The B16F10 Cell Receptor for a Metastasis-promoting Site on Laminin-1 Is a Heparan Sulfate/Chondroitin Sulfate-containing Proteoglycan Free

Laminin, a heterotrimeric glycoprotein that is a major constituent of the extracellular matrix, regulates the biological activity of a variety of cell types (1, 2). Laminin-1 has a profound effect on the metastatic phenotype of B16F10 melanoma cells, including increased in vitro proteinase production, migration, and invasion. Cells grown in the presence of laminin-1 form more tumors when injected into mice than cells cultured in the absence of laminin-1. In addition, cells that are adherent to laminin-1 form more tumors when injected into mice than either fibronectin-adherent cells or the parental cells (3, 4, 5, 6).

In an endeavor to identify biologically active sites on laminin-1, overlapping peptides spanning the α1 chain were generated and tested on various cells. These included the B16F10 mouse melanoma cell line, an established model for metastasis. A number of sites on laminin-1 have been found to affect the malignancy of B16F10 cells, each with different activity. Tyr-ile-gly-ser-arg (YIGSR, laminin β1 chain) reduces s.c. tumor growth, lung colonization, and angiogenesis (7, 8, 9). A M_r 32,000/67,000 cell surface glycoprotein has been identified as the cellular receptor, and its levels correlate with malignancy (10). In contrast, ile-lys-val-ala-val (IKVAV, laminin α1 chain) increases s.c. tumors, lung colonization, protease activity, and angiogenesis (11). A peptide from the laminin α1 COOH-terminal globular domain, AG73 (LQVQLSIRT), promotes the metastatic phenotype in B16F10 cells (12). These cells normally form lung tumors, with a few tumors arising in other organs. However, when coinjected with the AG73 peptide, the number of lung tumors increased, and unexpectedly, liver tumors were observed in ∼50% of the mice. Furthermore, when cells were adhesion-selected to AG73, liver metastases were still observed after tail vein injection in the absence of the peptide (13). This sequence has been found to also increase adhesion, migration, and protease production, but it has no effect on angiogenesis (12, 13, 14). These data identify the AG73 laminin α1 chain sequence as a potent site active in malignancy.

Here we identify the cellular receptor on B16F10 cells for the AG73 peptide. On the basis of affinity chromatography, glycosidase digestion, and competition experiments with various GAGs,² we found that a cell surface heparan sulfate/chondroitin sulfate B-containing proteoglycan binds to AG73 via the GAG side chains. Finally, we show that B16F10 cells lacking these GAG side chains no longer metastasize to the liver, but lung metastases are not affected. These data define the receptor for a site on laminin that is important in the malignant phenotype of B16F10 cells.

B16F10 melanoma cells (15) were cultured in DMEM (Life Technologies, Inc., Rockville, MD), containing 10% fetal bovine serum (HyClone, Logan, UT), 100 units/ml penicillin, 100 μg/ml streptomycin, and Nonessential Amino Acids Solution (Life Technologies, Inc.). The cells were maintained at 37°C in a humidified 5% CO₂/95% air atmosphere.

The AG73 peptide (RKRLQVQLSIRT) and the scrambled AG73T control peptide (LQQRRSVLRTKI) were manually synthesized and purified by high-performance liquid chromatography as described previously (16).

