Interactions of Fibrillin-1 with Heparin/Heparan Sulfate, Implications for Microfibrillar Assembly*

Fibrillin-1 is a major constituent of the 10–12 nm extracellular microfibrils. Here we identify, characterize, and localize heparin/heparan sulfate-binding sites in fibrillin-1 and report on the role of such glycosaminoglycans in the assembly of fibrillin-1. By using different binding assays, we localize two calcium-independent heparin-binding sites to the N-terminal (Arg45–Thr450) and C-terminal (Asp1528–Arg2731) domains of fibrillin-1. A calcium-dependent-binding site was localized to the central (Asp1028–Thr1486) region of fibrillin-1. Heparin binding to these sites can be inhibited by a highly sulfated and iduronated form of heparan sulfate but not by chondroitin 4-sulfate, chondroitin 6-sulfate, and dermatan sulfate, demonstrating that the heparin binding regions represent binding domains for heparan sulfate. When heparin or heparan sulfate was added to cultures of skin fibroblasts, the assembly of fibrillin-1 into a microfibrillar network was significantly reduced. Western blot analysis demonstrated that this effect was not due to a reduced amount of fibrillin-1 secreted into the culture medium. Inhibition of the attachment of glycosaminoglycans to core proteins of proteoglycans by β-d-xylosides resulted in a significant reduction of the fibrillin-1 network. These studies suggest that binding of fibrillin-1 to proteoglycan-associated heparan sulfate chains is an important step in the assembly of microfibrils.

The assembly of fibrillins into complex supramolecular tissue microfibrils is a multistep process. Several steps within the continuum of microfibrillar assembly have been described. One of the first steps in the assembly process is the oligomerization of fibrillin-1 into disulfide-bonded multimers, which occurs within a few hours after the secretion of fibrillin-1 from cells (13). It has not been demonstrated whether this process occurs in the extracellular space without the participation of cells or whether it is a cell-mediated process perhaps involving integrin receptors (14 -16) or other cell-surface molecules. In cell culture, a fibrillin-containing network forms over a few days (17). However, it takes several weeks of culture for fibroblasts to produce the typical bead-on-a-string structures, discernible by electron microscopy (18). Tissue microfibrils do not appear as bead-on-a-string structures but as simple thread-like filaments (3). The molecular basis for the morphological differences during the formation of microfibrils is obscure. Controversial models have been proposed for the arrangement of fibrillin-1 within assembled microfibrils, based on cross-linking patterns (19), on the solution structure of epidermal growth factor (EGF) 1 like modules (20), on the ultrastructural morphology of microfibrils (21), on mutation analysis of fibrillin-1 (22), on the localization of monoclonal antibody epitopes (23), or based on findings by atomic force microscopy (24) or electron tomography (25).
Many mutations in the fibrillin-1 protein lead to a heritable autosomal dominant disorder, the Marfan syndrome (Online Mendelian Inheritance in Man number 154700), with major complications in the skeletal, the cardiovascular, and the ocular system (26). The mutant fibrillin-1 is thought to exert a dominant negative effect on the structure or stability of microfibrils that precipitates the defects characteristic to Marfan syndrome, although the precise molecular pathogenetic pathway is not known. Central to these discussions are issues related to the effect of mutant fibrillin-1 molecules on the assembly of microfibrils. Are mutant fibrillin-1 molecules incorporated into microfibrils, where they may destabilize the microfibrils by proteolytic degradation (27)? Alternatively, mutant molecules may disrupt microfibril assembly. Pulse-chase experiments have revealed that many fibroblast strains isolated from patients with Marfan syndrome deposit reduced amounts of fibrillin into the extracellular matrix, suggesting that many mutations in fibrillin-1 impair the ability of the molecules to assemble (28,29).
Here, we report on the occurrence, the identification, the characterization, and the localization of heparin/heparan sulfate-binding sites in fibrillin-1. The interaction of fibrillin-1 with heparin/heparan sulfate has an important role in the assembly of microfibrils.

