Analysis of Factor XIII Substrate Specificity Using Recombinant Human Factor XIII and Tissue Transglutaminase Chimeras*

Human factor XIII (FXIII) and tissue transglutaminase (tTG) are homologous proteins. FXIII requires thrombin for activation and cross-links the γ chains of fibrin(ogen) more efficiently than the Aα chains. On the other hand, tTG is thrombin-independent and forms predominantly Aα and Aα-γ chain complexes. Previous work from this laboratory demonstrated that amino acid residues within exon 7 of FXIII were important for catalysis (Hettasch, J. M., and Greenberg, C. S. (1994) J. Biol. Chem. 269, 28309–28313). To determine to what extent the primary amino acid sequence within exon 7 defines substrate specificity, exon 7 of FXIII was replaced with the corresponding exon of tTG using gene splicing by overlap extension. Other work from this laboratory (Achyuthan, K. E., Slaughter, T. F., Santiago, M. A., Enghild, J. J., and Greenberg, C. S. (1993) J. Biol. Chem.268, 21284–21292) using synthetic peptides identified two other domains that might play a role in substrate recognition (located in exons 3 and 5). Therefore, recombinant chimeras of FXIII/tTG were also created in which these two exons were exchanged. FXIII, tTG, and chimeras 3, 5, and 7 were expressed in Escherichia coli, purified, and the nature of the fibrin cross-linking pattern of these five proteins was determined by immunoblot analysis. FXIII preferentially formed the γ-γ dimer, whereas tTG formed Aα-γ complexes. Chimera 7 formed Aα-γ complexes that resembled the cross-linking pattern of tTG. This finding demonstrates that the primary amino acid sequence of exon 7 of tTG confers some of the specificity for the Aα and Aα-γ cross-link pattern characteristic of tTG. Chimera 5 exhibited reduced cross-linking activity (50% of FXIII activity) but still retained preference for formation of the γ-γ dimer, whereas chimera 3 was not active. In conclusion, exchanging the primary amino acid sequence of the active site exon of human FXIII with that of human tTG modifies the enzyme such that the fibrin cross-linking pattern more closely resembles that of tTG (Aα and Aα-γ complexes) instead of FXIII (γ-γ dimers).

During the process of blood coagulation, plasma transglutaminase, also known as factor XIII (FXIII), 1
In contrast, the tissue transglutaminase (tTG) preferentially cross-links the A␣ chains and an A␣-␥ complex of fibrin (20,23,24). Immunoelectrophoretic analysis of cross-linked fibrin(ogen) can readily distinguish whether the cross-linked products resulted from the action of tTG or FXIII (20,23). This methodology has been used successfully to distinguish tTG and FXIII cross-linked fibrin(ogen) complexes in atherosclerotic aortic intimas and vascular graft pseudo-intimas (25,26).
The purpose of the present study was to localize the substrate recognition domains between these two homologous transglutaminases using recombinant chimeric FXIII/tTG molecules. Previous work from this laboratory demonstrated that amino acid residues within exon 7 of FXIII were important for catalysis (27). In addition, using synthetic peptides, we also identified two other exons (exons 3 and 5) that might play a role in substrate recognition (28). To determine whether the primary amino acid sequence in these exons was responsible for defining the characteristic cross-link pattern between FXIII and tTG, recombinant FXIII/tTG chimeras were created in which exons 3, 5, and 7 of FXIII were replaced with the corresponding amino acids from tTG. Replacing exon 7 (the exon that contains the active site) of FXIII with that of tTG resulted in a FXIII chimera that produced fibrin cross-link products characteristic of the tTG. This finding suggests that some of the specificity for the A␣ and A␣-␥ cross-link pattern characteristic of tTG resides in the primary amino acid sequence of exon 7.

