Demonstration of High Affinity Fibronectin Receptors on Rat Hepatocytes in Suspension*

A cell-binding peptide (M. = 86,000) which lacks the gelatin- and heparin-binding domains, was purified from trypsin-digested fibronectin. Preincubation of rat hepatocytes in suspension with the peptide, inhibited initial attachment of the cells to immobilized fibronectin while attachment to immobilized laminin and col- lagen was unaffected. '2"I-iabeled 86-kDa peptide bound to the cells in suspension at 4 "C in a time- dependent, saturable, and partially reversible reaction. Scatchard analysis of the binding data indicated a single class of receptors with a K d of 1.7 X lo-* M. The number of binding-sites was -2.8 X lO"/cell. Unlabeled 86-kDa peptide inhibited the binding of laaI- labeled 86-kDa peptide 30-fold more effectively than intact fibronectin. These results provide direct evi- dence for the presence of a domain in the fibronectin molecule which has or may acquire a high affinity for receptors involved in adhesion of hepatocytes to immobilized fibronectin.

A cell-binding peptide (M. = 86,000) which lacks the gelatin-and heparin-binding domains, was purified from trypsin-digested fibronectin. Preincubation of rat hepatocytes in suspension with the peptide, inhibited initial attachment of the cells to immobilized fibronectin while attachment to immobilized laminin and collagen was unaffected. '2"I-iabeled 86-kDa peptide bound to the cells in suspension at 4 "C in a timedependent, saturable, and partially reversible reaction. Scatchard analysis of the binding data indicated a single class of receptors with a K d of 1.7 X lo-* M.
The number of binding-sites was -2.8 X lO"/cell. Unlabeled 86-kDa peptide inhibited the binding of laaIlabeled 86-kDa peptide 30-fold more effectively than intact fibronectin. These results provide direct evidence for the presence of a domain in the fibronectin molecule which has or may acquire a high affinity for receptors involved in adhesion of hepatocytes to immobilized fibronectin.
The glycoprotein fibronectin is a mediator of cell-matrix interactions (1). It has affinity for matrix components such as collagens (2), glycosaminoglycans (31, and fibrin (4). These interactions have been studied in some detail and domains in the fibronectin molecule, responsible for the interactions, have been identified and isolated after proteolytic fragmentation of fibronectin. A large number of different cell types, including fibroblasts (5), hepatocytes (6), and platelets (7) have been shown to adhere to and spread out on substrates of immobilized fibronectin. A cell-binding tetrapeptide in the fibronectin molecule has been identified by use of a monoclonal antibody which inhibits cell attachment to fibronectin substrates (8,9). However, the cellular receptor for fibronectin has not been identified.
A problem in studies of fibronectin receptors is that soluble fibronectin binds very poorly to cells, making classical receptor-ligand studies difficult to perform. Furthermore, since fibronectin has binding sites for a number of different components it may be difficult to interpret binding data, particularly if binding to collagen-and proteoglycan-containing cell layers is studied. The explanations suggested for the lack of binding of soluble fibronectin to cells have been either (a) that cooperative interactions between immobilized fibronectin molecules and many cell surface receptors are required for stable binding (10) or (b) that fibronectin becomes activated by conformational changes during incorporation into extra-cellular matrices or by coating onto plastic surfaces (11). Recently, support for the latter theory was presented (12). The ability of soluble fibronectin to compete with immobilized fibronectin for binding to cellular receptors and thereby inhibit initial attachment of rat hepatocytes to fibronectincoated plastic dishes, was markedly increased by a limited trypsin digestion of the soluble fibronectin. The interaction of soluble fibronectin with hepatocytes in suspension could also be stimulated by addition of collagen or heparan sulfate, as measured as increased inhibition of initial attachment to fibronectin dishes. These results suggested that tryptic fibronectin fragments could be useful in studies of fibronectin receptors. In this investigation an 85-kDa fibronectin fragment, which lacks affinity for collagen and heparin, has been used to demonstrate the presence of a specific fibronectin receptor on rat hepatocytes. The binding of the fibronectin fragment to the cells in suspension has been characterized and compared with that of intact fibronectin.