EXPERIMENTAL PROCEDURES
Recombinant Proteins-Recombinant subdomains of human fibrillin-1 rF6, rF18 (23), rF6H, 2 rF23 (7), and rF45 (27) have been described in detail previously. The expression plasmid for rF6 has been designed to express the entire C-terminal half of fibrillin-1 (position 1487-2871; Ref. 23). However, it has been shown that the C-terminal unique domain of fibrillin-1 is proteolytically processed between position 2731 and 2732 by furin-type proteases (31). Consequently, rF6 spans amino acid residues 1487-2731 of fibrillin-1 and thus is almost identical to recombinant subdomain rF6H (position 1487-2725) 2 except for a hexahistidine tag at the C-terminal end of rF6H. Similarly, the subdomain rF23 has been shown to be processed between positions 44 and 45, resulting in a truncated N-terminal end (7).
To produce a new recombinant subdomain (rF51), spanning calciumbinding (cb) EGF 6 to cbEGF 10 of fibrillin-1, human fibrillin-1 cDNA (32) was amplified by polymerase chain reaction with sense oligonucleotide 5Ј-CGTAGCTAGCAGACATTAACGAGTGTGAAACCC-3Ј and antisense oligonucleotide 5Ј-ACCGCTCGAGCTATTAGTGATGGTGA-TGGTGATGAAGACAGATCCTTCCTGTGGC-3Ј introducing a restriction site for NheI at the 5Ј end and a restriction site for XhoI plus a sequence for a hexahistidine tag and a stop codon at the 3Ј end. The NheI-XhoI restricted 1048-base pair amplification product was ligated into the NheI-XhoI restricted plasmid pDNSP-rF16 2 in frame to the sequence for a signal peptide (23). The correct sequence was verified by DNA sequencing. The resulting plasmid was termed pDNSP-rF51 and was used to recombinantly express the polypeptide rF51 with the amino acid sequence Ala-Pro-Leu-Ala-Asp 613 -Leu 951 (His) 6 . The first four amino acid residues (Ala-Pro-Leu-Ala) result from the cloning strategy. The methods for transfection, selection of stable clones, and production of recombinant medium were described in detail previously (33). Purification of rF51 was performed as described for rF18 (23). Correct folding of rF51 was verified by binding to monoclonal antibody 201 that is dependent on correct disulfide bonds.
Polyclonal Antisera and Monoclonal Antibodies-A polyclonal antiserum (␣-rF6H) was produced according to standard procedures in rabbit using the recombinant C-terminal half of fibrillin-1 rF6H as antigen. 2 The specificity of this antiserum was verified by immunoblotting and enzyme-linked immunosorbent assays with rF6H, fibrillin-1 from cell culture medium, and other matrix proteins. The polyclonal antiserum B9543 was generated against an N-terminal half of fibrillin-1 and was characterized previously (34). The B9543 antiserum as well as monoclonal antibodies 201 and 84 were a generous gift from Prof. Lynn Y. Sakai, Shriners Hospitals for Children, Portland, OR.
Protein-Ligand Binding and Inhibition Assays-Affinity chromatography on heparin-Sepharose columns (HiTrap Heparin HP, 1 ml; Amersham Pharmacia Biotech) equilibrated in 20 mM Tris-HCl, pH 7.4, 50 mM NaCl, and either 2 mM CaCl 2 or 5 mM EDTA were typically performed at room temperature (ϳ20°C) with highly purified recombinant fibrillin-1 subdomains (100 -200 g) applied to the columns in equilibration buffer at a flow rate of 0.1 ml/min. After washing the columns with equilibration buffer, bound material was eluted with a linear NaCl gradient (0.05-1 M NaCl in 20 ml) in the same buffer at a flow rate of 0.5 ml generated by a Gradient Programmer GP-250 Plus and two P-500 pumps (Amersham Pharmacia Biotech). The flowthrough and the eluted volume was continuously fractionated in 0.7-ml aliquots by a Frac-100 collector (Amersham Pharmacia Biotech), whereas the amount of protein in each fraction was monitored at 280 nm using a Ultrospec 3000 spectrophotometer (Amersham Pharmacia Biotech). The NaCl concentrations in each individual fraction were determined by measurement of the conductivity and comparison to standards using a Microprocessor Conductivity Meter (LF3000; WTW, Germany). The correlation of the amount of individual recombinant fibrillin-1 subdomains with the absorbance recorded at 280 nm was verified by SDS-gel electrophoresis and Coomassie Blue staining.