Materials
All restriction enzymes, T4 DNA ligase, Taq polymerase, and streptavidin alkaline phosphatase were purchased from Life Technologies, Inc. FXIII-free human fibrinogen was obtained from American Diagnostica, Inc. (Greenwich, CT). Human ␣-thrombin was the generous gift of Dr. J. W. Fenton II (New York State Department of Health, Albany). Oligonucleotides were synthesized by Biosynthesis, Inc. (Lewisville, TX). 5-(Biotinamido)pentylamine (BP) was obtained from Pierce. SDS-polyacrylamide gel electrophoresis molecular weight mark-* This research was supported by Grants HL26309, HL28391, and HL38245 from the NHLBI, National Institutes of Health (to C. S. G.) and a Focus Giving Award from Johnson and Johnson (to C. S. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

Methods
Construction of Chimeras-The construction of the plasmids for FXIII (pG-FXIII) and tTG (pG-tTG) have been described previously (29). The plasmids containing the cDNA for wild type FXIII and tTG were created using the pGEX-3X vector (Pharmacia Biotech Inc.). This vector provides a means for expressing these two transglutaminases as glutathione S-transferase (GST) fusion proteins to simplify purification. The cDNAs for three different FXIII/tTG chimeras were generated using a polymerase chain reaction-based method known as gene splicing by overlap extension (30,31). The plasmids pG-tTG and pG-FXIII, synthetic oligonucleotides, and the polymerase chain reaction were used to exchange precisely segments of FXIII with corresponding regions of tTG. These three exons were postulated to be important for substrate specificity, and the amino acid substitutions were based on sequence alignment of tTG and FXIII. Three different FXIII/tTG chimeras were generated: (i) an exon 3 swap (designated chimera 3), where amino acid residues 43-105 of FXIII were replaced with residues 1-62 of tTG; (ii) an exon 5 swap (designated chimera 5), where amino acid residues 190 -229 of FXIII were replaced with residues 144 -183 of tTG; and (iii) an exon 7 swap (designated chimera 7), where amino acid residues 266 -323 of FXIII were replaced with residues 227-285 of tTG. The aligned sequences for these three exons are illustrated in Fig. 1A, and schematics of the FXIII/tTG chimeras are represented in Fig. 1B. After the chimeric regions were generated by splice overlap extension, FIG. 1. Panel A, sequence alignment of exons 3, 5, and 7 of human FXIII and human tTG. The amino acid residues that are identical are boxed. Panel B, schematic illustration of the FXIII/tTG chimeras. These three recombinant proteins were created by exchanging residues 43-105 of FXIII with residues 1-62 of tTG (chimera 3), residues 190 -229 of FXIII with residues 144 -183 of tTG (chimera 5), or residues 266 -323 of FXIII with residues 227-285 of tTG (chimera 7) using splice overlap extension.
the NcoI-SacI fragment from pG-FXIII was replaced with the NcoI-SacI fragment from chimera 3 and chimera 5 to generate pG-chimera 3 and pG-chimera 5, respectively. The pG-chimera 7 construct was created by replacing the NcoI-StuI fragment of pG-FXIII with the NcoI-StuI fragment of chimera 7. DNA sequencing (32,33) was performed on the regions that were generated by the polymerase chain reaction to verify that the nucleotide sequence was not altered during this manipulation.
Expression and Purification of GST-FXIII Chimeras-The generation of pG-chimera 3, pG-chimera 5, and pG-chimera 7 described above creates a system for expressing the FXIII/tTG chimeras as GST fusion proteins. These GST fusion proteins were purified using a glutathioneagarose affinity resin as described previously (34). Briefly, the Escherichia coli strain JM105 containing the plasmid for FXIII, tTG, or the 3 FXIII/tTG chimeras was grown in 2 liters of enriched medium. The E. coli were chilled on ice, harvested by centrifugation, and resuspended in 200 ml of cold buffer A (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 15% glycerol, 1 mM EDTA/dithiothreitol, 10 g/ml aprotinin, 0.2 mM phenylmethylsulfonyl fluoride). The resuspended cells were then sonicated seven times for 10 s with 20-s rest between bursts. The lysates were centrifuged at 22,000 ϫ g for 20 min at 4°C. After centrifugation, the supernatant was applied to a glutathione-agarose column (8 ϫ 2.6 cm) that was equilibrated with buffer A. The column was then washed with 20 bed volumes of buffer A containing 0.5% Triton X-100 followed by 20 bed volumes of buffer A without detergent. The GST fusion proteins were eluted with 50 mM Tris-HCl, pH 7.5, 15% glycerol, 10 mM glutathione. The fractions containing the recombinant protein were concentrated in an Amicon ultrafiltration cell and then dialyzed extensively against 20 mM Tris acetate, pH 8.0, 1 mM dithiothreitol, 1 mM EDTA, 20% glycerol. Protein concentration was determined using the Bio-Rad protein assay with bovine serum albumin as the standard. The recombinant proteins were aliquoted, frozen, and stored at Ϫ80°C until assayed for activity.
