Interactions between factor XIII and the αC region of fibrinogen

Fibrinogen is a 340 000-Da glycoprotein composed of 2 sets of disulphide linked nonidentical polypeptide chains; Aα, Bβ, and γ.^1,2  Thrombin catalyzes the polymerization of fibrinogen to fibrin by sequentially cleaving fibrinopeptide A and fibrinopeptide B (FpB), initiating lateral aggregation of protofibrils and fiber formation.^3-5  Factor XIII (FXIII) is a 325.8-kDa heterotetramer composed of 2 identical globular A subunits noncovalently bound to 2 FXIII-B subunits.^6-8  FXIII-A₂B₂ is converted to its active form by the thrombin catalyzed hydrolysis of the Arg37-Gly38 peptide bond at the N-terminal of the FXIII-A subunit.⁹  In the presence of calcium, the thrombin-cleaved FXIII-A₂B₂ complex dissociates, yielding FXIII-B₂ and activated FXIII-A₂.^10-12  Calcium has been shown to cause small but significant conformational changes in FXIII-A during activation, exposing potential exosites within FXIII-A.^13-15  Activated FXIII-A₂ stabilizes the forming protofibril by introducing ϵ-amino(γ-Glutamyl)Lysine cross-links between carboxyl terminal portions of adjacent fibrin γ chains, before lateral association of the protofibril.¹⁶  Cleavage of FpB and subsequent release of the αC regions from the central E region initiates lateral aggregation of protofibrils and enables activated FXIII-A₂ to cross-link adjacent αC, stabilizing the developing fiber and making it more resistant to fibrinolysis.^17,18 

Interactions between fibrin(ogen) and FXIII-A₂B₂ are well documented; Greenberg and Shuman¹⁹  demonstrated that nonactivated plasma FXIII-A₂B₂ bound specifically to fibrinogen by the FXIII-A₂ subunits with an equilibrium constant (K_d) of 10nM. Greenberg et al²⁰  also noted that binding of FXIII-A₂B₂ was unaffected by fibrinogen polymerization, suggesting that the interaction must occur before thrombin cleavage of fibrinopeptide A and FpB. Hornyak and Shafer²¹  compared activated, nonactivated platelet FXIII-A and nonactivated plasma FXIII-A₂B₂ for binding to fibrin clots (K_d of 2.1μM, 14μM, and 200nM, respectively). In addition, Hornyak and Shafer examined the effect of fibrin on the activation of platelet FXIII-A, a phenomenon previously described by Credo et al.^22,23  Hornyak and Shafer found that fibrin did not promote activation of platelet FXIII-A alone, but it did enhance FXIII-A₂B₂ activation, suggesting that this effect might be mediated by the dissociation of the B chains.²¹  Immunoblotting of fibrinogen plasmin degradation products identified binding regions in the Aα and Bβ chains for platelet FXIII-A. Binding was not observed on the fibrinogen γ chains.²⁴  Interestingly, Siebenlist et al suggested that plasma FXIII preferentially bound to fibrinogen molecules containing the variant γ′ because fibrinogen containing γAγ′ copurified with the nonactivated FXIII-A₂B₂, and that this interaction takes place by the B subunit of FXIII.²⁵ 

The binding region on fibrin for placental FXIII-A was first localized by Procyk et al.²⁶  Procyk used several antibodies specific to various regions of fibrinogen [T2G1 (anti-Bβ15-21), 1D4/x1-f (anti-Aα389-402), 4-2/x1-f (anti-γ392-406), Fd4-7B3 (anti-γ[Fragment D]), 1C2-2 (anti-Aα529-539)] to identify the key regions thought to be involved in FXIII-A binding. The results have shown that the antibodies against the Bβ chain and the C-terminal region of the γ chains [anti-Bβ15-21, anti-γ392-406, and anti-γ(Fragment D)] did not appreciably affect the binding of activated FXIII-A. The binding was however substantially lowered by an anti-Aα389-402 antibody and furthermore by cyanogen bromide fragment Hi2-DSK (Aα241-476), suggesting that the location of the FXIII-A binding site was within αC region Aα389-402. In agreement with Hornyak and Shafer,²¹  Procyk et al²⁶  showed that the activation of FXIII-A by thrombin in the presence of calcium was necessary to enable a binding interaction. This αC binding region is consistent with previous work by Credo et al,²³  in which αC residues 242-424 were found to enhance FXIII-A₂B₂ activation. Therefore, it would seem reasonable to speculate that binding of activated FXIII-A or FXIII-A₂B₂ or both would occur within this αC region.

