Identification of Two Amino Acids within the EIIIA (ED-A) Segment of Fibronectin Constituting the Epitope for Two Function-blocking Monoclonal Antibodies*

Alternative splicing of the fibronectin gene transcript gives rise to a group of adhesive glycoproteins showing restricted spatial and temporal expression during embryonic development, tumor growth, and tissue repair. Alternative splicing occurs in three segments termed EIIIB, EIIIA, and V. The EIIIA (or ED-A) segment of fibronectin is expressed prominently but transiently in healing wounds coincident with fibroblast expression of an activation marker, smooth muscle cell α-actin. A monoclonal antibody (IST-9) to the EIIIA segment blocks transforming growth factor-β-mediated smooth muscle cell α-actin expression by fibroblasts in culture. A second monoclonal antibody (DH1) blocks chondrocyte condensation in chicken embryos. We find that IST-9 and DH1 react with human, rat, and chicken but not with mouse or frog EIIIA, suggesting that His44 may be important for antibody binding. A series of deletion mutants of rat EIIIA, constructed as glutathione S-transferase fusion proteins, do not react with either IST-9, DH1, or a third monoclonal antibody (3E2). Mutations of pairs of amino acids to alanine have little effect, except for either (Val34Thr35) or (Tyr36Ser37), which are located in a β strand upstream from His44. For these double mutants, the binding to all three monoclonal antibodies is markedly reduced. By contrast, single mutants at Thr35, Tyr36, or Ser37 retain full activity, suggesting that the epitope for these antibodies is determined in part by conformation. Alanine-scanning mutagenesis of rat EIIIA demonstrates the importance of Ile43 and His44 for binding. Mutation of frog EIIIA (normally Val43Lys44) to rat (Ile43His44) is sufficient to restore fully IST-9 binding and much of the activity of DH1 and 3E2. Our findings demonstrate that the function-blocking antibodies, IST-9 and DH1, bind to the Ile43 and His44 residues in a conformationally dependent fashion, implicating the loop region encompassing both residues as critical for mediating EIIIA function.

The fibronectins (FNs) 1 comprise a group of extracellular matrix proteins that mediate cell adhesion, migration, prolif-eration, and differentiation (1). FNs play significant roles in embryonic development and are prominent components of the provisional matrix following tissue injury in adults (2,3). The fundamental importance of the FNs is substantiated by the observation that homozygous mutations in either the FN gene or in the ␣ 5 integrin, a FN-specific receptor, are lethal (4,5). The FNs are disulfide-linked, dimeric glycoproteins with structural domains that bind cells, collagen, proteoglycans, and fibrin. Each FN consists of homologous repeats, either type I, II, or III. Individual type III repeats within FN exhibit high sequence similarity between species (greater than 90% identity (6)). Despite variations in protein sequence identity between different type III repeats within FN (20 -40% (6)), these repeats show a high degree of structural homology (7)(8)(9)(10)(11). X-ray crystallographic studies demonstrate that each type III repeat consists of two ␤ sheets, made up of four strands (G, F, C, CЈ) and three strands (A, B, E) respectively, folded into a ␤ sandwich (7). This structural arrangement is also conserved in other proteins, including growth hormone (12), tenascin (13), neuroglian (14), tissue factor (15), and chitinases (16).
Diversity in the FNs occurs by alternative splicing in two type III repeats termed EIIIA (or ED-A) and EIIIB (or ED-B) and one non-homologous repeat called V (or IIICS) (1). The EIIIA and EIIIB segments are either entirely included or excluded, whereas the V region may be included, excluded, or partially included in FN. An additional splicing variant lacking the V region, the 10th type I repeat (I 10 ), and the 15th type III repeat (III 15 ) has recently been reported (17). Despite the present understanding of the structures within type III repeats and the mechanisms controlling the alternative splicing of FN mRNA (18 -26), the roles of the alternatively spliced domains on overall FN protein structure and cellular function remain unclear. Alterations in FN structure, which occur as a consequence of splicing, may influence the function of adjacent domains. For example, insertion of an alternatively spliced domain may change the conformation of the adjacent type III domains by re-adjusting interdomain rotations and tilts (7,13). This is supported by the observation that a small rotation occurs between FN-III 9 and FN-III 10 and places the RGD loop of FN-III 10 and the synergy site of FN-III 9 on the same surface of the FN molecules (7). Alternatively, the inclusion of an extra type III domain could have longer range effects on the overall conformation of the FN molecule by rotating the N-terminal portion of the FN molecule relative to the C-terminus. Such a structural change could enhance the accessibility of a functional domain, such as the RGD loop, in the cell-binding domain or of sites involved in fibril assembly (8,27,28). Finally, an alternatively spliced segment may interact directly with cells.