MATERIALS AND METHODS
Trypsin (TPCW-treated, type XIII), bovine serum albumin, phenylmethylsulfonyl fluoride, EDTA, dithiothreitok Bis-Tris, and guanidinium chloride were purchased from Sigma. DEAE-Sephacel, heparin-Sepharose, gelatin-Sepharose, and Percoll were obtained from Pharmacia, Uppsala, Sweden. TSK 3000 columns were from LKB, Bromma, Sweden. Na'=I (carrier-free) was purchased from the Radiochemical Centre, Amersham, England and Iodo-Beads from Pierce Chemical Co. Fibronectin was purified from human plasma according to the method of Vuento and Vaheri (13). Neutral salt-soluble collagen from rat skin (type I and 111) and laminin were kind gifts from Dr. K. Rubin, University of Uppsala, Sweden, and Dr. R. Timpl, Max Planck Institut, West Germany, respectively.
Purifiqatwn of the Cell-binding 85-kDa Fibronectin Fragment-Fibronectin (3 mg/ml) in 10 mM Tris-HC1 buffer, pH 7.5, containing 0.14 M NaCl and 0.02% NaNs was digested with trypsin (7 Fg/ml) for 90 min at 37 "C. The digestion was terminated by addition of phenylmethylsulfonyl fluoride to a final concentration of 0.4 mM. The digest was dialyzed against 50 mM Tris-HC1 buffer, pH 7.0, containing 0.1 M NaCl, 10 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, and 0.02% NaN3, and sequentially applied to columns of heparin-Sepharose (1 ml of gel/lO mg of fibronectin digest) and gelatin-Sepharose (1 ml of ge1/2 mg of fibronectin digest) in the same buffer. During a second passage of the unbound fraction through the columns no further material was retained, demonstrating that no heparin-or gelatin-binding peptides remained in this fraction. The unbound material was dialyzed against 10 mM Bis-Tris buffer, pH 6.0, containing 50 mM NaCl and 0.02% NaN3, and applied to a DEAE-Sephacel column (50 mg of protein/lO ml of gel). The DEAE column was eluted with a linear gradient (100 ml) of NaCl from 50 to 300 mM in this buffer (Fig. 1). The fractions which contained peptides that The abbreviations used are: TPCK, L-1-tosylamido-2-phenylethyl chloromethyl ketone; Bis-Tris, bis(2-hydroxyethyl)iminotris(hydrox-ymethy1)methane; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate. was applied to a 10-ml column of DEAE-Sephacel in 10 mM Bis-Tris buffer, p H 6.0, containing 50 mM NaCI and 0.02% NaN3. The arrow indicates the start of washing of the column after the sample had been applied, and the dotted line depicts the salt gradient. Fractions of 2.5 ml were collected and analyzed for absorbance at 280 nm (0) and ability to promote attachment and spreading of rat hepatocytes (not shown in the figure), as described under "Materials and Methods." Fractions were pooled as indicated (hatched area).  Fig. 1 were concentrated and dialyzed as described under "Materials and Methods." Samples of 1 mg were applied to a TSK 3000 column, eluted at a flow rate of 0.2 ml/min with 20 mM Tris-HCI buffer, pH 7.0, containing 4 M guanidinium chloride, and 2 mM dithothreitol. Absorbance a t 280 nm was measured and fractions of 0.2 ml were analyzed for ability to promote attachment and spreading of rat hepatocytes (not shown in the figure). Fractions were pooled as indicated (hatched area).
induced attachment and spreading of rat hepatocytes (see below for assay conditions) were pooled and concentrated to 20 mg/ml by vacuum dialysis. After dialysis against 4 M guanidinium chloride, 2 mM dithiothreitol, in 20 mM Tris-HCI buffer, pH 7.0, the peptides were chromatographed on a TSK 3000 column (volume, 40 ml; length, 100 cm) in the same buffer (Fig. 2). The first peak contained the adhesion promoting material which in SDS electrophoresis migrated as a single peptide with an apparent M, of 85,000 (Fig. 3). The 85-kDa fraction was pooled and dialyzed back to phosphate-buffered saline (10 mM PO,, 0.14 M NaCI, pH 7.4) by stepwise reducing the concentration of guanidinium chloride from 4, 3, 2, 1, 0.5 to 0 M in the presence of 2 mM dithiothreitol. Lastly, dithiothreitol was dialyzed away. This treatment was necessary for the peptide to refold into a soluble, monomeric, trypsin resistant, and thus presumably native form. The overall yield of the 85-kDa peptide was 11 mg from 100 mg of fibronectin, equivalent to 28% if one copy of 85-kDa peptide/ 220 kDa is assumed.
Electrophoresis-Electrophoresis was performed on polyacrylamide gradient gels in SDS according to the method of Blobel and Dobberstein (14).
Iodination of the 85-kDa Fragment-Labeling with ['251]iodine was performed by the chloramine-T method (15) using Iodo-Beads. The 85-kDa fragment was labeled to a specific activity of lo' cpm/pg. The integrity of the labeled peptide was ascertained by SDS electrophoresis followed by autoradiography.