Solid phase binding assays of fibrillin-1 subdomains to heparin were performed on 96-well plates (MaxiSorp; Nalge Nunc International). Since plastic surfaces cannot be coated with soluble heparin or other glycosaminoglycans by adsorption due to their high negative charge, heparin coupled to bovine serum albumin (BSA-heparin; Sigma) was used for coating. Albumin adsorbs readily to plastic surfaces. The wells were incubated for 16 h at 4°C with 100 l of BSA-heparin (20 g/ml) in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl. All subsequent steps were performed at room temperature (ϳ20°C). The wells were washed 3 times with 20 mM Tris-HCl, pH 7.4, 50 mM NaCl containing 0.05% (v/v) Tween 20 and either 2 mM CaCl 2 or 10 mM EDTA (washing buffer). Blocking of nonspecific binding sites was achieved with incubation of the wells for 1-2 h with 100 l of 20 mM Tris-HCl, pH 7.4, 50 mM NaCl, 5% (w/v) non-fat dry milk and either 2 mM CaCl 2 or 10 mM EDTA (binding buffer). The wells were incubated with serial dilutions (1:2; 100 l each) of the fibrillin-1 subdomains starting at 75 g/ml for 2 h, washed 3 times with washing buffer, and incubated for 1 h with 100 l of polyclonal antiserum B9543 (1:250, diluted in binding buffer) for rF18, rF23, rF45, and rF51 or ␣-rF6H (1:250, diluted in binding buffer) for rF6H. After washing 3 times, the wells were incubated with 100 l of goat anti-rabbit horseradish peroxidase conjugate (1:800 diluted in binding buffer; Bio-Rad) for 1 h and washed again. The color reaction of the assay was performed with 1 mg/ml 5-aminosalicylic acid (Sigma) in 20 mM phosphate buffer, pH 6.8, containing 0.045% (v/v) H 2 O 2 (100 l/well) for 3-4 min and stopped by adding 2 M NaOH (100 l/well). Color yields were determined at 490 nm using a Microplate EL310 Autoreader (Bio-Tek Instruments).
For inhibition assays, an identical procedure was employed except the following alterations. The recombinant subdomains were used at fixed concentrations of 20 g/ml for rF6H and rF18, or, depending on the experiment, 20 -30 g/ml rF23 in order to produce an absorption at 490 nm of ϳ0.5-0.8 after 3-4 min of color reaction. The binding and washing buffers used always contained 2 mM CaCl 2 . Heparin, chondroitin 4-sulfate, chondroitin 6-sulfate, dermatan sulfate, and BSAheparin (all from Sigma), as well as heparan sulfate 2 and heparan sulfate 6 were added as inhibitors in serial dilutions (1:3) at concentrations indicated in individual experiments. Heparan sulfate 2 and 6 preparations were kindly provided by Prof. Anders Malmström, University of Lund, Sweden. These preparations were characterized in detail previously as HS2 and HS6, respectively (35).
Cell Culture Experiments-All cell culture experiments were performed with normal human skin fibroblasts grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal calf serum at 37°C in a 5% CO 2 atmosphere.
For inhibition studies with glycosaminoglycans, confluent fibroblasts were trypsinized and seeded at 7.5 ϫ 10 4 cells/well of a 8-well chamber slide (Permanox; Nalge Nunc International) in a total volume of 0.5 ml together with serial dilutions of heparin, heparan sulfate 6, and BSAheparin at concentrations indicated in individual experiments. During the entire cultivation period, no differences in the morphology of the cells or in the cell numbers were observed. After 5 days, the cells were washed with phosphate-buffered saline (PBS), fixed with 70% (v/v) methanol, 30% (v/v) acetone for 5 min, and rehydrated in PBS. After blocking of nonspecific binding sites with normal goat serum (1:10 diluted in PBS; Dako) for 30 min, the wells were incubated with ␣-rF6H antiserum against fibrillin-1 (1:250 diluted in PBS) for 1 h followed by washing three times with PBS. After incubation with a goat anti-rabbit fluorescein conjugate (diluted 1:200 in PBS; Jackson ImmunoResearch), the fibrillin-1 network was analyzed by fluorescence microscopy with a Axioplan2 microscope (Zeiss). Digital images were recorded using a 3CCD color video camera (Sony) and the AxioVision 2.0 software (Zeiss).
