Multiple Heparan Sulfate Chains Are Required for Optimal Syndecan-1 Function*

Syndecans have three highly conserved sites available for heparan sulfate attachment. To determine if all three sites are required for normal function, a series of mutated syndecans having two, one, or no heparan sulfate chains were expressed in ARH-77 cells. Previously, we demonstrated that expression of wild-type syndecan-1 on these myeloma cells mediates cell-matrix and cell-cell adhesion and inhibits cell invasion into collagen gels. Here we show that to optimally mediate each of these activities, all three sites of heparan sulfate attachment are required. Generally, an increasing loss of syndecan-1 function occurs as the number of heparan sulfate attachment sites decreases. This loss of function is not the result of a decrease in either the total amount of cell surface heparan sulfate or syndecan-1 core protein. In regard to cell invasion, cells expressing syndecan-1 bearing a single heparan sulfate attachment site exhibit a hierarchy of function based upon the position of the site within the core protein; the presence of an available attachment site at serine 47 confers the greatest level of activity, while serine 37 contributes little to syndecan-1 function. However, when all three heparan sulfate chains are present, significantly greater biological activity is observed than is predicted by the sum of the activities occurring when the chains act individually. This synergy provides a functional basis for the evolutionary conservation of the three heparan sulfate attachment sites on syndecans and supports the idea that molecular heterogeneity, which is characteristic of proteoglycans, contributes to their functional diversity.

Syndecans have three highly conserved sites available for heparan sulfate attachment. To determine if all three sites are required for normal function, a series of mutated syndecans having two, one, or no heparan sulfate chains were expressed in ARH-77 cells. Previously, we demonstrated that expression of wild-type syndecan-1 on these myeloma cells mediates cell-matrix and cell-cell adhesion and inhibits cell invasion into collagen gels. Here we show that to optimally mediate each of these activities, all three sites of heparan sulfate attachment are required. Generally, an increasing loss of syndecan-1 function occurs as the number of heparan sulfate attachment sites decreases. This loss of function is not the result of a decrease in either the total amount of cell surface heparan sulfate or syndecan-1 core protein. In regard to cell invasion, cells expressing syndecan-1 bearing a single heparan sulfate attachment site exhibit a hierarchy of function based upon the position of the site within the core protein; the presence of an available attachment site at serine 47 confers the greatest level of activity, while serine 37 contributes little to syndecan-1 function. However, when all three heparan sulfate chains are present, significantly greater biological activity is observed than is predicted by the sum of the activities occurring when the chains act individually. This synergy provides a functional basis for the evolutionary conservation of the three heparan sulfate attachment sites on syndecans and supports the idea that molecular heterogeneity, which is characteristic of proteoglycans, contributes to their functional diversity.
Heparan sulfate is the most ubiquitous glycosaminoglycan (GAG) 1 on cell surfaces. These long, highly diverse carbohydrate polymers are negatively charged and are most often found covalently linked to protein in the form of proteoglycans. Heparan sulfate binds to many extracellular effector proteins including insoluble extracellular matrix molecules, soluble peptide growth factors, and cell adhesion molecules (1). It is through these interactions that heparan sulfate mediates or assists in diverse biological responses including cell-cell and cell-matrix adhesion, apoptosis, growth factor regulation, pro-teolysis, and angiogenesis. Thus, heparan sulfate influences cell behaviors that are critical during development, homeostasis, and disease progression.
The syndecans, a multigene family of proteoglycans, constitute the predominant source of heparan sulfate at the cell surface. On SDS-polyacrylamide gels, the syndecans run as broad smears indicating extensive molecular heterogeneity (2,3). Detailed analysis of syndecan-4 expressed in L cells demonstrates that it is present in several isoforms, as pure heparan sulfate proteoglycans, as various mixtures of heparan sulfate/ chondroitin sulfate hybrids, and as pure chondroitin sulfate proteoglycans (4). In addition, a molecular polymorphism has been described between simple and stratified epithelial tissues due to differences in the number and size of GAG chains (5). Although the functional significance of syndecan heterogeneity remains poorly understood, it has been widely speculated that structural variations in GAG chain number and size contribute to the multifunctional capacity of these proteoglycans.