The FNs display a wide range of physiological functions, which have been mapped to specific segments of FN. In some instances, the reacting sequences have been localized to short stretches of amino acids. For example, synthetic peptides that include the Arg-Gly-Asp (RGD) sequence from the FN-III 10 block interactions between FN and integrins (29,30). Although identified first in FN, the RGD sequence has been identified in numerous proteins and mediates cell adhesion (31). However, key peptide sequences often function best in the context of the whole type III repeat. For example, a short stretch of amino acid residues (Pro-His-Ser-Arg-Asn, PHSRN) in the 9th type III domain (FN-III 9 ), termed the synergy site, of FN has been found to enhance cell adhesion to the RGD sequence in FN (32)(33)(34)(35).
The function of the EIIIA and EIIIB sequences is largely unknown. These extra domains are present at specific stages of embryonic development and organogenesis (36 -41), whereas most normal adult tissues express much lower amounts of EIIIA and EIIIB (42). However, during specific pathological conditions, such as wound healing (43,44), lung, liver, and kidney fibrosis (45)(46)(47), vascular intimal proliferation (48,49) and cardiac transplantation (50), the expression of EIIIA and EIIIB domains is significantly up-regulated. Several lines of evidence, including studies utilizing the mAb IST-9 to block function, have shown that the EIIIA segment may play roles in promoting cell adhesion, regulating cell proliferation, and in promoting the differentiation of lipocytes and fibroblasts into myofibroblasts (27,47,(51)(52)(53)(54). Another EIIIA-specific mAb, DH1, has recently been shown to block chondrogenesis in chicken embryos (55). The differentiation of myofibroblasts is observed during morphogenetic processes, wound healing, organ fibrosis, and the stromal reaction to carcinomas (56 -58). Uncontrolled myofibroblast differentiation has been suggested to be the leading cause of several fibrotic diseases as well (59). Under pathological conditions, TGF-␤1 has been shown to be a potent inducer of the myofibroblast phenotype and can increase the expression of collagen, FN, and certain integrins by fibroblasts (60 -62). Tissue fibrosis may result from the disregulation of TGF-␤ expression (63). Recent data demonstrate a potential role for the EIIIA segment of FN in the regulation of TGF-␤'s action on dermal fibroblasts (53).
The observation that an EIIIA-specific mAb (IST-9) and a soluble form of the EIIIA segment markedly reduce TGF-␤mediated myofibroblast differentiation suggests that structural features of EIIIA are important in fibrosis. Here, we present an epitope map of EIIIA for two function-blocking mAbs, IST-9 and DH1, and another EIIIA-specific mAb, 3E2. To do so, we carried out systematic mutation of bacterial EIIIA fusion proteins, including deletion mapping and alanine-scanning mutagenesis. Our results clearly identify two amino acid residues of the EIIIA domain that are located at a loop region between two ␤ strands and constitute the IST-9 and DH1 epitopes. These data provide the foundation for the identification of a functional motif in the EIIIA domain. Various deletion constructs of rat EIIIA cDNA were generated by using PCR. The amplification was performed in a 25-l reaction volume containing 1 ng of EIIIA-pGEX-2T plasmid DNA, 10 mM Tris-HCl, pH 9.0, 50 mM KCl, 0.1% Triton X-100, 1.5 mM MgCl 2 , 200 M each dNTP, 0.5 M sense strand and antisense strand primer, and 2.5 units of Taq DNA polymerase (Promega, Madison, WI). The sequences of 5Ј and 3Ј primers are listed in Table I. Constructs were then subcloned and propagated as described above. Deletion of EIIIA cDNA was confirmed by DNA sequencing analysis.

Materials-Restriction
Site-directed Mutagenesis of the EIIIA Expression Plasmid-Point mutations were selectively introduced into the wild type rat EIIIA expression construct, EIIIA-pGEX-2T, by a procedure called unique site elimination (66) using the U.S.E. Mutagenesis kit. Synthesized oligonucleotides (300 pmol each; Table I) were phosphorylated in a 30-l reaction mixture containing One-Phor-All Buffer PLUS (10 mM Tris acetate, 10 mM magnesium acetate, 50 mM potassium acetate, pH 7.5), 1 mM ATP, and 10 units of T4 polynucleotide kinase. Phosphorylation reactions were incubated (37°C, 30 min), terminated by heating to 65°C for 10 min, and used directly in the mutagenesis reactions.