For inhibition studies with ␤-D-xylosides, the fibroblasts were first grown to confluency in the presence of 1 mM 4-methylumbelliferyl-␤-Dxylopyranoside (Sigma) and 1 mM p-nitrophenyl ␤-D-xylopyranoside (Sigma). The cells were trypsinized and seeded at 5 ϫ 10 4 cells/well of an 8-well chamber slide in a total volume of 0.5 ml together with 0.125 mM 4-methylumbelliferyl-␤-D-xylopyranoside and 0.125 mM pnitrophenyl ␤-D-xylopyranoside. Since the ␤-D-xylosides were dissolved in dimethyl sulfoxide, the final concentration of dimethyl sulfoxide in the cell culture wells was 0.05% (v/v). To monitor potential toxic effects, the control without ␤-D-xylosides was supplemented with 0.06% (v/v) dimethyl sulfoxide. The formation of the fibrillin-1 network was analyzed as described above.
Detection of Secreted Fibrillin-1 and Fibronectin in Cell Culture-Normal human skin fibroblasts were seeded at 5 ϫ 10 5 cells/3.8-cm 2 plate (12-well plates; Nalge Nunc International) in the presence of 0 -4 mg/ml heparin. The cells were grown confluent for 1 day, washed twice with PBS, and incubated with 2 ml serum-free culture medium including 0 -4 mg/ml heparin for 3 days. In the presence and absence of heparin, no differences in the morphologies of the cells were observed. After harvesting and filtrating (0.22 m pore size) the conditioned medium, the proteins were precipitated from 1-ml aliquots with 10% (w/v) trichloroacetic acid. The non-reduced samples were analyzed by a standard Western blotting procedure using monoclonal antibodies (ϳ10 g/ml) 201 to detect fibrillin-1 and 84 to detect fibronectin. (15) that denatured (6 M guanidine HCl) and reduced fibrillin, isolated from fetal bovine ligamentum nuchae, bound to heparin-Sepharose. Since it was not clear whether this binding represented an authentic binding interaction or whether it was conferred by the reduced and denatured state of fibrillin, binding of native authentic fibrillin-1 to heparin was tested. Conditioned medium produced by skin fibroblasts was incubated with immobilized heparin in a solid phase assay. A specific antiserum was used to detect fibrillin-1, which bound in a dose-dependent manner to heparin (Fig. 1). This result clearly established that authentic non-denatured fibrillin-1 has binding affinity to heparin.

Binding of Fibrillin-1 to Heparin and Identification of Binding Sites-Previously, it has been shown by others
To examine the fibrillin-1/heparin interaction in more detail, overlapping recombinant subdomains of fibrillin-1 were tested in heparin-Sepharose affinity experiments and in solid phase assays with immobilized heparin. An overview of the recombinant subdomains of fibrillin-1 used in this study is shown in Fig. 2. In heparin-Sepharose affinity binding experiments various fibrillin-1 subdomains showed distinctive binding characteristics (Fig. 3). The N-terminal subdomain rF23 (position 45-489) bound to heparin in the presence of calcium and in the presence of EDTA. NaCl concentrations to elute bound rF23 were in the range of 170 -480 mM NaCl in the calcium-containing buffer and ϳ300 -370 mM NaCl in the EDTA-containing buffer. Fragment rF45 (position 451-1027) showed no binding in the presence of calcium and only very minor binding in the presence of EDTA. Fragment rF51 (position 613-951) did not bind under any condition (not shown). The central fragment rF18 (position 910 -1527) bound to heparin-Sepharose in the presence of calcium. However, in the presence of EDTA, binding of rF18 was virtually abolished. About 375-470 mM NaCl displaced rF18 from the heparin-Sepharose under calcium conditions. The C-terminal half rF6 (position 1487-2731) bound to some extent (ϳ34 -35%) in the presence of calcium as well as in the presence of EDTA. About 110 -310 mM NaCl was necessary to displace rF6 from the heparin-Sepharose in calcium-containing buffer and ϳ200 -375 mM NaCl in EDTA-containing buffer. In summary, the NaCl concentrations used to displace the fibrillin-1 subdomains from heparin-Sepharose were above the physiological ionic strength indicating that these interactions potentially occur in tissues.
A solid phase binding assay was used to further confirm these binding data, whereby various recombinant fibrillin-1 subdomains were used in the soluble phase with immobilized heparin (Fig. 4). Saturable binding profiles in the presence of calcium were observed for rF6H, rF18, and rF23, whereas rF45 and rF51 showed no or only very minor binding (Fig. 4A). When the binding tests were performed in the presence of EDTA, the binding of rF18 to heparin was significantly reduced, whereas binding of rF23 and rF6H did remain on similar levels (Fig. 4B).
