Identification of Protein-disulfide Isomerase Activity in Fibronectin*

Assembly and degradation of fibronectin-containing extracellular matrices are dynamic processes that are up-regulated during wound healing, embryogenesis, and metastasis. Although several of the early steps leading to fibronectin deposition have been identified, the mechanisms leading to the accumulation of fibronectin in disulfide-stabilized multimers are largely unknown. Disulfide-stabilized fibronectin multimers are thought to arise through intra- or intermolecular disulfide exchange. Several proteins involved in disulfide exchange reactions contain the sequence Cys-X-X-Cys in their active sites, including thioredoxin and protein-disulfide isomerase. The twelfth type I module of fibronectin (I12) contains a Cys-X-X-Cys motif, suggesting that fibronectin may have the intrinsic ability to catalyze disulfide bond rearrangement. Using an established protein refolding assay, we demonstrate here that fibronectin has protein-disulfide isomerase activity and that this activity is localized to the carboxyl-terminal type I module I12. I12 was as active on an equal molar basis as intact fibronectin, indicating that most of the protein-disulfide isomerase activity of fibronectin is localized to I12. Moreover, the protein-disulfide isomerase activity of fibronectin appears to be partially cryptic since limited proteolysis of I10–I12 increased its isomerase activity and dramatically enhanced the rate of RNase refolding. This is the first demonstration that fibronectin contains protein-disulfide isomerase activity and suggests that cross-linking of fibronectin in the extracellular matrix may be catalyzed by a disulfide isomerase activity contained within the fibronectin molecule.

Multimeric fibronectin is a major constituent of extracellular matrices found throughout the body and plays a role in a wide variety of biological events, including maintenance of endothelial cell integrity, platelet adhesion, and cell migration during blood vessel repair (1)(2)(3). Fibronectin circulates at high concentrations in plasma as a soluble dimeric molecule and exists in an insoluble multimeric form in the extracellular matrix of loose connective tissue, granulation tissue, and basement membranes (1)(2)(3). This insoluble multimeric form of fibronectin is thought to be the primary functional form of the molecule (1,(3)(4)(5), mediating adhesive and migratory events associated with wound repair, neovascularization, and embryonic development.
Deposition of fibronectin into the extracellular matrix is a cell-mediated, multistep process that involves the binding of soluble fibronectin to specific sites on the surface of substrateattached cells that have been termed matrix assembly sites (6). Subsequent homophilic binding interactions between fibronectin molecules lead to the deposition of high molecular mass, disulfide-stabilized multimers into the extracellular matrix (7)(8)(9)(10)(11). Fibronectin that is deposited in tissues in vivo and in the matrix of cells cultured in vitro is in the form of high molecular mass multimers (1,(12)(13)(14). This multimeric fibronectin can be converted to monomeric fibronectin upon treatment with disulfide-reducing agents (12)(13)(14), suggesting that fibronectin in the extracellular matrix is stabilized by disulfide cross-linking. It has been postulated that this cross-linking event occurs by a disulfide exchange mechanism involving type I or II modules in the 70-kDa amino terminus of the molecule (15). However, the regions involved in disulfide cross-linking of fibronectin in the extracellular matrix have not been identified (16). In addition, the mechanisms of this cross-linking are unknown.
A number of proteins involved in disulfide exchange reactions, including protein-disulfide isomerase and thioredoxin (17)(18)(19)(20), contain the sequence Cys-X-X-Cys in their active sites. The fifth and sixth cysteines in the twelfth type I module of fibronectin (I 12 ) are arranged in a similar sequence, Cys-Asp-Asn-Cys (21), suggesting that fibronectin may contain a disulfide isomerase activity. To determine whether fibronectin contains an intrinsic protein-disulfide isomerase activity, we assayed fibronectin as well as proteolytic and recombinant fragments of fibronectin for their ability to reactivate reduced and denatured RNase. Using an established protein refolding assay (22,23), we found that fibronectin contains a proteindisulfide isomerase-and thioredoxin-like activity and that this activity is localized to I 12 . This is the first demonstration that fibronectin contains protein-disulfide isomerase activity and suggests that fibronectin may catalyze disulfide bond rearrangement during its incorporation into the extracellular matrix.