A distinct structural feature of syndecans is a conserved region of three GAG attachment sites near the N terminus of their core protein. Not only are these sites conserved among all four members of the syndecan family in mammals but also in syndecans from chick, Xenopus, and Drosophila (2, 6 -8). These sites are predominantly substituted with heparan sulfate chains, but in some syndecans they are also capable of bearing chondroitin sulfate (4,9). The high degree of conservation of the three N-terminal GAG attachment sites together with the finding that the individual sites are substituted with heparan sulfate more often when multiple sites are present (10) implies that all three sites are essential for syndecan's biological activity.
To address this question, a series of mutated syndecan-1 molecules having various combinations of their N-terminal GAG attachment sites deleted were expressed in ARH-77 human B lymphoid cells. Cells bearing mutated syndecan-1 were then tested in adhesion and invasion assays. We have previously demonstrated that parental ARH-77 cells, which express very low levels of cell surface heparan sulfate proteoglycan, do not aggregate, do not bind to type I collagen, and readily invade type I collagen gels. Following transfection with a cDNA for wild-type syndecan-1, these behaviors are reversed; the cells aggregate, bind to type I collagen, and are inhibited from invading collagen gels (11,12). In the present study, cells bearing syndecan-1 with two heparan sulfate attachment sites (a single deletion mutation) were analyzed to show what changes in syndecan-1 activity occur when heparan sulfate chains at specific sites are deleted. Likewise, cells bearing syndecans with only a single GAG attachment site (a double deletion mutation) were analyzed to examine the biological activity contributed by an individual heparan sulfate chain. We find that for syndecan-1 to optimally influence these cell behaviors, all three N-terminal sites and their heparan sulfate chains are required.
Cells expressing syndecan-1 having less than three heparan sulfate attachment sites show a reduction in biological activity that in general is relative to the number of chains present. These results provide a basis for the conservation of all three attachment sites on syndecans and imply that the structural heterogeneity common to proteoglycans serves as a mechanism for modulating their function.

EXPERIMENTAL PROCEDURES
Cell Culture and Mutagenesis-The EcoRI-HindIII fragment of clone 4 -19b (kindly provided by Dr. M. Bernfield), containing the full coding region of murine syndecan-1 (13), was excised from the pGEM-3Z vector and ligated into the EcoRI-HindIII sites of pBluescript II KS(ϩ) (Stratagene, La Jolla, CA). This was used as the template for oligonucleotidedirected mutagenesis (Transformer TM site-directed mutagenesis kit, CLONTECH, Palo Alto, CA), which was used to generate seven mutagenic constructs of syndecan-1 in which serines 37, 45, and 47 were converted to alanine in all possible combinations (Fig. 1). Following mutagenesis, the syndecan-1 coding region was excised at the BamHI-SalI sites and ligated into pCDNA3 (Invitrogen, San Diego, CA) at the BamHI-XhoI sites. All mutagenic constructs were confirmed by manual sequence analysis using the dideoxy method with the T7 Sequenase Quick Denature plasmid sequencing kit (Amersham Pharmacia Biotech).
ARH-77 human B lymphoid cells were obtained from the American Type Culture Collection (Rockville, MD). These cells express little, if any, endogenous syndecan-1 (14). ARH-77 cells transfected with a pM-AMneo vector containing a murine syndecan-1 cDNA (clone B3P3; Ref. 11) were used as a wild-type syndecan-1-expressing cell line in all assays. Cells were maintained in culture as described previously (11). ARH-77 cells were transfected with the vector containing mutant syndecan-1 constructs or vector alone (neomycin control) by liposomemediated transfection according to the manufacturer's instructions (Life Technologies, Inc.). Cells expressing the syndecan-1 molecules were sorted by "panning" using a monoclonal antibody 281.1 (that recognizes the syndecan-1 core protein) immobilized to 100-mm tissue culture dishes. Upon release of cells from the plates, expression of syndecan-1 was confirmed by immunofluorescence staining.