Taking the overlapping regions of the recombinant polypeptides into account, three regions with heparin binding affinity within the fibrillin-1 molecule have been identified as follows: one calcium-dependent binding region in the center of the molecule (position 1028 -1486) and two calcium-independent binding regions, one at the N-terminal end (position 45-450) and one within the C-terminal half (position 1528 -2731).
Inhibition Studies with Glycosaminoglycans-Although heparin is not a component of extracellular matrices, it initially was used in this study because of its structural similarity with other glycosaminoglycans that can be found in the extracellular matrix. Subsequently, it was tested whether glycosaminoglycans such as various forms of chondroitin sulfates and heparan sulfates can compete with heparin binding of the fibrillin-1 subdomains rF6H, rF18, and rF23 (Fig. 5). Heparin itself inhibited the interaction between all three subdomains with immobilized heparin (Fig. 5A). These data demonstrate that the observed interactions are specific. The amounts of heparin that resulted in 50% inhibition (IC 50 value) were in a range of 770 -1730 g/ml (51-115 M). Chondroitin-4-sulfate (Fig. 5B), dermatan sulfate (Fig. 5C), and chondroitin 6-sulfate (Fig. 5D) had none or very minor inhibitory effects on the interaction of all three recombinant fibrillin-1 subdomains with immobilized heparin. On the other hand, inhibition with the highly sulfated and iduronated heparan sulfate 6 resulted in inhibition of the heparin interaction with all three fibrillin-1 subdomains (Fig. 5E). The IC 50 values for heparan sulfate 6 were in a range between 235 and 720 g/ml (12-36 M), slightly lower than what has been observed with heparin as inhibitor. The high sulfation and/or iduronation pattern of heparan sulfate 6 was necessary for binding since the lower sulfated and iduronated heparan sulfate 2 did not have any inhibitory activity (Fig. 5F). These experiments demonstrate that the identified heparin binding regions in fibrillin-1 represent unique binding sites for a highly sulfated form of heparan sulfate but not for various forms of chondroitin sulfates. However, these results do not exclude the possibility that binding sites other than those for heparin/heparan sulfate exist in the fibrillin-1 molecule for binding to chondroitin sulfates.
It is known that clustering of binding sites within a narrow physical region can enhance binding strengths by several magnitudes (36). Glycosaminoglycans attached to the protein core of proteoglycans are almost always clustered. To mimic the clustered glycosaminoglycan pattern on proteoglycans, we used heparin coupled to bovine serum albumin (ϳ4 -5 heparin chains per molecule albumin) in inhibition assays instead of soluble heparin (Fig. 6). For the interactions between the recombinant fragments rF6H, rF18, rF23 and heparin, the IC 50 values of clustered heparin (BSA-heparin) were in the range of 0.07-0.77 g/ml (2-26 nM heparin), which is 4400 -25,500-fold less as compared with non-clustered soluble heparin (Fig. 5A  and Fig. 6). These data demonstrate that the affinity of heparin significantly increases when the molecules are presented as clusters to the fibrillin-1 subdomains, as compared with nonclustered heparin molecules. It further suggests that potential binding of fibrillin-1 to glycosaminoglycan chains of proteoglycans are of high affinity.
Inhibition of Fibrillin-1 Assembly by Glycosaminoglycans-In order to study potential effects of glycosaminoglycans on fibrillin-1 assembly in cell culture, the culture medium of skin fibroblasts was supplemented with various concentrations of either soluble heparin (Fig. 7, A-D) or with heparan sulfate 6 ( Fig. 7, E-H). After 5 days, fibrillin-1 assembly was evaluated by immunofluorescence with a specific fibrillin-1 antiserum. Addition of increasing concentrations of heparin (0 -0.5 mg/ml) or heparan sulfate 6 (0 -0.23 mg/ml) resulted in a dose-dependent reduction of the fibrillin-1 network. The network was re-  Fig. 7D) heparin or 0.23 mg/ml (11.5 M; Fig. 7H) heparan sulfate 6. These inhibiting concentrations of heparin and heparan sulfate 6 correlated well with the concentrations used in solid phase inhibition assays (Fig. 5, A  and E). The observed reduction of the fibrillin-1 network could be either due to a disturbed assembly mechanism or alternatively to epitope masking by bound heparin or heparan sulfate to fibrillin-1. Therefore, similar inhibition experiments were performed, and a variety of monoclonal antibodies and polyclonal antisera against different regions of the fibrillin-1 molecule were used to visualize the fibrillin-1 network (data not shown). For all antibodies and antisera used, the results were identical to the one described above, clearly demonstrating that the observed effect was not due to epitope masking. When clustered heparin (BSA-heparin) was used to inhibit fibrillin-1 network formation, then a much higher inhibitory effect was observed (Fig. 8). The presence of 3.1 g/ml (ϳ92 pM heparin) BSA-heparin resulted in about 90% inhibition of the fibrillin-1 network (Fig. 8B), whereas 12.5 g/ml (ϳ372 pM heparin) BSAheparin almost completely abolished the fibrillin-1 network (Fig. 8C). Again, the inhibitory potency of BSA-heparin correlated well with inhibition experiments using a solid phase assay (Fig. 6).
