Alternative splicing of the IIICS domain in fibronectin governs the role of the heparin II domain in fibrillogenesis and cell spreading.

The Heparin (Hep) II-binding domain of fibronectin regulates the formation of focal adhesions and actin stress fibers and hence plays an important role in cell spreading, migration, and fibronectin fibrillogenesis. Using human skin fibroblast cultures, we demonstrate that alternative splicing of the neighboring IIICS domain may regulate the activities of the Hep II domain in cell spreading and fibronectin fibrillogenesis. Recombinant Hep II domains, adjacent to either the IIICS domain or the H89 splice variant that contains the amino-terminal sequence of the IIICS domain, blocked fibronectin fibrillogenesis and required sulfated proteoglycans to mediate cell spreading. If the Hep II domain was adjacent to either the H0 or H95 splice variants, which both lack the amino terminus of the IIICS domain, fibrillogenesis was not inhibited and cell spreading was independent of a sulfated proteoglycan-mediated mechanism. The effect of the splice variants on the Hep II domain could be mimicked using a Hep II domain that contained only 6 amino acids from the III(15) repeat or 10 amino acids from the IIICS domain suggesting that sequences proximal to the III(14) repeat determined the role of the Hep II domain in these processes. We propose that alternative splicing of the IIICS domain modulates interactions between heparan sulfate proteoglycans and the Hep II domain and that this serves as a mechanism to control the biological activities of fibronectin.

As a major protein in the extracellular matrix, fibronectin provides positional cues to help direct a wide variety of biological processes including cell migration, adhesion, cell cycle progression, cell differentiation, and apoptosis (1, 2). These biological activities of fibronectin are mediated via interactions with various members of the integrin family and cell surface proteoglycans. It has been known for some time that these biological activities of fibronectin are contained within discrete structural domains and that neighboring domains within fibronectin can either suppress or enhance a particular activity. For example, the heparin (Hep) 1 II-binding domain of fibronectin can suppress the expression of metalloproteinases induced by the central cell-binding domain of fibronectin (3,4), indicating that the various domains of fibronectin can act cooperatively to regulate the biological activity of fibronectin.
During transcription, four regions of fibronectin are alternatively spliced to generate up to 20 different isoforms of fibronectin (2, 5). Two of the alternatively spliced sites known as the extra type III repeats (EIIIA and EIIIB) flank the central cell-binding domain of fibronectin and alternative splicing results in either the exclusion, or inclusion of these domains. A third site of alternative splicing occurs within the IIICS domain (also know as the variable region). Alternative splicing of this domain results in either the inclusion, or exclusion, of this domain as well as the removal of only part of the amino and/or carboxyl termini of the domain (6). The fourth site of alternative splicing is located at the carboxyl termini of fibronectin and results in the removal of the III 15 and I 10 repeats (7).
Alternative splicing of fibronectin is tissue-specific. Fetal tissues and tumors express a higher percentage of fibronectins with the EIIIA and EIIIB repeats. Expression of the EIIIA and EIIIB isoforms is also increased during wound healing (8 -11). Fibronectin isoforms that lack the III 15 and I 10 repeats are found exclusively in cartilage (7) and 50% of fibronectins in the plasma may lack the IIICS domain (12).
Alternative splicing can alter the activity of fibronectin by introducing a new activity that is contained within that spliced site. For example, alternative splicing of the IIICS domain will either add, or remove, integrin-and proteoglycan-binding sites thereby affecting cell adhesion or migration (13)(14)(15)(16). In human periodontal ligament fibroblasts, inclusion of the entire IIICS domain modulates the activity of FAK and regulates apoptosis (17). Inclusion of the EIIIA repeat regulates the expression of metalloproteinases in joint connective tissues, mediates the induction of the myofibroblastic phenotype by transforming growth factor-␤1, and activates lipocytes (18 -20).
The IIICS domain and the EIIIA and EIIIB repeats also influence the biological activities of neighboring domains. Thus, the IIICS domain acts cooperatively with the Hep II domain of fibronectin to control the invasive phenotype of human oral squamous cell carcinoma (21). When the alternatively spliced EIIIA domain is included in the translated product, the cell adhesion function of the neighboring integrin-binding domain in the III 10 repeat is enhanced (22). In addition, the role of III 10 in cell cycle progression and mitogenic signal transduction is enhanced when the EIIIA repeat is included (23). Presumably insertion of the EIIIA domain enhances these functions because it introduces a global conformational change into fibronectin that further exposes the integrin-binding site in the III 10 repeat. A similar function may exist for the EIIIB repeat which when included within fragments from the central cellbinding domain enhances cell adhesion and spreading (24).
