Plasmin-mediated proteolysis of human factor IXa in the presence of calcium/phospholipid: Conversion of procoagulant factor IXa to a fibrinolytic enhancer

Summary

Background

Factor (F) IX/IXa inactivation by plasmin has been studied; however, whether plasmin converts FIXa to a fibrinolytic enhancer is not known.

Objective

Investigate plasmin proteolysis site(s) in FIXa that inactivates and transforms it into a fibrinolytic enhancer.

Methods

NH₂-terminal sequencing, mass spec analysis and functional assays.

Results

Plasmin in the presence of Ca²⁺/phospholipid (PL) rapidly cleaved FIXaβ at Lys316↓Gly317 to yield FIXaγ followed by a slow cleavage at Lys413↓Leu414 to yield FIXaδ. FIXaγ/FIXaδ migrated indistinguishably from FIXaβ in nondenaturing gel system indicating that C-terminal residues 317–415/317–413 of heavy chain remain noncovalently associated with FIXaγ/FIXaδ. However, as compared to FIXaβ, FIXaγ or FIXaγ/FIXaδ (25–75 mixture, 8-hour/24-hour incubation analysis by Mass Spec) was impaired ~10-fold in hydrolyzing synthetic substrate CBS 31.39 (CH3-SO2-D-Leu-Gly-Arg-pNA), ~30-fold (~5-fold higher K_m, ~6-fold lower k_cat) in activating FX in a system containing Ca²⁺/PL, and ~650-fold in a system containing Ca²⁺/PL and FVIIIa. Further, FIXaγ or FIXaγ/FIXaδ bound FVIIIa with ~60-fold reduced affinity as compared to FIXaβ. Additionally, in ligand blots, plasminogen or Diisopropylfluorophosphate-inhibited plasmin (DIP-plasmin) bound FIXaγ and FIXaδ but not FIXaβ. This interaction was prevented by ε-aminocaproic acid or carboxypeptidase B treatment suggesting that plasminogen/DIP-plasmin binds to FIXaγ/FIXaδ through newly generated C-terminal Lys316 and Lys413. Importantly, FIXaγ/FIXaδ mixture but not FIXaγ enhanced tissue plasminogen activator (tPA)-mediated plasminogen activation in a concentration dependent manner. Similarly, FIXaγ/FIXaδ mixture but not FIXaγ enhanced tPA-induced clot lysis in FIX-depleted plasma.

Conclusion

Plasmin cleavage at Lys316↓Gly317 abrogates FIXaβ coagulant activity, whereas additional cleavage at Lys413↓Leu414 converts it into a fibrinolytic enhancer.

Introduction

Factor (F) IX is a single chain vitamin K-dependent glycoprotein that circulates in blood with Mr ~57,000 [1,2]. It plays a crucial role in blood coagulation since its deficiency results in a hemorrhagic disorder, known as hemophilia B. During hemostasis, FXIa/Ca²⁺ or FVIIa/tissue factor (TF)/Ca²⁺ cleaves FIX at the Arg145-Ala146 and Arg180-Val181 peptide bonds with resultant formation of a serine protease, FIXaβ, and release of a 35-residue activation peptide [2–4]. FIXaβ is comprised of a light chain (residues 1–145) and a heavy chain (residues 181–415) that are held together by a disulfide bond [5]. The light chain contains the N-terminal γ-carboxyglutamic acid domain (Gla) followed by two epidermal growth factor (EGF)-like domains; whereas, the heavy chain contains the C-terminal serine protease domain, which features the catalytic triad, His221 (c57 in chymotrypsin), Asp269 (c102 in chymotrypsin), and Ser365 (c195 in chymotrypsin). In the coagulation cascade, FIXaβ activates factor X (FX) to FXa in the presence of factor VIIIa (FVIIIa), Ca²⁺, and phospholipid (PL) provided by platelets [5].

Plasmin is a fibrinolytic protease comprised of a heavy chain with five Kringle domains and a light chain containing the catalytic domain [6,7]. Plasmin localizes to the fibrin clot site using its Kringle domains to bind to the exposed lysine residues to carry out fibrinolysis [6,7]. Notably, several studies have shown that plasmin proteolytically inactivates several coagulation proteins including FX [8,9], FXa [9–13], FIX [14], FVa [15,16] and FVIIIa [17,18]; during this process in many instances new C-terminal lysine residues are generated. These studies suggest that plasmin not only inactivates the clotting proteins, but also creates additional lysine binding sites for its Kringle domains at the clot site to enhance fibrinolysis.

Plasmin cleavage of FX/FXa has been extensively studied [8–13,19]. Circulating FX consists of a light chain (residues 1–139) and a heavy chain (residues 143–448) linked by a single disulfide bond; the tripeptide Arg-Lys-Arg that connects the light and heavy chains is proteolytically removed during the biosynthetic process [20]. Activation of FX by FIXa/Ca²⁺ or FVIIa/TF/Ca²⁺ involves cleavage at Arg194-Ile195 peptide bond releasing a 52-residue peptide and formation of FXaα. In contrast to autoproteolysis of FXaα that generates FXaβ with C-terminal Arg327 in the heavy chain, the proteolysis by plasmin in the presence of Ca²⁺/PL generates C-terminal Lys435 in FXaβ [10]. FXaβ generated by autoproteolysis or by plasmin has normal coagulant activity. Additional plasmin cleavage at Lys330↓Gly331 in the plasmin generated FXaβ or intact FXaα with C-terminal Lys448 yields FXaγ, which acts as a fibrinolytic enhancer [13]. Thus, FXaγ with C-terminal Lys330 and Lys435 or Lys448 facilitates fibrinolysis [13].