Preparation of Cell Attachment Substrates-Microtiter wells (Nunc, Denmark) were coated with 50 pl of fibronectin, fibronectin peptides, laminin (50 pg/ml), or collagen (100 pg/ml) in phosphatebuffered saline by adsorption a t 37 "C for 2 h. The protein solution was removed and the wells were incubated with 50 pl of phosphatebuffered saline containing 1.5% bovine serum albumin for 30 min a t 22 "C to block remaining protein binding sites on the plastic surface. This solution was removed before seeding of the cells.
Cell Attachment Assay-Hepatocytes were isolated from male Sprague-Dawley rats after perfusion of the liver in situ with collagenase as described (16). Cells (5 X IO') in 50 pl of buffer 3 (see Ref 16; 0.137 M NaCI, 4.7 mM KCI, 0.65 mM MgSO, X 7 H20, 1.2 mM CaCI2 X 2 H20, 10 mM HEPES) containing 1.5% bovine serum albumin, were seeded in each well and incubated a t 37 "C in humidified air. After incubations for indicated times the microtiter plates were washed and the number of cells attached determined as described (17). In short, the cells were lysed in a solution of Triton X-100, and the activity in the lysate of the enzyme hexosaminidase was determined. Cell spreading was studied in the presence of cycloheximide (25 pg/ml) as described (18).
Assay for Binding of lZI-lubeled 85-kDa Peptide to Hepatocytes-Cells (5 X 106/ml) in buffer 3 containing 1.5% bovine serum albumin were incubated end over end with '*'I-labeled 85-kDa peptide (IO6  . 4 (left). Inhibition of cell attachment by the 85-kDa fragment. Rat hepatocytes (1 X 106/ml) were preincubated at 4 "C for 20 min in the presence of the indicated amounts of purified 85-kDa fragment, and then seeded into fibronectin-coated wells under conditions described under "Materials and Methods." After incubation for 6 min at 37 "C the wells were washed and the number of cells attached determined. Per cent inhibition was calculated as: 100-100 X (number of cells attached in the presence of 85-kDa fragment/number of cells attached in the absence of 85-kDa fragment). The values shown represent averages of two parallel incubations.
FIG . 5 (center). Time course of binding of "'1-labeled 85-kDa fragment to rat hepatocytes in suspension. Cells were incubated with '251-labeled 85-kDa fragment at 4 "C and specific binding of radioactivity determined at the indicated times as described under "Materials and Methods" (0). In a parallel incubation, unlabeled 85-kDa fragment was added to give a concentration of 10 pg/ml at the time point indicated by the arrow (0). FIG. 6 (right). Inhibition of binding of 1a61-labeled 85-kDa fragment to hepatocyt-by fibronectin and the 85-kDa fragment. Cells were incubated with '9-labeled 85-kDa fragment at 4 "C for 2 h in the presence of indicated amounts of fibronectin (0) and 85-kDa fragment (0), respectively, and the bound radioactivity determined as described under "Materials and Methods." cpm/ml; lo7 cpm/pg unless otherwise indicated) at 4 "C for indicated times. The incubations were terminated by removing samples of 100 pl which were centrifuged at 1500 X g through 10 ml of 20% Percoll in buffer 3. The supernatant was aspirated off until 300 pl remained over the pellet. After freezing at -20 "C, the bottom tip of the tubes were cut off and counted in a Packard model 5260 Auto-Gamma Scintillation spectrometer. Unspecific binding (-0.9% of added radioactivity) was determined in the presence of 50 pg unlabeled 85-kDa fragment/mL The values shown in the figures represent averages of duplicate determinations after subtraction of unspecific binding.

RESULTS
Trypsin-digested fibronectin added to the medium has previously been shown to inhibit initial cell attachment to fibronectin-coated dishes more efficiently than intact fibronectin (12). Such a digest was fractionated as described under "Materials and Methods" to obtain a cell-binding peptide devoid of gelatin-and heparin-binding domains. Cell-binding activity was associated with an 85-kDa peptide (not shown) which was accumulating during the incubation of fibronectin with trypsin (Fig. 3, A and B). Since the 85-kDa peptide does not bind to gelatin or heparin it must originate from the central region of the fibronectin arms (-70-160 kDa from the Nterminal) (19-21). Tryptic fibronectin fragments apparently identical to this 85-kDa fragment have been described by others (22,23), but have not been obtained free from contaminating peptides. In the present study, denaturing conditions were used in order to purify the 85-kDa peptide?