The Hep II-binding domain is sandwiched between the alternatively spliced EIIIA repeat and the IIICS domain and is therefore a likely candidate to be affected by alternative splicing. It plays an important role in regulating cell adhesion, migration, fibronectin fibrillogenesis, signal transduction events, and the organization of focal adhesions and the actin cytoskeleton (17,21,(25)(26)(27)(28)(29). The Hep II domain contains binding sites for ␣ 4 ␤ 1 /␤ 7 integrins (30) and members of the syndecan family (28,31). In this present study, we examined whether alternative splicing of the IIICS domain affects the biological activity of the Hep II domain. Using recombinant proteins that contain the Hep II domain and four splice variants of the IIICS domain (H0, H120, H89, or H95), we show that alternative splicing of the IIICS domain regulates the ability of soluble Hep II domains to block fibronectin fibrillogenesis and utilize sulfated proteoglycans to promote cell spreading. Alternative splicing of the IIICS domain did not compromise the ability of the Hep II domain to bind to either the amino terminus of fibronectin or the fibroblast cell surface. Alternative splicing, therefore, appears to be a mechanism to control the biological functions of the Hep II domain, possibly by modulating cell signaling pathways.

MATERIALS AND METHODS
Cell Binding Assay-Neonatal human skin fibroblasts used in the binding assays were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. All binding assays were done as previously described (29). Freshly trypsinized cells were resuspended in DMEM containing 25 mM HEPES, 2 mg/ml bovine serum albumin (BSA), 100 units/ml penicillin G, 5 M streptomycin sulfate, and 25 g/ml cycloheximide and then incubated for 1 h at room temperature before being plated onto a 96-well plate at a final density of 3 ϫ 10 4 cells/well. Prior to labeling with 125 I-ligand, cells were allowed to adhere and spread for 3 h at 37°C. Fibroblasts were then incubated with 125 I-labeled fibronectin or 70-kDa fragments for 1 h at 37°C. All microtiter wells were precoated for 1 h with 40 g/ml of 160-kDa fragments at room temperature and subsequently blocked with 2 mg/ml BSA at room temperature for 3 h or overnight at 4°C.
Matrix Assembly Assay-Matrix assembly assays were performed as previously described (29). Cycloheximide-treated fibroblasts (3 ϫ 10 4 cells/well) in serum-free media were plated into 96-well plates precoated with 160-kDa fragment of fibronectin and blocked with BSA as described above. After 3 h, the media was removed and the cells were incubated for 1 h at 37°C with 125 I-fibronectin in serum-free media in the presence or absence of H0, H120, or Hep IIa. The cultures were washed, and the wells were then broken apart and counted in a ␥-counter.
Cell Spreading Assays-Confluent cultures of human skin fibroblasts were treated with 30 mM sodium chlorate and 10 mM NaCl for 48 h in sulfate-free media and 10% sulfate free serum as described (32,33). As a control, some chlorate-treated cultures were incubated with 10 mM sodium sulfate for 24 h. Cells were then trypsinized and resuspended cells were treated with 25 g/ml cycloheximide for 1.5 h to block endogenous fibronectin synthesis. Fibroblasts (1.0 ϫ 10 4 cells/well) were plated onto 0.28-cm 2 wells coated with molar equivalents of H0 (4.3 g/ml), Hep IIa repeats (3.4 g/ml), H120 (6.6 g/ml), H89 (5.4 g/ml), or H95 (5.5 g/ml). Wells were coated for 1 h at room temperature and blocked with heat denatured 2 mg/ml BSA (29). The cells were then allowed to attach and spread for 1.5 h at 37°C. The wells were washed with phosphate-buffered saline to remove unattached cells. The attached cells were fixed with 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4, for 30 min and viewed under a phase microscope. Quantification of the degree of spreading was done by randomly counting the number of spread versus round cells. At least 100 cells/well of triplicate wells were counted.
Competition binding assays were performed in DMEM containing 25 mM HEPES and 2 mg/ml BSA as previously described (29). Briefly, microtiter wells precoated with 10 g/ml 70-kDa fragment were incubated with 125 I-labeled His-III 14 (1.6 ϫ 10 5 cpm/pmol) in the presence or absence of unlabeled proteins for 1 h at 37°C. Wells were then washed three times, separated, and counted. Nonspecific binding was measured in wells coated with 2 mg/ml BSA.