One possibility how heparin or heparan sulfate could inhibit fibrillin-1 assembly in cell culture would be by inhibiting secretion of fibrillin-1 through an unknown mechanism. To test for this possibility, skin fibroblasts were cultured in the presence of increasing concentrations of heparin (0 -4 mg/ml), and the conditioned medium was then analyzed by immunoblotting for the presence of fibrillin-1 in the culture medium (Fig. 9). Even in the presence of high concentrations (4 mg/ml) of heparin, the amount of secreted fibrillin-1 was identical as compared with the control (Fig. 9A). Also for another heparin binding extracellular matrix protein, fibronectin, no changes in secretion have been observed in conditioned medium including heparin (Fig. 9B). This result clearly suggests that the inhibi-tory mechanism observed in the cell culture assembly assay must be due to binding of heparin or heparan sulfate to fibrillin-1 molecules.
Based on the experiments described, we hypothesized that interaction of fibrillin-1 with heparan sulfate containing proteoglycans are necessary to mediate or nucleate fibrillin-1 assembly. Accordingly, in cell culture the added heparin, heparan sulfate, or BSA-heparin would compete with binding of fibrillin-1 to these heparan sulfate chains and would thus result in a loss of the fibrillin-1 network. We set out to test this hypothesis by cultivating skin fibroblasts in the presence of ␤-D-xylosides, which are chemical analogues of xylose, the first monosaccharide in glycosaminoglycan biosynthesis that is covalently attached to a serine residue in the core protein of proteoglycans. Thus, ␤-D-xylosides reduce or abolish the attachment of glycosaminoglycans to proteoglycans by competing with the authentic substrates. In the presence of 0.125 mM 4-methylumbelliferyl-␤-D-xylopyranoside and 0.125 mM p-nitrophenyl-␤-D-xylopyranoside, the fibrillin-1 network produced by skin fibroblast was almost completely abolished (Fig. 10). These experiments suggest that glycosaminoglycans of proteoglycans are involved in fibrillin-1 assembly. DISCUSSION Microfibrils, 10 -12 nm in diameter, are supramolecular aggregates in the extracellular matrix with fibrillin-1 as a major backbone protein. The complete composition of microfibrils as well as the molecular mechanism of fibrillin-1 assembly from monomers into complex multimeric structures are not known. Here we identify heparin/heparan sulfate as a binding ligand of fibrillin-1 and demonstrate that this interaction plays an important role in the assembly of fibrillin-1.