Samis et al. studied plasmin-mediated cleavage of FIX in the absence of PL and found that plasmin cleaved FIX at Arg145 and Arg180, which are the sites for FIX activation [14]. These authors also found that plasmin cleaved FIX at three additional sites Lys43, Lys316 (c148), and Arg318 (c150) generating an inactive molecule [14]. However, Samis et al. did not examine whether the inactive FIX molecule also enhanced fibrinolysis. In this study, we investigated whether plasmin cleavage of FIXaβ in the presence of Ca²⁺/PL downregulates coagulation and enhances fibrinolysis. The extensive data presented here reveal that plasmin cleaves FIXaβ rapidly at Lys316-Gly317 (c148-c149) to yield FIXaγ, which is devoid of coagulant activity; FIXaγ is then cleaved at Lys413-Leu414 to yield FIXaδ, which promotes fibrinolysis.

Materials and methods

Reagents

Benzoyl-Ile-Glu-Gly-Arg-p-nitroanilide (S-2222), H-D-Ile-Pro-Arg-p-nitoanilide (S-2288), and H-D-Val-Leu-Arg-p-nitroanilide (S-2251) were from DiaPharma (West Chester, Ohio). CH₃-SO₂-D-Leu-Gly-Arg-p-nitroanilide (CBS 31.39) was from Diagnostica Stago. Dansyl-Glu-Gly-Arg-chloromethyl ketone (dEGR-ck) and diisopropylfluorophosphate (DFP) were from Calbiochem (San Diego, CA). Carboxypeptidase B (CpB) was from Roche and ε-aminocaproic acid (EACA) from ICN. Phosphatidylcholine (PC), phosphatidylserine (PS), phosphatidylethanolamine (PE), recombinant hirudin, polyoxyethylenesorbitan monolaurate (Tween 20), and fatty acid free bovine serum albumin (BSA) were from Sigma (St. Louis, MO). PL vesicles containing 60% PC, 20% PS and 20% PE were prepared as described [21]. Enhanced chemiluminescence (ECL) detection reagents were from Amersham Pharmacia Biotech Inc. Normal pooled plasma (NPP) and congenital FIX-deficient plasma used in coagulation assays were from George King Bio-Medical, Inc. (Overland Park, KS) and activated partial thromboplastin time reagent was from Baxter Diagnostics, Inc. (Deerfield, IL). Human immunochemically-depleted FIX plasma used in clot fibrinolysis assays was from Haemotologic Technologies (Essex Junction, Vermont).

Proteins

Purified human FX, FXIa, Glu-plasminogen, Russel viper venom (RVV-X) and α-thrombin were purchased from Enzyme Research Laboratories (South Bend, IN). Plasmin and goat anti-human plasminogen were from Haemotologic Technologies. The two-chain tissue plasminogen activator (tPA, Alteplase) was obtained from Genentech (South San Francisco). Donkey anti-goat IgG conjugated to peroxidase was from Sigma. Wild-type recombinant FIX was expressed in HEK 293 cells and purified using a FIX A-7 mAb column followed by a Mono-Q column as described [22,23]. The FIX had ∼12 Gla residues/mol as measured by the procedure of Price et al [24] and appeared homogeneous on both reduced and non-reduced SDS-PAGE with Mr ∼57,000. FVIII was activated to FVIIIa using 1 nM human α-thrombin in the presence of 0.1% (w/v) BSA and 5 mM CaCl₂ in TBS pH 7.5 (50 mM Tris–HCl, 150 mM NaCl, pH 7.5) at 37°C for two minutes as described [22,23,25] followed by the addition of recombinant hirudin to inactivate thrombin.

Characterization of plasmin cleaved FIXaβ in presence of calcium and PL

FIX was activated to FIXaβ as described in the supplementary material. FIXaβ (1 ml, 100 μg/ml) samples were then incubated with 3 μg/ml plasmin, 240 μM PL (60%PC/20%PS/20%PE) and 5 mM CaCl₂ in TBS pH 7.5 at 37°C for various time points (1,1.5,2,3,4,5,10,15 minutes and 0.5,1,2,4,8,12,24 hours). SDS-PAGE (Supplementary material) was performed to determine proteolysis in FIXaβ. Plasmin in each reaction mixture was inactivated using 1 mM DFP and Ca²⁺ was chelated using 10 mM EDTA. Plasmin cleaved FIXaβ was then put over a Centricon100 concentrator to remove PL, and concentrated using a Centricon30 concentrator. Each sample was then subjected to NH₂-terminal amino acid sequencing and mass spec analysis.

Amino acid sequencing

FIXaβ before and after proteolysis were run on 12% reduced SDS-PAGE and blotted to Immobilon-P polyvinylidene difluoride (PVDF) membrane in Towbin buffer (25 mM Tris-base, 192 mM glycine, 20% methanol). Membrane was stained with 0.1% Coomassie brilliant blue R in 50% methanol, 7% acetic acid, and destained in 50% methanol, 7% acetic acid. The membrane was then submitted to Midwest Analytical (St. Louis, MO) for NH₂-terminal amino acid sequencing of the bands.