Matrix Assembly Immunofluorescence Microscopy-The ability of the recombinant fibronectin domains to inhibit fibronectin fibrillogenesis was assayed as previously described (29). Human fibroblasts were plated at confluence (5 ϫ 10 4 cells/well) onto Teflon TM -coated 12-well glass slides that were precoated with 20 g/ml of the 160-kDa cellbinding fragment of fibronectin for 2 h. Cells were allowed to attach for 1 h in DMEM with 10% fetal bovine serum, before the medium was replaced with serum-free media (DMEM, 25 mM HEPES, pH 7.4, 2 mg/ml BSA, 25 g/ml cycloheximide, 100 units/ml penicillin G, and 5 M streptomycin sulfate). After 3 h, this medium was replaced and cells were incubated overnight at 37°C with serum-free medium containing 1 g/ml (2 nM) or 3 g/ml (6 nM) human plasma fibronectin with or without recombinant proteins. Cell layers were then washed two times with phosphate-buffered saline and fixed for 30 min at room temperature with 4% paraformaldehyde, 0.1 M sodium phosphate buffer, pH 7.4. Polyclonal anti-fibronectin serum and donkey anti-rabbit IgG conjugated to rhodamine were used to label fibronectin fibrils as described (34). Both antibodies were diluted 1:100 with 1% BSA, phosphatebuffered saline. Labeled fibronectin fibrils were analyzed by epifluorescence with a Nikon Optiphot microscope and digitized images were collected with a Photometrics Image Point CCD TM camera and the Image Pro Plus TM version 1.3 program.
The Hep IIb construct containing the first 6 amino acids of the III 15 repeat was constructed using the GST expression vector. Primers used to construct this module were: sense primer, 5Ј-CCGGAATTCGCTAT-TCTTGCACCAACTGAC and antisense primer, 5Ј-CCGCTCGAGCTA-AGAGAGAGCTTCTTGTCCTGT. The restriction enzyme sites (EcoRI and XhoI), which were engineered into the PCR products to facilitate cloning, are in bold. Underlined bases indicate the stop codons.
Since recombinant III 14 repeats were thrombin-sensitive, recombinant III 14 repeats were generated as histidine-tagged fusion proteins. Histidine-tagged recombinant III 12, III 14 , and III 13-14 repeats were constructed using the pET 28aϩ expression vector (Novagen, WI). The same primers that were used to make GST-III 12 were used to make His-III 12 . The primers used to amplify the His-III 14 repeat were: sense primer, 5Ј-GAATTCGCCATTGATGCACCATCC and antisense primer, 5Ј-CTCGAGCTATGGAAGGGTTACCAGTTG. Primers used to construct the His-III 13-14 were: sense primer, 5Ј-GAATTCAATGTCAGC-CCACCAAGAAGG and antisense primer, 5Ј-CTCGAGCTATGGAAG-GGTTACCAGTTG. The His-III 14 recombinant protein spans amino acids Ala 1871 through Pro 1970 and the His-III 13-14 recombinant protein spans amino acids Asn 1782 through Pro 1970 . The restriction enzyme sites (EcoRI and XhoI), which were engineered into the PCR products to facilitate cloning, are in bold. Underlined bases indicate the stop codons. The protein expression was induced with 1 mM isopropyl-thio-␤-D-galactopyranoside and histidine-tagged proteins were purified using nickel chromatography (Qiagen, Chatsworth, CA) and imidazole elution according to the manufacturer's instructions.
Expression constructs encoding the recombinant fusion proteins GST-III 12-14 IIICS and GST-IIICS (40) were generous gifts from Dr. James McCarthy (University of Minnesota, Minneapolis, MN). Expression constructs encoding the recombinant fusion proteins GST-III [12][13][14][15] with or without the H120, H95, and H89 splice variants of the IIICS region were made as previously described (41). These proteins were purified and the GST tag was removed with thrombin as described (29). The H0 and H120 proteins were further purified on a heparin-agarose column equilibrated with 10 mM Tris, pH 7.4, 50 mM NaCl. Proteins were eluted with 10 mM Tris, pH 7.4, 500 mM NaCl. The H120 protein was additionally purified using an anti-IIICS monoclonal antibody (10G6) coupled to CNBr-Sepharose (Amersham Biosciences, Inc.).