Previously, it has been shown by others (15) that denatured and reduced fibrillin binds to heparin. It was not clear, however, whether this binding was conferred by the denatured state of the fibrillin molecules or whether it is a property of native fibrillin. In this study, we demonstrated that authentic fibrillin-1 produced by skin fibroblasts indeed binds to heparin. By using overlapping recombinant fibrillin-1 polypeptides, three heparin binding regions have been identified. Two calcium-independent binding regions were found at either the N-terminal end between position 45 and 450 or in the Cterminal half between position 1528 and 2731 of the fibrillin-1 molecule, whereas a third calcium-dependent binding region was identified in the longest stretch of cbEGF modules in the center of the molecule between positions 1028 and 1486. The binding of heparin to proteins is often mediated through clusters of basic residues often composed by lysine or arginine residues (37). A cluster of basic amino acid residues (Gly-Lys-Lys-Gly-Lys-Thr) is located between positions 1313 and 1318 in the last loop of the cbEGF module 17. By analogy to the threedimensional structure of cbEGF module 32 (20), this sequence is expected to be prominently exposed on the surface of the molecule where it would be available for the interaction with heparin. Such a binding site could explain heparin binding to the central region of the fibrillin-1 molecule. Removal of calcium ions from the binding buffer resulted in a significant reduction of bound heparin to rF18. It has been demonstrated that the cbEGF modules in fibrillin-1 (23,38,39) as well as heparin (40) bind calcium. Thus, it is possible that this interaction is mediated partially through calcium ions bound to both cbEGF module(s) and heparin. Basic clusters can be composed by residues that are farther apart within the linear amino acid sequence of the polypeptide. For example heparin binding to the fibronectin module III-13 requires six basic discontinuous residues to form a cationic cradle (41). In the absence of threedimensional structural information, it is thus often not feasible to predict the exact location of heparin-binding sites.  6. Inhibition of the fibrillin-1 heparin interaction with clustered heparin in a solid phase binding assay. Recombinant fibrillin-1 polypeptides were used as soluble ligands at constant concentrations of 20 g/ml rF6H (squares), 20 g/ml rF18 (circles), and 20 g/ml rF23 (triangles) with immobilized heparin. Heparin covalently coupled to bovine serum albumin (4 -5 heparin molecules/molecule albumin) was used as an inhibitor at concentrations indicated. The range of BSA-heparin concentration resulting in 50% inhibition for all three polypeptides is marked with a gray bar. Binding is indicated in percent of binding without inhibitor.
Binding of the recombinant fibrillin-1 fragments to heparin could be inhibited by heparan sulfate but not by chondroitin 4-sulfate, dermatan sulfate, or chondroitin 6-sulfate. These results clearly established that the heparin-binding sites in fibrillin-1 represent heparan sulfate-binding sites. The high sulfated and iduronated heparan sulfate 6 showed inhibitory activity, whereas the low sulfated and iduronated heparan sulfate 2 did not inhibit the fibrillin-1-heparin interaction. Heparan sulfate is composed of a repeated disaccharide structure (-4-glucuronic acid-␤1-4 N-acetyl glucosamine-␣1-) n , which is modified to various degrees by N-deacetylation and N-sulfation of the hexosamine residue, by C 5 -epimerization of D-glucuronic acid to L-iduronic acid, and by additional O-sulfa-tion on both sugars. It has been demonstrated that heparan sulfate 2 contains about 30% primarily non-sulfated iduronic acid, whereas heparan sulfate 6 contains about 65% almost completely 2-O-sulfated iduronic acid (35). Consequently, the heparan sulfate-binding sites in fibrillin-1 have a selective specificity for sulfated, L-iduronate-rich heparan sulfate. Typically, individual heparan sulfate chains are organized in clustered regions of low and high sulfation, whereby the precise patterns of those clusters are largely unknown (for review see Ref. 42). It may be possible that binding of fibrillin-1 molecules along heparan sulfate chains directs the molecules in a proper alignment for the formation of dimers and multimers, the first steps in fibrillin assembly (13) (see also below). It has been demonstrated by degradation experiments with chondroitin ABC lyase that chondroitin sulfate proteoglycans are associated with fibrillin and microfibrils (10,11). Furthermore, immunoprecipitation studies suggested that the chondroitin sulfate containing proteoglycan decorin interacts with fibrillin-1 (12). It may be possible that interactions of such proteoglycans with fibrillin are mediated through chondroitin sulfate chains, which would require proper binding sites on the fibrillin-1 molecule. In the experimental set shown here, we did not analyze binding of fibrillin-1 to chondroitin 4-sulfate, dermatan sulfate, or chondroitin 6-sulfate and therefore cannot exclude that these glycosaminoglycans have binding sites in fibrillin-1 different from those for heparin/heparan sulfate. Based on the experiments described, we hypothesize that the observed interaction of fibrillin-1 with heparin/heparan sulfate reflects interactions of fibrillin-1 with heparan sulfate-associated proteoglycans in the extracellular matrix or on the surface of cells. Glycosaminoglycan chains on proteoglycans are frequently clustered in close proximity to each other. In order to mimic this structural property, we have used "clustered" heparin on albumin in inhibition experiments. Clustered heparin had a several thousand-fold higher inhibitory capacity on the fibrillin-1/heparin interaction indicating a much higher affinity as compared with soluble heparin. These data suggest that potential binding to clustered glycosaminoglycan chains on proteoglycans might be of high affinity. It has been demonstrated for other macromolecules that clustering or oligomerization of binding epitopes often increases binding strengths as compared with the monomers (36).