Preincubation of hepatocytes with the purified 85-kDa fragment in suspension, inhibited initial attachment to dishes coated with intact fibronectin. Almost complete inhibition (95%) of attachment to fibronectin dishes was obtained at a concentration of 50 pg/ml of the 85-kDa fragment (Fig. 4). These data indicate that the 85-kDa fragment contains all structures in fibronectin to which hepatocytes can attach. The effect of the 85-kDa peptide in solution on cell attachment was specific in that only attachment to immobilized fibronectin was inhibited while attachment of hepatocytes to immobilized laminin or collagen was unaffected (Table I).
The 85-kDa peptide immobilized on plastic dishes was as effective as intact fibronectin in promoting both attachment and spreading of hepatocytes (data not shown). This is in accordance with the results of Hayashi et al. (22) who studied the ability of fibroblasts to attach and spread out on dishes coated with the partly purified 85-kDa peptide. '261-labeled 85-kDa fragment incubated with hepatocytes in suspension at 4 "C bound to the cells in a time-dependent reaction which was completed after 1-2 h (Fig. 5). Previously bound lZ5I-labeled 85-kDa fragment was partially displaced (45%) by addition of excess unlabeled 85-kDa fragment (10 pg/ml) (Fig. 5). However, complete reversal of the binding of radioactivity could not be achieved even if the amount of unlabeled 85-kDa fragment was further increased (data not shown). The presence of unlabeled 85-kDa fragment or intact fibronectin from the start of the incubation, inhibited the binding of '251-labeled 85-kDa peptide in a dose-dependent manner. On a molar basis -30-fold more fibronectin (counted on number of monomers of M, = 220,000) than 85-kDa fragment was required to achieve 50% inhibition of the bimling of '251-labeled 85-kDa fragment to the hepatocytes (Fig.  6). When the cells were incubated with increasing amounts of '251-labeled 85-kDa fragment a saturation curve of binding was obtained (Fig. 7A). Half-maximal binding occurred at 15-20 nM (1.3-1.7 pg/ml) and saturation was achieved at 60-100 nM. A Scatchard plot of the binding data, indicated a single class of receptors. The apparent K d was 1.7 X lo-' M and the average number of binding sites was 2.8 X 105/cell. When unfractionated peptides of trypsin-digested fibronectin was '251-labeled and incubated with the hepatocytes, the 85-kDa peptide was selectively enriched in the cell-associated fraction (Fig. 8). This suggests that the 85-kDa peptide has the highest affinity for the fibronectin receptors among the trypsin-generated fibronectin fragments.

DISCUSSION
Recently several reports have indicated that the shape of fibronectin molecules may vary from globular to extended forms depending on ionic strength and pH of the solute (24-27). Also immobilized fibronectin attains different conformations, depending on the nature of the surface to which it is coated (24). Changes in absorption spectra induced by binding of collagen (27) or heparin (28) to fibronectin have previously been reported, and interpreted to reflect conformational changes in fibronectin. The increased binding of soluble fibronectin to hepatocytes in the presence of collagen or heparan sulfate/heparin or after limited trypsin digestion of the fibronectin suggests that such conformational changes in fibronectin are functionally important (12). Fibrillogenesis of fibronectin provides another example, since a conformar tional change must precede the disulfide bridging between thiol groups, as these are not exposed in native soluble fibronectin (29, 30).
