Divalent Cation Modulation of Fibronectin Binding to Heparin and to DNA*

Fibronectin is an adhesive glycoprotein that binds to heparin and to DNA. The binding of tryptic fragments of human plasma fibronectin to these ligands is found to be highly dependent on the concentration of divalent cations. We have identified 3 types of binding to hepa- rin. 1) Calcium-sensitive binding is inhibited by CaCL, but not by M&lz or by MnClZ. The NHz-terminal 31,000- dalton fragment (fragment 23) has this type of binding, which is half-maximally inhibited by 3 to 4 m~ CaC12. 2) Divalent cation-sensitive binding is exhibited by a 75,000-dalton fragment (fragment 13); its binding is inhibited by all 3 divalent cations. 3) Divalent cation-insensitive binding is characteristic of a 95,000-dalton fragment (fragment 10) and larger fragments. These 3 fragments (fragments 10, 13, and 23) are not disulfide-bonded to other fragments. Specific tryptic fragments of fibronectin also bind readily to native DNA in the presence of EDTA, but the binding of all fragments is abolished by the presence of 10 m~ CaClz or MgClZ. Our results indicate that the binding of specific domains of fibronectin to heparin or to DNA can be modulated by divalent cations.

Fibronectin is an adhesive glycoprotein that binds to heparin and to DNA. The binding of tryptic fragments of human plasma fibronectin to these ligands is found to be highly dependent on the concentration of divalent cations. We have identified 3 types of binding to heparin. 1) Calcium-sensitive binding is inhibited by CaCL, but not by M&lz or by MnClZ. The NHz-terminal 31,000dalton fragment (fragment 23) has this type of binding, which is half-maximally inhibited by 3 to 4 m~ CaC12.
2) Divalent cation-sensitive binding is exhibited by a 75,000-dalton fragment (fragment 13); its binding is inhibited by all 3 divalent cations. 3) Divalent cationinsensitive binding is characteristic of a 95,000-dalton fragment (fragment 10) and larger fragments. These 3 fragments (fragments 10, 13, and 23) are not disulfidebonded to other fragments. Specific tryptic fragments of fibronectin also bind readily to native DNA in the presence of EDTA, but the binding of all fragments is abolished by the presence of 10 m~ CaClz or MgClZ. Our results indicate that the binding of specific domains of fibronectin to heparin or to DNA can be modulated by divalent cations.
Fibronectins are adhesive, high molecular weight glycoproteins located on the cell surface, in extracellular matrices, or in plasma. They are thought to function in cell-substratum and cell-cell adhesion, maintenance of normal morphology of cells, cell migration, wound healing, phagocytosis and reticuloendothelial clearance, and blood coagulation (1-8).
The initial molecular event in these functions appears to be specific binding of fibronectin to a second biological macromolecule such as collagen, heparin and other glycosaminoglycans, fibrin, or plasma membrane components (1-8). Subsequent covalent cross-linking to some of these molecules can occur by a transglutaminase reaction (9)(10)(11). Fibronectin can also bind to 2 intracellular molecules, DNA (12) and actin (13,14), although the physiological significance of these interactions is not yet known. These multiple ligand-binding activities have been localized to specific protease-resistant domains on the fibronectin molecule; at least 5 to 6 structural and functional domains have been identifed {15).
The complex of fibronectin and heparin may play an important role in binding collagen and in phagocytosis during reticuloendothelial clearance of colloids. Heparin promotes the binding of fibronectin to collagen (16)(17)(18). In addition, fibronectin and heparin stimulate the phagocytosis of gelatin by liver slices and murine macrophages (7,(19)(20)(21) and the binding of type I11 collagen to macrophages (22), and they also induce a cryoprecipitate with fibrinogen in plasma (23).
Fibronectin binding to heparin occurs with moderately high affinity (KO = W 7 to IO-' M ) , but Scatchard analysis shows complex binding that suggests the existence of at least 2 components (24). Fibronectin binds to DNA with significantly lower affinity (KO = 5 X M) (12). We have recently purified a heparin-binding domain of fibronectin (24, 25). However, 1 or more additional heparin-binding sites have been identified by Hakomori and co-workers (26,27), Richter et al. (28), and us (29). The characteristics and functional relationships of each of these binding sites remain unclear. In this study, we describe differences between the binding of specific fibronectin fragments to heparin and to DNA, and describe 3 classes of sensitivity to divalent cations. The identification of these differences should facilitate the purification of each separate heparin-binding site, and their existence suggests a possible means of ionic modulation of fibronectin interactions with other macromolecules.