Mutagenesis mixtures consisted of 0.025 pmol of plasmid DNA, 1.25 pmol of U.S.E. selection primer (Table I), 1.25 pmol of target mutagenic primer, and One-Phor-All Buffer PLUS in a total volume of 20 l. Following incubation at 100°C for 5 min, the reaction mixtures were cooled in ice for 5 min and then incubated at room temperature for 30 min. Subsequently, 7 l of Nucleotide Mix (2.86 mM each dATP, dCTP, dGTP, and dTTP, 4.34 mM ATP, 1.43ϫ One-Phor-All Buffer PLUS) and 3 l of Reaction Mix (0.83-1.17 kilounits/ml FPL T4 DNA ligase, 0.83-1.67 kilounits/ml FPL T4 DNA polymerase, and 0.2-0.28 mg/ml T4 Gene 32 protein) were added into the reaction mixtures. The reaction mixtures were further incubated at 37°C for 1 h, followed by heating at 85°C for 15 min. Before being transformed into bacterial host cells, the mutagenesis reaction mixture was digested with 10 units of PstI in a final volume of 50 l, and transformation was carried out as described by the manufacturer. Transformed cells were incubated in 4 ml of L-broth with 100 g/ml of ampicillin at 37°C overnight with shaking at 250 rpm. Plasmid DNAs were prepared from the overnight cultures using QIAprep 8 Miniprep kit (Qiagen) and were subjected to a second restriction digestion by PstI. One g of digested DNA from the second round of restriction enzyme selection was transformed into Escherichia coli competent cells, followed by plating of the transformed cells onto LB plates containing 100 g/ml ampicillin. The plates were incubated at 37°C overnight, and individual transformant colonies were selected to prepare plasmids for sequencing to verify the presence of the desired point mutations.
DNA sequencing was performed on subcloned fragments in multicopy plasmids, PCR amplimer fragments using Taq polymerase in a dideoxy dye-terminator reaction (67). The sequencing reactions were resolved on an Applied Biosystems ABI 377 Sequencer (DNA Sequencing Core Facility, Department of Molecular Biology, Massachusetts General Hospital). The sequencing results were assembled and analyzed using the GCG Software Package (version 9.0, the University of Wisconsin and Genetics Computer Group), which includes "Pileup," "Bestfit," and "Pretty." Production and Purification of Bacterially Expressed EIIIA Proteins-A 500-ml L-broth culture containing 100 g/ml ampicillin was inoculated with 50 ml of an overnight culture of the recombinant E. coli strain and grown at 37°C for 2 h. Protein expression was then induced by the addition of isopropyl-␤-D-thiogalactoside to a final concentration of 1 mM. A protease inhibitor, AEBSF, was included in the induction culture medium with a final concentration of 1 mM. After 4 h induction, cells were harvested by centrifugation at 5,000 ϫ g for 30 min. The cell pellets were washed with PBS and used for protein purification or stored at Ϫ80°C until ready to use.
For protein purification, cell pellets were resuspended in 10 ml of PBS with 1 mM AEBSF (PBS/AEBSF) and sonicated on ice using a Sonifier 450 (Branson Ultrasonics Corp., Danbury, CT) with microtip at full power for 1 min. Subsequently, 100 l of Triton X-100 was then added into the sonicated suspension, and incubation at 4°C was carried out for 30 min. Cell debris was removed by centrifugation at 14,000 ϫ g for 30 min. The clarified supernatant was collected and mixed with 1 ml of glutathione-agarose (50% slurry pre-equilibrated with PBS/ AEBSF) at 4°C for 2 h. Protein-bound agarose beads were collected by centrifugation at 1,000 ϫ g for 1 min and washed with 10 ml of PBS/AEBSF 5 times. Washed beads were mixed with 1 ml of elution buffer (25 mM glutathione, 120 mM NaCl, and 100 mM Tris-HCl, pH 8.0) at 4°C for 10 min to elute the GST fusion proteins, followed by centrifugation at 1,000 ϫ g for 1 min. The elution step was repeated twice, and the eluted fractions were pooled and dialyzed against PBS. Purified proteins were quantitated using the BCA protein assay reagent and stored at Ϫ80°C.

SDS-Polyacrylamide Gel Electrophoresis and Western Blot
Analysis-Purified recombinant EIIIA proteins (10 g or as specified elsewhere) were mixed with an equal volume of 2ϫ SDS sample buffer (125 mM Tris-HCl, pH 6.8, 4% SDS, 0.005% bromphenol blue, 20% glycerol, 2% dithiothreitol, and 5% ␤-mercaptoethanol) and boiled at 100°C for 5 min. Samples were then loaded onto a pre-cast Tris glycine polyacrylamide gel (10%) (Novex, San Diego, CA) and resolved with 25 mA/gel in an XCell II Mini-Cell system (Novex) containing SDS running buffer (24 mM Tris base, 192 mM glycine, and 0.1% SDS) (68). Gels were then processed either for Western blot analysis or stained using Gel Code Blue for 1 h at room temperature, followed by extensive washes in distilled water.