Affi-gel 10 (Bio-Rad, Hercules, CA) peptide and laminin affinity columns (1 ml) were prepared according to the manufacturer’s instructions, and run as described previously (17). AG73 affinity columns were run in parallel with the AG73T-negative control column. The columns were equilibrated in running buffer containing 6.0 m urea, 1% Triton X-100, and Complete protease inhibitor mixture (Boehringer Mannheim, Indianapolis, IN) in Tris-buffered saline (pH 7.4). For each experiment, three 150-mm plates of subconfluent cells were surface-biotinylated using sulfo-NHS-biotin (Pierce, Rockford, IL). A crude cell membrane fraction was prepared by hypo-osmotic lysis in 20 mm Tris (pH 7.4) containing 10 mm KCl, 0.1% β-mercaptoethanol, and 1 mm EDTA. After dounce homogenization, the nuclei were removed by centrifugation at 1,500 × g for 5 min. The NaCl concentration of the remaining supernatant was increased to 150 mm, and the cell membranes were pelleted at 50,000 × g for 30 min. The cell membrane pellet was solubilized in Tris-buffered saline containing 2 ml of 8.0 m urea, 1% Triton X-100, 0.5 m KCl, and Complete protease inhibitor mixture. The insoluble material was removed by centrifugation at 14,000 × g for 20 min, and the volume was increased to 10 ml with running buffer. A 500-μl aliquot of the crude cell membrane fraction (∼700 μg total protein) was incubated with either the peptide or a control peptide affinity column for 2 h at 4°C. The columns were washed with running buffer (60 ml) and then sequentially eluted with 2-ml aliquots of running buffer containing either 20 mm EDTA, 250 mm NaCl, 1.0 m NaCl, or 2.0 m NaCl. Material in the eluted fractions was precipitated with acetone, washed in 80% ethanol, and air dried. Proteins were separated by SDS-PAGE (4–20% gels) and transferred to nitrocellulose filters (Novex, San Diego, CA). The filters were blocked in 5% nonfat milk in PBS-T (0.1% Tween 20), washed, incubated with streptavidin-horseradish peroxidase in PBS-T for 1 h, and then washed again three times for 10 min each in PBS-T. The biotinylated material was visualized by enhanced chemiluminescence (Amersham, Arlington Heights, IL). This experiment was repeated at least three times.

To enrich for the receptor, twice the amount of solubilized membrane preparation was passed over a 5-ml HiTrap Q ion exchange column (Amersham Pharmacia Biotech, Inc., Piscataway, NJ). After washing the column with 130 ml of running buffer, proteins were eluted with 10 ml of increasing NaCl concentrations (0.2–2 m). All of the fractions were dialyzed against H₂O. Before passing this material over the laminin and AG73 affinity columns, 12 m urea in Tris-buffered saline (pH 7.4) without Triton X-100 but with Complete protease inhibitor was added.

Material eluted from the peptide affinity columns with 1.0 m NaCl was precipitated with acetone and digested for 4 h with chondroitinase ABC (1.0 units/ml), heparitinase (0.1 units/ml), or both together containing Complete protease inhibitor mixture. The digested material was separated by 4–20% SDS-PAGE, transferred to nitrocellulose filters, and processed as above. This assay was performed at least three times.

U-bottomed 96-well plates (Dynex Technologies, Inc., Chantilly, VA) were used because more cells attached faster than to flat-bottomed plates. These plates were coated with 0.1 μg of peptide in 50 μl of H₂O and dried overnight at room temperature. The wells were blocked with DMEM-3% BSA for 1 h at 37°C and then washed three times with DMEM-0.1% BSA. Cells were washed two times with DMEM-0.1% BSA and removed from the plate with Versene (Life Technologies, Inc.). Cells (30,000) in 50 μl of DMEM-0.1% BSA were added per well and incubated at 37°C for 30 min. The medium with unattached cells was removed from the wells. The adherent cells were stained for 10 min with crystal violet and washed twice with water. The cells were lysed with 50 μl of 10% SDS, and the absorbance (600 nm) was measured. Wells containing peptide alone were also stained to determine the background. Each assay was performed in triplicate at least four times.

Plates were prepared as above; however, before addition of the cells, the wells were preincubated with 5 μg of the GAGs in 25 μl of DMEM-0.1% BSA for 30 min at 37°C to optimize adherence inhibition. Heparin, heparan sulfate, chondroitin sulfates A, B, and C, hyaluronic acid (Sigma, St. Louis, MO), chemically modified heparin and heparan sulfate (desulfated N-acetylated, desulfated N-sulfated, and undesulfated N-acetylated; Seikagaku, Rockville, MD), and heparin fragments ranging from 2–20 mers (a gift from Dr. A. Marolewski, RepliGen, Needham, MA) were tested. Heparin, heparan sulfate, and chondroitin sulfate B were used at 0.1–5.0 μg/well in the dose-response experiment with the IC₅₀ calculated by Prism (GraphPad Software, Inc., San Diego, CA). Cells (30,000) were added in 25 μl of DMEM-0.1% BSA to bring the total volume to 50 μl. The assays were performed in triplicate at least four times.