Materials and Methods
Preparation of Fibronectin-Fibronectin was purified from human plasma by gelatin and heparin affinity column chromatography at room temperature (23 "C) unless otherwise specified. Human plasma from the National Institutes of Health Blood Bank (500 ml) was supplemented with 12.5 ml of 0.2 M EDTA and 2.5 ml of 0.2 M phenylmethanesulfonyl fluoride (Sigma) freshly prepared in 956 ethanol. After centrifugation at 9OOO X g for 15 min at 4 "C, the supernatant material was incubated at 37 "C for about 10 min, deaerated for 5 min, and applied to a precolumn of Sepharose CL-4B (bed volume of 20 m l ) which had been equilibrated with 0.15 M NaCl, 5 m~ EDTA, 10 ZIIM Tris-HC1 (pH 7.4). The flow-through fractions were applied to a gelatin-Sepharose affinity column (bed volume of 20 m l ) which had been washed with 40 ml of 4 M urea, 10 m~ Tris-HCl (pH 7.4), followed by 60 ml of 0.15 NaCl, 5 mM EDTA, 10 mM Tris-HC1 (pH 7.4). After washing with 40 ml of 0.5 M NaC1, 10 mM Tris-HC1 (pH 7.4), followed by 60 ml of 0.15 M NaCl, 5 m~ EDTA, 10 mM Tris-HC1 (pH 7.4), the gelatin-Sepharose column was eluted by 4 M urea in 10 mM Tris-HC1 (pH 7.4). The fvst 12 ml were discarded. The next 20 ml were collected and directly applied to a heparin-Sepharose affinity column (bed volume of 20 ml) which had been washed with 40 ml of 0.5 M NaCl, 10 m~ Tris-HC1 (pH 7.4), followed by 60 ml of 0.15 M NaC1, 5 m~ EDTA, 10 mM Tris-HC1 (pH 7.4). After washing with 60 ml of 0.15 M NaCl, 5 mM EDTA, 10 m~ Tris-HC1 (pH 7.4), the heparin-Sepharose column was eluted by 0.5 M NaCl in 10 mM Tris-HC1 (pH 7.4). Fibronectin eluted from the heparin affinity column was concentrated by addition of solid ammonium sulfate to 40% saturation, centrifugation, and dialysis against 10 mM Tris-HC1 (pH 7.0), 0.15 M NaCl at 4 "C, and was stored in polypropylene tubes (Nunc) in liquid N2 at 4 to 19 mg/ml.
The gelatin and heparin affinity columns can be used repeatedly, and were stored at 4 "C after the addition of 0.02% sodium azide. The Fibronectin Binding to heparin affinity chromatography step was necessary to remove trace contaminants that remain after gelatin affinity chromatography (30,31). About 85 mg of pure fibronectin were obtained from 500 ml of human plasma.
Trypsin Digestion-Fibronectin (3 mg/ml) was digested by L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington, 238 units/mg) in 1 mh5 CaC12, 30 mM NaCl, 50 mM Tris-HC1 (pH 7.0) in polypropylene tubes (Nunc or Falcon) at 30 "C at the specified enzyme-substrate ratios. A t specified intervals, digestion was terminated by the addition of 1/200 volume of 0.2 M phenylmethanesulfonyl fluoride freshly prepared in 95% ethanol and immersion in an ice bath. The same amount of phenylmethanesulfonyl fluoride was added 1 h later. The digest was frozen in powdered dry ice and stored at -80 "C.
Binding Analysis by Affinity Chromatography-Tryptic digests of fibronectin (80 pg) in 0.2 ml of Tris-NaC1 (50 mM Tris-HC1 (pH 7.0), 0.1 M NaCl) containing 10 mM EDTA or divalent cations were applied to heparin-Sepharose affinity columns (0.7 X 4 cm) (Bio-Rad polypropylene Econo-Columns or Isolab Quik-Sep columns) of 0.25-ml bed volume that were prewashed with 0.5 ml of 0.5 M NaC1, 50 mM Tris-HC1 (pH 7.0) followed by 2 X 0.5 ml of Tris-NaC1 containing 10 mM EDTA or divalent cations. The columns were then washed with 2 X 0.5 ml of the above solutions and eluted as described in the figures by 2 X 0.5 ml of Tris-NaC1 containing 10 mM MgClz or 10 m~ CaC12, or alternatively by 0.5 M NaC1, 50 mM Tris-HCl (pH 7.0) to elute a l l bound fragments. Analysis of 10 samples usually required about 30 min from application of samples to elution of binding fragments. The eluates were collected in glass test tubes (borosilicate glass, 10 x 75 mm, Kimble) and were adjusted to a final concentration of 9% (w/v) trichloroacetic acid at 0 to 4 "C. After 2 h or overnight, fragments were collected by centrifugation at 25,000 X g for 5 min at 4 "C and analyzed for protein or subjected to SDS'-polyacrylamide gel electrophoresis after neutralization and solubilization using 40 p1 of SDS sample buffer containing 2% SDS, 10 m~ sodium phosphate (pH 7.0) 10% (w/v) glycerol, 0.025 N NaOH, freshly prepared 0.1 M dithiothreitol, 0.0015% bromphenol blue.