Intact protein mass spectrometry

Forty μg of FIXaβ and plasmin cleaved-FIXaβ were denatured using SDS-PAGE loading buffer. Proteins were recovered from SDS-PAGE loading buffer using chloroform/methanol precipitation [26] and dissolved in 70% formic acid prior to immediate LC-MS analysis. LC-MS used a polymeric stationary phase (PLRP/S at 40˚C) as described previously [27] in buffers containing 0.1 % trifluoroacetic acid. Positive ion electrospray-ionization mass spectrometry was performed using an ion-trap mass spectrometer (LTQ; Thermo, San Jose). Data were analyzed using qualbrowser software (Thermo, San Jose) and MagTran freeware. The data (Results section) indicate that 1,1.5,2,3,4,5,10 and 15 min incubations yield FIXaβ cleaved only at Lys316-Gly317 at 3 μg/mL plasmin; this molecule is termed FIXaγ. The FIXaγ is further cleaved at Lys413-Leu414 (~20% at 1 hour, 30% at 2 hour, ~40% at 4 hour and ~75% at 8, 12 and 24 hour); the molecule cleaved at Lys316-Gly317 and at Lys413-Leu414 is termed FIXaδ. The second cleavage was only observed in ~75% of the FIXaγ molecules as determined by mass spec analysis and was used in all comparative analyses. A higher concentration of plasmin increased the rate of FIXaδ formation but maximum conversion was still ~75% (see results). Resolution of intact mass spectrometry procedure used to determine the ratio of FIXaγ and FIXaδ was >90%.

Measurement of FX activation

FX activation by FIXaβ, FIXaγ or FIXaγ/FIXaδ mixture was measured under two conditions: One, in the presence of Ca²⁺/PL; and two, in the presence of Ca²⁺/PL and FVIIIa [23,25]. In the absence of FVIIIa, each reaction contained 20 nM FIXaβ, 25 μM PL, 5 mM CaCl₂, and FX ranging from 25–6400 nM. The activations were carried out for 2.5–20 minutes. The K_m and k_cat values were obtained using the program GraFit from Erithacus Software. Since plasmin cleaved-FIXaβ is essentially devoid of clotting activity, only one condition for activation of FX was used in the presence of FVIIIa. Four identical reaction mixtures each containing 0.05 nM FIXa (FIXaβ or FIXaγ or FIXaγ/FIXaδ mixture), 10 μM PL, 5 mM CaCl₂, 0.07 nM FVIIIa and 15 nM FX were setup and activation was carried out at 37 °C for 0.5, 1.0, 1.5 or 2 min. The buffer used in all experiments (plus/minus FVIIIa) was TBS/BSA pH 7.5 and the reaction volume was 50 μl. The reactions were carried out at 37 °C for different lengths of time, after which 2 μl of 0.5 M EDTA was added to stop the reaction. Then 100 μl of TBS/BSA pH 7.5 and 10 μl of S-2222 (final concentration 100 μM) were added. From this mixture, 150 μl was placed in a well on an Immulon 4 HBX flat-bottom 96-well microtiter plate and the p-nitroaniline release was measured (ΔA₄₀₅/minute) for up to 30 minutes. FXa generated was calculated from a standard curve constructed using FXa prepared by insolubilized RVV-X.

Binding of FIXaβ, FIXaγ and FIXaγ/FIXaδ to FVIIIa

Apparent K_d (K_dapp) binding values for FIXaβ, FIXaγ or FIXaγ/FIXaδ mixture to FVIIIa were determined by the ability of dEGR-IXaβ, dEGR-IXaγ or dEGR-FIXaγ/dEGR-FIXaδ mixture to inhibit FVIIIa potentiation of FX activation by FIXaβ [23,25]. Reaction mixtures contained 0.2 nM FIXaβ, 0.07 nM FVIIIa, 480 nM FX, 10 μM PL, and various concentrations of dEGR-IXaβ, dEGR-IXaγ or dEGR-FIXaγ/dEGR-FIXaδ mixture in TBS/BSA pH 7.5 containing 5 mM CaCl₂. The IC₅₀ (concentration of inhibitor required for 50% inhibition) was determined by fitting the data to IC₅₀ four-parameter logistic equation of Halfman [28]: where y is the rate

$$y=\frac{a}{1+{\left(x/\mathrm{I}{\mathrm{C}}_{50}\right)}^{\mathrm{s}}}+\mathrm{background}$$ (Eq. 1)

of FXa generation in the presence of a given concentration of dEGR-IXaβ, dEGR-IXaγ or dEGR-FIXaγ/dEGR-FIXaδ mixture represented by x, a is the maximum rate of FXa formation in the absence of dEGR-IXa species, and s is the slope factor. Each point was weighted equally and the data were fitted to equation 1 using the non-linear regression analysis program GraFit from Erithcus Software. To obtain K_dapp values for the interaction of dEGR-IXa proteins with FVIIIa, we used the following equation as described by Cheng and Prusoff [29] and further elaborated by Craig [30]:

$$K_d=\frac{\mathrm{I}{\mathrm{C}}_{50}}{1+(A/{\mathrm{EC}}_{50})}$$ (Eq. 2)

where A is the concentration of FIXaβ and EC₅₀ is the concentration of FIXaβ that gives a 50% maximum response in the absence of the competitor at a specified concentration of FX used in the experiment.