Fibronectin and Fibronectin Fragments-Human fibronectin was purified from a fibronectin and fibrinogen-rich by-product of factor VIII production as described (24). The 160/180-kDa proteolytic fragment of fibronectin was generated by limited trypsin digestion of intact fibronectin essentially as described (25). The 40-kDa (cathepsin and trypsin) gelatin-binding and 70-kDa (cathepsin) amino-terminal fragments of fibronectin were the generous gifts of Drs. Paula McKeown-Longo and Denise Hocking (Albany Medical College, Albany, NY) and were made as described (6,26). The 110-and 19-kDa thermolysin fragments of fibronectin were the generous gifts of Dr. Ken Ingham (American Red Cross, Bethesda, MD) and were prepared as described (27)(28)(29). A schematic diagram of various fibronectin fragments used in this study is shown in Fig. 3 (inset).
Generation of I 12 /pVL1392/His 6 -The baculovirus expression vector pVL1392 (Invitrogen, Carlsbad, CA) was modified so that genes cloned into the multiple cloning site would be produced with six histidines at the carboxyl terminus of the protein. The complementary oligonucleotides 5Ј-AATTCTTCACCATCACCATCACCATTGATCAG-3Ј and 5Ј-GATCCTGATCAATGGTGATGGTGATGGTGAAG-3Ј (encoding the histidine tag and containing EcoRI and BamHI overhangs) were hybridized to each other and then ligated to pVL1392 that was previously digested with EcoRI and BamHI. I 12 was amplified with the sense primer 5Ј-CCCAGATCTACTCTCCTCCCATCCACTCAAG-3Ј and the antisense primer 5Ј-CCCGAATTCCAGCCCCAGGTCTGCGGCAG-3Ј using I 12 /GE-1 as a template (30). The sense primer has a BglII restriction enzyme site at the 5Ј-end (underlined letters), and the antisense primer has an EcoRI restriction enzyme site at the 5Ј-end (underlined letters). The base in boldface (C) was added to the antisense primer following the EcoRI site to maintain the correct reading frame. Polymerase chain reaction-amplified DNA was subcloned into the baculovirus vector pVL1392/His 6 . Cloning I 12 upstream of the EcoRI site resulted in the addition of Gly-Ile-Leu before the six-histidine tag. Amplified DNA contains sequences coding for the fibronectin signal sequence and the I 12 coding sequence. DNA was sequenced (31) to ensure that no DNA mutations were introduced during polymerase chain reaction amplification and to verify the sequence of the modified pVL1392 vector.
Baculovirus Expression-I 12 /pVL1392/His 6 was cotransfected into insect cells with Baculogold DNA (Pharmingen, San Diego, CA) following the manufacturer's instructions. Recombinant viruses were prepared using established methods (32,33). SF21 insect cells were grown under serum-free conditions using SF900-II (Life Technologies, Inc.). Conditioned medium containing recombinant I 12 was applied to an SP Sephadex C-25 column equilibrated with 60 mM NaCl and 100 mM Tris, pH 6.2. The column was washed with 250 mM NaCl and 100 mM Tris, pH 6.2, and then eluted with 350 mM NaCl and 100 mM Tris, pH 6.2. Purified I 12 was dialyzed into phosphate-buffered saline or into 30 mM NaCl, 1 mM EDTA, and 0.1 M Tris prior to use in the RNase refolding assay. Purity of proteins was assessed using a discontinuous Tricine 1 / SDS-polyacrylamide gel electrophoresis system according to the method of Schagger and von Jagow (34) and visualized with either 0.025% Serva blue G (Serva, Paramus, NJ) or silver nitrate (35).
Production and Reactivation of Reduced and Denatured RNase-RNase was reduced and denatured essentially as described by Pigiet and Schuster (22). Briefly, RNase A (30 mg) was incubated overnight in 6 M guanidine HCl containing 0.15 M dithiothreitol and 0.1 M Tris, pH 8.6, at room temperature. RNase was then purified on a Sephadex G-25 column equilibrated with 0.1% acetic acid. Peak samples were pooled, nitrogen-sparged, and then stored at Ϫ80°C until use. Reduced and denatured RNase at a final concentration of 400 g/ml (30 M) was mixed with various proteins in a reaction mixture containing 0.1 M Tris, pH 7.4, and 1 mM EDTA as described (22,23). The extent of RNase reactivation was determined by removing aliquots of the reaction mixture at various time intervals and measuring RNase activity as described (23,36). The absorbance change at 284 nm was recorded in a final assay mixture that included 1.4 M RNase and 0.44 mM cytidine 2Ј:3Ј-cyclic monophosphate in 0.1 M MOPS, pH 7.0.