Some recombinant and proteolytic proteins were concentrated by solvent extraction with flakes of polyethylene glycol followed by dialysis against phosphate-buffered saline or by lyophilization following dialysis against 0.2 M ammonium bicarbonate. Lyophilized proteins were reconstituted in phosphate-buffered saline or DMEM, 25 mM HEPES. All radioiodinated proteins were iodinated with carrier-free Na 125 I using the chloramine-T method (42). [12][13][14] Repeats to Serve as Soluble Competitors of Fibronectin-Previous studies had shown that proteolytic fragments of fibronectin that contain the III 12-14 repeats of the Hep II domain inhibit the incorporation of exogenously added plasma fibronectin into fibrils (29). To determine whether alternative splicing of the IIICS domain affects the biological activity of the Hep II domain, cultures were incubated with recombinant proteins containing the III 12-14 repeats of the Hep II domain and different splice variants of the IIICS domain ( Fig. 1). Fibril formation was then examined by immunofluorescence microscopy. As shown in Fig. 2A, cycloheximidetreated fibroblast cultures assemble exogenously added plasma fibronectin into fibrils (arrowheads). If the Hep IIa protein, which contains the III 12-14 repeats and the first 10 amino acids of the IIICS domain, was added to the cultures, the assembly of exogenous plasma fibronectin into fibrils was inhibited (Fig.  2B). This biological activity of the Hep II domain in fibrillogenesis was regulated by the splice pattern of the IIICS domain. If cycloheximide-treated fibroblast cultures were incubated with H120 proteins that contain the Hep II domain and the entire 120 amino acids of the IIICS domain, fibronectin fibrillogenesis was blocked (compare Fig. 2, A and C). The only fibronectin observed in these cultures were aggregates of the plasma fibronectin (arrows). In contrast, if the H0 protein that lacks the IIICS domain was used, fibronectin fibril formation was not inhibited (Fig. 2D, arrowheads). The effect of the H120 protein on fibrillogenesis was not due to the biological activity of the IIICS domain, since the IIICS domain alone did not block fibrillogenesis (Fig. 2E). The effect of the IIICS domain on the activity of Hep II domain required that the IIICS domain be tandem with the Hep II domain. If the entire IIICS domain was added simultaneously with the H0 variant, fibronectin formation was still not inhibited (Fig. 2F). This suggests that the sequences adjacent to the III 14 repeat influence the ability the Hep II domain to inhibit fibronectin fibril formation.

Alternative Splicing of the IIICS Domain Regulates the Ability of the III
To demonstrate that the amino acid sequences immediately carboxyl-terminal to the III 14 repeat can affect the biological activity of the Hep II domain during fibrillogenesis, the immunofluorescence microscopy assays were repeated using recombinant H89 and H95 proteins (Fig. 1). As shown in Fig. 2G, if the III 14 repeat of the Hep II domain was adjacent to the H89 splice variant, which has the first 89 amino acids of the IIICS domain, the Hep II domain inhibited fibrillogenesis. However, if the H95 splice variant that lacked the first 25 amino acids of the IIICS domain was adjacent to the Hep II domain, the Hep II domain had no effect on fibrillogenesis (Fig. 2H). Therefore, the ability of the Hep II domain to block fibrillogenesis is not simply due to the lack or presence of the IIICS domain, but rather is a result of the sequences that are tandem to the III 12-14 repeats.
The III 15 repeat did not appear to have any affect on the activity of the IIICS domain in the H120 variant. Recombinant proteins containing the Hep II domain, the entire IIICS domain, but lacking the III 15 repeat (Fig. 1), behaved like the H120 splice variant and also blocked fibrillogenesis (Fig. 2I). This further indicates that it was the presence of the sequences adjacent to the III 14 repeat that modulated the activity of the Hep II domain.
To further demonstrate that alternative splicing can affect the role of the Hep II domain in fibrillogenesis, a biochemical matrix assembly assay using 125 I-fibronectin was employed (43). In this assay, cycloheximide-treated fibroblasts were incubated with 125 I-fibronectin in the presence or absence of the H120 and H0 splice variants. A 70-kDa fragment shown previously to block fibrillogenesis was used as a positive control (29,44). As shown in Fig. 3, equimolar concentrations (10 Ϫ6 M) of the 70-kDa fragment and the H120 recombinant protein block the binding of 125 I-fibronectin to cycloheximide-treated cell layers by ϳ40%. However, when molar equivalents of the H0 protein were used, binding of 125 I-fibronectin to the cells was enhanced rather than blocked. This further suggests that the splice pattern of the IIICS domain differentially affected the activity of the Hep II domain in fibrillogenesis.
Alternative Splicing Does Not Affect the 70-kDa Binding Activity of the Hep II Domain-It has been generally assumed that many of the fibronectin domains that inhibit fibrillogenesis do so because they block binding of the amino terminus of fibronectin to the cell surface (29,45). Previous studies have indicated that the binding of the amino-terminal of fibronectin to the cell surface is the initial critical step in fibrillogenesis (44). To determine whether alternative splicing of the IIICS domain affected the role of the Hep II domain in fibrillogenesis by altering the amino-terminal binding activity of the III 12-14 repeats, we first identified the amino-terminal-binding site in the III 12-14 repeats.