Matrigel (Becton Dickinson, Bedford, MA; 133 μl) was coated on 48-well flat-bottomed plates (Costar Corporation, Cambridge, MA) and allowed to gel. The peptides were added at a concentration of 200 μg/ml immediately after seeding 40,000 cells/well in 465 μl of serum-free medium. After an overnight incubation, the medium was gently removed from each well. Approximately 300 μl of Diff-Quik fixative (Dade International, Miami, FL) was added per well and gently removed after 30 s. This was repeated with Diff-Quik solution II diluted 1:1 in H₂O. Each assay was done in triplicate at least four times with similar results.

The assay was performed as described above; however, heparin, heparan sulfate, or chondroitin sulfate B were added at the same concentration as the peptide (200 μg/ml) just before addition of the peptide. These GAGs were also added to the assay in decreasing concentrations (100, 50, 40, 30, 25, 20, 15, 10, and 5 μg/ml) to determine the lowest concentration at which the effects of the peptide were reversed. This assay was performed in triplicate at least four times with similar results.

Cells were resuspended at 600,000 cells/ml in DMEM-0.1% BSA, and enzymes were added at concentrations of 0.05 units/ml of heparitinase or 0.02 units/ml of either heparinase or chondroitinase (Seikagaku). Cells were incubated on a rotator at 37°C for 90 min. After enzyme treatment, the cells were used in a cell adhesion assay as described above or pelleted and resuspended in PBS at 250,000 cells/ml and immediately injected into mice as described below. The assay was done at least six times for adherence and twice for the in vivo work.

Cells were treated as described above to remove surface GAGs using heparitinase and chondroitinase ABC. The cells were pelleted, resuspended to the same density (600,000 cells/ml) in DMEM-0.1% BSA, and used in an adherence assay after 0, 30, 60, 90, or 120 min of incubation at 37°C to allow for GAG regeneration. Control cells were treated identically except that they were maintained in the presence of the enzymes. This experiment was performed at least four times.

B16F10 melanoma cells were treated with a combination of heparitinase and chondroitinase ABC, as described above, to remove surface GAGs. Removal of the GAGs was monitored by a cell adhesion assay to AG73, as described above. After glycosidase treatment, 50,000 B16F10 cells were injected into the tail veins of C57BL6/Nsd mice (Harlan, Indianapolis, IN). As a control, an equal number of mice were injected with cells handled similarly in the absence of the glycosidases. The AG73 peptide was dissolved in PBS at a concentration of 5 mg/ml. One mg of peptide per animal was injected i.p. immediately after the tail vein injection. An equal number of mice injected with the treated and untreated cells had PBS i.p. injected as a control. The mice were sacrificed at either day 14 or 19 after injection, and the livers and lungs were examined for tumors. This experiment was performed two times with 10 mice/group.

The surface proteins of B16F10 melanoma cells were biotinylated, and a crude cell membrane preparation was made to identify the cell surface receptor for the AG73 peptide. This material was run on affinity columns made with either AG73 or AG73T, the scrambled peptide. This preparation was also enriched for the receptor by passage over an ion exchange column before being run over a laminin affinity column. No proteins bound to the scrambled peptide column (Fig. 1,B). Multiple protein bands appeared in the EDTA-eluted fraction from the AG73 column (Fig. 1,A, Lane 3). Because EDTA did not consistently inhibit B16F10 cell attachment to AG73,³ these contaminating proteins may bind to the true receptor or may form fairly weak Ca²⁺ interactions with the peptide. Higher affinity bound proteins were released with NaCl. The major protein eluted from the AG73 and laminin columns with 1 and 2 m NaCl appeared as a high molecular weight smear of approximately M_r 150,000–200,000, suggestive of proteoglycans, which typically run as a smear because of variations in glycosylation (Fig. 1,A, Lanes 5 and 6; laminin data not shown). Therefore, this material was digested with glycosidases, which resulted in a molecular weight shift of the protein to M_r ∼90,000 (Fig. 1,C). Treatment with heparitinase, chondroitinase ABC, or a combination of both (Fig. 1 C, Lanes 3, 4, and 2, respectively), all revealed this similar size core protein, indicating that both heparan sulfate and chondroitin sulfate are present in this proteoglycan. However, it appears that the receptor is more highly glycosylated with chondroitin sulfates than heparan sulfate, because only a small amount of the core protein was detected after heparitinase treatment alone. The high molecular weight bands are likely material that did not enter the gel optimally.