Transglutaminase Assay-FXIIIa activity was assayed using a microtiter plate assay described previously (35). Briefly, microtiter plates (Costar Corp., Cambridge MA) were coated with 200 l of N,N-dimethylcasein (20 mg/ml) for 1 h at room temperature followed by blocking with 3% bovine serum albumin in TBST (100 mM Tris, pH 8.5, 150 mM NaCl, 0.05% Tween 20). FXIII, chimera 7, chimera 5, or chimera 3 (0.3-32 g/ml) was added to the wells followed by a solution (50 l) containing 100 mM Tris, pH 8.5, 20 mM CaCl 2 , and 40 units/ml thrombin. The microtiter plate was incubated at 37°C for 30 min to activate FXIII. After activation, 50 l of a solution containing 100 mM Tris, 40 mM dithiothreitol, and 2 mM BP was added to the wells of the microtiter plate. After the addition of BP, the plate was incubated at 37°C for 40 min to allow FXIIIa to cross-link BP into the dimethylcasein. The microtiter plates were washed once with TBST containing 1 mM EDTA, followed by three washes with TBST. The plates were then incubated for 1 h at room temperature with 100 l of streptavidin-alkaline phosphatase diluted 1:500 in 3% bovine serum albumin/TBST. Plates were then washed with TBST, and 200 l of the substrate p-nitrophenyl phosphate (1 mg/ml in 100 mM Tris, pH 9.8, 100 mM NaCl, 5 mM MgCl 2 ) was added. The rate of color development was monitored for 15 min at 15-s intervals at 405 nm in a V max kinetic microplate reader (Molecular Devices, Menlo Park, CA). In experiments where fibrin was used as the substrate, fibrinogen (10 g/ml) was used to coat the microtiter plate wells (1 h at room temperature), and then 20 units/ml thrombin was added to convert the fibrinogen to fibrin. The thrombin was then washed out with TBST, and the assay was performed as described above for dimethylcasein.
Binding Analysis of FXIIIa/tTG Chimeras to Fibrin Clots-FXIII, chimera 7, chimera 5, or chimera 3 (9 nM) was incubated with FXIII-free fibrinogen (1 mg/ml) in the presence of 20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 10 mM CaCl2, and 20 units/ml ␣-thrombin. Clots were allowed to form for 1 h and then squeezed mechanically to release the clot liquor. Gel electrophoresis of clot supernatants was performed on a 4 -10% polyacrylamide gradient using the buffer system of Laemmli (36). After electrophoresis, the proteins were transferred to nitrocellulose (0.2 m) as described by Towbin et al. (37). After the transfer was complete, the nitrocellulose membrane was blocked for 1 h with 3% nonfat milk dissolved in 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% Tween 20. The FXIIIa or FXIIIa/tTG chimeric antigen was detected by incubation for 1 h with a polyclonal antibody against the human FXIII A chain subunit (Calbiochem), followed by incubation for 1 h with goat anti-rabbit IgG conjugated to alkaline phosphatase (Bio-Rad). The FXIIIa or FXIIIa/ tTG chimeric antigen was visualized by precipitation of the chromogenic substrates nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate. The binding of wild type FXIII and the FXIII/tTG chimeras to fibrin clots was measured by quantitating the disappearance of FXIIIa or FXIIIa/tTG chimeric antigen from the clot supernatants using a Hoefer GS 300 densitometer.
