a mixture of dibutyl phthalate/olive oil (4:1) and separated from unbound radiolabeled SLC by centrifugation (8,000 ϫ g, 4°C, 5 min). The binding of 125 I-labeled SLC was counted in a ␥ counter. All determinations were made in triplicate.

RESULTS
Versican Binds Various Chemokines-The binding of versican to 18 different chemokines was examined by dot blot analysis. Various chemokines were spotted onto a nitrocellulose membrane, biotinylated versican was added, and after washing, its binding was determined. Anti-versican core protein mAb 2B1 and anti-CS mAb CS56 were also spotted on the membrane to serve as positive controls for versican binding. As shown in Fig. 1A, biotinylated versican bound the CXC chemokines ␥ interferon-inducible protein 10 (IP-10), platelet factor 4, and SDF-1␤, the C chemokine lymphotactin, and the CC chemokines MCP-1, MCP-2, RANTES, liver and activationregulated chemokine (LARC), and SLC. In contrast, biotinylated versican failed to bind interleukin-8 or MIP-1␣, which are known to bind HS (29 -31).
We then performed a reverse assay, where the ability of immobilized versican to bind soluble chemokines was tested by ELISA. As shown in Fig. 1B, versican bound SLC and MCP-1 in a dose-dependent manner. Similarly, immobilized versican also bound SDF-1␤ and lymphotactin in a dose-dependent manner (data not shown).
Versican Binds Chemokines in a GAG-dependent Manner-To examine whether the chemokine binding is mediated by the GAG chains on versican, the effects of GAG-degrading enzymes on the binding of SLC to versican were examined. As shown in Fig. 2, top panel, both chondroitinase ABC and chondroitinase B inhibited the binding of versican to SLC, whereas neither of these enzymes affected the reactivity of versican with anti-versican core protein mAb 2B1 (Fig. 2, middle panel).
In addition, chondroitinase ABC as well as chondroitinase B almost completely abrogated the reactivity of versican with the anti-CS mAb CS56, whereas keratanase failed to alter the reactivity of versican with mAb CS56 (Fig. 2, bottom panel). These results suggest that the GAGs on versican play an important role in chemokine binding. However, the possibility remains that the core protein of versican is also involved in SLC binding, since treatment with chondroitinases did not completely inhibit the binding of SLC to versican.
To further investigate the involvement of GAGs in the chemokine binding, we examined whether exogenously added soluble GAGs could inhibit the binding of chemokines to versican. Interestingly, CS B, CS E, or HS almost completely abolished the binding of most of the chemokines examined, whereas CS C and CS D inhibited the binding of a restricted set of chemokines, and chondroitin and CS A were only slightly inhibitory (Fig. 3A). Given that CS B, CS E, and HS were extremely potent inhibitors of the chemokine binding to versican, we speculated that these GAGs might have particularly high chemokine binding abilities compared with other types of GAGs. To further examine this issue, various GAGs were immobilized on plastic wells, and the binding of SLC was determined by ELISA. As shown in Fig. 3B, SLC bound strongly to CS B, CS E, and HS in a dose-dependent manner but bound other GAGs only weakly or not at all, consistent with the results shown in Fig. 3A. SLC did not bind to a nonsulfated type of GAG, chondroitin. Collectively, these results indicate that the GAG chains of versican are instrumental in chemokine binding and that the sulfation of GAG chains may affect the binding.
Versican Down-regulates the ␣ 4 ␤ 7 Integrin-mediated Cell Adhesion Induced by SLC-We next investigated the possible functional consequences of the versican-chemokine interaction. First, we examined whether versican could influence the SLCinduced activation of integrins using ␣ 4 ␤ 7 integrin-and CCR7expressing TK-1 mouse T cells. SLC, which we found to bind versican strongly, can facilitate the adhesion of ␣ 4 ␤ 7 integrinpositive cells to their ligand, MAdCAM-1 (32). To examine the effect of versican on this chemokine action, SLC was coimmobilized in wells with MAdCAM-1-IgG alone or with MAdCAM-1-IgG and versican. To evaluate nonspecific binding, human IgG was used instead of MAdCAM-1-IgG. As shown in Fig. 4A, left panel, SLC enhanced the binding of TK-1 cells to MAd-CAM-1-IgG, showing a bell-shaped dose-response pattern in the absence of versican. This up-regulated adhesion was apparently mediated by the interaction between ␣ 4 ␤ 7 integrin and MAdCAM-1, since the effect was completely abrogated by anti-␤7 mAb and anti-MAdCAM-1 mAb (data not shown). By contrast, SLC immobilized on versican up-regulated the cell adhesion only marginally, if at all (Fig. 4A, right panel). The lack of strong up-regulation of TK-1 adhesion by SLC complexed with versican was not simply due to the reduced availability of SLC to TK-1, because the actual amount of SLC bound to the versican-coated wells would have been sufficient to up-regulate cell adhesion, at least when SLC was immobilized at a concentration of 1 g/ml or higher (see Fig. 4B and its legend).