Western Blotting-Culture media conditioned by cells expressing syndecan-1 were brought to a final concentration of 2 M urea and 50 mM sodium acetate. Syndecan-1 was isolated by ion exchange and affinity chromatography as described previously (15). Equivalent aliquots of proteoglycan, still bound to the beads, were either 1) untreated or 2) digested twice with chondroitinase ABC (50 milliunits/ml) (Seikagaku, Tokyo, Japan). Proteoglycans were then eluted and analyzed on 4 -12% acrylamide, Tris/glycine gel (Novex, San Diego, CA) and transferred to cationic membranes (Gene-Trans membrane; Plasco Inc., Woburn, MA), which were then probed with 125 I-281.2 antibody and visualized on a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Quantification of Proteoglycans-For determination of the ratio of HS per syndecan-1 core protein, 35 SO 4 -labeled syndecan-1 was isolated by extracting cells with 20 mM Tris-HCl, pH 7.4, containing 1% Triton X-100, 0.15 M NaCl, 5 mM N-ethylmaleimide, 5 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride on ice for 30 min. After centrifugation, supernatants were incubated with antibody 281.2 bound to Sepharose beads. Bound syndecan-1 was then digested first with chondroitinase ABC (as described above) and then heparitinase (Seikagaku) (1 milliunit/ml). After each digestion, released counts were quantified by liquid scintillation counting. After all digestions were completed, the core proteins were eluted from the beads and transferred to Zeta-Probe membranes (Bio-Rad) using a dot-blotting apparatus (Millipore, Bedford, MA). The membranes were probed with a 125 I-281.2 antibody overnight at 4°C. Following extensive washing, individual dots were cut from the membranes and quantified by ␥-counting.
For comparison of the total heparan sulfate present on the surface of the various transfectants, immunofluorescence staining was performed using an antibody that recognizes native heparan sulfate chains (Ref. 16; clone 10E4, Seikagaku), a biotinylated secondary antibody (Vector Laboratories, Burlingame, CA), and avidin conjugated to fluorescein isothiocyanate (Vector). Mean fluorescence intensities were obtained by flow cytometry. A comparison of the total amount of syndecan-1 core protein expressed on the various transfectants was also performed by flow cytometry using fluorescein isothiocyanate-labeled 281.2.
Assay for Collagen Binding-Cell binding to collagen was assayed as described previously (15). Briefly, 8.0 ϫ 10 4 cells were incubated for 30 min at room temperature in 96-well plates (polyvinyl U-bottomed plates; Dynatech, Chantilly, VA) that had been coated with either type I collagen (1 mg/ml) or bovine serum albumin (1 mg/ml). The plate was centrifuged at 120 ϫ g for 20 min. Following centrifugation, cells were fixed and then stained with 4% trypan blue in phosphate-buffered saline. In this assay, unbound cells form a densely stained pellet while bound cells remain as a uniform coating on the well surface.
Assay for Cell Invasion-The percentage of invasive cells was determined as described previously (11). Briefly, 5.0 ϫ 10 4 cells were placed on the surface of a hydrated type I collagen gel (0.5 mg/ml; Collaborative Biomedical Products, Bedford, MA). Forty-eight hours later a limited trypsin digestion was employed to release the attached "noninvasive" cells, and then complete collagenase digestion was employed to release and recover the invasive cells. The number of noninvasive versus invasive cells was quantified using a Coulter counter.
To determine if the level of syndecan-1 core protein influences the invasiveness of cells, transfectants were stained with fluorescein isothiocyanate-labeled 281.2, and cells having equivalent amounts of syndecan-1 core protein were harvested by fluorescence-activated cell sorting. Sorted cells were then analyzed in the collagen gel assay as described above.