As shown in Fig. 4, the III 14 repeat of the Hep II domain appeared to contain the majority of activity responsible for inhibiting fibrillogenesis, since fibronectin fibrillogenesis was always reduced compared with control cultures when cycloheximide-treated fibroblast cultures were incubated with recombinant proteins containing the III 14 repeat of fibronectin (III 12-14, His-III [13][14] , and His-III 14 ). In contrast, neither the III 12 nor III 13 repeats have a substantial affect on the formation of fibronectin fibrils (Fig. 4, C and D).
Solid phase binding assays show that the III 14 repeat also contains the major amino-terminal-binding site in the Hep II domain, since recombinant proteins containing either the III 14 repeat or the III 13-14 repeats efficiently bound a 70-kDa amino-terminal fragment of fibronectin (Fig. 5A, q and Ⅺ). Binding of the III 14 and III [13][14] repeats to the 70-kDa fragments occurs in a dose-dependent fashion. Both proteins appear to bind equally well to the 70-kDa fragments, as no differences were detected between the ability of the His-III 14 and His-III [13][14] to bind to the 70-kDa fragments throughout the range tested. Most of this binding activity appeared to be due to interactions with the III 14 repeat. This is illustrated by the fact that at the highest concentration tested (10 4 fmol), the recombinant III 14 repeats bind about 54 times better than the III 13 repeat (Fig. 5A, ‚) to adsorbed 70-kDa fragments. This suggests that while the III 13 repeats may contain binding sites for the amino terminus of fibronectin, the primary binding site is located within the III 14 repeat of the Hep II domain. At all concentrations tested, neither the His-III 12 repeat (Fig. 5A, OE) nor the III 12 repeat (data not shown) demonstrated any detectable binding to the adsorbed 70-kDa fragment. Binding of the His-III 14 repeat to 70-kDa fragments was not due to the histidine tag since the His-III 12 repeat did not bind at any of the concentrations tested.
Competitive binding assays with 125 I-labeled His-III 14 confirmed the specific binding interaction between the III 14 repeat and the 70-kDa fragment. As shown in Fig. 5B, recombinant proteins containing the III 14 repeat were the best competitors for binding of 125 I-His-III 14 to adsorbed 70-kDa fragments (Fig.  5B). At a concentration of 0.27 M, the His-III 14 , His-III 13-14 , and III 12-14 recombinant proteins each competed for 60 -70% of the 125 I-His-III 14 binding to the 70-kDa fragment. The competition was dose dependent since 2.7 M of the recombinant proteins containing the III 14 repeat competed for greater than 80% of the 125 I-His-III 14 binding to the 70-kDa fragment. This further indicates that the majority of the 70-kDa binding activity of the III 12-14 repeats is mediated through the III 14 repeat. In contrast, neither the His-III 12 nor the III 13 recombinant proteins had any effect on 125 I-labeled His-III 14 binding to adsorbed 70-kDa fragments. His-III 13-14 repeats (F). Only proteins containing the type III 14 repeat were able to reduce the amount of fibronectin assembled into a matrix. All cultures were fixed and labeled with a rabbit anti-fibronectin serum followed by an anti-rabbit IgG conjugated to rhodamine as described under "Materials and Methods." Bar represents 50 m.
Surprisingly, despite the fact that the III 14 repeat contains the major amino-terminal binding activity and is adjacent to the IIICS in fibronectin, alternative splicing of the IIICS domain did not affect the 70-kDa binding activity of the Hep II domain. As shown in Fig. 6, there is no significant difference in the ability of the H120 and H0 proteins to bind adsorbed 70-kDa fragments. Both these recombinant proteins bind to adsorbed 70-kDa fragments as well as the Hep IIa protein (Fig.  6, ). The binding interaction with the 70-kDa fragments was specific for the III 12-14 repeats, since the IIICS domain did not demonstrate binding to adsorbed 70-kDa fragments (Fig. 6, q). This suggests that alternative splicing of the IIICS domain has no effect on the binding interaction between the Hep II domain and the amino terminus of fibronectin.
Effect of Alternative Splicing of the IIICS Domain on the Cell Binding Activity of the Hep II Domain-Since the splice pattern of the IIICS domain had no affect on the ability of the Hep II domain to bind to the amino terminus of fibronectin, we investigated whether the splice pattern had any affect on the ability of the Hep II domain to interact with the cell surface. Previous studies have shown that the Hep II domain contains binding sites for members of the syndecan and integrin families (28,30,31). In addition, interactions with members of these families (in particular syndecan-2 or ␣ 4 ␤ 1 integrins) have the capacity to modulate fibronectin fibrillogenesis (46,47).
