Identification and kinetics analysis of a novel heparin-binding site (KEDK) in human tenascin-C.

The interaction between tenascin-C (TN-C), a multi-subunit extracellular matrix protein, and heparin was examined using a surface plasmon resonance-based technique on a Biacore system. The aims of the present study were to examine the affinity of fibronectin type III repeats of TN-C fragments (TNIII) for heparin, to investigate the role of the TNIII4 domains in the binding of TN-C to heparin, and to delineate a sequence of amino acids within the TNIII4 domain, which mediates cooperative heparin binding. At a physiological salt concentration, and pH 7.4, TNIII3-5 binds to heparin with high affinity (K(D) = 30 nm). However, a major heparin-binding site in TNIII5 produces a modest affinity binding at a K(D) near 4 microm, and a second site in TNIII4 enhances the binding by several orders of magnitude, although it was far too weak to produce an observable binding of TNIII4 by itself. Moreover, mutagenesis of the KEDK sequence in the TNIII4 domain resulted in the significant reduction of heparin-binding affinity. In addition, residues in the KEDK sequences are conserved in TN-C throughout mammalian evolution. Thus the structure-based sequence alignment, mutagenesis, and sequence conservation data together reveal a KEDK sequence in TNIII4 suggestive of a minor heparin-binding site. Finally, we demonstrate that TNIII4 contains binding sites for heparin sulfate proteoglycan and enhances the heparin sulfate proteoglycan-dependent human gingival fibroblast adhesion to TNIII5, thus providing the biological significance of heparin-binding site of TNIII4. These results suggest that the heparin-binding sites may traverse TNIII4-5 and thus require KEDK in TNIII4 for optimal heparin-binding.

The interaction between tenascin-C (TN-C), a multisubunit extracellular matrix protein, and heparin was examined using a surface plasmon resonance-based technique on a Biacore system. The aims of the present study were to examine the affinity of fibronectin type III repeats of TN-C fragments (TNIII) for heparin, to investigate the role of the TNIII4 domains in the binding of TN-C to heparin, and to delineate a sequence of amino acids within the TNIII4 domain, which mediates cooperative heparin binding. At a physiological salt concentration, and pH 7.4, TNIII3-5 binds to heparin with high affinity (K D ‫؍‬ 30 nM). However, a major heparin-binding site in TNIII5 produces a modest affinity binding at a K D near 4 M, and a second site in TNIII4 enhances the binding by several orders of magnitude, although it was far too weak to produce an observable binding of TNIII4 by itself. Moreover, mutagenesis of the KEDK sequence in the TNIII4 domain resulted in the significant reduction of heparin-binding affinity. In addition, residues in the KEDK sequences are conserved in TN-C throughout mammalian evolution. Thus the structure-based sequence alignment, mutagenesis, and sequence conservation data together reveal a KEDK sequence in TNIII4 suggestive of a minor heparin-binding site. Finally, we demonstrate that TNIII4 contains binding sites for heparin sulfate proteoglycan and enhances the heparin sulfate proteoglycan-dependent human gingival fibroblast adhesion to TNIII5, thus providing the biological significance of heparin-binding site of TNIII4. These results suggest that the heparin-binding sites may traverse TNIII4 -5 and thus require KEDK in TNIII4 for optimal heparin-binding.
Tenascin-C (TN-C) 1 is a large extracellular matrix glycoprotein composed of a linear arrangement of domains homologous to epidermal growth factor, fibronectin type III domains (FNIII), and fibrinogen (1,2). TN-C has an oligomeric structure made up of six identical subunits in a radially arranged sixarmed structure called a hexabrachion (3). Each subunit exists as a string of small globular domains, including 8 -15 FNIII domains (4,5). FNIII is found in a wide variety of proteins including cell surface receptors and cell adhesion molecules (6). The core structure of the FNIII repeats is conserved forming a ␤-sandwich with four ␤-strands on one side and three on the other (7,8). The third FNIII repeat of TN-C (TNIII3), which contains an Arg-Gly-Asp (RGD) tripeptide sequence, has been shown to interact with cells via a variety of integrins, including ␣v␤3, ␣v␤6, ␣9␤1, and ␣2␤1 (9 -14). In addition, TN-C binds to a number of other cell surface proteins such as CTB-proteoglycan (15), neurocan, phosphacan (16), and syndecan-1 (17). Syndecan-1 and TN-C are co-expressed transiently in the organ primordial at the interface between mesenchyme and epithelium (18,19). Syndecan-1 binding to TN-C depends on the heparan sulfate side chains of the proteoglycan (17).