Affinity Co-electrophoresis (ACE)-35 SO 4 -Labeled syndecan-1 was isolated by extracting cells as described above. After centrifugation, the supernatant was brought to 6 M urea and 50 mM sodium acetate, pH 6.0, and subjected to DEAE and 281.2 affinity chromatography (15). Proteoglycans were eluted from the affinity beads with 0.5% (w/v) triethylamine, neutralized with 0.5 volumes of 1 M Tris-HCl, pH 7.0, and desalted over an excellulose column (Pierce) into ACE running buffer (0.1 M sodium acetate, 50 mM sodium MOPSO (Fluka Biochemika, Ronkonkoma, NY), pH 7.0). Approximately 50,000 cpm of labeled material was analyzed on a 1% agarose gel having nine parallel lanes of agarose containing decreasing amounts of type I collagen (15,17). After electrophoresis, the gel was soaked in 5% acetic acid for 4 h and dried overnight with forced air at 42°C, and the labeled material was imaged using a PhosphorImager.
Assay for Cell Aggregation-Cells were placed into wells of a 24-well plate in aggregation buffer (Hank's balanced salt solution, 1 mM calcium chloride, 1 mM magnesium sulfate, 1% bovine serum albumin, and 10 mM HEPES buffer) and rotated on a gyratory shaker (100 revolutions/min) at 37°C for 60 min (12). The cells were then removed and placed onto a glass slide, and the number of cells in aggregates lying within a central rectangular area of 6.2 mm 2 were counted. Aggregates were defined as containing four or more cells.

Production of Cells Bearing Syndecan-1 Having Mutated
Heparan Sulfate Attachment Sites-To determine if all three heparan sulfate attachment sites are essential for syndecan-1 biological activity, we generated a series of seven constructs by performing oligonucleotide-directed mutagenesis on a murine syndecan-1 cDNA. Serine residues at position 37, 45, and/or 47 were changed to alanine, thereby preventing the addition of GAG chains at those sites. Mutations were confirmed by DNA sequencing, and each construct was separately transfected into the human B lymphoid cell line ARH-77. The mutated syndecan-1 molecules are capable of bearing no, one, or two heparan sulfate chains at the N-terminal attachment sites, whereas wild-type syndecan-1 is capable of expressing three heparan sulfate chains (Fig. 1). The two GAG attachment sites within the syndecan-1 core protein located at serines 210 and 220 (adjacent to the cell membrane) were not deleted, because they are reported to bear only chondroitin sulfate chains (18). The biological activity of syndecan-1 in the invasion, adhesion, and aggregation assays resides in its heparan sulfate chains (11,12,19). Thus, although chondroitin sulfate chains are present on the syndecans, they probably do not influence the functions being assessed in the present study.
Characterization of Mutants-Deletion of GAG attachment sites should yield syndecan-1 molecules having a smaller modal molecular size than wild-type syndecan-1. To confirm this, syndecan-1 purified from the media of cells bearing the mutated proteoglycans was examined by Western blotting. The relative mobilities reflect the deletion of one, two, or all three N-terminal GAG attachment sites and their associated chains ( Fig. 2A). Wild-type syndecan-1 appears as a broad smear characteristic of proteoglycans bearing multiple GAG chains. The triple deletion mutant (TDM) also appears as a smear because it bears GAG chains at serines 210 and 220. Upon heparitinase digestion, all of the syndecans migrate with a similar pattern and as relatively small molecules (Fig. 2B). This indicates that, as previously reported (18), the majority of N-terminal attachment sites on syndecan-1 are substituted with heparan sulfate chains. Otherwise, if substantial chondroitin sulfate was present at each of the three sites, the proteoglycan remaining after heparitinase digestion would show a pattern similar to that observed for intact proteoglycans ( Fig. 2A). The possible exception is the syndecan-1 having a single attachment site at serine 37. This proteoglycan has an intact size very similar to the TDM, suggesting that the serine 37 attachment site may not always bear a GAG chain.
To further confirm successful deletion of GAG attachment sites, we analyzed the levels of cell surface heparan sulfate per syndecan-1 core protein. As expected, with loss of GAG attachment sites, a reduction in the amount of heparan sulfate per core protein occurs relative to wild-type syndecan-1 (Fig. 3A). Even when a single site is deleted (leaving two attachment sites), a dramatic effect on the amount of heparan sulfate per core protein is seen, with all single site deletion mutants having at least a 50% decrease in heparan sulfate per core protein relative to wild-type syndecan-1. When sites at 37 and 45 or sites 37 and 47 are mutated (leaving single attachment sites at position 47 or 45, respectively), the heparan sulfate per core protein ratios are similar to those of mutants having only a single GAG chain deleted. However, when sites at 45 and 47 are absent (mutant with a single attachment site at 37), very low levels of heparan sulfate per core protein are detected.