Cell surface binding interactions with the Hep II domain were quantified using 125 I-labeled proteins containing the Hep II domain with and without an adjacent IIICS domain. These binding assays indicated that alternative splicing of the IIICS domain did not affect binding of the Hep II domain to cells pre-spread on the 160-kDa fragment of fibronectin. As shown in Fig. 7A, at 10 Ϫ6 M the H0 proteins which contain the Hep II domain but lack a neighboring IIICS domain bind with similar apparent affinities to fibroblasts cultures as recombinant proteins which contain both the Hep II domain and the aminoterminal sequences of the IIICS domain (H120 and Hep IIa). Binding of the H120 splice variants to the cell surface was not mediated by the IIICS domain, since the IIICS domain (Fig.  7A) did not demonstrate significant binding to the fibroblast cell surface. Furthermore, both the recombinant III 14 and III [13][14] repeats bind the cell surface suggesting that the binding was mediated by at least the III 14 repeat. Whether the III 13 repeat plays a role was not determined. The III 12 repeat was not involved since it failed to show any significant binding to the cell surface (Fig. 7B).
The presence of the IIICS, however, does modulate interactions between cell surface-sulfated proteoglycans and the Hep II domain during cell spreading. To demonstrate this, fibroblasts were treated with chlorate to prevent sulfation of cell surface proteoglycans. Chlorate-treated cells were then plated on dishes pre-coated with either H0 or the Hep IIa proteins to determine the affect of alternative splicing on binding interactions between sulfated proteoglycans and the Hep II domain during cell spreading. As shown in Fig. 8A, fibroblasts efficiently spread on the Hep IIa protein. This cell spreading is mediated by interactions with sulfated proteoglycans, since chlorate-treated cells were able to attach but not spread on the Hep IIa protein. Approximately 80% of the cells on the Hep IIa protein remained round (Fig. 8B). This effect on cell spreading was reversible and cultures incubated with sulfate recovered their ability to spread on the Hep IIa protein.
As in fibrillogenesis, the sequences proximal to the III 14 repeat affects the role of the Hep II domain during cell spreading. Whereas cell spreading on the Hep IIa protein is mediated by interactions with sulfated proteoglycans, cell spreading on the H0 variant that contained the III [12][13][14][15] repeats was independent of sulfated proteoglycans. Approximately 80% of the chlorate-treated cells were able to spread on H0. In contrast, only 20% of the chlorate-treated cells could spread on the Hep IIa protein (Fig. 8B). This suggests that alternative splicing of the IIICS domain may affect interactions between sulfated proteoglycans, especially heparan sulfate proteoglycans, and the Hep II domain.
To test this hypothesis, chlorate-treated cells were also plated on the H120 splice variant that contained the entire IIICS domain. As shown in Fig. 8A, chlorate-treated cells spread poorly on the H120 splice variant compared with cells plated on the H0 splice variant and almost 80% of the cells were round. In contrast, cells were well spread on the H0 splice variant. Cell spreading on the H120 splice variant was therefore similar to cell spreading on the Hep IIa protein and is dependent on interactions with sulfated proteoglycans. A similar result was obtained when chlorate-treated fibroblasts were plated on the H89 splice variants (data not shown). In contrast, when chlorate-treated fibroblasts were plated on the H95 splice variant, the fibroblasts spread well indicating that this variant behaved like the H0 splice variant and cell spreading was now independent of interactions with sulfated proteoglycans (data not shown). Curiously, even though similar numbers of cells treated with chlorate and sulfate could spread on the H0, H120, and Hep IIa splice variants, cells treated with chlorate and sulfate did not appear to spread as well on the H120 splice variant as they did on H0 or Hep IIa splice variants (Fig. 8, A  and B). The reason for this is unclear.