Plasminogen/plasmin binding to FIXaβ, FIXaγ or FIXaγ/FIXaδ mixture using ligand binding Western blots

Nonreduced samples were run on 12% SDS-PAGE and blotted to PVDF membranes in Towbin buffer. The membranes were blocked using 5% milk in TBS pH 7.5, 0.05% Tween 20 at room temperature for ~3 hours. The membranes were then washed two times in TBS pH 7.5, 0.01% Tween 20 and once in TBS pH 7.5. In one experiment, the membrane was incubated with Glu-plasminogen or DIP-Plasmin (3 μg/mL) in TBS pH 7.5, 1% milk over night at 4° C. In a second experiment, the PVDF membrane was incubated with 100 U/mL CpB in TBS pH 7.5 for 30 minutes at 37°C and washed in TBS pH 7.5 prior to incubation with DIP-plasmin in TBS pH 7.5, 1% milk over night at 4° C. In a third experiment, the PVDF membrane was incubated with DIP-Plasmin (3 μg/mL) and 2 mg/mL EACA in TBS pH 7.5, 1% milk at room temperature for 30 minutes followed by incubation overnight at 4°C. The membranes were then washed two times in TBS pH 7.5, and incubated with polyclonal goat anti-human plasminogen antibody (15 μg/mL) in TBS pH 7.5, 0.03% BSA for two hours at room temperature. The membranes were then washed one time with TBS pH 7.5, 0.01% Tween 20 and one time with TBS pH 7.5 and incubated with peroxidase-conjugated donkey anti-goat IgG (1:200) in TBS pH 7.5, 0.03% BSA for two hours at room temperature. Membranes were then washed once with TBS, 0.01% Tween 20 followed by once with TBS pH 7.5 and developed with the Amersham ECL kit.

Fibrinolysis (Clot Lysis) Assay

The method of Sperzel and Huetter [31] was followed with minor modifications as outlined earlier [32]. In initial experiments, RVV-X was used to initiate fibrin formation in immunochemically-depleted FIX plasma, and lysis of the formed clot was induced by simultaneous addition of tPA. Briefly, 225 μL FIX-depleted plasma was mixed with 12.5 μL of PL (60%PC/20%PS/20%PE, 100 μM) and 2.5 μL of TBS/BSA. Ten μL of each test compound (FIXaβ, FIXaγ and FIXaγ/FIXaδ mixture) or saline control was added to 240 μL of above FIX-depleted plasma mixture. Two hundred-twenty five μL of this mixture was then added to 25 μL of RVV-X and tPA in TBS/BSA containing 100 mM CaCl₂. In the 250 μL final volume, concentration of RVV-X was 10 ng/mL and tPA was 0.25 μg/mL. Next, RVV-X was used to initiate fibrin formation using NPP. Here, 225 μL of NPP was mixed with 12.5 μL of PL and 12.5 μL of TBS/BSA. Two hundred-twenty five μL of this mixture was added to 25 μL of RVV-X and tPA in TBS/BSA containing 100 mM CaCl₂. In the 250 μL final volume, the concentration of RVV-X was 10 ng/mL and that of tPA was 0.25 μg/mL or 0.5 μg/mL. In additional experiments, FXIa was used to initiate fibrin formation using NPP and mixtures of NPP and FIX-depleted plasma (50:50 and 20:80). Here, 225 μL of plasma was mixed with 12.5 μL of PL and 12.5 μL of TBS/BSA. Two hundred-twenty five μL of this mixture was added to 25 μL of FXIa and tPA in TBS/BSA containing 100 mM CaCl₂. In the 250 μL final volume, the concentration of FXIa was 64 ng/mL and tPA was 0.5 μg/mL. Clot formation and lysis in each experiment was monitored with a Molecular Devices microplate reader (SpectraMAX 190) measuring the optical density at 405 nm. Under control conditions (zero tPA), OD405 increased immediately indicating clotting, followed by an extremely slow decrease representing fibrinolysis. Because clotting was almost complete between 5 to 8 minutes, fibrinolysis induced by tPA was evaluated as a relative decrease of OD405 up to 120 min.

Results

Plasmin cleavage of FIXaβ

Plasmin-mediated cleavage of FIXaβ in the presence of Ca²⁺/PL was analyzed by FIXa activity, SDS-PAGE, NH₂-terminal amino acid sequencing and mass spec analysis. After 10 minutes, only ~5% of the original clotting activity remained and after 15 minutes, <2% of the FIXaβ activity remained (Fig. 1A). Based upon the NH₂-terminal amino acid sequencing data (see below), the loss of clotting activity corresponded to the generation of two fragments termed Hγ_N and Hγ_C on reduced SDS-PAGE (Fig. 1B). Based upon the mass spec data presented later in the manuscript, continued incubation of FIXaβ with plasmin resulted in cleavage of Hγ_C at Lys413↓Leu414 to yield Hδ_C (Fig. 1C). Note that Hγ_C (Gly317-Thr415) and Hδ_C (Gly317-Lys413) migrate indistinguishably in non-reduced SDS-PAGE.

Fig 1. Plasmin cleavage of FIXaβ.

A, Loss of FIXa clotting activity. FIXaβ (100 μg/ml) was incubated with plasmin (3 μg/ml) in the presence of 5 mM Ca²⁺ and 240 μM PL at 37°C. Samples were removed at indicated times, diluted 50 to 10,000-fold and measured for FIXa clotting activity as described in Materials and methods. Percent FIXa clotting activity remaining at different time points is depicted. B, Reduced SDS-PAGE analysis. Samples were removed at indicated times from the reaction mixture outlined in ′A′ above and 5μg of protein was loaded in each lane. Based upon NH₂-terminal sequence and the mass spec data, the heavy (Hβ) and light chains (L) of FIXaβ as well as the heavy chain cleavage fragments of FIXaγ (Hγ_N and Hγ_C) up to 15 min are indicated. Molecular weight (MW) markers as well as the N-terminal sequence of each band is provided. C, Non-reduced SDS-PAGE analysis. Samples were removed between 30 min to 24 hour from the reaction mixture outlined in ′A′ above and 5 μg of protein was loaded in each lane. Two discrete bands were visualized, one corresponding to light chain disulfide linked to Hγ_N (L-S-S-Hγ_N) and the second corresponding to Hγ_C/Hδ_C (based upon mass spec analysis).