Heparin/heparan sulfate and heparan sulfate proteoglycans (HSPGs) are a group of macromolecules that are characterized by long unique carbohydrate chains attached to a protein core (20,21). They exist as components of the extracellular matrix or as membrane-associated proteins and participate in diverse biochemical and physiological processes via interactions with a variety of proteins including extracellular matrix proteins (22)(23)(24). Heparin inhibits the adhesion of peripheral neurites to TN-C in a culture (25,26). To understand TN-C function, it is important to identify the specific domains of the protein responsible for binding to glycosaminoglycans. Two heparinbinding sites have been identified in TN-C (27,28), one comprising the fifth FNIII repeat (TNIII5) and the other comprising the COOH-terminal fibrinogen-like domain.
Heparin/heparin sulfate binding seems to be a common feature of extracellular matrix molecules, including TN-C. TN-C binds to heparin at physiological salt concentrations, although the affinity of TN-C binding to heparin has not been demonstrated specifically until now (29). The present study was undertaken to examine more specifically the interaction of TN-C with heparin using surface plasmon resonance (SPR) and to delineate the putative heparin-binding site that mediates this interaction. Until today, the characterization of binding sites on heparin that interact with the FNIII domains has largely focused on the TNIII5 domain, although it is not clear whether or not these are the only motifs required for heparin interaction. In this study we investigate the specific role of the TNIII4 domain in heparin binding by characterizing wild-type and mutant TNIII proteins through the use of SPR direct binding studies. Here we present evidences suggesting that the secondary heparin-binding site in TNIII4 contributes to overall TNIII binding to heparin.

EXPERIMENTAL PROCEDURES
Materials-Biotinylated heparin (BiHep) was from Sigma. Biosensor BiacoreX and sensor chip CM5 (disposable sensor surface covered with a carboxymethyl dextran layer for protein immobilization) were from Biacore (Uppsala, Sweden). HBS buffer (10 mM Hepes, pH 7.4, contain-* This work was supported by the Korea Science and Engineering Foundation through the Intellectual Biointerface Engineering Center (IBEC) at Seoul National University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Immobilization of Heparin-In these analyses, streptavidin was coupled to a CM5 sensor chip using the amine coupling kit (Biacore) with the procedures specified by the manufacturer. There are two flow cells/ sensor chip for the BiacoreX system. The streptavidin was immobilized in the first flow cell (FC-1) to 20,000 resonance units. Bovine serum albumin was immobilized in the second flow cell (FC-2) to 950 resonance units as a negative control for background correction. BiHep was immobilized to the streptavidin-coupled CM5 sensor chip according to supplier's instructions. The flow cell with the sensor surface was kept at 25°C.
Biacore Binding Analysis-Evaluation of kinetic parameters (k a and k d ) was performed by using a SPR-based BiacoreX biosensor. To determine the binding kinetics of each TN-C proteins, a series of samples with varied concentrations was injected during the association phase for 5 min (40 l/min). The dissociation phase was carried out over a period of 10 min. The flow cell could be regenerated with a 3-min injection of 1.0 M NaCl, which removed the TN-C proteins while leaving the heparin intact. The kinetic analysis of sensograms from the interaction of various analytes with the immobilized BiHep was based on the rate equation, dR/dt ϭ k a CR max Ϫ (k a C ϩ k d )R t , where dR/dt is the rate of change of the SPR signal (resonance units) due to analyte interaction with immobilized BiHep at time t seconds, k a and k d are the associationand dissociation-rate constants, respectively, C is the concentration of analyte and R max is the maximum analyte binding capacity to BiHep in resonance units. Both association-and dissociation-rate constants were determined from the analysis of sensograms using the Biaevaluation software, version 3.0. All binding curves were corrected for background and bulk refractive index contribution by subtraction of the reference flow cells. Models were fitted both globally across the data sets and for a single concentration.