Sequences Proximal to the Carboxyl Terminus of the III 14 Repeat Regulate Its Activity in Fibrillogenesis and Cell Spreading-To demonstrate further that the sequences adjacent to the carboxyl terminus of the III 14 repeat in the Hep II domain regulates the activity of the Hep II domain in fibrillogenesis and cell spreading, a second recombinant III 12-14 module (Hep IIb) was constructed. The Hep IIb protein differed from the Hep IIa protein in that it contained the first 6 amino acids of the III 15 repeat instead of the first 10 amino acids of the IIICS domain (Fig. 9). As shown in Fig. 10, the Hep IIb protein has the opposite activity of the Hep IIa protein. Thus, in cycloheximide-treated fibroblast cultures the Hep IIa protein blocked fibronectin fibrillogenesis, whereas the Hep IIb protein did not (compare panels in Fig. 10, B and C). Cell spreading on the two Hep II proteins was also different. In chlorate-treated cultures, only 20% of chlorate-treated fibroblasts could spread on the Hep IIa protein (Fig. 8B). In contrast, over half (59%) of the fibroblasts could spread on the Hep IIb protein. Since this was the same percentage of fibroblasts as untreated fibroblasts that could spread on the Hep IIb protein, this indicates that cell spreading on Hep IIb protein was not affected by the chlorate treatment (compare Fig. 10, D and E). Thus, cell spreading on Hep IIb protein was similar to that on the H0 and H95 splice variants and was independent of interactions with sulfated proteoglycans. Therefore, the amino acids carboxyl-terminal to the III 14 repeat determine the role of Hep II domain in fibrillogenesis and cell spreading. DISCUSSION In this paper we show that alternative splicing of the IIICS domain can affect the biological activity of Hep II domain in fibronectin fibrillogenesis and cell spreading. When splice variants (H120 and H89), which contain the amino terminus of the IIICS were adjacent to the III 14 repeat of the Hep II domain, the Hep II domain was able to block fibrillogenesis and cell spreading was dependent on interactions with the sulfated proteoglycans. However, if the splice variants (H0 and H95) that lacked the amino terminus of the IIICS domain were adjacent to the III 14 repeat, the Hep II domain failed to block fibronectin fibrillogenesis and cell spreading was independent of sulfated proteoglycans. This affect was due to alternative splicing varying which amino acids were proximal to the III 14 repeat, since the Hep IIb protein that contained only the first 6 amino acids of the III 15 repeat behaved like the H0 and H95 variants and the Hep IIa protein that contained the first 10 amino acids of the IIICS domain supported identical biological activities as did the H120 and H89 variants. Not all biological activities of the Hep II domain were affected by alternative splicing of the IIICS domain. Neither the ability of the III [12][13][14] repeats to bind to the fibroblast cell surface (28,30) nor the ability to bind the amino terminus of fibronectin was affected (29). This suggests that alternative splicing of the IIICS domain is a mechanism by which specific biological activities of the Hep II domain are regulated.
It is doubtful that alternative splicing of the IIICS domain or the 10 amino acids from the IIICS domain in the truncated Hep IIa domain altered the biological activities of the Hep II domain because it introduced different global conformational changes in the III [12][13][14] repeats. CD spectra of the Hep IIa domain and the H0 variants were similar, indicating that the absence of the IIICS domain or the presence of just 10 amino acids from the IIICS domain did not significantly affect the folding of the III 12-14 repeats (data not shown). Furthermore, not all the biological activities of the Hep II domain were affected by alternative splicing of the IIICS domain. All splice variants tested bound equally well to the amino terminus of fibronectin. This is unlikely to occur if major conformational changes had been introduced into the Hep II domain.
The IIICS splice variants and the amino acids at the carboxyl termini of the Hep IIa and IIb proteins could conceivably affect the biological activity of the Hep II domain by influencing interactions between the individual repeats of the Hep II domain. Thermal denaturation studies have indicated that there are strong cooperative interactions between the III [13][14] repeats and that the III 13-14 repeats act as a structural unit, even in the presence of 6 M urea (48). These interactions apparently influence the biological activity of the Hep II domain. Individual III 13 and III 14 modules bind heparin much less strongly than the parent fragments containing them (49) and as shown in this study the amino-terminal binding activity of the intact III [12][13][14] repeats and their effectiveness as an inhibitor of fibrillogenesis differed from that of the individual III 12 , IIII 13 , and III 14 repeats. Interestingly, these interactions had the same affect on the heparin binding activity of the III 14 repeat and its ability to inhibit fibrillogenesis. That is, these activities were always stronger when the repeats were associated within the intact Hep II domain. In the case of the heparin binding activities of the III 13 and III 14 repeats this may be due to the precise juxtapositioning of the III 13 and III 14 repeats required for the binding of heparin (39). A similar situation may also exist for the role of the III 14 repeat in fibrillogenesis, since fibrillogenesis has been proposed to involve the heparin binding activity of the Hep II domain (46,50). Curiously, these interactions had the opposite affect on the amino-terminal binding activity of the Hep II domain. Why these interactions would have the opposite affect on the amino-terminal binding activity of the Hep II domain is difficult to reconcile without more information on the structural requirements needed for the binding of the amino terminus to the III 12-14 repeats.