For Western blot analysis, separated proteins were transferred electrophoretically to a polyvinylidene difluoride membrane (Bio-Rad) for 2 h at 32 V using Xcell II Blot Module (Novex) filled with Transfer Buffer (190 mM Tris, 25 mM glycine, and 20% methanol). The membrane was treated with Blocking Buffer (5% non-fat dry milk, 0.05% Tween 20 in PBS) at room temperature (overnight). Following brief washes, the membrane was reacted with an EIIIA-specific monoclonal antibody (1:500 diluted in SuperBlock) for 1 h at room temperature and washed with PBST (0.05% Tween 20 in PBS) three times (5 min each wash). Subsequently, the membrane was incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (Pierce) (1:5000 in SuperBlock), followed by three washes in PBST. The immunoblot was then incubated with Supersignal chemiluminescence substrate for 10 min and exposed to a phosphor cassette. Images of the blots and gels were processed with the Molecular Image System GS-525 and the Fluor-S MultiImager, respectively, using Multi-Analysis software version 1.1 (Bio-Rad).
Enzyme-linked Immunosorbent Assays-The reactivity of mutant EIIIA-GST fusion proteins was tested by enzyme-linked immunosorbent assays (ELISA). Microtiter plates (96-well) were coated with 100 l/well of 10 g/ml purified EIIIA-GST proteins in coating buffer (100 mM NaHCO 3 , pH 8.6) at 4°C in a humidified chamber overnight. The plates were briefly rinsed in washing buffer (0.1% Tween 20 in PBS), blocked with 300 l/well of blocking buffer (3% bovine serum albumin in coating buffer) at 37°C for 1 h, and rinsed again in washing buffer. Serial dilutions (1:60 -1:2 ϫ 10 10 ) of monoclonal antibodies (mAbs) were prepared in washing buffer. Diluted mAbs (100 l) were added to the wells, and the reactions were incubated at room temperature for 1 h. Wells were washed as above, incubated (1 h, room temperature) with diluted horseradish peroxidase-conjugated goat anti-mouse IgG (1:5000 in washing buffer, 100 l/well), washed again, and incubated (30 min,  room temperature) with substrate solution (200 l/well, 0.4 mg/ml o-phenylenediamine dihydrochloride, 0.4 mg/ml urea hydrogen peroxide, and 50 mM phosphate-citrate buffer). The absorbance of individual reactions was then measured (420 nm, ThermoMax microplate reader, Molecular Devices). Titer evaluations of antibody dilutions were done by logarithmic curve fitting. The x value corresponding to 50% of the highest absorbance along the sigmoidal curve is defined as titer for the reactivity of mAbs to specific EIIIA proteins.

Protein Sequence Comparison and Antibody Reactivities
Suggest That the C-CЈ-E Segment of EIIIA Encompassing the His 44 Residue Constitutes Epitopes-Protein sequences, derived from mRNAs, for human, mouse, rat, chicken, and frog EIIIA segments show extensive sequence similarity (Fig. 1A). All are 90 amino acids in length, and the consensus sequence for these five species is 70% conserved. The EIIIA protein sequences for mouse, rat, chicken, and frog display 96.7, 94.4, 85.6, and 80% identity, respectively, to the human EIIIA protein (Fig. 1B). All sequences conform to a domain structure in which seven ␤-strands (denoted by A, B, C, CЈ, E, F, and G) are conserved in the type III repeat crystal structure (7).
The mAb IST-9, raised against human cellular fibronectin (cFN), specifically recognizes the EIIIA segment in rat and human cFN (64). This mAb exhibits function-blocking activities in these species (47,53). We tested the reactivity of IST-9 and 3E2 against the EIIIA segment in chicken, frog, and mouse FN by either immunofluorescence, immunoblotting, or ELISA. IST-9 reacted with chicken cFN (data not shown), but not appreciably with either mouse or frog EIIIA (see below). Likewise DH1, which binds chicken FN (55), also does not react with mouse or frog EIIIA (see below). 3E2 reacts with chicken (data not shown) and mouse (41) but not frog EIIIA (see below). Comparisons of these reactivities with the protein sequences ( Fig. 1) revealed that an amino acid residue (His 44 ) within the EIIIA segment is conserved in rat, human, and chicken but is not in either mouse or frog (arrow, Fig. 1). As a preliminary test of reactivity with mouse EIIIA, we prepared a mutation that replaces the conserved His 44 residue of rat EIIIA segment with the arginine found in mouse EIIIA. This resulted in a significant reduction of IST-9 reactivity (Fig. 2). A potential polymorphism at residue Glu 54 suggests that this residue might also play a role. Based on the sequence comparison and our immunological evidence, we hypothesized that the conformational domain C-CЈ-E of rat EIIIA encompassing the His 44 residue is crucial for constituting the IST-9 epitope.