NH₂-terminal amino acid sequencing of plasmin cleaved FIXaβ

FIXaβ was incubated with plasmin in the presence of Ca²⁺ and PL for 1,1.5,2,3,4,5,10,15 minutes and 0.5,1,2,4,8,12,24 hours at 37°C. Each sample was run on reduced SDS-PAGE, transferred to the PVDF membrane and subjected to NH₂-terminal amino acid sequencing as outlined in Materials and methods. The sequencing data revealed that plasmin cleaved FIXaβ heavy chain at Lys316↓Gly317. Two new bands were visualized following cleavage of the heavy chain. One band with the NH₂-terminal sequence VVGGE, which is the NH₂-terminal fragment of the heavy chain, termed Hγ_N. The other band with the NH₂-terminal sequence GRSAL, which is the C-terminal fragment of the heavy chain, termed Hγ_C (Fig. 1B). Thus, based upon the NH₂-terminal sequencing and SDS-PAGE, cleavage of the FIXaβ heavy chain only at the Lys316↓Gly317 peptide bond was observed (Fig. 1B).

Mass spec analysis of plasmin cleaved FIXaβ at different time points

Plasmin cleavage at Lys316↓Gly317 in FIXaβ heavy chain yields FIXaγ with Hγ_C fragment (Gly317-Thr415) that has an average theoretical molecular mass of 11031.63 Da (or 11027.6 Da containing the C336-C350 and C361-C389 disulfide bonds), which corresponds to sequence GRSALVLQYLRVPLVDRATCLRSTKFTIYNNMFCAGFHEGGRDSCQGDSGGPHVTEVEGTSFLTGIISWGEECAMKGKYGIYTKVSRYVNWIKEKTKLT. In the mass spec analysis of samples up to 15 minutes of incubation, the Hγ_C of FIXaγ heavy chain was “Instrument Reported” as a molecular ion at m/z = 1839.3, the 6-charge ion that is dominant for this peptide with measured average mass of 11029.8 Da. These data are consistent with absence of any additional proteolysis in the Hγ_C up to 15 minutes of incubation.

After removal of the two carboxy-terminal residues (Leu314 and Thr415) by subsequent cleavage of FIXaγ at Lys413↓Leu314, the resulting Hδ_C fragment (Gly317-Lys413) has the theoretical molecular mass of 10813.3 Da with the two disulfide bonds intact. This ion was “Instrument Reported” as a molecular ion at m/z = 1803.1 Da, the 6-charge ion that is dominant for this peptide with measured average mass of 10812.6 Da. This quantification assumes that the two species (Hγ_C and Hδ_C) have the same ionization efficiency, which is reasonable especially considering that there is very little change in the charge-state distribution between the two species. Fraction of the second species (Hδ_C) was ~20% at 1 hour, 30% at 2 hour, ~40% at 4 hour and ~75% at 8, 12 and 24 hour. The “Instrument Reported” data at 8 hour time point is shown in Fig. 2A. Based upon the NH₂-terminal sequencing and the mass spec data, it would appear that formation of FIXaγ (cleavage at Lys316↓Gly317 in FIXaβ) is rapid followed by a slow cleavage at Lys413↓Leu414 in FIXaγ to yield FIXaδ. Since proteolysis at Lys413-Leu414 of FIXaγ with 3 μg/mL (34 nM) plasmin was slow, we performed additional experiments using 6μg/mL (68 nM) plasmin. Under these conditions FIXaγ was cleaved faster at Lys413-Leu414 (~12% at 15 min, ~20% at 30 min, ~35% at 1hour, ~45% at 2 hour,~70% at 4 hour and ~75% at 8,12 and 24 hour). As expected, the rate of FIXaδ formation was increased ~2-fold when plasmin concentration was doubled. However, at each plasmin concentration, the maximum FIXaδ formed at 24-hour was ~75% indicating strong product inhibition, which prevents complete conversion to FIXaδ. These cleavage sites are schematically depicted in Fig. 2B. Notably, both species (FIXaγ and FIXaδ) are devoid of coagulant activity.

Fig 2. Plasmin cleavage sites in FIXaβ and their influence on native gel electrophoresis.

A. Mass spec data of the Hγ_C chain upon extended incubation with plasmin. Shown are the extracted ion chromatograms for the uncleaved and cleaved Hγ_C incubated with plasmin for 8 hours. The uncleaved Hγ_C of FIXaγ molecule (average mass 11029.8 Da reported by 6-charge ion at m/z = 1839.3) versus the cleaved Hγ_C of FIXaδ (average mass 11812.6 Da reported by the 6-charge ion at m/z 1803.1) are depicted. At this data point, it was estimated that 1.53 e3 of the uncleaved product remains compared to 4.8 e3 of the cleaved species. The y-axis has arbitrary units and is dependent upon the detector response. Based on the mass spec data, the Hγ_C is cleaved between residues Lys413 and Leu414. B. Schematic representation of the cleavage sites in FIXaβ heavy chain (Hβ) observed in the present study. Light chain (L) containing γ-carboxy glutamic acid (Gla) domain, two epidermal growth factor (EGF) like domains and the heavy chain (Hβ) containing the protease domain is depicted. The two cleavage sites are marked by black arrows at Lys316-Gly317 (312RVFHK↓GR318) and at Lys413-Leu414 (409KEKTK↓LT415). C, Native gel electrophoresis of FIX, FIXaβ and FIXaγ/δ (8-hour incubation time point). Gels were run according to Williams and Reisfeld [33] and 4 μg of protein was loaded in each lane.