No contaminating RNase activity was detected in any of the purified proteins used in the RNase refolding assay, as assessed by the inability of purified proteins to hydrolyze cytidine 2Ј:3Ј-cyclic monophosphate in the absence of added RNase (data not shown). It is also unlikely that purified proteins contained small amounts of contaminating redox reagents since the addition of 5 M to 2 mM reduced or 4 -100 M oxidized dithiothreitol did not accelerate the rate of RNase refolding in comparison with controls (data not shown).
Statistical Analysis-Statistical analysis was performed using a oneway analysis of variance. Significant differences between groups were identified using the Tukey HSD multiple comparison post hoc test. Statistics were performed using Statistica software (StatSoft, Tulsa, OK).
Determination of Enzymatic Parameters-Plots of substrate concentration versus initial velocity were fit to the Michaelis-Menten equation using the Enzfitter program (Biosoft, Ferguson, MO) or DeltaGraph (Deltapoint, Monterey, CA). Lineweaver-Burk plots thus derived were used to determine V max and K m values. To determine V max and K m , the standard RNase refolding assay (described above) was performed in the presence of fibronectin (3 M), protein-disulfide isomerase (1.5 M), or thioredoxin (30 M) using RNase concentrations ranging from 15 to 75 M. At various time intervals, aliquots of the reaction mixture were removed, and the amount of active RNase was determined by measuring the change in absorbance at 284 nm over time (as described above). A standard curve was generated in which the change in absorbance at 284 nm over time was recorded for various concentrations of native RNase. Initial velocities for the Michaelis-Menten plots (M native RNase generated per min) were calculated from the slope of the line derived from the plot of native RNase concentration versus time.
19-and 40-kDa Fragment Digestion with ␣-Chymotrypsin-The 19or 40-kDa fragment was incubated with ␣-chymotrypsin at a protein ratio of 1:200 (w/w) for 1 h at 37°C. The reaction was terminated with the addition of 2 mM phenylmethylsulfonyl fluoride (37). Samples of digested proteins were removed and analyzed by Tricine/SDS-polyacrylamide gel electrophoresis prior to use in the RNase refolding assay.

Fibronectin Contains Protein-disulfide Isomerase Activity-
Soluble protomeric fibronectin is believed to be stabilized in the extracellular matrix as a result of disulfide cross-linking. Fibronectin contains a Cys-X-X-Cys motif in I 12 (21). This motif has been identified as the active-site sequence in other proteins that exhibit disulfide isomerase activity (17)(18)(19)(20). To determine whether fibronectin has disulfide isomerase activity, we used an established protein refolding assay. This assay has been widely used to assess the isomerase activity of thioredoxin (22,23,38), protein-disulfide isomerase (38), and other proteins (23,39) and measures the ability of proteins to catalyze the refolding of reduced and denatured RNase (36). During the process of RNase refolding, non-native as well as native disulfide bonds form (22, 40 -43). Therefore, both disulfide oxidation as well as isomerization reactions must occur to achieve the native folded state.
Intact fibronectin at a concentration of 4 M enhanced the refolding of RNase at a rate similar to that of 1 M proteindisulfide isomerase (Fig. 1) and 30 M thioredoxin (data not shown). A control protein, ovalbumin, which contains disulfide bonds as well as free sulfhydryl residues, but lacks the Cys-X-X-Cys motif, showed little activity in this assay (Fig. 1). The ability of fibronectin to catalyze RNase refolding was concentration-dependent. Intact fibronectin at a concentration of 3 M had ϳ84% of the activity of 4 M fibronectin; 2 M fibronectin had ϳ74% of the activity (data not shown). To directly compare the ability of fibronectin, protein-disulfide isomerase, and thioredoxin to catalyze the refolding of reduced and denatured RNase, k cat and K m values were determined. The ability of fibronectin to catalyze RNase refolding followed typical Michaelis-Menten kinetics (Fig. 2B, inset). The reactions catalyzed by protein-disulfide isomerase ( Fig. 2A, inset) and thioredoxin (data not shown) also followed typical Michaelis-Menten kinetics. Lineweaver-Burk plots ( Fig. 2 and data not shown) were used to determine the V max , K m , and k cat /K m values for each of these proteins (Table I). As shown in Table I, fibronectin is ϳ9-fold more active than thioredoxin and 9-fold less active than protein-disulfide isomerase in catalyzing RNase refolding.