The aim of the current study was to characterize the interactions between FXIII and the fibrin(ogen) αC region to better understand the role of the αC region in regulating FXIII activation. This was performed using recombinantly expressed truncations of the αC region α233-425 and investigating binding interactions with (1) nonactivated recombinant FXIII-A (rFXIII-A), (2) activated rFXIII-A, (3) nonactivated plasma FXIII-A₂B₂, and (4) thrombin-cleaved FXIII-A₂B₂ in the presence and absence of calcium by surface plasmon resonance (SPR). In addition, site-directed mutagenesis of the αC binding region 389-402 has enabled us to identify for the first time a key amino acid residue involved in the binding of activated rFXIII-A. Furthermore we have confirmed (1) its specificity for activated rFXIII-A, (2) that the binding is specific for the calcium-induced conformational change observed during rFXIII-A activation, and (3) the binding is independent of rFXIII-A cross-linking activity. Finally, we have identified a novel high-affinity FXIII-A₂B₂ binding site on the αC region of fibrinogen.

Nine human fibrinogen αC fragments: α233-425, α233-403, α233-388, α233-375, α233-341, α233-290, α233-265, α289-425, and α371-425 (termed α fragment 1-9, respectively) were recombinantly expressed in Escherichia coli with the use of pGEX-6P-1 glutathione-S-transferase (GST) gene fusion system (GE Healthcare). Variants of the αC region 371-425 (Pro389Ala, Asp390Ala, Trp391Ala, Gly392Ala, Phe394Ala, Glu396Ala, Gly399Ala, Ser402Ala) were produced by introducing site-directed point mutations of highly conserved residues with the use of the QuickChangeII Site Directed Mutagenesis Kit (Stratagene) carried out according to the manufacturers instructions. DNA sequencing confirmed the αC fragments and the mutations. See supplemental Data for further details (available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

Full-length rFXIII-A was amplified by polymerase chain reaction from rFXIII-A construct pGF13A2, a kind gift from Dr C.S. Greenberg, Duke University Medical Center.²⁷  DNA sequencing confirmed the region of interest to be consistent with the FXIII-A subunit. The polymerase chain reaction product was digested with BglII and NotI, cloned into pGEX-6P-1, and transformed into DH5α E coli for screening. A recombinant FXIII-A double thrombin cleavage variant R37A/K513A was produced by introducing site-directed point mutations of wild-type rFXIII-A residues with the use of the QuickChangeII Site Directed Mutagenesis Kit carried out according to the manufacturers' instructions. DNA sequencing confirmed the mutations coding for R37A and K513A.

All GST-tagged recombinant proteins (GSTαC fragments 1-9, GSTα fragment 9 variants, GST–FXIII-A and GST–FXIII-A R37A/K513A) were expressed with the use of BL21-Gold E coli, following the method detailed in the supplemental material. GST fusion proteins were purified by GST-affinity chromatography with the use of an AKTAprime system, according to the recommended method by GE Healthcare. Details of the cell lysis, protein purification, and GST PreScission cleavage protocols used in this investigation can be found in supplemental Data.

Fibrogammin P (CSL Behring) was used as a source of plasma FXIII-A₂B₂ after isolation from other additives by gel filtration chromatography with the use of a Biocad sprint automated chromatography system (Perspective Biosystems) as detailed previously by Standeven et al.²⁸  For additional details see supplemental Data.

Plasminogen-free fibrinogen from human plasma (Calbiochem) was prepared by ammonium sulfate precipitation to remove bound FXIII as described by Siebenlist et al.²⁵  For additional details see supplemental Data. The FXIII-free fibrinogen was screened by Western blotting and biotin-labeled pentylamine incorporation FXIII-A activity assay to confirm the removal of any contaminating FXIII. The fibrinogen concentration was determined at A₂₈₀nm with the use of the extinction coefficient 1.51 for a solution of 1 mg/mL.