The finding that heparan sulfate per syndecan-1 core protein is lower on all of the mutants than on wild-type syndecan-1 suggests that the levels of total cell surface heparan sulfate on the mutants may also be substantially lower. However, this is not the case. Loss of a single heparan sulfate attachment site at any of the three positions does not drastically lower the level of cell surface heparan sulfate (Fig. 3B). However, similar to what was seen above for heparan sulfate per core protein, heparan sulfate levels following the deletion of two attachment sites is dependent on the position of the chains deleted. When sites at positions 37 and 45 or positions 37 and 47 are mutated, relatively high levels of heparan sulfate are still present at the cell surface. In contrast, when sites at 45 and 47 are absent, relatively low levels of heparan sulfate are detected (Fig. 3B). Taken together with data in Figs. 2 and 3A, it appears likely that when only a single attachment site at position 37 is present, it is poorly substituted with heparan sulfate chains. TDMbearing cells also express low levels of cell surface heparan sulfate, possibly due to the presence of some heparan sulfate on the syndecan-1 GAG attachment sites proximal to the cell membrane (positions 210 and 220).
Thus, the majority of cells bearing mutant or wild-type syndecan-1 have similar levels of overall cell surface heparan sulfate (Fig. 3B), but the mutants have lower amounts of heparan sulfate per core protein (Fig. 3A). This suggests that mutant syndecan-1 is being expressed at higher levels on the cell surface than wild-type syndecan-1. Analysis of live cells by flow cytometry using an antibody to the syndecan-1 core protein demonstrates that this is indeed the case (Fig. 3C). Interestingly, TDM-bearing cells have strikingly high levels of core protein, indicating that the heparan sulfate composition of syndecan-1 may regulate the level of its core protein on the cell surface.
Deletion of Heparan Sulfate Attachment Sites Diminishes Syndecan-1-mediated Cell Adhesion to Collagen-Cells transfected with wild-type syndecan-1 bind tightly to collagen and therefore do not pellet upon centrifugation (Fig. 4). In contrast, neomycin-and TDM-transfected cells form a tight pellet indicating lack of strong binding to collagen. This demonstrates that syndecan-1 is responsible for cell adhesion to collagen and that the heparan sulfate chains present within the N-terminal region of the core protein are required for this binding. Treatment of cells with heparitinase prior to the assay abolishes cell binding, also indicating that adhesion is mediated through the heparan sulfate chains (data not shown; see Ref. 19). Cells bearing single or double GAG deletions exhibit an intermediate range of adhesion; when one of the three chains is deleted, only a very slight rim of pelleting is present, and when two of the three chains are deleted, a broad ring of pelleted cells is present but much less distinct than the pellet seen with neomycin-or TDM-transfected cells. Analysis of cells bearing the other heparan sulfate attachment mutants indicates that among all three single deletion mutants the extent of cell pelleting was similar and that among all three double deletion mutants the extent of pelleting was similar (data not shown). It is interesting that cells bearing the mutant having a single attachment site at position 37 and the TDM have almost identical amounts of heparan sulfate per core protein and total heparan sulfate on the cell surface, yet the cells differ in their behavior on collagen. This finding underscores the importance of having heparan sulfate chains positioned in the N-terminal region of the core protein. Taken together, these data show that all three heparan sulfate attachment sites are essential for optimal syndecan-1-mediated binding of cells to collagen and that the extent of binding is determined by the number, not the position, of the heparan sulfate chains.