Construction of Expression Vectors and Site-directed Mutagenesis-
Human TN-C cDNAs were amplified from adult human cDNA by employing standard PCR amplification methods. PCR primers were designed to recognize type III modules (TNIII3-5), TN3 forward primers 5Ј-AGCCATATGGCTGTTCCTCCTCCCACTGACCTG-3Ј, TN3 reverse primers 5Ј-GCCGAATTCATGTTGTGAAGGTCTC-3Ј, TN4 reverse primers 5Ј-GGAATTCCTCACTTGGGCGTGTCCAAC-3Ј, and TN5 reverse primers 5Ј-TCGAATTCTATGTTCGGTAATTAAT-3Ј. PCR-amplified cDNA products were digested with KpnI and EcoRI, separated using a PCR Purification kit (Qiagen, Chatsworth, CA), and ligated into the pBADHisA expression vector (Invitrogen). Site-directed mutagenesis was performed using the QuikChange kit method (Stratagene, La Jolla, CA). The following TN-C residues were individually substituted with glutamine: Lys 751 , Lys 754 , Lys 1057 , and Arg 1059 . All expression constructions and mutations were verified by DNA sequencing. The recombinant fusion proteins containing a poly-His tag were expressed and purified using a Ni 2ϩ affinity column under denaturing conditions according to the manufacturer's protocol (Invitrogen).
Cell Adhesion Assay-Human gingival fibroblasts (HGFs) were cultured in ␣-minimal essential medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Invitrogen). HGFs were serum-starved for 24 h, harvested by 0.02% trypsin, 1 mM EDTA, resuspended in ␣-minimal essential medium, washed three times with ␣-minimal essential medium containing 100 g/ml soybean trypsin inhibitor and 1% bovine serum albumin, and plated at 5 ϫ 10 4 cells/well in ␣-minimal essential medium. Twenty-four-well plates were coated with recombinant proteins overnight at 4°C and then blocked with 1% (w/v) bovine serum albumin in phosphate-buffered saline for 30 min and rinsed with phosphate-buffered saline. Suspensions of control cells were maintained in bovine serum albumin-coated plates for the indicated time. After 60 min at 37°C adherent cells were washed twice with phosphate-buffered saline, fixed with 3% paraformaldehyde (Sigma), and stained with 0.25% (w/v) crystal violet (Sigma) in 2% (v/v) ethanol/water. After extensive washings with distilled water, plates were allowed to dry. The absorbance was read and measured at 570 nm, and nonspecific adhesion was determined in wells coated with 1% bovine serum albumin as negative control.

RESULTS
Kinetic Study of TNIII Domains-To assess the respective participation of the various reported TN-C heparin-binding domains in its interaction with heparin, we prepared a series of recombinant fragments of human TN-C spanning FNIII (TNIII) domains expressed in bacteria. The recombinant TN-C fragments were expressed as His-tagged fusion proteins at the COOH terminus to provide convenient purification. In addition, five neighboring amino acids were included to facilitate protein folding. Recombinant TNIII domains were purified near homogeneity with metal affinity resin as determined by SDS-PAGE analysis (Fig. 1).
Direct binding studies using SPR-based Biacore technology were performed. In TN-C, TNIII domain 5 was identified as a major heparin-binding region using isolated recombinant domains described previously (28). In agreement with the previous results, our direct binding studies using SPR-based Biacore technology clearly demonstrated that the TNIII5 was the major heparin-binding domain. The k d for recombinant TNIII3-5 and TNIII4 -5 exhibited a very high affinity for heparin under physiological conditions (30 and 40 nM for TNIII3-5 and TNIII4 -5, respectively) using BiHep immobilized to the streptavidin-coupled CM5 sensor chip (Fig. 2). As shown in Table I, both TNIII3-5 and TNIII4 -5 exhibited a fast association rate with heparin (k a TNIII3-5, 8.64 ϫ 10 3 ϫ M Ϫ1 s Ϫ1 ; k a TNIII4 -5, 4.46 ϫ 10 3 ϫ M Ϫ1 s Ϫ1 ) biosensor surfaces. The resultant complexes between TN-C-heparin were stable as illustrated by slow dissociation rates (k d TNIII3-5, 0.29 ϫ 10 Ϫ3 ϫ s Ϫ1 ; k d TNIII4 -5, 0.2 ϫ 10 Ϫ3 ϫ s Ϫ1 ).
Interestingly, TNIII5 exhibited a decreased affinity (K D ϭ 4.25 M) for heparin compared with TNIII3-5 and TNIII4 -5 (160-and 106-fold, respectively). However, both TN3 and TN4 bind heparin biosensor surfaces so poorly that association and dissociation kinetics could not be quantitated (Table I). An important new finding in the present study is that the heparin binding of an expression protein containing only the TNIII5 domain was always much weaker than the binding of TNIII3-5 or TNIII4 -5, suggesting that the TNIII4 domain contributes significantly to complete heparin binding. How can the TNIII4 domain enhance the binding for heparin? The most likely explanation lies in the nature of cooperative binding. TNIII5 containing the primary heparin-binding site alone produces a modest affinity binding at a K D near 4 M; a second site in TNIII4 could enhance the binding by several orders of magnitude, even if it were far too weak to produce an observable binding of TNIII4 by itself.