Immunoblot Analysis of Cross-linked Fibrin-Previously established assay conditions designed to determine the rate of fibrin cross-linking by FXIIIa were used to analyze the cross-linking activity of the FXIII/ tTG chimeras (28). In these experiments, FXIII (30 nM), tTG (325 nM), or chimera 3, 5, or 7 (0.03-1.2 M) was incubated with FXIII-free fibrinogen (0.5 mg/ml) in the presence of 0.1 M Tris acetate, pH 7.4, 0.15 M NaCl, 0.1% polyethylene glycol 8000, 10 mM CaCl 2 , and 20 units/ml ␣-thrombin for 1 h at room temperature. The total assay volume was 300 l, and the reactions were stopped by the addition of SDS-polyacrylamide gel electrophoresis sample buffer containing 6 M urea, followed by boiling for 5 min. Gel electrophoresis of these samples (3.75 g of fibrinogen was loaded/well) was performed on a 4 -10% polyacrylamide gradient gel, and the proteins were transferred to nitrocellulose (0.2 m) as described above. After transfer was complete, the nitrocellulose was blocked with 3% nonfat dry milk dissolved in 20 mM Tris, pH 7.4, 150 mM NaCl, 0.5% Tween 20 for 1 h. To detect the cross-linked ␥ chain products, the nitrocellulose was incubated for 1 h with a monoclonal antibody to the human fibrinogen ␥ chain (4A5) followed by incubation for 1 h with goat anti-mouse IgG conjugated to alkaline phosphatase (Bio-Rad). Fibrinogen on the nitrocellulose was visualized by precipitation of the chromogenic substrates nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate. To detect the cross-linked A␣ chain products, a monoclonal antibody (F103) to the human fibrinogen A␣ chain was used in place of 4A5.

RESULTS
To gain insight into the molecular basis of transglutaminase substrate specificity, three recombinant chimeras of FXIII and tTG were expressed, purified, and biochemically characterized. We demonstrated previously that several amino acid residues around the active site cysteine in exon 7 were important for FXIII catalysis (27). In addition, our laboratory has identified two regions (one in exon 3 and the other in exon 5) as potential substrate recognition sites for transglutaminases using synthetic peptides (28). Fig. 1A illustrates the aligned amino acid sequence of these three exons from FXIII and tTG. The sequence identity between tTG and FXIII in the exon 3, exon 5, and exon 7 domains is 29, 53, and 55%, respectively, and is highlighted by the boxed regions. In this study, these three exons in FXIII were exchanged with the corresponding exons of tTG using a polymerase chain reaction technique known as gene splicing by overlap extension (for details of the exchange, see "Experimental Procedures"). Fig. 1B is a schematic diagram of the structure of the three recombinant FXIII/tTG chimeras that were created. Chimera 3 has amino acid residues 43-105 of FXIII replaced with residues 1-62 of tTG, chimera 5 has amino acid residues 190 -229 of FXIII exchanged with residues 144 -183 of tTG, and chimera 7 has amino acid residues 266 -323 of FXIII interchanged with residues 227-285 of tTG.
The cDNAs for FXIII and tTG were subcloned into a commercially available vector, pGEX-3X as described (34,38). The cDNAs for the chimera proteins were created using gene splicing by overlap extension and subcloned into the pGEX-3X vector as described under "Experimental Procedures." This vector provides a means for expressing transglutaminases as GST fusion proteins, thereby simplifying the purification procedure. A representative Coomassie Blue-stained gel of these GSTtransglutaminases is illustrated in Fig. 2A. From the left, the first lane shows the location of molecular weight markers, and the second, third, fourth, fifth, and sixth lanes show GST-FXIII, GST-chimera 3, GST-chimera 5, GST-chimera 7, and GST-tTG, respectively, after elution from the glutathione affinity column.