Diagonal SDS-Polyacrylamide Gel Electrophoresis-Disulfide linkages between fragments were examined by 2-dimensional gel electrophoresis. To obtain a full range of fragments, 80 pg of each of 3 tryptic digests of fibronectin were combined after digestions at enzyme/substrate ratios and times of 0.002% for 30 min, 0.2% for 30 min, and 2% for 30 min. The mixture was electrophoresed in a 1.2mm thick SDS-polyacrylamide slab gel without reduction by dithiothreitol. After electrophoresis in the fmt dimension, a 1-cm gel strip was cut off with a circular blade and incubated in 50 ml of 0.1 M dithiothreitol, 0.5% SDS, 0.01 M sodium phosphate (pH 7.0) with gentle swirling at 23 "C for 40 min. The strip was placed horizontally on the top of a second dimensional SDS-polyacrylamide slab gel that was 1.6 mm thick. The strip was sealed in place with 1% agarose (Bio-Rad No. 162-0017) containing 0.1 M dithiothreitol, 0.5% SDS, 0.01 M sodium phosphate (pH 7.0), and 2 to 3 drops of 0.0075% bromphenol blue. Electrophoresis was performed at 6 mA until the bromphenol blue entered the resolving gel, and then at 40 to 50 mA.
Purification of Fragment 23-After a brief tryptic digestion, fragment 23 of fibronectin was purified by DEAE-cellulose column chromatography, essentially as described by Mosher and Proctor (9) and McDonald and Kelley (11).
Other Procedures-SDS-polyacrylamide gel electrophoresis using a discontinuous Laemmli buffer system (32,33) and peptide mapping using Staphylococcus aureus protease V8 (34) were performed as described previously (29). Heparin-Sepharose was prepared according to

Heparin and to DNA
The binding of these fibronectin fragments to heparin-Sepharose is markedly dependent on the concentration and type of divalent cation (Fig. 2). The overall binding of fragments decreases in the sequence 10 m~ EDTA > 10 mM MgC12 = 10 mM MnClz > 10 mM CaCl2. In the presence of 10 m~ EDTA, 19 of 27 fragments bind to heparin; fragments 12, 16-18, 22, 24, 28, and 29 do not bind (Fig. 2).
The presence of divalent cations results in specific inhibition of the binding of certain fragments to heparin. In the presence of either 10 m~ MgClz or MnCl,, 4 additional fragments are not bound by heparin ( Fig. 2; fragments 13,14,19, and 26). In addition, 4 other fragments (fragments 20,23,25, and 27) show decreased binding. In the presence of 10 m~ CaClZ, 2 additional fragments pass through the columns without binding (fragments 23 and 25), and 2 others are less tightly bound (fragments 15 and 20). Twelve fragments (fragments 2-4, 6-11, 16, 20, and 27) are insensitive to divalent cations and continue to bind to heparin in their presence (Fig. 2).
We examined whether these effects of divalent cations might result from simple charge effects by comparing the binding of these fragments to DNA, another negatively charged polymer that is known to bind fibronectin and specific fibronectin fragments (12, 29). Unexpectedly, none of the 27 tryptic fragments binds to native DNA-cellulose in the presence of either calcium or magnesium (Fig. 2). A number of fragments do bind in the presence of EDTA, a compound that was routinely included in previously published reports of fibronectin binding to DNA (Fig. 2). The fragments that bind to DNA in the presence of EDTA are similar to but not identical with the fragments binding to heparin in EDTA (Fig.  2). The presence of 10 m~ MnClZ inhibits the binding of most fragments, but fragments 7 to 10 still bind to the DNA. These patterns of tryptic fragments binding to DNA are quite different from those to heparin, e.g. in the different effects of MnC12 and MgC12. The modulation of binding by divalent cations is therefore probably not the result of simple charge effects.