The "Native" Conformation of Full-length Rat EIIIA Protein Is Required for Its IST-9 Reactivity-To generate sufficient material for mapping the IST-9 binding epitope, the wild type sequence of rat EIIIA was cloned into a bacterial expression  (7). The arrow denotes His 44 which is conserved in human, rat, and chicken EIIIA but not in mouse and Xenopus. Underlined amino acids, Asp 53 and Glu 54 , of mouse EIIIA sequence that we obtained differ from the Glu 53 -Asp 54 in the published mouse EIIIA sequence (82). B, a dendrogram of the aligned polypeptide sequences in A. The numbers represent the percent identity of the corresponding sequence relative to the human EIIIA. Optimized multiple sequence alignment was generated using the Pileup program.
vector, pGEX-2T, and then used to generate various deletion constructs of rat EIIIA by PCR (Fig. 3). All of these deletion mutants retain the hypothesized epitope sequence of domain C-CЈ-E. Wild type and deletion mutant constructs were expressed as GST fusion proteins in E. coli and purified by glutathione-affinity chromatography.
Antibody reactivities of wild type rat EIIIA and the derived deletion mutants were tested by ELISA, and the dilution yielding 50% binding (titer) for each mAb was determined. Three EIIIA-specific mAbs, IST-9, 3E2, and DH1, were included in these analyses. When reacted with wild type rat EIIIA, the titers of IST-9, 3E2, and DH1 were 5 ϫ 10 4 , 4 ϫ 10 3 , and 1 ϫ 10 4 , respectively (Fig. 4). However, unlike the strong reactivity exhibited by rat EIIIA toward these mAbs, none of the six deletion mutants displayed any detectable antibody reactivity (insets, Fig. 4). These results ruled out the possibility that the C-CЈ-E domain alone of the EIIIA segment was sufficient to constitute the antibody recognition epitope. Nevertheless, the amino acid sequence of C-CЈ-E domain could still be crucial for the reactivity with mAbs and may need to be maintained in a specific conformation. This antibody-reactive conformation may only react with these mAbs when the full-length EIIIA polypeptide sequence is intact.
Thr 35 , Tyr 36 , Ser 37 , Glu 40 , and Asp 41 Residues of Rat EIIIA Are Important for Maintaining an Optimal Conformation for Antibody Binding-To test whether and to what extent the C-CЈ-E domain of EIIIA protein is crucial for antibody reactivity, we generated a series of rat EIIIA double mutants in which two adjacent amino acids were simultaneously replaced by alanines and then tested for reactivity to mAbs by ELISA. In some preliminary screenings, antibody reactivities were determined by Western blotting. We found that most of the double mutants retain some or all of the antibody reactivity of the wild type rat EIIIA protein (Table II). Among these was a double mutant, rat EIIIA-D53A/E54A, representing a potential polymorphism in mouse EIIIA ( Fig. 1 and Table II). No loss of reactivity to IST-9 and 3E2 and a minimal effect on DH1 binding were observed (Fig. 5, Table II), indicating that His 44 rather than Glu 54 was most important for reactivity. However, some double mutants sharply reduced antibody reactivity (Fig.  5). One of these mutants, rat EIIIA-V34A/T35A, completely lost its ability to react with all three mAbs tested. Another double mutant, rat EIIIA-Y36A/S37A, displayed a complete loss of reactivity to IST-9 and 3E2 but still retained low (ϳ1%) reactivity to DH1, as compared with wild type EIIIA. An additional double mutant, rat EIIIA-E40A/D41A, showed a dramatic reduction in reactivity to IST-9 (3% of rat EIIIA-WT) and less significant effects on the reactivities of 3E2 and DH1. On the other hand, rat EIIIA-P50A/D51A had no significant effect on IST-9 reactivity and only minor effects on the reactivities of 3E2 and DH1.
The reduction in antibody reactivity exhibited by rat EIIIA-V34A/T35A and rat EIIIA-Y36A/S37A could result directly from disruption of the antibody-recognition epitope or indirectly by changing the tertiary structure of the EIIIA segment, which in turn alters the conformation of the epitope. To distinguish these possibilities, we prepared single mutants that contained only one alanine replacement at Thr 35 , Tyr 36 , Ser 37 , Glu 40 and Asp 41 , respectively. These were tested for their reactivities with mAbs. Despite our observation that the double mutants (rat EIIIA-V34A/T35A, rat EIIIA-Y36A/S37A, and rat EIIIA-E40A/D41A) showed either a complete loss or a dramatic reduction in antibody reactivity, we found that the single mutants, rat EIIIA-T35A, rat EIIIA-Y36A, rat EIIIA-S37A, rat EIIIA-E40A, and rat EIIIA-D41A, exhibited antibody reactivity levels comparable to those of wild type rat EIIIA (Fig. 6 and Table II), implicating the amino acid residues, Thr 35 , Tyr 36 , Ser 37 , Glu 40 and Asp 41 , in maintaining an optimal conformation of the epitope.