Sequence Alignment and Analysis-NMR and crystal structures reveal that the RGD sequence of FIII10 resides in a loop extending from the domain surface (30,31). The structurebased sequence alignment indicates that the TNIII3-5 domains share the same overall structure, with functional fragments occupying surface loops (Fig. 3A). It is notable that the loop regions of the alignment most often contain hydrophilic amino acids, because they are positioned on the surface of the protein. The sequence KGRHKSKPARVK in TNIII5 has been implicated in heparin binding activity previously (28). These six positively charged residues (contributed from a loop between F and G strands) project into the solvent. Thus, we focused on the loop regions in TNIII4 for a potential role in binding to heparin. With the above information taken collectively, the KEDK (amino acids 751-754) sequence possesses a predominantly positive surface charge that makes its relevance as a potential heparin-binding site very likely. Moreover, the KEDK loop on the fourth domain of TN-C has a similar high sequence conservation between different species ranging from chick, mouse, pig, and human (Fig. 3B). To further investigate the heparin binding activity of the KEDK sequence, we decided to produce recombinant TNIII3-5 proteins bearing double mutations replacing basic residues Lys 751 and Lys 754 by Gln.
KEDK Residues in TNIII4 Contribute to Heparin Binding Activity of TNIII5-The heparin binding activity of TNIII4 -5 mutants was assessed by SPR-based Biacore technology. The heparin binding activity of the TNIII4 -5 mutants in the puta-tive heparin-binding sites was compared with the TNIII4 -5 wild-type protein. The results from SPR-based Biacore are summarized in Table II. The K D for recombinant TNIII4 -5 K751Q/K754Q exhibited a reduced affinity (K D ϭ 29.2 M) for heparin compared with TNIII4 -5 (70-fold). As expected, mutations of the KGR sequence in the TNIII5 domain, TNIII4 -5 K1057Q and TNIII4 -5 R1059Q, result in complete inhibition of heparin binding (Table II). Based on these data, we conclude that the TNIII5 domain is the primary heparin-binding site, and residues at Arg 1059 and Lys 1057 in this domain are of critical importance in the interaction. The fact that mutations of the KEDK sequence in the TNIII4 domain reduced affinity for heparin strongly indicates that the complete heparin binding in TN-C relies on secondary heparin binding formed by KEDK residues. These results suggest that the KEDK residues in the loop region between F and G strands contribute to an optimal heparin binding activity of TN-C, and heparin-binding sites in TN-C may span both of the TNIII4 and TNIII5 domains.
Role of Heparin-binding Domain of TNIII4 in Cell Adhesion-To investigate the biological significance of the heparinbinding domain of TNIII4, we explored whether a heparinbinding property was relevant to the process of cell adhesion. HGFs were plated on dishes coated with a series of TNIII fragments. When HGF cell adhesion to a TNIII fragmentcoated dish was quantitated by the crystal violet staining of fixed cells, TNIII4 -5 exhibited significantly higher adhesive activity than TNIII5. Notably, the cell binding to TNIII4 -5 K751Q/K754Q was significantly reduced. Moreover, the adhesion of HGF to TNIII4 -5 was completely abrogated by the presence of 5 g/ml heparin in the plating medium. These results suggest that prior occupancy of the heparin-binding sites in TNIII4 -5 by soluble heparin might cause interference with its ability to bind to cell surface HSPGs. To test this possibility, cell surface HSPGs were removed from HGF by treatment of the cells with heparinase I, an enzyme that acts on highly sulfated HSPGs (32). Heparinase I digestion of HGF cell surfaces reduced the number of TNIII4 -5-binding sites by 75%. In control experiments, treatment of HGF with chondroitinase ABC failed to inhibit a TNIII4 -5-mediated adhesion of HGF showing that cell surface heparan sulfates, but not chondroitin sulfates, contribute to the cell-adhesive properties of TNIII4 -5 (Fig. 4). Taken together, these results support the conclusion that TNIII4 contains binding sites for HSPGs and  2. Biacore analysis of TNIII3-5, TNIII4 -5, and TNIII5 binding to immobilized heparin. Biosensograms are shown for the interaction of TNIII3-5 (A), TNIII4 -5 (B), and TNIII5 (C) with immobilized heparin as described under "Experimental Procedures." The apparent equilibrium dissociation constants estimated from these data are summarized in Table I. enhances the adhesive activity of TNIII5 to HSPGs through a cooperative mechanism.