Because the potential receptor contained both heparan sulfate and chondroitin sulfate side chains, these GAGs and others were tested for their ability to inhibit B16F10 melanoma cell attachment to AG73. Only heparin, heparan sulfate, and chondroitin sulfate B (Fig. 2,A, Lanes 1, 2, and 4, respectively) significantly reduced cell adhesion to the peptide; chondroitin sulfates A and C, and hyaluronic acid had no effect. Serial dilutions of the GAGs demonstrated that the concentration required to inhibit adherence by 50% followed the order heparin (IC₅₀, 1.09 μg/ml) < heparan sulfate (IC₅₀, 4.43 μg/ml) < chondroitin sulfate B (IC₅₀, 39.92 μg/ml; Fig. 2 B). These GAGs did not inhibit adherence to laminin (data not shown).

We tested whether the overall charge of the receptor might be important for mediating cell attachment to AG73, because heparin is more highly sulfated than heparan sulfate. It is also possible that the sulfation pattern plays a role in this interaction. Therefore, adherence inhibition was performed with chemically modified heparin and heparan sulfate. Only the unmodified forms of heparin and heparan sulfate and N-desulfated N-acetylated heparin inhibited cell attachment (Fig. 2,C, Lanes 1 and 4). The GAGs that were modified to remove more of the sulfate groups, completely desulfated N-acetylated and completely desulfated N-sulfated, had a minimal effect on cell attachment (Fig. 2,C, Lanes 2 and 3). Because N-desulfated N-acetylated heparin was able to prevent adhesion to AG73, it is likely that N-sulfation is not necessary for the receptor-peptide interaction. However, N-desulfated N-acetylated heparan sulfate did not significantly affect adherence, suggesting that the overall charge of the GAG chains may play a critical role (Fig. 2,C, Lane 4, open bar). Various sized heparin fragments were used in the adhesion inhibition assay to determine whether a specific minimal size was required for function, as observed with FGFs (18, 19, 20). Inhibition of cell adhesion was not observed unless a fragment of at least 16-mer size was included in the assay, supporting the idea that the charge and/or size of the B16F10 receptor GAG side chains is crucial for binding to AG73 (Fig. 2 D).

When grown on Matrigel, B16F10 cells form a network like most tumor cells (Fig. 3,A). The morphology was not affected by the scrambled peptide; however, this network was disrupted, and a monolayer of cells was observed after addition of the AG73 peptide (Fig. 3, B and C, respectively). Addition of an equal amount of the GAGs heparin, heparan sulfate, and chondroitin sulfate B alone did not affect this network (data not shown). When the GAGs were included with the AG73 peptide at this concentration, B16F10 cells formed the network demonstrating that these GAGs could block the effect of AG73 in an assay more complex than adhesion (data not shown). Decreasing amounts of the GAGs were added to the assay to determine the lowest concentration of the GAGs able to counteract the effects of the AG73 peptide. Heparin and heparan sulfate inhibited the effects of AG73 at a lower concentration than chondroitin sulfate B. A network was clearly observed when heparin or heparan sulfate was added at 20 μg/ml, whereas the same level of this formation was only apparent at 25 μg/ml for chondroitin sulfate B (Fig. 3, E, H, and J, respectively). At a lower concentration (15 μg/ml) none of the GAGs was able to counteract the effects of AG73 (Fig. 3, F, I, and L, respectively).