The glycosaminoglycans heparin and heparan sulfate 6 were used to test their influence on the ability of fibrillin-1 to assemble in cell culture. Both glycosaminoglycans completely inhibited the assembly at concentrations comparable to those used in the in vitro binding assays, whereas no effect was observed in the amount of fibrillin-1 secreted from the cells. With clustered heparin the same inhibitory effects were achieved at much lower concentrations, again correlating well with the inhibitory potency shown by in vitro experiments. These results can be interpreted in two ways. (i) The inhibiting glycosaminoglycans bind to sites on the fibrillin-1 molecules that are identical with or close to important assembly epitopes but are functionally not related to the assembly process. In this instance, the glycosaminoglycans would inhibit assembly by steric interference between ligands important for assembly. If this is true, then we can deduce important information from the presented data about the location of assembly epitopes. (ii) Binding of fibrillin-1 to heparan sulfate chains is a prerequisite for assembly. In this case, the glycosaminoglycans used in the inhibition experiments prevent binding of fibrillin-1 to these heparan sulfate chains, and the assembly process would not be initiated. Several lines of evidence exemplified below point to the second possibility. The assembly of fibrillin in cell culture can be disrupted by chlorate treatment which prevents sulfation of glycosaminoglycans and proteins (12). However, from those experiments it is not possible to distinguish whether sulfation of glycosaminoglycans or of proteins such as fibrillins and microfibril associated glycoprotein-1 are critical for the assembly. To shed light on this question, we have treated fibroblasts with ␤-D-xyloside derivatives, which reduce the amount of glycosaminoglycan chains in proteoglycans. Since ␤-D-xylosides are typically not completely specific in inhibition of just one type of glycosaminoglycan chain (43)(44)(45), a mixture of 4-methylumbelliferyl-␤-D-xylopyranoside and p-nitrophenyl-␤-D-xylopyranoside was used to obtain a maximum reduction of glycosaminoglycan chains on core proteins. Through this treatment, heparan sulfate and chondroitin sulfate are not biosynthesized on the protein cores of proteoglycans (45,46). The incorporation of fibrillin-1 into an extracellular network was significantly reduced in ␤-D-xyloside-treated fibroblasts, clearly indicating that glycosaminoglycans attached to the core protein of proteoglycans are involved in the microfibrillar assembly process. Certainly, detailed studies with more specific inhibition of glycosaminoglycans, for example by degradation with heparan sulfate-degrading enzymes, will be required. It is not clear at this stage whether proteoglycans secreted into the extracellular space or cell membrane-associated proteoglycans are necessary for microfibrillar assembly. Data are accumulating that the assembly of other extracellular proteins such as fibronectin, laminin, and thrombospondin are also dependent on the interaction with glycosaminoglycan chains of proteoglycans (43,46). Candidate proteoglycans located on the cell surface are members of the syndecan or the glypican families (for review see Ref. 47). In fact, for fibronectin and laminin it has been shown that syndecan-2 plays an important function in the assembly process (48). Binding of fibrillin-1 to cell-surface proteoglycan(s) could promote several intriguing functions in its assembly process. (i) It is possible that the initiation of fibrillin-1 assembly requires a high local concentration, which would be achieved by binding to glycosaminoglycan chains of proteoglycans. (ii) Since fibrillin-1 only binds to highly sulfated and iduronated regions within a glycosaminoglycan chain, the patterns of high and low sulfated regions could determine a spatial arrangement of fibrillin-1 necessary to facilitate fibrillin-1 selfinteractions or for disulfide bond formation, which is known as one of the initial steps in fibrillin-1 assembly (13). (iii) Binding of fibrillin-1 to glycosaminoglycans potentially confers conformational changes to the fibrillin-1 protein necessary to expose epitopes for assembly.
Relatively fast on and off rates for protein binding to heparan sulfate chains are ideal to support the proposed functions (30). For example the glycosaminoglycan chains could provide surfaces upon which fibrillin-1 molecules quickly find each other in order to concentrate, to align in the proper register, and to change its conformation. Once the supported step in the assembly process has been "catalyzed," the fibrillin-1 molecules or multimers could be released immediately into the extracellular matrix.