There are some minor lower molecular weight bands in the chimera 3, 5, and 7 lanes; immunoblot analysis ( Fig. 2B) with a polyclonal antibody to FXIII indicates that these bands are proteolytic fragments of the recombinant proteins. The exchange of amino acid residues between the tTG and FXIII may create a molecule that is more susceptible to proteolytic degra-dation. Densitometric scanning of the Coomassie Blue-stained gel indicates that the GST-transglutaminase preparations contain at least 85% full-length protein.
The activity of the purified recombinant FXIII chimeras was first evaluated by examining the incorporation of a small biotinylated primary amine (BP) into dimethylcasein. Fig. 3A compares the activities of FXIIIa and chimeras 7, 5, and 3. The incorporation of BP into dimethylcasein was rapid as the protein concentration was increased from 1.8 -12 nM for FXIII and chimera 7. In contrast, the incorporation of BP into dimethylcasein by chimera 5 proceeded at a slower rate and plateaued at a rate of milli-optical density units/min that was 35% that of the wild type enzyme. In addition, the protein concentration of chimera 5 required to achieve this activity was 5-fold (60 nM) greater than that of FXIIIa (12 nM). The activity of chimera 5 was not enhanced even when the concentration of this enzyme was increased 16-fold (to 192 nM). Chimera 3 (1.8 -192 nM) was not active in this assay. Similar results were obtained when the incorporation of BP into fibrin was examined (Fig. 3B). Chimera 7 had the same degree of activity as FXIIIa. The enzymatic activity of chimera 5 was reduced by 70%, and chimera 3 was not active in this assay. Quantitative immunoblot analysis revealed that FXIII and the three chimeras were cleaved 75-85% by 20 units/ml ␣-thrombin, demonstrating that the differ- To evaluate the ability of recombinant chimera 7 and chimera 5 to recognize soluble fibrinogen as a substrate, the capacity of fibrinogen to inhibit the incorporation of BP into dimethylcasein was examined. In these experiments, preactivated FXIIIa (3 nM), chimera 7 (3 nM), or chimera 5 (30 nM) was incubated with 0.5 mM BP in the absence or presence of increasing concentrations of soluble fibrinogen (0.02-1.25 mg/ ml). Fig. 4 illustrates the inhibition curves obtained for FXIII and these two recombinant FXIII/tTG chimeras. Fibrinogen blocked BP incorporation into dimethylcasein with an IC 50 of 0.4 mg/ml for all three recombinant proteins. These findings demonstrate that chimera 7 and chimera 5 are capable of binding and interacting with soluble fibrinogen with the same affinity as FXIIIa.
The next series of experiments examined the capacity of chimera 3 to bind substrate. Because chimera 3 was inactive, this study was designed to examine whether chimera 3 could act as a competitive inhibitor of FXIIIa. In these experiments, FXIIIa (3 nM) was incubated with 0.5 mM BP in the absence or presenceof7.5-240nMchimera3. Fig.5illustratesaconcentrationdependent inhibition of FXIIIa activity by chimera 3. When the concentration of chimera 3 (240 nM) was increased 80-fold over FXIIIa (3 nM), there was a 60% decrease in the activity of the wild type enzyme. Similar results were obtained with a catalytically inactive mutant of FXIII. This inactive form of FXIII has been characterized previously (27) and was created by converting the active site cysteine (residue 314) to alanine. This point mutation produces a catalytically inactive molecule that still has the capacity to bind fibrin. The FXIII C314A mutant produced a dose-dependent inhibition of FXIIIa (3 nM) over the same concentration range (7.5-240 nM) as chimera 3. When the concentration of FXIII C314A (240 nM) was increased 80-fold over the wild type enzyme (3 nM), FXIIIa activity was reduced by 50%. These findings demonstrate that chimera 3 is capable of recognizing and binding substrate and suggest that even though chimera 3 is catalytically inactive the molecular structure of the protein is still intact.
To evaluate the ability of these chimeras to recognize large macromolecular substrates, the extent to which these recombinant chimeras bound fibrin was examined. Table I summarizes the fibrin binding data and illustrates that chimera 7, chimera 5, and chimera 3 bind fibrin to the same extent as FXIIIa.