To confirm that GAGs play a role in B16F10 cell adhesion to AG73, surface GAGs were removed with glycosidases before addition of the cells to the adhesion assay. Cell attachment was decreased after treatment with either heparitinase or a combination of heparitinase, heparinase, and chondroitinase ABC (Fig. 4, Lanes 2 and 6). Chondroitinase ABC slightly inhibited adhesion, whereas heparinase and chondroitinase AC did not (Fig. 4, Lanes 4, 3, and 5, respectively), supporting the hypothesis that the AG73 receptor contains chondroitin sulfate B, which influences cell adhesion. Complete inhibition was not observed, in part because suboptimal enzyme digestion conditions were necessary for use with live cells. Viability of the treated cells was monitored by attachment to laminin and to fibronectin, which was unaffected by cell surface GAG removal (data not shown). These data confirm the functional role of heparan sulfate and chondroitin sulfate B in interactions with the AG73 peptide.

It was important to establish how long it takes for B16F10 cells to replace these GAGs. Cells were treated with the glycosidases heparitinase and chondroitinase ABC to remove the surface GAGs. After resuspension in the absence of the glycosidases, these cells were used in the adherence assay at various time points (0–120 min) to allow for cell surface GAG regeneration. Although adherence was not of statistical significance for the various time points, it appeared that cell surface GAG regeneration was complete between 60 and 90 min postglycosidase treatment (Fig. 5). Again, optimal conditions for GAG removal were not possible because of treatment of live cells. At each time point, control cells were treated identically without removal of the glycosidases.

B16F10 melanoma cells form mainly lung tumors when i.v. injected into mice, and they also form liver tumors in ∼50% of mice if the AG73 peptide is i.p. injected immediately after the tumor cell injection (12). We determined whether cell surface GAGs had any effect on tumor formation and location in the presence of AG73. Cells were treated with a combination of heparitinase and chondroitinase ABC. Untreated control cells were processed identically without addition of the enzymes. Immediately after tail vein injection of the cells, mice received either PBS or AG73 by i.p. injection. After 2 weeks, the livers and lungs were examined. All of the mice had lung metastases; however, no differences in the number of tumors were observed with the cells pretreated either in the presence or absence of glycosidases (data not shown). Normally, ∼10% of control mice form liver tumors. Thirty percent (6 of 20) of the mice injected with untreated control cells had liver tumors, and 45% (9 of 20) of the mice receiving untreated cells plus i.p. injected AG73 had liver tumors (Table 1). When the cells were treated with glycosidases, 25% (5 of 20) of the mice had liver tumors in the presence and absence of i.p.-injected AG73. A similar pattern was noted for the number of tumors in each group. These data demonstrate that surface removal of the heparan sulfate and chondroitin sulfate GAGs reduces AG73-induced liver tumors to background levels.

It has been shown previously that B16F10 cells form mainly lung tumors but also liver tumors if the laminin peptide AG73 is coinjected or introduced i.p (12). Because the B16F10 receptor-ligand interaction with AG73 clearly induces a metastatic phenotype, the identification of the receptor is a critical first step to understanding the mechanisms involved. In this paper, evidence is presented that the B16F10 receptor for AG73 is a heparan sulfate/chondroitin sulfate-containing proteoglycan. On the basis of in vitro and in vivo experiments, heparan sulfate and chondroitin sulfate B appear to be the functional portion of the proteoglycan for the interaction of these cells with the peptide. Typically, the GAG side chains of proteoglycans, not the core protein, mediate interactions with other molecules (21). The GAGs can interact with numerous extracellular ligands, including adhesion molecules, extracellular proteases, protease inhibitors, growth factors, and extracellular matrix proteins. Through these interactions, proteoglycans function in many cell events, such as migration and proliferation, as well as play a role in many pathways, including signaling during cell adhesion (21, 22).

An AG73 affinity column isolated a protein that appeared as a high molecular weight smear after SDS-PAGE, typical of proteoglycans because of variations in the amount of glycosylation of the protein. The M_r of the eluted material corresponded well with that of dystroglycan, and the M_r of the core protein resulting from glycosidase treatment also matched this protein (23). However, using a specific antibody (a kind gift of Dr. Kevin Campbell), we found that the receptor was not dystroglycan (data not shown). Antibodies to syndecan-1, a heparan sulfate proteoglycan identified previously as a cell surface receptor for AG73 on a human submandibular gland cell line, and glypican also failed to recognize the receptor (17). The nature of the core protein, which likely functions to localize the proteoglycans on the cell surface and may also be involved in signaling, is not known at this time.