To localize the disulfide isomerase activity within fibronectin, proteolytic fragments of fibronectin encompassing the entire fibronectin molecule (Fig. 3, inset) were assayed for activity. Fragments of fibronectin that demonstrated the most activity in this assay were the 160/180-and 19-kDa fragments, which contain the carboxyl-terminal type I modules I 10 -I 12 (Fig. 3). The 40-and 70-kDa fragments, which contain disulfides, but do not contain the Cys-X-X-Cys motif, had activity only slightly higher than the control protein, ovalbumin (Fig.  3). The central 110-kDa cell-binding fragment of fibronectin, which contains one free sulfhydryl residue, but no disulfide bonds (21), had little activity in this assay (Fig. 3). These data indicate that the protein-disulfide isomerase activity of fi-bronectin is localized to the 19-kDa fragment that contains I 10 -I 12 .
I 12 Demonstrates Disulfide Isomerase Activity in the RNase Refolding Assay-I 12 is located within the 19-kDa fragment and contains the Cys-X-X-Cys motif. Therefore, to determine whether the disulfide isomerase activity of the 19-kDa fragment could be further localized to I 12 , recombinant I 12 was generated using a baculovirus expression system and tested in the RNase refolding assay. As shown in Fig. 4, I 12 at a concentration of 4 M was more active than 4 M intact fibronectin and as active as 1 M protein-disulfide isomerase. These data suggest that most or all of fibronectin's protein-disulfide isomerase activity is localized to I 12 .
Regulation of the Protein-disulfide Isomerase Activity of Fibronectin-Deposition of fibronectin into the extracellular matrix is a highly regulated process (44 -48). If fibronectin's protein-disulfide isomerase activity is involved in cross-linking of fibronectin in the extracellular matrix, then it would be expected that its protein-disulfide isomerase activity would also be tightly regulated. One mechanism by which this activity may be regulated is through conformation-induced activation. In such a model, isomerase activity would be partially masked in the native soluble fibronectin molecule and enhanced by conformational changes induced upon binding of fibronectin to cell surfaces. If fibronectin's protein-disulfide isomerase activity is partially cryptic, limited proteolysis might enhance its protein-disulfide isomerase activity by generating smaller fragments whose conformations are distinct from those in the intact molecule. Others have shown that proteolytic fragments of proteins can have enhanced or novel activities when compared with native proteins (49,50). For example, the 40-kDa gelatinbinding fragment and the amino-and carboxyl-terminal fibrinbinding fragments of fibronectin have enhanced chemotactic activity when compared with the activity of intact fibronectin (49,50). In addition, proteolytic fragments of plasminogen (51) and collagen XVIII (52) have anti-angiogenic properties not associated with the native molecules.