The cross-linking activity of purified FXIII-A₂B₂, rFXIII-A, and rFXIII-A R37A/K513A variant was determined with the use of a modified version of a 5-(biotinamido)pentylamine incorporation assay.²⁹  Modifications to the method can be found in supplemental Data.

Recombinant FXIII-A and FXIII-A₂B₂ were activated with the use of 5 U/mL human α-thrombin with 1.5mM calcium for 2 hours at 37°C, unless stated otherwise. The sample was centrifuged at 13 500 rpm for 5 minutes to remove precipitate, and the concentration was determined at A₂₈₀nm with the use of the extinction coefficient 1.58 for a solution of 1 mg/mL. When removal of thrombin was necessary, biotinylated thrombin (Merck) was used at 5 U/mL for 2 hours at 37°C. Biotinylated thrombin was removed with the use of streptavidin agarose according to the manufacturer's protocol.

In all cases SPR was performed with the use of a Biacore 3000 platform, and data were evaluated with BIAcore 3000 BIAevaluation 4.1 software (GE Healthcare).

With the use of SPR-calcium running buffer [20mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), 140mM NaCl, 1.5mM CaCl₂, 0.05% (vol/vol) surfactant P20, pH 7.4] ∼ 2500 response units (RUs) of goat anti-GST antibody (GE Healthcare) was immobilized directly onto a CM5 sensor chip (flow cell 2) by amine coupling according to the manufacturer's instructions. A second flow cell (flow cell 1) was immobilized with the same antibody but was used as a blank reference cell. GSTα fragment 1-9 (3.6μM) was injected for 300 seconds at a flow rate of 20 μL/min over flow cell (Fc) 2 to reach a capture level of ∼ 500 RU. rGST (100 μg/mL; Sigma-Aldrich) was injected over Fc1 and 2 for 300 seconds at a flow rate of 20 μL/min to block any remaining anti-GST binding sites. PreScission cleaved rFXIII-A was dialyzed for 16 hours at 4°C into SPR-calcium running buffer. Dialyzed rFXIII-A was thrombin activated and centrifuged, and the concentration was determined as described above. A 2-fold serial dilution of activated rFXIII-A (7.8-1000nM) was injected in ascending order for 60 seconds at a flow rate of 30 μL/min over Fc1 and Fc2. Regeneration was achieved with buffer flow after stabilization for 300 seconds. Because of the rapid binding of activated rFXIII-A to the αC, it was not possible to perform kinetic analysis. The K_d was therefore obtained with a predefined steady state affinity model with the use of a blank reference (Fc2-1, Fc4-3) and buffer subtracted data. This interaction was also performed in the reverse orientation to confirm the K_d. With the use of SPR-calcium running buffer 5000 RU of activated rFXIII-A was directly immobilized onto a CM5 sensor chip as detailed above. αC fragments 1, 3, and 9 (cleaved from the GST tag) were dialyzed into SPR-calcium running buffer. A 2-fold serial dilution of 0.01-10μM αC fragment was injected in ascending order for 60 seconds at a flow rate of 30 μL/min over the immobilized activated rFXIII-A. Regeneration was achieved with buffer flow after stabilization for 300 seconds (n = 3).

αC fragment 9 variants were cleaved from the GST tag and dialyzed into SPR-calcium running buffer. A fixed concentration of 1μM was injected over immobilized activated rFXIII-A for 60 seconds at a flow rate of 30 μL/min. The binding response of each variant at 55 seconds, after buffer subtraction, was taken and plotted against the wild-type α fragment 9 binding response for comparison (n = 3).

Approximately 500 RU of GSTα fragment 1 or GSTα fragment 9 were captured on a CM5 sensor chip as described above. Nonactivated FXIII-A₂B₂ and thrombin-cleaved FXIII-A₂B₂ activated in the presence and absence of calcium were tested for binding to the captured GSTα fragment 1 and 9. Two aliquots of purified FXIII-A₂B₂ were dialyzed for 16 hours at 4°C into SPR-EDTA (ethylenediaminetetraacetic acid) buffer [20mM HEPES, 140mM NaCl, 5mM EDTA, 0.05% (vol/vol) surfactant P20, pH 7.4]. A third aliquot was dialyzed into SPR-calcium buffer pH 7.4. The FXIII-A₂B₂ dialyzed into SPR-calcium buffer, and 1 of 2 FXIII-A₂B₂ aliquots dialyzed into SPR-EDTA buffer were thrombin cleaved as described above. A 2-fold serial dilution (0.7-200nM) of nonactivated FXIII-A₂B₂ or thrombin-cleaved FXIII-A₂B₂ were injected for 60 seconds over the captured GSTα fragment 1 or GSTα fragment 9 at a flow rate of 30 μL/min. Removal of bound FXIII-A₂B₂ was achieved with the use of two 50-second pulses of 2M NaCl at a flow rate of 30 μL/min. The surface was reequilibrated in running buffer for 300 seconds.