Ile 43 and His 44 of Rat EIIIA Segment Are Critical for IST-9 Binding-Site-directed mutagenesis of the rat EIIIA segment was conducted to generate a panel of single mutants. These mutants were tested for reactivity to IST-9, 3E2, and DH1 ( Fig.  6 and Table II). In one mutant, rat EIIIA-H44R, an arginine residue was substituted for histidine to mimic the Arg 44 (H44R) in the mouse EIIIA segment. A striking loss of reactivity for IST-9 and DH1 was observed (solid and striped bars, Fig. 6). By contrast, reactivity with 3E2 was retained (gray bar, Fig. 6), consistent with the reported use of this mAb in immunohistochemistry of mouse tissues (41).
When alanine-scanning mutagenesis was carried out from Arg 33 through Pro 48 , the majority of single mutants retained reactivity to IST-9, 3E2, and DH1 ( Fig. 6 and Table II). As observed for H44R, when alanine was substituted for arginine (H44A), a marked decrease in the reactivity of mAbs IST-9 and DH1 was observed. By contrast, 3E2 retained reactivity to H44A as it did to H44R. When the adjacent isoleucine residue was mutated to alanine (I43A), IST-9 reactivity was markedly reduced. These results indicated that Ile 43 and His 44 were required for IST-9 and DH1 binding. These data also indicated that IST-9, 3E2, and DH1 recognized distinct but overlapping epitopes.
Ile 43 and His 44 Substitution in Frog EIIIA Protein Restore Antibody Binding-To establish whether or not Ile 43 and His 44 are sufficient to constitute the IST-9 binding epitope, we sought to introduce these two residues into comparable positions of a homologous type III repeat that does not react with IST-9. We found that neither IST-9, 3E2, nor DH1 reacted with frog EIIIA (Fig. 7B). Although frog EIIIA does not react with these mAbs, it displays an 80% homology in protein sequence to both human and rat EIIIA (Figs. 1B and 7A). Moreover, it is well established that type III repeats are highly related structurally (7). We prepared a double (V43I/K44H) and a single (K44H) mutant and observed that IST-9 reacted with these mutants at levels comparable to, or higher than, wild type rat EIIIA (solid bar, Fig. 7B). Thus, either Ile 43 and His 44 together or frog Val 43 with His 44 are sufficient to reconstitute IST-9 reactivity. These results taken with those for rat I43A (Fig. 6) indicate that specific amino acids at position 43 in conjunction with His at position 44 are required for IST-9 binding. On the other hand, both the double (V43I/K44H) and single (K44H) mutants exhibited about 50% of the binding to 3E2 and DH1 relative to wild type rat EIIIA (gray bar and striped bar, Fig. 7B). In   FIG. 2. Immunochemical studies of the EIIIA segment of FN. Western blot of bacterial EIIIA fusion proteins reacting with IST-9. Lane 1, wild type rat EIIIA protein; lane 2, a mutant rat EIIIA protein (rat EIIIA-H44R) in which the conserved His 44 residue found in rat FN was replaced by an arginine present in mouse FN. Proteins (3 g per lane) were loaded, resolved, and blotted as described under "Experimental Procedures." The His to Arg mutation significantly reduced the IST-9 reactivity of the rat EIIIA-H44R mutant protein.
summary, Ile 43 and His 44 in conjunction with the conformation of the EIIIA segment, likely involving the C domain, are critical to the IST-9 epitope and contribute to the epitopes for 3E2 and DH1 (Fig. 8). DISCUSSION We report here that two key amino acids, Ile 43 and His 44 , are a necessary part of the epitope for a function-blocking monoclonal antibody (IST-9) that reacts with the EIIIA segment of human, rat, and chicken FN. The Ile 43 and His 44 residues lie in a loop between two ␤ strands (C and CЈ). The overall conformation of the EIIIA segment, and particularly that conferred by the C domain, exerts important effects on the IST-9 epitope (Fig. 8). The epitopes for DH1 and 3E2 are also conformationally dependent and appear to reside in the same loop. The key residues for DH1 and 3E2 binding overlap with, and yet are distinct from, those for IST-9. We infer that the loop between the C and CЈ ␤ strands is critical to the mechanism by which the EIIIA segment regulates cell function.