Therefore, even though the exchange of exon 5 and exon 3 of tTG into the corresponding region of FXIII has altered the catalytic activity of FXIIIa, the ability of these chimeras to recognize and bind fibrin substrate has not changed; this implies that the overall structure of the molecule has not been altered by the exchange of these amino acid residues.
Fibrin cross-linking analysis of several FXIII mutants (34) as well as information available from the x-ray crystal structure (39,40) of the zymogen of FXIII suggest that the ␤-barrel domain in the carboxyl terminus of the FXIII A chain may be important for aligning macromolecular substrates into the active site for cross-linking (40). Therefore, to probe further the molecular basis of transglutaminase substrate recognition, the next series of experiments examined the effect of removing the ␤-barrel domain on the activity of the FXIII/tTG chimeras. Fig.  6 illustrates the activity profile of FXIII and the three FXIII/ tTG chimeras after truncating these molecules at amino acid residue Lys 513 . The incorporation of BP into dimethylcasein by FXIII ⌬K513 reached the same extent as full-length FXIII when the concentration was increased 10-fold over the fulllength enzyme. In contrast, the incorporation of BP into dimethylcasein by chimera 7 ⌬K513 proceeded at a slower rate and plateaued at a rate of milli-optical density units/min that was 50% that of FXIII ⌬K513. In addition, the protein concentration of chimera 5 required to achieve this activity was 5-fold (60 nM) that of wild type (12 nM), and the activity was not enhanced even when the concentration of chimera 5 was increased to 192 nM. Chimera 3 (1.8 -120 nM) was not active in this assay.
The next series of experiments was designed to determine the fibrin cross-linking pattern of the three recombinant chimeras. In this study, FXIII-free fibrinogen (0.5 mg/ml) was incubated with FXIIIa (30 nM), tTG (325 nM), chimera 3 (1.  phoresis sample buffer. Gel electrophoresis on a 4 -10% gradient was performed on these cross-linked fibrin clots followed by immunoblot analysis using a monoclonal antibody to the ␥ chain of fibrinogen (Fig. 7A) or a monoclonal antibody to the A␣ chain of fibrinogen (Fig. 7B). FXIIIa predominantly crosslinked the ␥ chains of fibrinogen and the tTG cross-linked A␣ and A␣-␥ complexes. The fibrin cross-linking pattern of chimera 7 resembled that of the tTG. Immunoblot analysis revealed that chimera 7 produced A␣ and A␣-␥ complexes. Chimera 7 was not as efficient at cross-linking fibrin as tTG, requiring a concentration 8-fold higher than that of tTG to produce the cross-linked products. In contrast, the fibrin crosslinking pattern of chimera 5 resembled that of FXIIIa, predominantly cross-linking the ␥ chains. Chimera 5 was also not as efficient as FXIIIa, requiring a concentration 40-fold higher than that of FXIIIa to produce a substantial amount of ␥-␥ dimer. Chimera 3 was not able to cross-link either the A␣ or ␥ chain. A small degree of cross-linking can be seen in the samples incubated in the presence of EDTA. This cross-linked material is present in the fibrinogen sample and is caused by activation of FXIII during the isolation of fibrinogen from plasma. Fig. 8 illustrates the ␣ carbon backbone of FXIII and highlights the regions of FXIII which were replaced by corresponding regions of tTG. The diagrams also show the relationship of the catalytic triad to the altered portions of the FXIII/tTG chimeras. These images were created using the atomic coordinates of the x-ray crystal structure (39,40) of FXIII deposited in the Protein Data Bank at Brookhaven National Laboratory and the graphics program RASMOL. DISCUSSION The mechanism by which transglutaminases recognize their protein substrates remains unknown. Most of the current understanding of the basis of substrate specificity has been defined using small molecular weight molecules (41)(42)(43), and there is little evidence available regarding the interaction of macromolecular substrates with the enzyme. Even though FXIII and tTG are homologous proteins, there are significant differences in their biologic properties. FXIII requires thrombin for activation (1), and the active form of the enzyme is a dimer (44). The activity of tTG is thrombin-independent, and the active form is a monomer (41). In addition, the predominant FXIII cross-link product of fibrin is the ␥-␥ dimer (17,18,45), whereas that of the tTG is A␣-␥ and A␣ complexes (20,23,24).