Alternative splicing of the IIICS domain may be a mechanism to influence signaling events between cell surface heparan surface proteoglycans and the Hep II domain of fibronectin. The Hep II domain contains at least two sites where interactions with heparan sulfate proteoglycans can occur; one within the III 13 repeat and another within the III 14 repeat (51)(52)(53). Both of these heparin-binding sites modulate the formation of focal adhesions and stress fibers (25,28,54) and therefore play an important signaling role in regulating cell spreading. Alternative splicing of the IIICS domain affected this role, since the H0 and H89 variants as well as the Hep IIb protein could promote cell spreading in the absence of cell surface heparan sulfate proteoglycans, whereas the Hep IIa protein and the variants H95 and H120 splice variants could not. This effect was not due to differences in the ability of H0 splice variant and the Hep IIa proteins to bind to the cell surface. In direct binding assays with radiolabeled ligands, H0, H120, and Hep IIa proteins bound equally well to the cell surface. This suggests that alternative splicing of the IIICS domain modulates the ability of the Hep II domain to interact with heparan sulfate proteoglycans and that alternative splicing of the IIICS domain may be a mechanism to restrict, or limit, the receptor-mediated signaling pathways used by the Hep II domain to control cell spreading so as to avoid unwarranted interactions.
The cell surface-binding site used by the H0 variant to promote cell spreading in this study is not known. Several reports have now indicated that adhesion and spreading may involve cooperative interactions between cell surface proteoglycans and integrins, especially ␣ 4 ␤ 1 integrins (41,55,56). Alternative splicing appears to affect cooperative interactions between the ␣ 4 ␤ 1 integrin and heparan sulfate proteoglycan-binding sites in the IIICS domain. In studies with A375-SM melanoma cells, both cell attachment and cell spreading on the H120 and H89 splice variants was almost exclusively dependent on ␣ 4 ␤ 1 integrins, whereas cell attachment and spreading on the H0 and H95 splice variants involved cooperative interactions between ␣ 5 ␤ 1 , ␣ 4 ␤ 1 , and cell-surface proteoglycans (41). The Hep II domain is similar to the IIICS domain in that it also has an ␣ 4 ␤ 1 integrin binding sequence (15,30,39,57) and heparinbinding sites (51)(52)(53)56). Thus, it is conceivable that alternative splicing is regulating the ability of the Hep II domain to promote cooperative interactions between heparan sulfate proteoglycans and ␣ 4 ␤ 1 integrins. In the case of the Hep IIa protein and the H120 and H89 splice variants, cell spreading may utilize cooperative interactions between the cell surface proteoglycan and ␣ 4 ␤ 1 integrin-binding domains. Whereas, cell spreading on the Hep IIb protein, and the H0 and H95 splice variants may only require interactions with integrin signaling pathways.
It is clear from these studies that the amino-terminal binding activity of the Hep II domain may not be the major role of these repeats in fibrillogenesis as previously suggested (29). Alternative splicing did not affect the ability of the Hep II domain to bind to the amino terminus of fibronectin, but it did affect the ability of the Hep II domain to inhibit fibrillogenesis. Thus, there was not a direct correlation between the aminoterminal binding activity of the Hep II domain and the ability to block fibrillogenesis. The III 14 and III [13][14] repeats, which had the highest affinity for the amino terminus of fibronectin, were less effective at inhibiting fibril formation than the Hep IIa protein which contained the III 12-14 repeats.
The splice variants that enable the Hep II domain to block fibrillogenesis also require interactions with sulfated proteoglycans to promote cell spreading. Therefore, the Hep II domain may be inhibiting fibrillogenesis not by binding to fibronectin directly but through signaling pathways initiated via interactions with cell surface proteoglycans. Indeed syndecan-2, a heparan sulfate proteoglycan found in fibroblasts that binds to the Hep II domain has been shown to have a role in fibrillogenesis (46,50). When Chinese hamster ovary cells were transfected with a full-length syndecan-2, fibronectin fibrillogenesis was enhanced, but when the cells were transfected with syndecan-2 containing a truncated cytoplasmic domain, matrix assembly was blocked. Syndecan-2 appears to have a regulatory role in fibrillogenesis, since the transfection of the truncated syndecan-2 can override the induction of fibril formation via activated integrins (46). Thus, the Hep II domain may be mediating fibrillogenesis via a syndecan-mediated sig-naling pathway and not via a direct fibronectin-fibronectin binding interaction as previously suggested (29).
In conclusion, alternative splicing of the IIICS domain not only affects cell-mediated adhesion events via the IIICS domain but it also affects how isoforms of fibronectin utilize the Hep II domain in fibrillogenesis and cell spreading. Fibronectins that lack the IIICS domain, as in half of plasma fibronectin dimers (12), would be unable to utilize the III 12-14 repeats to promote fibril assembly and heparan sulfate-dependent cell spreading. In contrast, cellular fibronectins that contain the amino terminus of the IIICS domain would utilize the Hep II domain to promote the assembly of fibronectin fibrils and cell spreading via heparan sulfate proteoglycans. Alternative splicing of the IIICS domain may therefore be a mechanism to generate differences in the architecture of fibronectin matrices to support the unique structure and signaling role of the extracellular matrix in different tissues.