Purified fibrinogen (100 RU) was immobilized directly onto a CM5 sensor chip (Fc2) by amine coupling according to the manufacturer's instructions. A blank reference surface (Fc1) was activated and deactivated as described above. The immobilized fibrinogen was converted to fibrin as previously described.³⁰  Purified FXIII-A₂B₂ was dialyzed at 4°C for 16 hours into SPR-calcium running buffer and thrombin activated. A 2-fold serial dilution of thrombin and calcium-activated FXIII-A₂B₂ (0.7-200nM) was injected over the immobilized fibrin at a flow rate of 30 μL/min for 120 seconds. Removal of bound FXIII-A₂B₂ was achieved with the use of two 40-second pulses of 5mM NaOH containing 100mM NaCl at a flow rate of 30 μL/min. The surface was reequilibrated in running buffer for 300 seconds. Similar method was applied for analyses of the interactions between nonactivated FXIII-A₂B₂ (in the absence of thrombin or calcium) and FXIII-A₂B₂ cleaved by thrombin in the presence of EDTA (without calcium) with fibrinogen and fibrin. The SPR buffers contained EDTA or calcium to be consistent with the conditions used for the cleavage of FXIII by thrombin. Nonactivated or thrombin-only activated FXIII-A₂B₂ was never exposed to calcium to prevent nonproteolytic activation thought to occur in the presence of calcium alone. Data were fitted to a bivalent analyte model because the 1:1 Langmuir binding model resulted in poor curve fitting (χ² < 5 and χ² > 20, respectively). The bivalent analyte model was chosen to take into account the FXIII-A₂ and FXIII-B₂ dimers in the analyte FXIII-A₂B₂. However, the K_d obtained for this interaction is an estimate only because fibrinogen is also a dimer that, when bound to FXIII-A₂B₂, would result in a complex multivalent interaction that would be difficult to fit to a kinetics model. All SPR investigations were tested for mass transfer with the use of a predefined Mass Transfer control wizard (BIAcore 3000 software Version 4.1). The results confirmed that in all cases the interaction rates observed were not limited by mass transfer (data not shown).

A CM5 sensor chip surface was used to capture GSTα fragment 1 with the use of the same method as described above. GST-cleaved αC fragments 1-9 and rFXIII-A were dialyzed for 16 hours at 4°C into SPR-calcium running buffer. To determine the binding response from thrombin-activated rFXIII-A alone, 125nM was injected for 60 seconds over captured GSTα fragment 1 at a flow rate of 30 μL/min. The binding response was classified as 100% binding of activated rFXIII-A to the captured GSTα fragment 1. Thrombin-activated rFXIII-A (125nM) was preincubated for 1 minute at 25°C with the cleaved αC fragments 1-9 at increasing molar concentrations of 12.5nM, 125nM, 1250nM, and 6250nM. The mix of activated rFXIII-A with αC fragment 1-9 was injected for 60 seconds at a flow rate of 30 μL/min over the captured GSTα fragment 1. The percentage binding response of the activated rFXIII-A binding to the captured GSTα fragment 1 was plotted against the molar concentration of the αC fragment used in the preincubation with activated rFXIII-A. GST-cleaved α fragment 9 was also used as a competitive inhibitor to ascertain whether the binding of activated FXIII-A₂B₂ to the captured GSTα fragment 1, and directly immobilized fibrin, could be inhibited. PreScission-cleaved α fragment 9 was preincubated at increasing molar concentrations (0μM, 0.012μM, 0.125μM, 1.25μM, and 12.5μM where 0μM refers to activated FXIII-A₂B₂ in the absence of α fragment 9 competitor) for 60 seconds with 125nM thrombin-activated FXIII-A₂B₂ in the presence of calcium. The sample was injected for 60 seconds at a flow rate of 30 μL/min over captured GSTα fragment 1 and immobilized fibrin. The binding response in response units was converted to percentage binding of FXIII-A₂B₂ to the surface; captured GSTα fragment 1 or immobilized fibrin was used for comparison (n = 3).