Deletion of Heparan Sulfate Attachment Sites Promotes Cell Invasion-As expected, cells expressing the TDM were highly invasive in collagen gels as compared with cells expressing wild-type syndecan-1, which do not invade (Fig. 5A). This confirms that the heparan sulfate chains at these sites are required for inhibition of invasion and that the chondroitin sulfates at serines 210 and 220 probably do not contribute to this inhibitory activity. Deletion of a single attachment site at either serine 37 or 45 results in only a small (16 and 9%, respectively) increase in cell invasion when compared with wild-type syndecan-1. However, removal of serine 47 results in a significant (35%) increase in invasion, suggesting that this site is the most influential in regard to the anti-invasive effect of syndecan-1. Data from the double deletion mutants also supports this, because cells bearing syndecan-1 with only the serine 47 attachment site were much less invasive (39%) than mutants having a single attachment site at either serine 37 or 45 (82 and 69%, respectively). While these results demonstrate that all three heparan sulfate chains are required for optimal inhibition of cell invasion by syndecan-1, having a heparan sulfate chain attached at serine 47 clearly is the most influential single site on syndecan-1 for regulating this behavior.
Closer analysis of the data in Fig. 5A reveals that the resulting degree of invasion of cells bearing the TDM (105% invasion as a percentage of control) is greater than would be predicted by adding the percentage increase in invasion of the three single deletion mutations (16% ϩ 9% ϩ 35% ϭ 59%). This predicted value is significantly lower than the actual value (p Ͻ 0.05) and supports the conclusion that all three GAG attachment sites are required for optimal syndecan-1 functional activity.
As a control, cells having equal amounts of syndecan-1 core protein were harvested and analyzed for their invasion into gels. Results in Fig. 5B show that the invasive behavior of these cells is similar to that observed with unsorted cells (Fig. 5A), indicating that differences in cell invasion are not due to differences in the levels of syndecan-1 at the cell surface. To determine if the basis for variability in the invasive behavior between the mutants is due to differences in their affinity for type I collagen, we analyzed several mutated proteoglycans by ACE. Wild-type syndecan-1, syndecan-1 with an attachment deletion at position 47, and syndecan-1 with deletions at positions 37 and 47 all have similar affinities for collagen (Fig. 6), although they vary greatly in their ability to inhibit cell invasion (Fig. 5A). Both a low and high affinity population are present in all of the samples, with the high affinity fraction containing approximately 50 -55% of the proteoglycan. Of the mutants examined, all forms bearing at least one heparan sulfate chain show a shift in proteoglycan mobility at a collagen concentration of 40 -160 nM, with the K d falling somewhere in this range. The similarity in affinity among the GAG attachment mutants and wild-type syndecan-1 is consistent with previous work showing that commercial heparin, individual heparan sulfate chains of syndecan-1, and intact syndecan-1 all have similar affinities for type I collagen (15,20). Thus, variability in the invasive behavior of cells bearing the various mutants is not likely to be due to differences in individual proteoglycan affinity for type I collagen.
Deletion of Heparan Sulfate Attachment Sites Diminishes Cell-Cell Adhesion-Expression of wild-type syndecan-1 on ARH-77 cells induces cell-cell adhesion in a heparan sulfate-dependent manner (12). Overall, deletion of a single attachment site decreases the ability of syndecan-1 to mediate cellcell adhesion, deletion of two chains decreases adhesion further, and removal of all three sites abolishes cell-cell adhesion (Fig. 7). Interestingly, over half of the adhesive ability of syndecan-1 was still present when only a single heparan sulfate attachment site was available at position 45 or 47. This is in contrast to the poor adhesion that occurs when only a single attachment site is available at position 37. Comparison of mutant behavior in aggregation assays with the data in Fig. 3B indicates that the extent of syndecan-1-mediated cell-cell adhesion may be determined by the amount of heparan sulfate present at the cell surface.

DISCUSSION
Because the three N-terminal GAG attachment sites of syndecans are highly conserved, a series of mutated syndecans having two, one, or none of these sites were expressed in ARH-77 cells, and their behavior was analyzed in a series of functional assays. Generally, loss of one heparan sulfate chain results in a slight to moderate decrease in cell binding to collagen, inhibition of invasion, and cell-cell adhesion; loss of two chains results in a more substantial decrease in these functions; and elimination of all three chains results in total loss of function (Table I). We conclude that the three N-terminal GAG attachment sites on syndecan-1 are essential for maximal heparan sulfate-mediated biological activity. This correlation provides a functional basis for the evolutionary conservation of these three heparan sulfate attachment sites found throughout the syndecan family of proteoglycans.