Nondenaturing gel electrophoresis of FIX and FIXa

Nondenaturing gel electrophoresis (Supplementary material) of FIX, FIXaβ and FIXaγ/FIXaδ mixture at 8-hour time point is shown in Fig. 2C. FIX migrated slower than FIXaβ or FIXaγ/FIXaδ. Importantly, FIXaβ and FIXaγ/FIXaδ mixture migrated similarly. These data indicate that the activation peptide does not stay associated with FIXaβ and is released upon activation of FIX. However, the 317–415 or 317–413 fragments generated by plasmin cleavage in FIXaβ likely remain noncovalently associated with FIXaγ or FIXaδ.

CBS 31.39 amidolytic activity of FIXaβ, FIXaγ and FIXaγ/FIXaδ mixture

Next, we compared hydrolysis of synthetic substrate CBS 31.39 by FIXβ, FIXaγ and FIXaγ/FIXaδ (25/75) mixture. Method details are provided in the supplementary material and the data are presented in Table 1. Addition of Ca²⁺ to FIXaβ resulted in ~6.7-fold increase in K_m and ~6.4-fold increase in k_cat, whereas essentially no effect on the K_m or k_cat was observed for FIXaγ or FIXaγ/FIXaδ mixture. Overall, FIXaγ or FIXaγ/FIXaδ mixture had a decreased k_cat in the presence and absence of Ca²⁺ and ~10-fold reduced k_cat/K_m as compared to FIXaβ. Since k_cat/K_m values for FIXaγ and FIXaγ/FIXaδ mixture are similar, it is safe to conclude that FIXaγ and FIXaδ hydrolyze CBS 31.39 with comparable efficiency.

Table 1.

Kinetic parameters for the hydrolysis of CBS 31.39 by different FIXa species

Protein	Km (mM)	kcat (min−1)	kcat/Km (min−1 mM−1)
FIXaβ (+Ca2+)	4.31±0.35	189±12	43.9
FIXaβ (−Ca2+)	0.64±0.07	29.4±2.2	45.9
FIXaγ (+Ca2+)	2.94±0.32	11.3±1.1	3.8
FIXaγ (−Ca2+)	2.55±0.26	10.5±1.0	4.1
FIXaγ/δ (+Ca2+)	3.01±0.37	11.5±1.3	3.8
FIXaγ/δ (−Ca2+)	3.25±0.42	11.9±1.4	3.6

Concentration of each reagent in the reaction mixture was: 107 nM FIXa with either 5 mM CaCl₂ or 1 mM EDTA and varying concentrations of CBS 31.39.

FX activation in the presence and absence of FVIIIa

Next, we investigated activation of FX by FIXaγ or FIXaγ/FIXaδ (25/75) mixture. In the presence of Ca²⁺/PL only, FIXaγ activated FX with K_m ~467 nM as compared to ~96 nM for FIXaβ (Fig. 3A). FIXaγ also had ~6-fold reduced k_cat (1.9±0.2 × 10⁻³ min⁻¹) as compared to FIXaβ (11.1±0.9 × 10⁻³ min⁻¹) in activating FX. Limited data presented using FIXaγ/FIXaδ behaved similar to FIXaγ in activating FX in the presence of Ca²⁺/PL (Fig. 3, triangles). Thus, as compared FIXaβ, FIXaγ or FIXaδ had ~30-fold decrease in k_cat/K_m in activation of FX in the absence of FVIIIa. In the presence of FVIIIa, under the kinetic condition used, FIXaγ or FIXaγ/FIXaδ mixture was ~650–fold less active than FIXaβ (Fig. 3B). Thus, FIXaγ or FIXaδ is impaired in hydrolyzing the synthetic substrate as well as the physiologic substrate FX.

Fig 3. FX activation by FIXaβ, FIXaγ and FIXaγ/FIXaδ (25/75) mixture in the absence and presence of FVIIIa.

A, Activation of FX in the absence of FVIIIa. Each reaction mixture contained 20 nM FIXaβ or FIXaγ or FIXaγ/FIXaδ (25/75) mixture, 25 μM PL, 5 mM CaCl₂ and various concentrations of FX. Activation was carried out at 37°C and FXa activity was measured by hydrolysis of synthetic substrate S-2222. K_m and k_cat were determined as described in Materials and methods. B, Activation of FX in the presence of FVIIIa. Each reaction mixture contained 0.05 nM of FIXa, 10 μM PL, 0.07 nM FVIIIa, 5 mM CaCl₂ and 15 nM FX. Activation was carried out at 37°C from 0.5 to 2 min, and FXa activity was measured by hydrolysis of synthetic substrate S-2222 as described in Materials and methods. In both A and B, closed circles represent FIXaβ, open circles represent FIXaγ and triangles represent FIXaγ/FIXaδ (25/75) mixture. The data for FIXaβ or FIXaγ represent average of two experiments. The limited data for the FIXaγ/FIXaδ mixture are presented to verify that FIXaδ behaves similar to FIXaγ.