We then examined whether the apparent inhibitory effect of versican was mediated by its GAG chains. For this purpose, we used GAGs instead of versican in the above-mentioned assay and compared their effects with that of versican. As shown in Fig. 4C, the SLC⅐CS B complex induced the adhesion of TK-1 cells to immobilized MAdCAM-1-IgG only minimally, similar to the SLC-versican complex, whereas the SLC⅐HS complex strongly up-regulated the cell adhesion (Fig. 4C), in agreement with previous reports that the chemokine-HS complexes effectively activate integrins on lymphocytes (7,8). Little or no chemokine-induced cell adhesion was observed in CS A-coated wells, probably because of the inability of CS A to bind SLC (Fig. 3B). In addition, inhibitory effects of versican and CS B appear specific for SLC-induced activation of integrins, since versican or CS B did not inhibit phorbol 12-myristate 13-acetate-induced TK-1 adhesion (data not shown). These results demonstrate that a certain type of CS, similar or identical to CS B, negatively regulates SLC-induced cell adhesion, suggest-ing that versican exerts its suppressive effect on chemokine function at least partly through its GAG chains. In a separate study, we showed that versican is indeed modified at least partly with CS B (15).

Versican or a Certain Type of GAG Can Down-regulate the Chemokine-induced Intracellular Ca 2ϩ
Response-We next evaluated another effect of versican on chemokine functions by examining the intracellular Ca 2ϩ response elicited by SLC and SDF-1␤ in L1.2/CCR7 (22) and Jurkat cells (33), respectively. As shown in Fig. 5A, leukotriene D4, SLC, or SDF-1␤ induced a rapid and transient intracellular Ca 2ϩ response. However, versican inhibited the chemokine-induced intracellular Ca 2ϩ response in a dose-dependent manner. Versican did not affect the leukotriene D4-induced intracellular Ca 2ϩ response (Fig.  5A). When GAGs were used instead of versican, CS B was again inhibitory on SLC-induced intracellular Ca 2ϩ response, whereas CS A and HS were not (Fig. 5B). These results further suggest that versican inhibits the ability of chemokines to elicit intracellular signals, and it does so, at least in part, via its GAG chains.
How Does Versican Inhibit the Effects of SLC?-To investigate the mechanism by which versican inhibits chemokine functions, we examined whether versican affects the binding of SLC to CCR7-expressing cells. As shown in Fig. 6A, the binding of 10 ng/ml 125 I-SLC to L1.2/CCR7 cells was inhibited by a 200-fold excess amount of unlabeled SLC, which reduced the binding to the background level. In contrast, SLC binding was not significantly inhibited by even 10 g/ml versican, even though the intracellular Ca 2ϩ response induced by 1 g/ml SLC in L1.2/CCR7 cells was inhibited by 0.4 g/ml versican (Fig. 5). Taking the molecular size of SLC (15 kDa) and versican (1, 600 kDa) into account, this means that the intracellular Ca 2ϩ response was inhibited at a molar ratio of 1:0.0037 (SLC:versican), whereas the binding of SLC to its receptor was uninhibited, even at a molar ratio of 1:9.37 (SLC:versican). Thus, the down-regulation of chemokine function by versican is unlikely to be due to the inhibition of SLC binding to its receptor. We next examined if GAGs affect the binding of SLC to L1.2/CCR7 cells. The binding of 10 ng/ml 125 I-SLC to L1.2/CCR7 cells was inhibited by 50 g/ml but not by lower doses of CS B or HS (5 and 0.5 g/ml), whereas, as shown in Fig. 5, the intracellular Ca 2ϩ response induced by 1 g/ml SLC was inhibited by 50 g/ml of CS B. Taking the molecular size of CS B (11-25 kDa) into consideration, this result implies that the SLC-induced intracellular Ca 2ϩ response was inhibited by CS B at a molar ratio of 1:30.0 -68.2 (SLC:CS B), whereas the binding of 125 I-SLC to L1.2/CCR7 cells was inhibited only at much higher molar ratios (1:3,000 -6,820).