As has been demonstrated with syndecan-4 and with syndecan-1 chimeras (4, 10), we find that all three N-terminal GAG  6. Syndecans with one, two, or three heparan sulfate attachment sites have similar affinities for type I collagen. 35 SO 4 -Labeled syndecan-1 was isolated from cell extracts and analyzed by affinity co-electrophoresis for binding to type I collagen within a range of concentrations. At high collagen concentrations, 50 -55% of the radiolabeled material is retained near the top of the well in all the samples (as determined by analysis using a PhosphorImager). This represents the fraction of the proteoglycan binding to collagen with high affinity. The low affinity fraction is present near the bottom of the lanes. attachment sites can bear heparan sulfate chains and that all three sites are essential for ensuring the maximal expression of heparan sulfate per core protein (Fig. 3A). Also, as with both syndecan-1 chimeras and perlecan, our data indicate that coupling occurs across the heparan sulfate sites of wild-type syndecan-1, because the three sites acting together bear more heparan sulfate than predicted by the behavior of each site acting independently (10,21). Our findings lend functional import to this synergistic phenomenon that occurs across GAG attachment sites; i.e. as the number of heparan sulfate attachment sites increases, the overall biological activity of syndecan-1 increases.
In addition, we also find that individual heparan sulfate attachment sites do not influence all functional activities equally. Removal of attachment sites at positions 45 and 47 (leaving a single attachment site at position 37) results in a syndecan-1 with a very low heparan sulfate per core protein ratio and low levels of total cell surface heparan sulfate. This suggests that when attachment sites are absent at both positions 45 and 47, the site at position 37 is poorly substituted with heparan sulfate. This form of syndecan-1 does not inhibit invasion well, nor is it very effective in mediating cell aggregation (Figs. 5A and 7). Thus, the site at position 37 appears to be the least influential in regulating syndecan-1 functional activity. Another striking observation is the apparent role of attachment position 47, which has the greatest impact among the three attachment sites on inhibiting invasion as demonstrated with both single and double heparan sulfate chain deletions (Fig. 5A).
In contrast, adhesion of cells to collagen is not substantially influenced by heparan sulfate chain position; rather, the number of heparan sulfate chains is the most important determining factor. Surprisingly, in regard to cell-cell adhesion, syndecan-1 bearing only a single heparan sulfate chain at position 45 or 47 can promote extensive aggregation. These results on cell adhesion to collagen and cell-cell adhesion together support an important conclusion, that even when relatively low amounts of heparan sulfate-bearing syndecan-1 are present on the cell surface they can have substantial impact on cell adhesion. This is possibly due to mobility of syndecan-1 in the plane of the membrane, which allows focal concentration of the heparan sulfate-bearing molecules at sites where they interact with ligands.
The behavior of cells bearing the TDM supports several important conclusions. First, the presence of heparan sulfate on the N-terminal GAG attachment sites is absolutely required for syndecan-1-mediated cell adhesion to collagen, inhibition of cell invasion, and cell-cell adhesion. Second, chondroitin sulfate on the attachment sites near the plasma membrane does not mediate cell adhesive activity in any of the assays utilized in this study. Third, at least some heparan sulfate may be present on the attachment sites near the plasma membrane (positions 210 and 220). Low levels of heparan sulfate have also been found in chimeric proteoglycans containing these syndecan-1 GAG attachment sites (10). However, we do not know if heparan sulfate is present at these sites on wild-type syndecan-1 transfectants or if substitution of heparan sulfate at these sites is a consequence of removing the three N-terminal attachment sites.