Nonactivated rFXIII-A was dialyzed for 16 hours at 4°C into SPR-calcium running buffer, and a separate rFXIII-A aliquot was dialyzed into SPR-EDTA buffer. Both rFXIII-A aliquots were thrombin activated, and binding analysis was performed with the use of captured GSTα fragment 1 as previously described, using SPR-calcium running buffer for the rFXIII-A sample activated in the presence of thrombin and calcium, and a second SPR-EDTA running buffer for rFXIII-A activated with thrombin only in the presence of EDTA. The rFXIII-A samples were injected for 60 seconds over the captured GSTα fragment 1 at a flow rate of 30 μL/min.

rFXIII-A was dialyzed for 16 hours at 4°C into SPR-calcium running buffer. The rFXIII-A was thrombin activated and incubated with 50mM iodoacetamide (Sigma-Aldrich) for 15 minutes at 37°C. An aliquot of activated rFXIII-A in the absence of iodoacetamide was used for comparison. Before performing the binding study, 50mM iodoacetamide was injected for 60 seconds at a flow rate of 30 μL/min over the captured GSTα fragment 1. The iodoacetamide alone did not result in a binding response. Activated rFXIII-A (0.5μM) in the presence and absence of 50mM iodoacetamide was injected at a flow rate of 30 μL/min for 60 seconds over captured GSTα fragment 1.

Recombinant GSTα fragments 1-9 and αC fragments 1-9 cleaved from the GST tag were visualized with the use of sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE; Figure 1A-C) and exhibited major bands of the expected molecular mass as shown in Table 1. SDS-PAGE analysis of thrombin-activated rFXIII-A clearly showed a decrease in molecular mass by ∼ 4 kDa, resulting from the N-terminal cleavage of the activation peptide by thrombin (Figure 1D lanes 1-2). This effect was not observed in the rFXIII-A R37A/K513A variant (Figure 1D lanes 3-4) as predicted because R37 and K513 have been identified as thrombin cleavage sites.^9,31  The FXIII-A pentylamine incorporation assay indicated that the wild-type rFXIII-A and FXIII-A₂B₂ were active in contrast to variant rFXIII-A R37A/K513A which showed minimal cross-linking activity (Figure 1E).

Binding of activated rFXIII-A was observed for GSTα fragments 1, 2, 8, and 9 (α233-425, α233-403, α289-425, and α371-425, respectively). Negligible binding was observed for αC fragments 3-7 (α233-388, α233-375, α233-341, α233-290, and α233-265, respectively), which was too low to undertake formal analysis (data not shown). Figure 2 shows the binding of activated rFXIII-A to GSTα fragment 1 and GSTα fragment 9. Applying a steady state affinity model showed that GSTα fragment 1 (α233-425) bound to activated rFXIII-A with a K_d of 2.35 ± 0.09μM, whereas GSTα fragment 2 (α233-403) bound with a K_d of 2.64 ± 1.21μM. Competitive inhibition studies that used αC fragments 1 and 2 confirmed that the binding region for activated rFXIII-A was within αC 233-425. Complete inhibition of binding to captured GSTα fragment 1 could not be achieved with the use of α fragments 3-7, supporting the SPR data that the αC region 233-388 is not sufficient to mediate the interaction with activated rFXIII-A (Figure 2C). Rather, these results suggest the presence of essential determinants within the C-terminal portion of residues α389-403.

To further localize the αC binding region for activated rFXIII-A, amino-terminal truncations α fragment 8 (α289-425) and 9 (α371-425) were used. Activated rFXIII-A bound GSTα fragment 8 with a K_d of 1.81 ± 0.45μM and GSTα fragment 9 with a K_d of 3.19 ± 0.57μM (Figure 2B; Table 1). Competitive inhibition studies that used α fragments 8 and 9 showed complete inhibition of rFXIII-A binding to captured GSTα fragment 1, comparable to the inhibition observed with the use of cleaved α fragment 1 (Figure 2D).