DISCUSSION
It is generally accepted that proteoglycans interact with small molecular weight humoral mediators such as chemokines that are produced in the same tissue. These interactions protect the small molecules from degradation, help them form concentration gradients in situ, and help to present them to specific receptors on the cell surface (1)(2)(3)(4)(5). However, it is not known whether proteoglycans always regulate chemokine function positively, let alone the exact identity of the proteoglycans involved, and it is also unclear whether this interaction is a prerequisite for chemokine function. In the present study, we showed that a large CS proteoglycan, versican, which is derived from a human renal carcinoma cell line, binds a variety of chemokines through its CS side chains. This binding was inhibited by the addition of soluble CS B, CS E, or HS or by treatment with chondroitinases. We also showed that, in the absence of a core protein, certain GAG chains such as CS B, CS E, and HS could directly interact with a CC chemokine, SLC. Furthermore, we found that versican or certain types of CS can inhibit the function of SLC; i.e. SLC that was complexed with versican or CS B was significantly less efficient than SLC alone in inducing ␣ 4 ␤ 7 integrin activation as well as the intracellular Ca 2ϩ response, and the extent of inhibition by versican or CS B was dose-dependent. In contrast, SLC that was complexed with HS was as efficient as SLC alone in its chemokine activities. These findings suggest that the interaction between proteoglycans and chemokines results in different biological consequences, depending on the proteoglycans involved, and that certain proteoglycans or their GAGs alone may negatively regulate the function of chemokines in certain situations.
Although CS proteoglycans bear negatively charged GAG chains that can interact with positively charged residues of chemokines, a nonspecific electrostatic interaction may not be the sole factor determining the interaction between CS proteoglycans and chemokines. The results shown in Fig. 3 that nonsulfated chondroitin failed to interact with SLC indicates that sulfation of the GAG chains is important in the interaction. However, CS E but not CS D could interact with SLC even though both are composed of disulfated disaccharides, indicating that sulfation at a specific position is critical. In addition, CS B and CS E, but not CS A, interacted with SLC even though all of these GAGs have 4-sulfation on N-acetylgalactosamine residues, suggesting that 4-sulfation on N-acetylgalactosamine residues is not sufficient but that sulfation of another position(s) and/or the structure of uronic acid residues is critical for chemokine binding.
Based on the observations that binding of the CC chemokine SLC to versican was inhibited by chondroitinase B, which specifically cleaves the CS B chain, and that CS B itself bound SLC directly, it is likely that CS B or a GAG containing a CS B-type structure on versican is actually involved in the chemokine binding. This hypothesis is supported by our GAG composition analysis using FACE (fluorophore-assisted carbohydrate electrophoresis) and immunological analysis using anti-GAG mAbs, which indicated that versican is modified with at least CS B and CS C (15). However, we do not know currently whether versican is modified with CS E or a similar moiety, and further structural analysis will be required to understand the exact carbohydrate structure responsible for SLC binding. The mechanism by which versican or a specific GAG chain (versican/GAG) inhibits SLC function remains unclear, but several possibilities can be considered: (i) versican/GAG inhibits SLC binding to specific chemokine receptors; (ii) versican/ GAG competes with endogenously expressed cell-surface proteoglycans that may help present SLC to its receptors; (iii) versican/GAG transduces a negative signal by binding to molecules different from chemokine receptors; (iv) versican/GAG allows SLC to bind its receptor even after forming a complex with SLC but inhibits an appropriate signal to be transduced.
The first possibility, that versican inhibits the binding of SLC to CCR7, appears unlikely since versican inhibited the SLC-induced intracellular Ca 2ϩ responses (Fig. 5) at a dose that did not affect the binding of SLC to its receptor (Fig. 6A). The observations that the SLC-induced intracellular Ca 2ϩ response was inhibited by CS B at a molar ratio of 1:30 -68 (Fig.  5) and that the binding of 125 I-SLC to CCR7-expressing cells The open circles (E) indicate the control in which SLC was directly coated on the plastic wells. Note that the amount of SLC bound to versican-coated wells at a concentration of 1 g/ml was comparable with that of SLC bound to uncoated wells at 0.3 g/ml, which could effectively up-regulate cell adhesion (see A, left panel). This indicates that the lack of strong up-regulation of TK-1 adhesion in the presence of versican shown in A (right panel) is not due to the insufficient availability of SLC to TK-1 cells, since little up-regulated adhesion was observed even when the SLC concentration was increased to 1 g/ml or higher in versican-coated wells (A). C, effects of GAGs (CS A, CS B, or HS) on the SLC-induced adhesion of TK-1 cells to MAdCAM-1. The experimental design was identical to that for A, except that 50 g/ml GAGs were used instead of versican. Adhesion of TK-1 cells to MAdCAM-1-IgG (100 ng/well (q)) or human IgG (100 ng/well (E)) was determined as described under "Experimental Procedures." Results were expressed as the percentage of bound cells, where the input cell number into each well was defined as 100%.