The preceding binding experiments do not indicate whether divalent cations modulate the overall binding of specific fragments to ligands or modulate only their initial recognition and binding interactions. To examine this question, tryptic fragments of fibronectin were bound to heparin-Sepharose in the presence of EDTA and were sequentially eluted by adding 10 mM MgClZ, 10 mM CaCl2, and then 0.5 M NaCl to the elution buffer.
Subsequent treatment with 10 m~ CaC12 results in a striking release of most of fragment 23 with a slight release of 9 others (fragments 2-6,13,14,25, and 27, Fig. 1D). The remainder of the fragments (fragments 1-11, 15, 19, 20, 23, 25, and 27) is eluted in 0.5 M NaCl (Fig. 1E). These patterns of elution by magnesium and calcium are generally the converse of the patterns of binding in Fig. 2, indicating that these cations inhibit all of the binding interactions of a fragment with heparin, rather than simply preventing its initial recognition and binding interaction with heparin. The release of small amounts of some fragments with each elution, as noted above, suggests weak binding or slightly overlapping specificities, since the elutions were performed at each step using 4 column volumes of all solutions. If the order of elution was reversed, 2.e. calcium before magnesium, the pattern of fragments eluted in calcium was the summation of the fragments eluted in Fig.   1, C and D (data not shown). Summarizing the major results  in these experiments, there are at least 3 classes of fibronectin fragments that bind to heparin: calcium-sensitive (e.g. fragment 23); divalent cation-sensitive (e.g. fragments 13 and 14); and divalent cation-insensitive (e. g. fragments 1-11 and 15).
Since the binding of fibronectin to a ligand can be inhibited by the reduction of disulfide bonds (37), the disulfide bonds of fragments were not reduced before ligand-binding analyses. Some fragments that appear to bind to heparin may therefore be adsorbed to affinity columns because they are disulfidebonded to other fragments that contain heparin-binding sites. The extent of disulfide bonding between fragments was ex- amined by 2-dimensional gel electrophoresis before and after reduction of disulfide bonds. Fig. 3 shows that most of each of the fragments of fibronectin is located along the diagonal, indicating the absence of detectable interfragment disulfide bonds. However, at least 6 fragments (fragments 19,20,24,27-29) are extensively disulfide bonded, and several others show evidence of disulfides in a subpopulation of fragments of that size (Fig. 3). It is therefore difficult to unambiguously interpret the binding of such fragments to affinity columns as evidence for a binding site. However, 3 major fragments analyzed in this study (fragments 10, 13, and 23) show no evidence for such complicating intersubunit disulfide bonds.
Since fragment 23 is unique in its sensitivity to calcium, we examined its binding at physiological concentrations of this cation. As in unfractionated digests, purified fragment 23 binds to heparin in 10 m EDTA-Tris-NaC1, but does not bind in 10 m CaC12-Tris-NaC1 (Fig. 4). A dose-response curve shows that the concentration of calcium required to inhibit the binding of fragment 23 by 50% is 3 to 4 m (Fig.  4), which is similar to the calcium concentration of 2 to 3 m in human blood. Parenthetically, it should be noted that even in 10 m EDTA, but with 0.1 M NaCl present, the binding of fragment 23 to heparin in these experiments is not strong. After very extensive washing (12 or more column volumes), most of fragment 23 is eventually washed off the heparin-Sepharose columns.

DISCUSSION
Our four major findings are: 1) different tryptic fragments of fibronectin have markedly different divalent cation sensitivities in binding to heparin, 2) the major types of binding interactions are calcium-sensitive, divalent cation-sensitive, or divalent cation-insensitive binding; 3) the binding to heparin of a calcium-sensitive fragment is inhibited at physiological concentrations of calcium; and 4) the binding of fibronectin fragments to DNA is prevented by divalent cations, and the pattern of sensitivity of specific fragments is different from the pattern for heparin binding.
Our results indicate that the binding of fibronectin to heparin is complex, in agreement with the binding characteristics found previously in studies of the intact molecule (23,24,38).