In addition to the differences observed in the cross-linking of fibrin, differences have also been seen when small peptides were used to define substrate specificity (41). Using a 15-amino acid peptide modeled after the amino acid sequence of ␤-casein, Gorman and Folk (46 -48) concluded that some of the differences in specificity between tTG and FXIIIa were caused by secondary interactions of the oligopeptide and the enzyme. Folk hypothesized that there was an extended active site of at least 9 -10 amino acid residues and suggested that these secondary interactions might increase the binding affinity of the peptide for the enzyme, produce a conformation that enhanced catalysis, or both (41). Gorman and Folk also found that a 13-amino acid peptide modeled after the sequence surrounding the glutamine in the fibrin ␥ chain was a poor substrate for FXIIIa (41,46). They postulated that recognition of fibrin as a substrate came from other domains in the enzyme and suggested that fibrin was held in position at the catalytic site by the teritary structure of the enzyme and substrate. This hypothesis is also supported by recent studies performed with keratinocyte transglutaminase. Using several recombinant deletion constructs of keratinocyte transglutaminase Kim et al. (49) found that residues 62-92 in the amino terminus of the enzyme modulated substrate specificity. However, these conclusions are drawn from studies performed with small glutamine peptide and small primary amines as substrates and may not reflect macromolecular substrate interactions.
Other studies examined the substrate requirements using a peptide-bound lysine residue as the amine donor instead of small primary amines (50,51). The amino acid residue preceding the donor lysine was found to modulate the recognition of this lysine as a cross-link site. Even though these studies provide important information regarding the influence of amino acid residues in close proximity to the glutamine or lysine to modulate the cross-linkability of that site, the macromolecular recognition domains that allow three large proteins to associate to produce a intermolecular ⑀-(␥-glutamyl)lysine bond remain undefined. The present study examines the molecular basis of fibrin substrate specificity using recombinant FXIII/tTG chimeras in which putative substrate recognition domains (exons 3, 5, and 7) of FXIII were replaced with the corresponding regions of tTG. The ability of these three FXIII/ tTG chimeras to cross-link fibrin was examined to determine whether these domains influence macromolecular substrate recognition.
Analysis of the fibrin cross-linking pattern produced by chimera 7 revealed the presence of A␣-␥ chain and A␣ chain complexes instead of the characteristic ␥-␥ dimer formed by FXIIIa. However, the concentration of recombinant chimera required to achieve this activity was 40-fold higher than FXIIIa and 8-fold higher than tTG, suggesting that chimera 7 was not as efficient as the wild type enzymes in cross-linking fibrin. On the other hand, when the activity of chimera 7 was compared with FXIIIa in experiments designed to measure the incorporation of a small primary amine into a macromolecular substrate (either fibrin or dimethylcasein), the enzymatic activity of chimera 7 was indistinguishable from that of FXIIIa. These results provide evidence that the primary amino acid sequence of the active site exon defines some of the macromolecular substrate specificity for the transglutaminases and suggest that it is the macromolecular substrate containing the peptidebound lysine that is influenced by this domain. The catalytic triad, identified by x-ray crystal structure analysis (39,40) and biochemical characterization of site-specific mutants (27,52), sits at the base of the active site exon, which is in the center of the catalytic core. Even though there is a high degree of sequence identity between tTG and FXIII in this region, some of the amino acid residues in this domain must alter the size and shape of the pocket such that tTG accommodates the A␣-␥ and A␣ complexes and FXIIIa the ␥-␥ dimer. The extended active site originally described by Folk (41) as 9 or 10 amino acids may actually be much larger. Furthermore, the exact residues involved in the A␣ cross-linking activity of the wild type and chimera 7 remain to be established. It is possible that introducing the active site exon of tTG into FXIII modifies the active site such that the residues cross-linked in the A␣-␥ complex by chimera 7 are different from those cross-linked by FXIIIa.