was not inhibited at this dose range (Fig. 6B) suggest that CS B may also exert its inhibitory function without inhibiting the binding of SLC to its receptor. Concerning the second possibility, that versican/GAG competes with endogenous proteoglycans that may help present SLC to its specific receptor CCR7 and, thus, inhibit SLC signaling, our experiments showed that SLC specifically bound to a single class of high affinity receptors expressed on L1.2/CCR7 cells with a K d of 2.26 nM (data not shown), indicating that endogenous cell-surface proteoglycans are not involved in SLC binding in this cell line. In addition, our preliminary experiments showed that the binding of SLC to L1.2/CCR7 cells was not altered by treating the cells with chondroitinase or heparitinase (data not shown). In line with this notion, it was recently reported that cell-surface GAG chain expression is not necessary for chemokines such as MIP-1␣, MIP-1␤, and RANTES to exert their biological effects (34). Thus, the second possibility also appears unlikely. Concerning the third possibility, that versican/GAG transduces a negative signal by binding to a nonchemokine receptor type molecule, neither the TK-1 cells nor the L1.2/CCR7 cells used in this study bound to versican immobilized in plastic wells (data not shown). Furthermore, these cells do not express L-selectin or the active form of CD44, to which ACHN-derived versican binds (14,15). Thus, versican on its own does not appear to bind to the cells used in this study, and hence, the third pos-sibility seems unlikely. The fourth possibility, that the SLC that is complexed with versican or certain GAGs is unable to transduce an appropriate chemokine signal after binding to CCR7, is currently difficult to exclude and may, at least in part, be involved in the mechanism of inhibition. Consistent with this idea, it was recently reported that soluble complexes of RANTES and GAG chains can bind to an appropriate chemokine receptor but cannot induce intracellular Ca 2ϩ signaling (35). Collectively, the available data suggest that versican/GAG does not inhibit SLC signaling through inhibiting SLC binding to CCR7. Rather, the inhibition of SLC signaling is probably at least partly due to inhibition of the chemokine signal after SLC binds its receptor. Still other possible mechanisms remain to be completely excluded.
Versican binds certain chemokines but not all of them. It is notable that versican tends to preferentially bind chemokines that attract mononuclear leukocytes. Preliminary analysis of the amino acid sequences of versican-reactive and nonreactive chemokines has failed to identify any conserved amino acid residues or peptide regions that may be involved in versican binding (data not shown). Other factors, such as the tertiary structure, may explain the differences in binding affinity.
Although versican is found in the skin, brain, kidney, and aorta (13), SLC is principally found in secondary lymphoid tissues (36). One might thus argue that this particular combination of proteoglycan and chemokine, i.e. versican and SLC, is not physiological. However, it should be pointed out that versican also binds SDF-1␤ that is expressed in normal human skin as well as in inflammatory infiltrates of autoimmune skin diseases (37). In the inflamed skin, a large number of mononuclear cells expressing the receptor for SDF-1␤, CXCR4, selec- tively accumulates where SDF-1␤ is localized (37), suggesting the possibility that SDF-1␤ is involved in the infiltration of mononuclear cells into the inflamed skin. To inhibit excessive infiltration of mononuclear cells, however, an excess amount of SDF-1␤ produced in inflammatory infiltrates may have to be eliminated locally. In this sense, our result showing that versican-SDF-1␤ interaction results in down-regulation of SDF-1␤ activity (Fig. 5A) is interesting, pointing to the possibility that versican expressed in the inflamed skin interacts with SDF-1␤ to modulate the infiltration of mononuclear cells by downregulating SDF-1␤ activity, although this has yet to be evidenced experimentally. In addition, several CS proteoglycans, i.e. L-selectin-reactive CS proteoglycans of 150 and Ͼ200 kDa (38) and endoglycan (39), have been reported to be expressed in lymph node high endothelial venules where SLC is produced (36). It would be interesting to examine whether these CS proteoglycans can interact with SLC to regulate the function of SLC in situ.
Given that only specific kinds of GAGs can bind chemokines (Fig. 3), it is likely that only versican species modified with specific GAGs can bind chemokines. We previously showed that versican derived from ACHN binds via its GAG chains, a leukocyte adhesion molecule, L-selectin (14), and that certain GAGs such as CS B, CS E, and HS bind L-selectin directly (15). This specificity of binding is quite similar to what we observed for chemokine binding (Fig. 3B); i.e. those GAGs that bind L-selectin also bind certain chemokines, and those GAGs that do not bind L-selectin fail to bind the chemokines. Thus, we speculate that versican species modified with L-selectin-reactive GAGs also bind chemokines, which may be of particular importance in certain in vivo situations such as in inflammatory responses, where L-selectin and chemokines are concomitantly involved. The biological significance of the interaction between versican and chemokines is an important subject for further investigation.