Emerging evidence suggests that alternatively spliced FNs, prominent in embryogenesis, are important in the adult tissue response to injury. The expression of FNs that include the EIIIA or EIIIB segments are present in rather low amounts in many adult tissues (42). A stereotypical pattern of response to injury is evident in which FNs that lack the EIIIA and EIIIB segments (i.e. plasma FN) are deposited first. Following tissue injury, the EIIIA and EIIIB segments are included in the FN mRNAs synthesized by wound cells. This occurs in skin (43,44), arteries (48,49,69), kidneys (70,71), liver (47), and heart (50,72,73). Such up-regulation occurs both as a consequence of increases in total FN and increases in the ratios of inclusion of segments into FN (50,74). Moreover, the temporal pattern of appearance of EIIIA and EIIIB differ, suggesting distinct roles for each segment (47,50). Splicing in a nonhomologous repeat, the V region, also occurs following injury. The V95 segment is included in most FNs, but variations occur in the inclusion of the cell adhesive portion, termed CS-1, during regeneration of peripheral nerves (75) and following cardiac transplantation (50).
In addition to data on expression, a growing body of evidence supports a functional role for the EIIIA segment in modulating the phenotype of cells. A mAb, IST-9, blocks the conversion of lipocytes to myofibroblasts (47). Another mAb, DH1, has been shown to block chondrogenesis (55). Recent studies also demonstrate that IST-9 blocks the stimulatory activity of TGF-␤ on smooth muscle cell ␣-actin expression during myofibroblast differentiation (53) and that TGF-␤ controls the expression of extracellular matrix molecules and certain integrins (60, 61, 76). Thus, disregulation in the TGF-␤ signaling pathway may result in tissue fibrosis (63,77,78). Because the EIIIA segment appears to be involved in the TGF-␤ signaling mechanism in myofibroblasts, identification of the IST-9 epitope on EIIIA provides an important first step to probing this mechanism.
In this report we observe that the Ile 43 and His 44 residues of EIIIA are necessary for IST-9 binding and partially constitute the reactive epitopes for DH1 and 3E2. We have shown that none of the three mAbs react with frog EIIIA unless the Lys 44 is replaced by a His residue found in human, rat, and chicken FN (Fig. 7). Replacement of the His 44 with Arg by site-directed mutagenesis in rat EIIIA (Fig. 6) or Lys which occurs in frog TABLE II Epitope mapping of three rat EIIIA-specific monoclonal antibodies using deleted and mutagenic rat EIIIA proteins 1 The antibody reactivities of mutant EIIIA proteins were determined by Western blotting and ELISA. Reactivity of each mutant is shown relative to the reactivity of EIIIA-WT protein (100%). ϩ, Ͼ20% reactivity; ϩ/Ϫ, 1-20% reactivity; Ϫ, Ͻ1% reactivity; nd, not determined. 2 The domains are defined by the crystallography of a human FN fragment (FN-III 7-10 ) elucidated by Leahy et al. (7). * Deletion constructs containing the indicated amino acids.
FIG. 5. Reaction of mAbs with rat EIIIA double mutants. Recombinant EIIIA protein was mutagenized simultaneously at two amino acid residues replacing with alanines. The specificity of mAbs for mutated rat EIIIA (rEIIIA) fusion proteins was determined by ELISA, and the titer of each mAb to mutant rat EIIIA was determined relative to wild type rat EIIIA (rEIIIA-WT). Solid, gray, and striped bars represent the relative titer of IST-9, 3E2, and DH1, respectively. Amino acids in the EIIIA protein are indicated by single-letter code followed by the residue number representing the mutated position relative to the beginning of the EIIIA segment.
FIG. 6. Alanine scanning mutagenesis of rat EIIIA and reactivity to mAbs. Mutagenized rat EIIIA (rEIIIA) fusion proteins were generated and purified as described under "Experimental Procedures." The replaced amino acid in each mutant is indicated. The titer values are given in percentage (%) using results obtained with wild type rat EIIIA (rEIIIA-WT) as the 100% reference value. Solid, gray, and striped bars indicate the relative titer of IST-9, 3E2, and DH1, respectively, as described in Fig. 5.  FN (Fig. 7) does not suffice for either IST-9 or DH1 binding. In contrast, mutation at His 44 made to either Arg or Ala does not significantly affect 3E2 binding (Fig. 6). Although full activity is restored to IST-9 by His 44 , partial reactivity is regained with DH1 and 3E2. The adjacent residue, Ile 43 , is important to all three mAb epitopes as well. Substitution of an Ala for Ile 43 blocks IST-9 binding completely and markedly reduces 3E2 and DH1 binding (Fig. 6). Reduction in binding was also observed when Asp 41 and Gly 42 were mutagenized to Ala. Gly 42 appeared to be quite important for the epitope for DH1. Taken together, these results indicate that the loop between the C and CЈ ␤-strands encompasses the epitope for these mAbs.