DISCUSSION
In the present study we used SPR-based Biacore technology to define the influence of the KEDK sequence of the TNIII4 domain upon the interaction of TN-C with heparin. In agreement with a previous finding, we observed that the TNIII5 fragment containing a major heparin-binding site binds to heparin under conditions of physiological pH and ionic strength. However, the affinity of TNIII5 alone for heparin is significantly lower than that of fragments containing both TNIII4 and TNIII5 domains. Therefore, the presence of the TNIII4 domain may enhance the heparin-binding affinity of TNIII5 by some cooperative mechanism. The simplest model would postulate two binding sites, one in TNIII4 and another in TNIII5, which can bind simultaneously to heparin. It is of interest to note that the cooperative contribution of the adjacent FNIII repeat(s) is often found in maximum heparin binding. A similar example of this cooperative heparin-binding with the adjacent FNIII repeat can be seen with the primary heparin-binding site in the 13th type III repeat of fibronectin (FNIII13) and a putative minor site ϳ60 Å away in the 14th type III repeat of fibronectin (FNIII14) (30). An additional similar example comes from the two contiguous heparin-binding sites in the 10th and 11th type III repeat of tenascin-X (TN-X) (33).
The overall architecture of FNIII repeat molecules may adopt a variety of shapes depending on the tilted angles of inter-repeats and the length of these inter-repeat regions. The previous study (7) proposed that the TNIII4 and TNIII5 repeats are tightly packed resulting in an extended and relatively straight strand. Our analysis of the structure-based sequence alignment of the TNIII repeats and FNIII repeat structure indicates that the sequence KEDK in the loop region between the F and G ␤-sheets of TNIII4 is on the same face of the molecules as the sequence KGR in the TNIII5, although two TNIII repeats are concatenated in a slight spiral. Interestingly, the heparin-binding site in TN-X has been predicted to reside in this loop region between the F and G ␤-sheets of the FNIII10 module and a linker region between FNIII10 and FNIII11. However, because of the linker-domain region, TN-X has more a flexible orientation between the FNIII10 and FNIII11, producing a cationic pocket.
Heparin-protein interactions are based on ionic contacts between positively charged residues (Arg and Lys) and the negatively charged groups of heparin (sulfate groups of glucosamine or the carboxylate group of iduronic acid) (34). Residues Lys 751 and Lys 754 in the TNIII4 domain were found to constitute the minor heparin-binding domain, with a double mutation of K751Q/K754Q resulting in the significant reduction of heparin-binding affinity. In addition, residues in the KEDK sequences are conserved in TN-X between different species ranging from a chick to a human. Thus the structure-based sequence alignment, mutagenesis, and sequence conservation data together suggest that the KEDK sequence in TNIII4 is essential for the high affinity binding to heparin. Because the KEDK sequence in TNIII4 is on the same face of the molecules as the KGR sequence in TNIII5, heparin-binding sites may traverse TNIII4 -5 and thus require the residues KEDK from the TNIII4 for optimal heparin binding.
The sequence KGRHKSKPARVK in TNIII5 has been implicated in heparin binding activity previously (28). In this study, we found that mutagenesis of Lys 1057 and Arg 1059 reduces heparin binding drastically. However, other basic residues cannot be excluded for the heparin-binding site. The simplest model would postulate two binding sites, one in TNIII4 and another in TNIII5, which can bind simultaneously to heparin. We have now confirmed that the secondary low affinity heparin-binding site in the TNIII4 domain is conserved ranging in chick, mouse, pig, and human. The conservation of this activity over the 300-million years that separate chickens and humans suggests that it is biologically important for the functioning of TN-C in tissues. Thus, we conclude that the TNIII4 domain contributes to optimal heparin binding and the KEDK sequence in TNIII4 is minor heparin-binding site. FIG. 4. TNIII4 mediates heparan sulfate-dependent HGF adhesion. 24 wells were individually coated with recombinant TNIII fragments (1 M) prior to the addition of HGFs that were pretreated at 37°C for 60 min with heparinase I (2 units/ml) or chondroitinase ABC (chond. ABC, 2 units/ml) or that were coincubated with heparin (5 g/ml) prior to plating. Data represent the mean of three separate determinations Ϯ S.D. and are representative of three experiments.