Mutations of GAG sites on syndecans have not as yet been related to any specific pathological state. Therefore, other than providing a functional explanation for the conservation of the three heparan sulfate attachment sites, what have we learned from this work? First, it has been shown both in vivo and in vitro that syndecan-1 exhibits a molecular polymorphism due to differences in the number and size of its heparan sulfate and chondroitin sulfate chains (5). Specifically, syndecan-1 present on simple epithelia has more heparan sulfate chains (two or three chains present) and larger heparan sulfate chains than does syndecan-1 on stratified epithelial cells (only one heparan sulfate chain). Because these cell types have different adhesive requirements (simple epithelia bind tightly to underlying extracellular matrix and under normal conditions are unable to migrate, while stratified epithelia form extensive cell-cell adhesive interactions and often migrate away from the extracellular matrix (e.g. stratified keratinocytes)), it was speculated that the different syndecan forms have distinct functions (5). Results from our present study fully support this notion. In cells expressing syndecan-1 mutants lacking all three heparan sulfate chains, cell migration (invasion) increases and adhesion to extracellular matrix (collagen) decreases relative to cells expressing wild-type syndecan-1. Moreover, when only a single heparan sulfate chain is present on syndecan-1 in our cells (similar to syndecan-1 on stratified epithelia), the cells are invasive and bind extracellular matrix poorly but still exhibit extensive heparan sulfate-dependent cell-cell adhesive interactions. Thus, it is likely that tissue-specific differences in the FIG. 7. Reducing the number of heparan sulfate chains on syndecan-1 weakens cellular aggregation. Cells expressing wildtype or mutated syndecan-1 were assayed for their ability to aggregate using a rotation-mediated aggregation assay (12). Values represent the means Ϯ S.E. (n ϭ 3) of the percentage of cells in aggregates.

TABLE I Summary of mutant syndecan-1 functional activity
Functional activity of syndecan-1 decreases as the number of heparan sulfate attachment sites is reduced, and removal of all three HS attachment sites abolishes syndecan-1 activity. Cell binding to collagen (Fig.  4) was graded visually on a scale ranging from tight cell adhesion (ϩϩϩ) to no adhesion (Ϫ). For invasion and aggregation assays, the mean activity of the transfectants expressing syndecan-1 having either two or one HS attachment site was calculated from data shown in Fig.  5A (invasion) and from Fig. 7 (aggregation). These are shown along with values obtained for cells expressing either wild-type syndecan-1 (three HS attachment sites) or TDM (no HS attachment sites). number of heparan sulfate chains present on syndecan-1 at least in part account for the differing adhesive capacities and behavior of simple and stratified epithelial cells. Second, syndecans exhibit extensive heterogeneity even when expressed within the same cell type as indicated by their broad smearing pattern on SDS-polyacrylamide gel electrophoresis. Studies of L cells transfected with wild-type syndecan-4 show the proteoglycans produced have one, two, or three heparan sulfate chains as well as varying numbers of chondroitin sulfate chains (and some isoforms have only chondroitin sulfate chains) (4). This finding, together with our present results, indicates that within some tissues, or even on a single cell surface, there are probably syndecans that, due to their differing number of heparan sulfate chains, have distinct functional capacities. This dichotomy of function among syndecans could be important, for example, on migrating cells; chondroitin sulfate-containing syndecan could localize to the leading edge of the cell (chondroitin sulfate proteoglycans have been shown to promote cell migration (22,23)) and heparan sulfate-containing syndecan to the trailing edge in order to anchor the cell. Were such a mechanism in place, the relative amount of heparan sulfatesyndecan versus chondroitin sulfate-syndecan could regulate the balance between migration and tight adhesion. Last, when heparan sulfate attachment sites were deleted on glypican, a glycosylphosphatidylinositol-anchored heparan sulfate proteoglycan, it was found that the molecules sorted to different cell surface compartments (i.e. apical, basolateral) based on the number of heparan sulfate chains present on the core protein (24). This finding led the authors to speculate that modulation in the number of heparan sulfate chains controls the subcellular expression and therefore the functional availability of glypican. Thus, modulation of cell behavior via molecular heterogeneity of cell surface proteoglycans may not be restricted solely to the syndecans; rather, it may be a widespread mechanism to facilitate the multifunctional nature of these molecules.