The existence of at least 3 classes of binding interaction suggests that fibronectin has at least 3 different heparinbinding sites. Two-dimensional SDS gel analysis in the absence and presence of reducing agent indicates that at least fragments 10, 13, and 23 can bind to heparin as monomers; these fragments show different cation sensitivities (divalent cation-insensitive, divalent cation-sensitive, and calcium-sensitive, respectively). Peptide mapping using S. aureus protease V8 digestion of proteins separated in an SDS-polyacrylamide gel shows that fragment 23 does not share any common peptides with fragments 10 and 13, but that there is considerable but not complete homology between peptides 10 and 13.2 The calcium-sensitive, heparin-binding site in fragment 23 is therefore distinct from other sites; however, fragments 10 and 13 are probably partially overlapping fragments. Although other interpretations are possible, the simplest interpretation of our data is that there are two additional heparinbinding sites in fragment 10 (Mr = 95,000), one of which is insensitive to divalent cations, accounting for the insensitivity of the whole fragment. After further cleavage by trypsin, only the divalent cation-sensitive site is retained in fragment 13 (M, = 75,000) and a divalent cation-insensitive site may be retained in fragment 27 (MI = 21,000), which is insensitive to inhibition by divalent cations. Further structural and sequencing studies are needed to elucidate these relationships.
This study clarifies apparently conflicting results in the previous literature concerning the number of heparin-binding sites on fibronectin (25,26,29,39), since the number of heparin-binding fragments is shown to vary depending upon the divalent cation composition of buffers. Fragment 23 (Mr = 31,000), shown in this study to be calcium-sensitive, corresponds to the NH&mnhal domain of fibronectin, i.e. the 32,000-dalton heparin-binding thermolysin fragment from hamster plasma fibronectin reported by Hakomori and coworkers (26, 27) and the 30,000-dalton heparin-binding cathepsin D/plasmin fragment from human plasma fibronectin reported by Richter et al. (28).
The divalent cation-insensitive heparin-binding site can be isolated in either the presence or the absence of calcium (24, 25,29) and has been localized to the carboxyl-terminal third of the polypeptide chain of fibronectin (24, 38). A 50,000dalton pronase fragment containing this domain has been purified and characterized (25). The divalent cation-sensitive heparin-binding site may be located in a structural domain at the COOH-terminal end of the 50,000-dalton fragment (15, 29), and the present study suggests that it can now be isolated by appropriate manipulation of the concentrations of divalent cations (Fig. 2, fragments 13 and 14).
Our results suggest that the interaction of fibronectin with heparin (and possibly with the closely related molecule heparan sulfate) may be affected by divalent cations in vivo. Since divalent cation-sensitive and insensitive sites are presumably present on the same polypeptide of intact fibronectin, heparin would be expected to bind by at least one site to fibronectin regardless of the ionic environment; consistent with this notion, the binding of fibronectin to heparin does occur in the presence or absence of divalent cations (24, data not shown). However, the modulation of binding, shown here to occur at physiological concentrations of extracellular calcium, could regulate the interaction of specific sites on fibronectin with specific extracellular glycosaminoglycans, and might alter the strength of binding or the nature of cross-links between extracellular matrix molecules.
Although fibronectin is known to bind to DNA, studies of this binding interaction have generally been performed in the Fibronectin Binding to Heparin and to DNA 5267 presence of EDTA (12). Specific fragments of human and chicken fibronectins bound to native DNA in EDTA (29,40). This study confirms these results, but also shows that such binding is abolished in the presence of calcium or magnesium ( Fig. 2). About 9 0 % of intact, purified fibronectin bound to a native DNA-cellulose column in the presence of EDTA, whereas 33 to 70% of the fibronectin bound in the presence of these ions.' The binding of intact fibronectin to DNA in the presence of divalent cations therefore appears weak, and proteolytic fragmentation appears to weaken the interaction further so that binding can occur only in the absence of divalent cations. It is therefore not clear whether fibronectin binds to DNA with sufficient strength under physiological conditions to be important biologically. It is of interest to consider the effects of divalent cations on the binding of fragments of plasma fibronectin to other ligands. In contrast to the results with heparin, we could find no specificity in the effects of calcium, magnesium, or manganese ions on the binding of any particular fragment in a mixture of fibronectin fragments to gelatin, actin, fibrin, or S. aureus. However, the total quantity of fragments binding to certain ligands was often slightly increased or decreased, with the amounts retained in the presence of various ions found to be in the following order: MgClz > MnC12 = CaClz > EDTA for gelatin-Sepharose (only minor differences among all 4), CaC12 = MgC12 = MnClz 2 EDTA for fibrin-Sepharose, and EDTA 2 MnClz ? MgC12 > CaC12 for both actin-Sepharose and S. aureus cells (data not shown). These results further suggest that the effects of divalent cations on fibronectin interactions with heparin reflect modulation of specific sites on fibronectin, since only certain fragments are affected in their binding to this particular ligand.