The adherence inhibition studies indicated that heparin was the most effective, then heparan sulfate, and finally chondroitin sulfate B. This suggested that charge of the GAG was important for preventing the peptide-receptor contact because heparin has the highest negative charge and chondroitin sulfate B the lowest. Although the glycosidase study suggested that the receptor proteoglycan was more enriched in chondroitin sulfate, the functional data demonstrate that the heparan sulfate side chains are the major biologically active components. Cell surface proteoglycans containing heparan sulfates are important in a number of biological responses. For example, various FGF isotypes bind to heparan sulfate proteoglycans of the extracellular matrix and to the FGF receptor to form a signaling unit for mitogenesis (24, 25, 26). Also, heparan sulfate-containing proteoglycans have been shown to cooperate with and function as coreceptors with integrins in a number of cell-matrix-mediated activities (27). In a human submandibular gland cell line, the interaction of AG73 and syndecan-1, a heparan sulfate proteoglycan, influences acinar development, but our studies clearly demonstrate a different receptor on B16F10 melanoma cells (17). The receptor identified on B16F10 melanoma cells has a smaller core protein, contains a chondroitin sulfate B side chain, and does not react with syndecan-1 antibody. These data suggest that different but related receptors on different cells can recognize the same sequence on laminin.

Tumor cells proliferate, invade, and form structures when plated on Matrigel (28). We found that AG73 blocked the network formation by B16F10 melanoma cells; the cells did not detach, but rather a monolayer was observed. It is clear that multiple ligands besides the AG73 site on laminin-1 are present in the more complex Matrigel matrix and include other binding sites on laminin-1 and on collagen IV (29). Therefore, the cells are likely binding to multiple sites in Matrigel. We conclude that the interaction of B16F10 cells with the AG73 sequence is important for network formation. Exogenous GAGs alone did not affect network formation either when preincubated with the cells or with the Matrigel. Certain GAGs did reverse the effect of AG73 on network formation, and it is clear from these studies that heparin, heparan sulfate, and chondroitin sulfate B are important functional ligands for the AG73 peptide.

Polyanionic substances, such as heparin, are known to inhibit the metastatic phenotype. Heparin has numerous effects in vivo that may contribute to this. For instance, heparin prevents tumor cell-platelet interaction via selectins, which, in turn, exposes the tumor cells to immune surveillance as well as impedes attachment to endothelial cells limiting extravasation (30). Tumor cells interact with heparan sulfate proteoglycans on endothelial cells, and heparin can compete for this attachment. Interestingly, we find that cell surface heparan sulfate proteoglycans are critical for liver metastasis. Heparin also inhibits all of the chemokines thus far tested. Many tumor cell lines express chemokines, and one common function is activation of integrin adhesion molecules, which are used to attach tumor cells to extracellular matrix proteins. Another effect of heparin is inhibition of heparanase activity. Tumor cells produce heparanases that degrade heparan sulfate proteoglycans, lysing basement membranes and thereby facilitating tumor invasion and metastasis. Similarly, heparin impairs matrix metalloproteinase activity that is linked to tumor cell disruption of the basement membrane and invasion of the extracellular matrix (31). All of these effects of heparin occur over a relatively short time period while the tumor cells are in the bloodstream, attaching to the endothelium, or penetrating the basement membrane and extracellular matrix to enter a secondary organ. Results from our in vivo study indicate the interaction and effect(s) of the AG73 peptide on B16F10 cells must occur rapidly because GAG regeneration was complete within 60–90 min. It is possible that AG73 acts on one or more of these heparin-inhibitable steps that are critical for metastasis. For instance, we have tested the effect of AG73 on the heparanase promoter and found there is a slight increase in activity.⁴

In summary, the present study demonstrated both in vitro and in vivo that the B16F10 receptor for a metastasis-promoting laminin α1 peptide is a proteoglycan, with the functional, interactive portion being the GAG side chains.

We thank Dr. A. Marolewski (RepliGen Corp.) for the sized heparin fragments, Dr. K. Campbell (University of Iowa College of Medicine, Iowa City, IA) for the dystroglycan antibodies, and Dr. D. Ron (Technion, Haifa, Israel) for the glypican Westerns. We also thank Harry Grant for critical reading of the manuscript.