Interestingly, when either the Pro 39 or the Glu 45 is replaced by alanine, the IST-9 reactivity is enhanced. Replacement with alanine at Pro 39 may relieve the steric restriction imposed by the proline residue and hence make the conformation of the His 44 -containing C-CЈ loop more reactive. On the other hand, the charged Glu 45 may provide electrochemical interaction with the adjacent His 44 residue, and this charge-charge interaction could be disrupted by the replacement of an uncharged residue like alanine, potentially making the Ile 43 and His 44 residues more accessible to IST-9. However, the enhancement of IST-9 reactivity due to mutations made to Pro 39 and Glu 45 residues was not observed with either 3E2 or DH1, indicating that rat EIIIA presents distinctive epitopes for each mAb.
The reactivities of rat EIIIA deletion constructs to the three mAbs demonstrate that these epitopes are only active in an intact polypeptide (Fig. 4). The GST moiety in the purified EIIIA fusion proteins does not appear to alter the conformation of rat EIIIA because comparable reactivities are obtained when the GST and EIIIA moieties are separated by thrombin cleavage (data not shown). The influence of molecular conformation on rat EIIIA reactivity is further demonstrated by the antibody reactivities of rat EIIIA double mutants. Mutation of pairs of amino acids to Ala, either Val 34 Thr 35 or Tyr 36 Ser 37 , located in a ␤-strand upstream from His 44 , markedly reduces antibody binding, while these as single mutants or other mutant pairs tested remain active (Fig. 5, Fig. 6, and Table II). It is likely that the double mutants at Val 34 Thr 35 or Tyr 36 Ser 37 have a greater impact on the conformation of the ␤-strand C, which in turn would result in a significant change in the conformation and antibody reactivity of the downstream C-CЈ loop.
The loop regions between ␤-strands of type III repeats often serve functional roles, such as cell adhesion and antibody recognition. Indeed, the type III repeat structure is highly conserved in tenascin (13), neuroglian (14), human growth hormone receptor (12), and the isolated FN-III 10 of FN (10, 79), despite relatively low levels of sequence identity among FN-III repeats. The most notable sequence differences are observed when comparing the loop regions of different type III repeats, suggesting that these loops may mediate functions as well as antibody epitopes (10,80).
In conclusion, we have demonstrated that the critical residues mediating antibody recognition by two function-blocking FIG. 7. Reaction of mAbs with the frog EIIIA fusion proteins. Wild type frog EIIIA (fEIIIA-WT) fusion protein with the sequence Val 43 -Lys 44 was mutated at these two residues (fEIIIA-V43I/K44H) to resemble rat EIIIA. Alternatively, fEIIIA-WT was mutated only at one position (fEIIIA-K44H). A, the sequence comparison between wild type rat EIIIA (rEIIIA-WT) and frog EIIIA is shown in A where the mutated residues are highlighted. Capital letters denote the conserved peptide sequences among human, mouse, rat, chicken, and Xenopus EIIIA proteins. B, the titer of mAbs to rEIIIA-WT, fEIIIA-WT and mutated fEIIIA fusion proteins. Solid, gray, and striped bars indicate the relative titer of IST-9, 3E2, and DH1, respectively. Titer values are determined by using the results obtained with rEIIIA-WT as 100% reference value.
FIG. 8. The epitopes for IST-9 and DH1 within EIIIA. The amino acid sequence of wild type rat EIIIA (rEIIIA-WT) is shown, and the corresponding anti-parallel ␤-strands are indicated by short heavy lines denoted by A, B, C, CЈ, E, F, and G, respectively, based on the x-ray crystallography of human fibronectin (7). The predicted epitopes of the EIIIA segment that react with two mAbs, IST-9 and DH1, are highlighted by the shaded box which encompasses the two critical residues (Ile 43 and His 44 ) denoted by an arrow. mAbs are located at the C-CЈ loop of the EIIIA domain. Because IST-9 is a function-blocking mAb, these data support a model in which the functional motif resides within the rat EIIIA domain in the proximity of the Ile 43 His 44 . It is not known yet if either the EIIIA segment interacts directly with a cell-surface receptor (53) or influences indirectly the activity of another segment of FN (27). In either case, the EIIIAϩFN-triggered signals may converge onto the TGF-␤-mediated signaling pathway and thus modulate the stimulatory activity of TGF-␤. The detailed characterization of the EIIIA structure will provide a basis for developing therapeutic strategies for modulating excessive scarring and tissue fibrosis.