To confirm the binding interaction, the experiment was performed in the reverse orientation by immobilizing activated rFXIII-A and injecting αC fragments 1, 9, and 3. The results showed that αC fragments 1 and 9 bound with a K_d of 3.91 ± 0.54μM and 3.56 ± 0.53μM, respectively. Negligible binding was observed for α fragment 3 (Figure 3). These data indicate that essential determinants of the fibrinogen αC binding region for activated rFXIII-A lie within αC residues 389-425 and are consistent with the localization of 389-403 deduced above. The localization is also in agreement with Procyk et al.²⁶ 

If the interaction between activated FXIII-A and the αC region is physiologically significant, it would be anticipated that key residues would be conserved between species. Alignment showed strong conservation of residues within 389-403 (Figure 4A). Site-directed mutagenesis of highly conserved amino acid residues within the αC region showed that a glutamic acid residue at position Aα396 is involved in the binding of activated rFXIII-A (Figure 4B), whereas substitution of 7 adjacent conserved residues was without effect. This suggests that FXIII-A is not binding to a linear binding site that is wholly contained within residues 389-403, but it is binding to an assembled determinant that also involves residues located between 371 and 389.

SPR binding analysis showed that wild-type nonactivated rFXIII-A and the inactivate variant R37A/K513A did not bind to the captured GSTα fragment 1 (Figure 5A). Furthermore, it was observed that the αC region 233-425 does not bind to rFXIII-A activated with thrombin only but also required the presence of calcium that induces a conformational change that exposes portions of FXIII-A that are not structurally exposed in the nonactivated form^13-15  (Figure 5B). Binding did not depend on catalytic activity because blocking the rFXIII-A active site Cys314 with iodoacetamide did not inhibit binding to the αC region 233-425, which occurred to the same degree as activated rFXIII-A in the absence of iodoacetamide (Figure 5C).

SPR was used to determine whether FXIII-A₂B₂ bound to GSTα fragment 1 (α233-425) and GSTα fragment 9 (α371-425). With the use of a bivalent analyte model the SPR results showed that GSTα fragment 1 (α233-425) bound to nonactivated FXIII-A₂B₂ (K_d1 = 7.3 ± 6.3nM), thrombin-cleaved FXIII-A₂B₂ (no calcium; K_d1 = 3.7 ± 0.3nM), and thrombin-cleaved FXIII-A₂B₂ activated in the presence of calcium (K_d1 = 21.9 ± 2.2nM; Table 2). In addition, the binding of the different activation states of FXIII-A₂B₂ were further localized to fibrinogen αC 371-425 [nonactivated FXIII-A₂B₂: K_d1 = 30.9 ± 23nM; thrombin-cleaved FXIII-A₂B₂ (no calcium): K_d1 = 5.4 ± 1.8nM; thrombin-cleaved FXIII-A₂B₂ activated in the presence of calcium: K_d1 = 7.3 ± 1.8nM] (Figure 6A; Table 2). The SPR sensorgrams for the binding of FXIII-A₂B₂ to GSTα fragment 1 were identical to the those displayed for the binding of FXIII-A₂B₂ to GSTα fragment 9 and therefore are not shown.

The various activation states of FXIII-A₂B₂ were also tested for binding to full-length fibrinogen and fibrin. The results showed that both fibrinogen and fibrin bound to nonactivated FXIII-A₂B₂ with a K_d1 of 11 ± 4.7nM and 35.2 ± 4nM, respectively, thrombin-cleaved FXIII-A₂B₂ (no calcium) with a K_d1 of 2.9 ± 2.1nM and 6.1 ± 1nM, and thrombin-cleaved FXIII-A₂B₂ activated in the presence of calcium with a K_d1 of 10.7 ± 0.9nM and 14.5 ± 3.1nM (Figure 6B-C; Table 2).

α Fragment 9 (α371-425), in the absence of the GST tag, was used as a competitive inhibitor to confirm the specificity of the interaction. The results showed that α fragment 9 inhibited the binding of thrombin-cleaved FXIII-A₂B₂ (activated in the presence of calcium) to the SPR-captured GSTα fragment 1 and full-length fibrin (Figure 7). Furthermore, these results confirm that the interaction between the α fragment 9 and activated FXIII-A₂B₂ occurs in solution in addition to binding to an immobilized surface.