In this study the increased affinity of fibronectin for cell surface receptors, induced by uactivation" of fibronectin, has been investigated by a more direct approach than previously. The major cell-binding peptide of trypsin-digested fibronectin was identified and purified. It had an apparent M, of 85,000 and did not contain binding sites for gelatin or heparin. In contrast to soluble '251-labeled fibronectin, which did not detectably bind to hepatocytes (12), the 85-kDa tryptic cellbinding fragment of fibronectin bound to the cells in suspension at 4 "C with high affinity (1.7 X lo-' M) in a timedependent, saturable, and specific manner. Furthermore, preincubation of the cells with the 85-kDa fragment in suspension inhibited the initial attachment of the cells to dishes coated with fibronectin. This result strongly suggests that the binding sites recognizing the 85-kDa peptide are identical

Specificity of inhibition of cell attachment by the 85-kDa fragment
Rat hepatocytes (1 X 10B/ml) were preincubated at 4 "C for 20 min in the absence and presence of purified 85-kDa fragment (50 pg/ml), and then seeded into wells coated with the indicated proteins as described under "Materials and Methods." After incubation for 12 min at 37 "C the wells were washed and the number of cells attached determined. The number of cells attached in the absence of 85-kDa fragment to wells coated with fibronectin is set to 100%. The values shown represent averages of two parallel incubations. with the receptors used by hepatocytes for adhesion to fibronectin matrices. In contrast, the cell-binding peptide isolated by Piersbacher et al. (31) did not detectably bind to cells in suspension. Neither did it affect cell attachment to fibronectin dishes at concentrations where the 85-kDa fragment is completely inhibitory (31). However, at millimolar concentrations of the tetra-peptide, cell attachment to fibronectin dishes was inhibited. From these data a K d of 5 X M for the interaction of the tetra-peptide and its receptor was calculated (9). The difference in affinity between the tetra-peptide and the 85-kDa fragment, respectively, for the fibronectin receptor (>104-fold) may indicate that the 85-kDa fragment contains additional structures which are required for optimal binding strength. The tetra-peptide is also present in collagen and a few other proteins (9). However, cell attachment to collagen (as well as to laminin) was not affected by the presence of the 85-kDa peptide in the medium indicating that also collagen contains additional cell-binding structures, which are recognized by specific collagen receptors.
Intact fibronectin at high concentrations did compete with lZ5I-labeled 85-kDa peptide for binding to the hepatocytes although 30-fold less effectively than the 85-kDa peptide. However, it is not clear whether this competition is due to a lower affinity of intact fibronectin for the receptor or to a small fraction of the fibronectin molecules being "activated" to the high-affinity form. A recent paper by McKeown-Longo and Mosher (32) can be interpreted in favor of the idea that activation of fibronectin is a cellular event that precedes the actual binding of fibronectin to its receptor. They could detect a binding of '%I-labeled fibronectin at 37 "C to cell layers of growing fibroblasts. A fraction of the bound fibronectin was found in a saturable, detergent-soluble pool with a calculated Kd of 3.6 X lo-' M. Thus it is quite possible that the hepatocyte receptor for the 85-kDa fragment and the detergent-soluble fibronectin receptor on the fibroblasts are homologous. In the study of McKeown-Longo and Mosher (32), transformed cells were shown to attach to fibronectin-coated dishes but failed to bind soluble '251-labeled fibronectin. The transformed cells apparently had functional fibronectin receptors but may have been defective in the activation of fibronectin to a form which can bind to its receptor. A possible candidate as cellular "activator" of fibronectin would be membrane-bound heparan sulfate proteoglycan (33, 34). A close co-distribution of cell surface heparan sulfate proteoglycan, fibronectin, and actin has been demonstrated on fibroblasts early in their spreading process (35). Furthermore, exogenously added heparan sulfate has been shown to stimulate the binding of fibronectin to  Fig. 3, and labeled with lZ5I as described under "Materials and Methods" ( l a n e I). The labeled peptide mixture was incubated with the hepatocytes for 3 h a t 4 "C and cell bound material separated from unbound material as described under "Materials and Methods." The cell bound radioactivity was solubilized for 15 min at 22 "C in 10 mM Tris-HC1 buffer, pH 8.0, containing 0.5% Triton X-100, 1 mM phenyimethylsulfonyl fluoride, 2 m M EDTA, 1 mM N-ethylmaleimide, and pepstatin (10 rg/ml) ( l a n e 2). The samples were prepared for electrophoresis as described in the legend to Fig. 3 and run on a 7-15% polyacrylamide gel. After drying, the gel was subjected to autoradiography using Kodak X-Omat x-ray film. The migration distance of molecular mass standards (kDa) are indicated.
hepatocytes (12). In agreement with this hypothesis, tumor cells generally synthesize heparan sulfate with a low sulfate content (36-38) and reduced affinity for fibronectin (37). Cell attachment and spreading to a performed "matrix" would not be affected by reduced sulfation of heparan sulfate since a fibronectin-heparan sulfate interaction apparently is not required for these events. This was demonstrated in this study to be the case for hepatocytes which in the presence of cycloheximide spread out on dishes coated with the 85-kDa peptide.
In conclusion, the results presented here support the concept that activation of fibronectin by conformational changes is important for cell binding. The purified cell-binding fragment of trypsin-digested fibronectin should be useful in studies of fibronectin receptor function on different cell types and possibly in purification of fibronectin receptors. The advantages of the 85-kDa fragment over intact fibronectin is (a) increased affinity for the receptor and ( b ) lack of affinity for gelatin and sulfated glycosaminoglycans, which makes the interaction with cell membrane components more specific and interpretable.