To determine whether fibronectin's protein-disulfide isomerase activity could be increased by limited proteolysis, the 19-kDa fragment of fibronectin (containing I 10 -12 ) was digested with chymotrypsin and assayed for its ability to catalyze RNase refolding. Analysis of the digested and undigested 19-kDa fragment by Tricine/SDS-polyacrylamide gel electrophoresis indicated that digestion of the 19-kDa fragment by chymotrypsin resulted in generation of fragments with apparent molecular , and a linear fit of the data was calculated using Delta-Graph. The initial velocity was calculated as described under "Experimental Procedures." V max was calculated from the reciprocal of the y intercept, and K m from the reciprocal of the x intercept. Insets,Michaelis-Menten plots. The data generated in the assay described above were fitted to the Michaelis-Menten equation as described under "Experimental Procedures." rdRNase, reduced and denatured RNase. masses of ϳ14, 12, and 10 kDa (data not shown). The chymotrypsin-generated fragments of the 19-kDa fragment at a concentration of 1 M had activity similar to 4 M intact 19-kDa fragment as measured by the RNase refolding assay (Fig. 5A). Moreover, chymotrypsin proteolysis of the 19-kDa fragment (4 M) resulted in a dramatic increase in the rate of RNase refolding ( Fig. 5A and Table II). Table II compares the time required for RNase to reach 50% of maximal refolding in the presence of various proteins or fibronectin fragments. Whereas the intact 19-kDa fragment required 66 h to refold 50% of the RNase, proteolytic 19-kDa fragments required only 18 h. In comparison, RNase treated with 4 M protein-disulfide isomerase reached 50% of its native refolded state in 28 h, whereas RNase treated with 30 M thioredoxin required 63 h. In the uncatalyzed reaction, Ͼ100 h was required. The increase in protein-disulfide isomerase activity after limited proteolysis was specific to the 19-kDa fragment since the ability of the 40-kDa gelatin-binding fragment of fibronectin to refold RNase did not change substantially upon limited digestion with chymotrypsin ( Fig. 5B and Table II). Chymotrypsin digestion of the 40-kDa fragment generated fragments of ϳ31, 28, 27, 16, and 11 kDa (data not shown). These data indicate that limited proteolysis of the 19-kDa fragment increases its protein-disulfide isomerase activity, most likely due to generation of a fragment whose conformation favorably exposes its protein-disulfide isomerase activity. , and ovalbumin (4 M) were tested for their ability to catalyze the refolding of reduced and denatured RNase. Reduced and denatured RNase was incubated with the designated proteins for ϳ100 h. RNase activity is expressed as -fold increase over uncatalyzed refolding of RNase, which has been normalized to 1. Similar results were seen when the 70-h time point was analyzed. The data shown are the means of three or more experiments. Error bars represent S.E. Values that are statistically different from ovalbumin (p Ͻ 0.05 when analyzed using the Tukey HSD test) are shown by an asterisk. Inset, fibronectin and fibronectin fragments. Modules of fibronectin are represented as rectangles (type I modules), ovals (type II modules), and squares (type III modules). The proteolytic fragments of fibronectin used in this paper and their apparent molecular masses are indicated, as are the location of fibronectin's free sulfhydryl residues (SH). Disulfide-containing subunits (types I and II) are stippled.

DISCUSSION
In this report, we have demonstrated that the extracellular matrix protein fibronectin contains an intrinsic protein-disulfide isomerase activity. This is the first demonstration of protein-disulfide isomerase activity in fibronectin and suggests that fibronectin may catalyze disulfide bond rearrangement during its incorporation into the extracellular matrix. Other covalent cross-linking events may also lead to fibronectin multimerization under certain circumstances (53,54). For example, fibronectin can be covalently cross-linked in the extracellular matrix via ⑀-(␥-glutamyl)lysyl bonds by activated factor XIII (53,54). It has also been suggested that fibronectin multimers may be stabilized by noncovalent interactions since attempts to identify cross-linked regions of fibronectin have not been successful (16). However, the ability of multimeric extra-cellular matrix fibronectin to be converted to monomeric fibronectin by treatment with disulfide-reducing agents (13,14) suggests that extracellular matrix fibronectin is stabilized predominantly by inter-or intramolecular disulfide exchange.
Our data support a model whereby incorporation of soluble fibronectin into the extracellular matrix involves a disulfide exchange mechanism catalyzed by an isomerase activity located within the fibronectin molecule. Disulfide-stabilized, multimeric fibronectin has been reported to be a functionally distinct form of fibronectin that has enhanced adhesive properties, is active in suppressing cell migration and tumor formation (55,56), and mediates enhanced binding of bacteria to host tissue (57). Recent evidence also suggests that the extracellular matrix multimeric form of fibronectin has growth-promoting properties not possessed by protomeric fibronectin (58) and that the effects of matrix fibronectin on cell growth are dependent on its exact molecular configuration (59). It has also been shown that treatment of cells with a fragment derived from the first type III module of fibronectin (III 1-C ) can lead to inhibition of fibronectin deposition or disruption of a preexisting fibronectin matrix and also results in inhibition of cell growth (60,61). Thus, the identification of protein-disulfide isomerase activity within the fibronectin molecule is important not only for elucidating the biochemical mechanisms that regulate fibronectin multimerization, but also in defining the functional consequence of this multimerization.