Binding of FVIIIa to FIXaγ/FIXaδ

FIXaγ/FIXaδ are defective in FX activation in the presence of Ca²⁺/PL, and addition of FVIIIa in the system leads to further impairment in FX activation. To explore these observations further, we evaluated K_d of interaction of FIXaγ and FIXaγ/FIXaδ mixture with FVIIIa. In this competition-based assay, dEGR-IXaβ and dEGR-IXaγ or dEGR-IXγ/dEGR-IXaδ were used to compete for FIXa binding to FVIIIa and inhibit FXa generation. K_dapp obtained under these conditions for dEGR-IXaβ was 0.7 nM and for dEGR-IXaγ it was 56 nM (Fig. 4). Limited data presented with dEGR-IXaγ/dEGR-IXaδ mixture support that dEGR-IXaδ behaves similar to dEGR-IXaγ. Thus, dEGR-IXaγ or dEGR-IXaδ is ~80-fold defective in binding to FVIIIa suggesting that the FVIIIa binding site is affected by plasmin cleavage primarily at the Lys316-Gly317 peptide bond.

Fig 4. Inhibition of FXa generation by dEGR-IXaβ and dEGR-IXaγ or dEGR-IXaγ/dEGR-IXaδ (25/75) mixture in the presence of FVIIIa.

Each reaction mixture contained 0.2 nM FIXa, 0.07 nM FVIIIa, 480 nM FX, 10 μM PL, 5 mM CaCl₂, and varying concentrations of dEGR-IXa proteins (dEGR-IXaβ or dEGR-IXaγ or dEGR-IXaγ/dEGR-IXaδ mixture) in TBS/BSA, pH 7.5. The percent of FXa generated at various concentrations of the competitor—dEGR-IXaβ or dEGR-IXaγ or dEGR-IXaγ/dEGR-IXaδ mixture is depicted. The curves represent best fit to the IC₅₀ four-parameter logistic equation (Eq. 1). Closed circles represent dEGR-IXaβ, open circles represent dEGR-FIXaγ and triangles represent dEGR-IXaγ/dEGR-IXaδ mixture. The data for dEGR-IXaβ or dEGR-IXaγ represent average of two experiments. Limited data for the dEGR-IXaγ/dEGR-IXaδ mixture are presented to verify that dEGR-IXaδ behaves similar to dEGR-IXaγ.

Plasmin binding to FIXaγ and FIXaδ using ligand blots

Ligand binding blots were performed in which plasmin, FIXaβ, FIXaγ and FIXaγ/FIXaδ mixture were first run on non-reduced SDS-PAGE and transferred to Immobilon-P PVDF membrane. The membrane was incubated with Glu-plasminogen or DIP-plasmin and bound plasminogen or plasmin was detected using anti-plasminogen polyclonal antibody. In this assay, Glu-plasminogen and DIP-plasmin bound to Hγ_N of FIXaγ and to Hδ_N (same fragment as Hγ_N) as well as to Hδ_C of FIXaδ but not to FIXaβ or Hγ_C of FIXaγ (Figs. 5A and 5B). In addition, when FIXaγ/FIXaδ blot was incubated with 100 units/mL CpB prior to incubation with DIP-Plasmin, the binding of DIP-Plasmin to Hγ_N of FIXaγ or FIXaδ and Hδ_C of FIXaδ was abrogated (Fig. 5C). Further, when FIXaγ/FIXaδ blot was incubated with the DIP-Plasmin in the presence of EACA, DIP-Plasmin binding to FIXaγ/FIXaδ was not observed (Fig. 5D). These results suggest that plasminogen/plasmin, through its Kringle domain(s), binds to the C-terminal Lys316 of FIXaγ/FIXaδ and C-terminal Lys413 of FIXaδ. Removal of these C-terminal lysine residues by CpB abolishes plasmin binding, as does the presence of EACA, a lysine analog.

Fig 5. Ligand blot assay of FIXaβ, FIXaγ and FIXaδ.

Plasmin, FIXaβ, FIXaγ and FIXaγ/FIXaδ mixture were electrophoresed using 12% non-reduced SDS-PAGE. The concentration of each protein (nanogram) loaded is indicated. The proteins were transferred to an Immobilon-P PVDF membrane and the membrane was blocked using 5% milk. A, The membrane was then incubated with 3 μg/mL Glu-plasminogen. To detect bound plasminogen, a goat polyclonal plasminogen antibody was used followed by a secondary anti-goat antibody. The blot was developed with the Amersham ECL kit. B, The membrane was incubated with 3 μg/mL DIP-plasmin (instead of Glu-plasminogen) and the bound DIP-plasmin was detected using plasminogen antibody as in panel ‘A’. C, The membrane was treated with 100 U/mL of CpB for 30 minutes at 37ºC prior to incubation with DIP-plasmin. Detection of bound DIP-plasmin was performed as in panel ′B′. D, The membrane was incubated with DIP-plasmin in the presence of EACA and detection of the bound DIP-plasmin was performed as in panel ′B′. Pm, plasmin; FIXa-L, light chain of FIXa; Hγ_N, NH₂-terminal of FIXaγ heavy chain; FIXa-Hδ_C, C-terminal of FIXaδ heavy chain.