These findings also do not exclude the possibility that a region in the ␤ sandwich domain contains a substrate recognition site. The absence of such a substrate binding domain may explain the need to increase the concentration of chimera 7 40-fold over the wild type enzyme to see the cross-linking pattern of fibrin. Alternatively, this reduced catalytic efficiency toward macromolecular substrates may be because FXIIIa is a dimer and tTG a monomer. To address this question, recombinant chimeras were created in which the carboxyl terminus of the protein was deleted. We reported previously that the Gly 38 -Lys 513 proteolytic fragment (53) and the ⌬K513 and ⌬K502 carboxyl-terminal deletion mutants (34) behave as monomers when analyzed by gel filtration chromatography. Removal of the ␤-barrel domain of chimera 7 did not substantially alter the enzymatic activity, suggesting that another domain is important for aligning the substrate at the active site to promote efficient catalysis. Exon 5 is located at the amino-terminal end of the catalytic FIG. 8. Location of the putative substrate recognition domains in the three-dimensional structure of FXIII. Fig. 7 illustrates the ␣ carbon backbone of the zymogen of FXIII. Panels A, B, and C illustrate the location of exon 3 (blue) in the ␤-barrel domain and exons 5 (green) and exon 7 (pink) in the catalytic core, respectively. The area represented in yellow in all three panels represents the catalytic core. core of FXIII. Exchanging exon 5 of FXIII with that of tTG resulted in a recombinant chimera that exhibited reduced catalytic activity in the fibrin cross-linking assay. Similar results were seen when the incorporation of the small primary amine was measured with either fibrin or dimethylcasein. Although less active than FXIII, chimera 5 revealed the same preference for the ␥ chain of fibrin as the wild type enzyme. These results suggest that productive contact between the enzyme and the substrate has been reduced and that this interaction resides with the glutamine-containing macromolecular substrate. The finding that chimera 5 binds fibrin as well as FXIIIa coupled with the observation that chimera 5 has reduced activity suggest that replacement of these amino acid residues reduces the productive interaction of the glutamine with the primary amine or lysine substrate. Chimera 5 is still a transglutaminase, but the rate at which this enzyme cross-links macromolecular substrates is reduced dramatically. Examination of the location of exon 5 in relation to the catalytic triad (Fig. 8B) suggests that replacement of these amino acid residues creates a distortion in the enzyme such that the glutamine in the macromolecular substrate is slightly misaligned at the catalytic center thereby reducing the rate of catalysis. Exon 3 is located in the region of the molecule identified as the amino-terminal ␤ sandwich by x-ray crystallography, and it has been suggested that there are interactions between the ␤ sandwich domain of one subunit and the catalytic core of the other subunit (39,40,54). Exchanging exon 3 of FXIII with that of tTG resulted in a recombinant molecule that was catalytically inactive in both the BP assay and the fibrin cross-link assay. There are at least two possible explanations for this result. The recombinant chimera may be incapable of recognizing substrates because it is not folded properly, or alternatively, replacement of this domain in FXIII may interfere with the intrasubunit contacts such that the glutamine-containing macromolecular substrate cannot be aligned appropriately at the catalytic site. The data presented in this paper favor the latter explanation because it was established that even though chimera 3 was catalytically inactive it was still able to bind fibrin and to act as a competitive inhibitor of FXIIIa activity. This suggests that chimera 3 is capable of recognizing macromolecular substrates to the same extent as FXIIIa and implies that chimera 3 is folded correctly.
In summary, the present study provides evidence that the primary amino acid sequence of exon 7 in the transglutaminases defines some the properties required for this enzyme to recognize macromolecular substrates. Further experiments with other recombinant chimeras will identify other domains that may be important for transglutaminase substrate recognition and activity.