In the RNase refolding assay, 4 M fibronectin had disulfide isomerase activity similar to that of 1 M protein-disulfide isomerase or 30 M thioredoxin (Figs. 1 and 2). Analysis of k cat /K m values indicated that fibronectin is ϳ9-fold more active than thioredoxin and 9-fold less active than protein-disulfide isomerase in catalyzing RNase refolding. The rates determined in this study for protein-disulfide isomerase are lower than those previously reported (19,62,63). However, in those previous reports, RNase refolding was performed in the presence of reduced and oxidized glutathione at pH 8, conditions substantially different from those used in our assays. Thus, these differences likely account for the differences in rates reported.  The RNase refolding assay used in this and other studies predominantly measures the ability of proteins to catalyze dithiol oxidation reactions (17). However, refolding of reduced and denatured RNase to a native state involves formation (oxidation) and rearrangement (isomerization) of non-native disulfide bonds until the final folded state is achieved (22, 40 -42), a process that is thought to be driven by the search for the most stable protein conformation (17). The assay buffer employed in this study does not require the addition of an external oxidative agent or the prior "activation" of protein catalysts by the addition of a reducing agent (64,65) and thus is likely to more closely approximate the environment outside the cells where fibronectin deposition into the extracellular matrix occurs.
Our data indicate that fibronectin's protein-disulfide isomerase activity is localized predominantly to the last type I module, I 12 (Fig. 4), which contains the Cys-X-X-Cys motif (21). I 12 was more active on a molar basis than intact fibronectin, indicating that most or all of fibronectin's isomerase activity is localized to I 12 . Localization of the protein-disulfide isomerase activity of fibronectin to I 12 suggests the possibility that the Cys-X-X-Cys motif of I 12 may be important in catalyzing disulfide cross-linking of fibronectin in the extracellular matrix since the Cys-X-X-Cys motif in protein-disulfide isomerase and thioredoxin is the active-site sequence required for catalyzing disulfide bond isomerization (17)(18)(19)(20). The Cys-X-X-Cys motif in von Willebrand factor has also been shown to be critical for the ability of von Willebrand factor to form disulfide-stabilized multimers (66).
We have previously shown that fibronectin containing mutations in I 12 or lacking I 12 can become incorporated into the extracellular matrix of cells containing a pre-established matrix (30). However, if the protein-disulfide isomerase activity of I 12 is critical for the formation of disulfide-stabilized fibronectin multimers, it is possible that the non-mutant fibronectin present in the assay provided this activity. In support of this, we and others have shown that fibronectin deletion mutants lacking the Arg-Gly-Asp sequence (67,68) or lacking a large internal portion (III 1 -I 12 ) (69) become incorporated into the extracellular matrix of cells containing a pre-established fibronectin matrix (67, 69), but are not incorporated into the extracellular matrix of cells lacking a pre-established fibronectin matrix (68,69).
The cryptic nature of fibronectin's protein-disulfide isomerase activity is consistent with the highly regulated nature of fibronectin matrix assembly (44 -48). Our data indicate that the protein-disulfide isomerase activity of fibronectin can be increased by limited proteolysis, a treatment that likely results in changes in protein conformation. Conformational changes have also been detected in fibronectin following binding of fibronectin to surfaces (50,70) or following alterations in pH and ionic strength (71,72). Conformational alterations in fibronectin have been shown to lead to exposure of cryptic binding sites (9,10). Conformational alteration of III 1 generates a binding site for the 70-kDa amino terminus of fibronectin (10). Similarly, conformational alteration of III 10 leads to exposure of a cryptic binding site for III 1 (9). Fibronectin that is incubated with a fragment of III 1 (III 1-C ) (55) or with conformationally altered III 10 (9) forms disulfide-stabilized multimers, presumably due to conformational changes in fibronectin induced by interaction with III 1-C or with altered III 10 . These data suggest a model in which the interaction of fibronectin with cell surfaces triggers a series of conformational changes leading to exposure of fibronectin-fibronectin interactive sites (9, 10, 55, 68, 73, 74) and activation of fibronectin's disulfide isomerase activity that may be important for fibronectin fibril formation. Future studies will be directed toward defining the interactions that regulate fibronectin's protein-disulfide isomerase activity and determining whether this activity mediates fibronectin cross-linking during fibronectin matrix assembly.