Previous studies have shown the importance of the αC region for FXIII activation. The aim of this study was to characterize the interactions between FXIII-A and FXIII-A₂B₂ with the fibrinogen αC region 233-425 that could contribute to this process. Using recombinant truncations of the fibrinogen αC region 233-425, we have identified a novel high-affinity interaction between FXIII-A₂B₂ and fibrinogen αC region 371-425 which was evident both in the zymogen and after activation with thrombin and calcium. In addition, we have confirmed a previously described low-affinity interaction between activated FXIII-A and the αC region of fibrinogen, but we have extended the analysis to confirm the domain involved and to implicate a key residue, Glu396. The low-affinity interaction was dependent on calcium, which induces a conformational change in the β-barrel 1 and the β-sandwich domains of thrombin-cleaved FXIII-A.^13-15  In addition, calcium could mediate the interaction with Glu396. However, the low-affinity interaction was independent of catalytic activity, showing that it did not arise from a transient covalent intermediate.

Low affinity and reversible binding of activated FXIII-A₂ to fibrinogen was reported by Procyk et al²⁶  with the use of antibodies specific for the αC region 389-402 and with the cyanogen bromide fragment Hi2-DSK (Aα241-476), suggesting that the location of the FXIII-A binding site was within αC region Aα389-402.²⁶  Our results obtained by mapping recombinant fragments implicate the same αC region and, in addition, present a key residue, Glu396, as being critical for binding activated FXIII-A. Furthermore, our data also imply that additional residues upstream (α371-389) may contain secondary stabilizing sites that cannot function in the absence of Glu396, because it is unlikely that only one amino acid residue could make up the contact site for activated FXIII-A. The role of this interaction is currently undefined, but it is notable that the αC region 389-402 is close to the key αC glutamine residues (Gln366, Gln328, and Gln237) required for FXIII-A cross-linking that aid in clot formation and stabilization. Mutagenesis of the key interacting residue Glu396 in full-length fibrinogen will enable the putative role of the interaction in FXIII-A–mediated cross-linking to be assessed and will be of value in designing structural studies to identify the other residues comprising this binding site.

Our results also confirm high-affinity interactions between nonactivated and activated FXIII-A₂B₂ and (1) a recombinant αC fragment of fibrinogen and (2) the αC region on fibrin. The tetrameric nature of the SPR analyte FXIII-A₂B₂, together with the dimeric structure of the ligand fibrin(ogen), which is thought to adopt a conformation similar to that of soluble fibrinogen when immobilized,³²  has the potential to result in a complex multivalent interaction. However, SPR data are consistent with K_d1 dissociation constants in the range of 35nM. These values are in line with the K_d (10nM) deduced by Greenberg and Shuman¹⁹  for the binding of nonactivated plasma FXIII-A₂B₂ to full-length fibrinogen, in this case monitoring the interaction on the surface of latex beads. Interactions between FXIII-A₂B₂ and FXIII-A₂ with fibrin(ogen) have also been shown by immunoblotting,²⁴  clot binding assays,²¹  and enzyme-linked immunoabsorbent assay.^20,33 

Greenberg et al have suggested that FXIII may circulate in plasma bound to the fibrinogen molecule, allowing the nonactivated FXIII to be present for cross-linking activity when required.¹⁹  Interestingly, in our investigation all forms of FXIII bound similarly to the αC region, suggesting that this region contributes to binding of FXIII to circulating fibrinogen, supporting the hypothesis proposed by Greenberg et al. However, Siebenlist et al²⁵  have proposed that circulating FXIII-A₂B₂ binds preferentially to a subset of fibrinogen molecules that contain an extended γ chain as a result of alternative splicing of the γ chain mRNA and which comprise ∼ 10% of fibrinogen γ chains in the plasma. Fractionation of fibrinogen by anion exchange chromatography into pools enriched or depleted in γ′ chains have shown preferential association of FXIII-A₂B₂ with the enriched pool.²⁵  However, it is notable that Gersh and Lord³⁴  have reported that fibrinogen of γ/γ, γ/γ′, and γ′/γ′ composition all bind FXIII-A₂B₂ with a K_d of 41nM, within the same order of magnitude as determined for the interaction between fibrinogen γ/γ′ and fibrin γ/γ′ with FXIII-A₂B₂ (K_d, 11nM and 35nM, respectively) observed in this investigation.

The existence of an interaction between FXIII-A₂B₂ and the αC region of fibrinogen is predicted by work from Credo et al²²  which reported that the presence of fibrinogen accelerates activation of FXIII, and more specifically the αC residues 242-424 enhanced the activation of FXIII to almost the same degree as full-length fibrinogen.²³  For this enhancement to occur an interaction between FXIII-A₂B₂ and the αC regions of fibrinogen must take place at an early stage of clot formation. Our results show (1) that the αC region 371-425 binds with similar affinity to FXIII-A₂B₂ to that displayed by full-length fibrinogen and (2) that the αC fragment 371-425 inhibited FXIII-A₂B₂ binding to fibrin. We were not able to unequivocally demonstrate FXIII-A₂B₂ binding to the αC region on full-length fibrinogen, as opposed to fibrin or the recombinant αC fragment. In this context, probably the best interpretation of currently available data are that γ′ chain binding is important to promote a carrier protein function for FXIII-A₂B₂ and that the αC region:FXIII-A₂B₂ interactions may have an important separate role during fibrin formation which may be to enhance FXIII activation. However, further functional studies of αC region:FXIII-A₂B₂ interactions will need to be carried out to confirm this hypothesis.

In FXIII-A₂B₂, sushi domains 1-9 of FXIII-B mostly cover FXIII-A,³⁵  making it probable that binding of the nonactivated FXIII-A₂B₂ to the fibrinogen αC region takes place by the B subunit. In ongoing studies, we have confirmed that the αC region binds to the B subunit (K.A.S., P.J.G., unpublished data, 2009). Thrombin, in the absence of calcium, cleaves the activation peptide from the amino-terminus of FXIII-A but does not dissociate the complex,⁹  so that the interaction between fibrin and FXIII-A₂B₂ might be expected to be maintained. Accordingly, in our hands thrombin cleavage of FXIII-A₂B₂ did not affect FXIII-A₂B₂ binding to the αC region 371-425. Finally, thrombin, in the presence of calcium, both cleaves the activation peptide from the amino-terminus of FXIII-A and causes dissociation of the FXIII-A₂B₂ complex. We observed that the binding interaction was preserved, showing that it did not depend on the presence of both chains. We assume that FXIII-B may remain in complex with fibrinogen through a relatively tight interaction with a slow dissociation rate constant while activated FXIII-A is released. The relatively weak interaction of liberated FXIII-A with the αC is consistent with the need for the enzyme to be mobile to perform its cross-linking functions.

The results of this study localize the binding of FXIII-A₂B₂ to a high-affinity interaction site on the αC region 371-425, in addition to a low-affinity αC binding site that involves key amino acid residue Glu396 for activated FXIII-A. Activated rFXIII-A binding is dependent on the conformational change induced by calcium and is independent of catalytic activity. These results support the view that interactions fundamental to FXIII-A₂B₂ activation for αC cross-linking, fibrin formation, and stabilization are taking place in this region. Further studies are required to evaluate the effect of these interactions on thrombus formation in health and disease.

We thank Dr C. S. Greenberg and Dr T. S. Lai for the FXIII-A expression vector, and Dr Jeff Keen, Proteomics Facility, University of Leeds, for performing the peptide mass fingerprinting.

This work was supported by the British Heart Foundation (program grant RG/08/004/25 292). K.A.S. received a Young Investigators Award, ISTH Boston 2009.

Contribution: K.A.S. carried out the laboratory work, assisted in study design, analyzed data, and wrote the manuscript; P.J.A. made the FXIII-A constructs; R.J.P. assisted in study design and data analysis and critically reviewed the manuscript; A.J.B. helped with data analysis and critically reviewed the manuscript; J.M.B. helped with laboratory work; R.A.S.A. supervised the work and assisted in study design; P.A.C. provided general advice during the study and critically reviewed the manuscript. H.P. supervised the work, assisted in study design and data analysis, and critically reviewed the paper; P.J.G. provided overall supervision, assisted in study design, and helped write the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Peter J. Grant, Division of Cardiovascular and Diabetes Research, The LIGHT Laboratories, University of Leeds, Leeds, LS2 9JT, United Kingdom; e-mail: p.j.grant@leeds.ac.uk.