Effect of FIXaγ and FIXaδ on tPA-mediated plasminogen activation

Both plasminogen and tPA use their Kringle domains to bind to the C-terminal lysine residues in fibrin for colocalization at the clot site. Glu-plasminogen exists in closed T-state conformation and adopts the open R-state conformation upon binding to EACA or the C-terminal lysine residues on cell surface protein receptors [34–37]. Intact Glu-plasminogen (T-state) in the absence of ω-amino acids of the lysine class e.g., EACA, is severely attenuated in its activation by plasminogen activators, but the R-state that exists upon binding to EACA or the C-terminal lysine residues in the pertinent protein receptors is highly activatable [36]. The Glu-plasmin initially formed by proteolysis at Arg561-Val562 rapidly removes the NH₂-terminal 77 residues in Glu-plasmin/Glu-plasminogen to yield Lys-plasmin/Lys-plasminogen, which exist in open R-conformation ± EACA [37]; the Lys-plasminogen is then activated to the final product Lys-plasmin.

Since Glu-plasminogen is the native circulating plasminogen, we studied enhancement of tPA-mediated activation of Glu-plasminogen by FIXaβ, FIXaγ and FIXaγ/FIXaδ (25/75) mixture. Detailed methods are given in the supplementary material. As shown in Fig 6, FIXaβ or FIXaγ did not potentiate the activation of Glu-plasminogen by tPA. In contrast, FIXaγ/FIXaδ potentiated tPA mediated plasminogen activation in a concentration dependent manner. In control experiments (Table S1), FIXaβ, FIXaγ or FIXaγ/FIXaδ mixture under similar conditions had no influence on the catalytic activity of tPA (synthetic substrate S-2288) or plasmin (synthetic substrate S-2251). Although we cannot exclude other mechanisms, it appears that the enhancement of tPA-mediated plasminogen activation by FIXaδ is likely mediated via the C-terminal lysine residues (Lys316 and Lys413) of FIXaδ which could allow both tPA and plasminogen/plasmin binding and colocalize them.

Fig 6. tPA-mediated activation of plasminogen by FIXaδ but not by FIXaβ or FIXaγ.

Each reaction mixture (100 μL) contained 5 mM CaCl₂, 1 μM Glu-plasminogen, and 10 nM tPA in TBS pH 7.5. In one set of experiments, reaction mixtures contained 10 nM (diamonds), 50 nM (squares) or 100 nM (triangles) dEGR-IXaγ/dEGR-IXaδ (25/75) mixture. In other experiments, reaction mixtures contained 100 nM dEGR-IXaβ (circles), 100 nM dEGR-IXaγ (hexagon) or 100 nM BSA (inverted triangles). The reaction mixtures were incubated for 5–30 minutes at 37°C and 5 μL aliquots were removed at a given time, diluted 30-fold in TBS/BSA pH 7.5 and placed on ice. Plasmin activity was measured using synthetic substrate S-2251 as outlined in Materials and methods. The plasmin generated was calculated from a standard curve measuring S-2251 hydrolysis by purified plasmin. The values represent mean of three different experiments.

Enhancement effect of FIXaδ on plasma clot lysis

Initial experiments were performed to study the fibrinolytic effect of dEGR-IXaβ, dEGR-IXaγ and dEGR-IXaδ in the tPA (0.25 μg/mL)-induced clot lysis in FIX-depleted plasma. Addition of RVV-X to FIX-depleted plasma in the absence of tPA caused fibrin formation with minimal lysis as reflected by an increase in OD405 (Fig 7A). Simultaneous addition of tPA caused dissolution of the fibrin clot induced by tPA mediated conversion of plasminogen to plasmin. Further, addition of dEGR-IXaβ or dEGR-IXaγ had essentially no effect on promoting clot lysis. In contrast, addition of dEGR-IXaγ/dEGR-IXaδ mixture enhanced tPA-induced plasma clot lysis. Since dEGR-IXaγ had no effect, the enhanced clot lysis seen with dEGR-IXaγ/dEGR-IXaδ mixture can be attributed primarily to dEGR-IXaδ, which could bind plasminogen and tPA simultaneously. The midpoint of clot lysis at 7.5 nM and 15 nM dEGR-IXaδ in the dEGR-IXaγ/dEGR-IXaδ mixture was ~28min and ~32min, respectively as compared to ~36min without dEGR-IXaδ. Consistent with these observations, when NPP (Gray line, Fig 7A) was used instead of FIX-depleted plasma, the fibrinolysis was similar to the FIX-depleted plasma. Next, we examined whether increasing the concentration of tPA from 0.25 μg/mL to 0.5 μg/mL would result in enhanced fibrinolysis by FIX/FIXa in the RVV-X activation system. However, fibrinolysis was similar in both NPP (mid-point of clot lysis ~19 min) and FIX-depleted plasma (mid-point of clot lysis ~20 min). These data indicate that FIX/FIXa does not contribute to fibrinolysis when clotting is initiated with RVV-X. The reason(s) why FIX/FIXa does not contribute to fibrinolysis in this system is unknown.

Fig 7. Effect of FIX on plasma clot lysis assay.

A, Effect of FIXaδ on the FIX-depleted plasma clot lysis. Clot formation was initiated with dilute RVV-X, and fibrinolysis was induced by simultaneous addition of 0.25 μg/mL tPA. Fibrin formation and lysis were monitored by measuring the optical density at 405 nm as outlined under Materials and methods. The effects of addition of dEGR-IXaγ/dEGR-IXaδ (25/75) mixture (10 nM and 20 nM), dEGR-IXaγ (100 nM) and dEGR-IXaβ (100 nM) are presented. Control experiments consisted of no tPA ± dEGR-IXaβ, dGER-IXaγ or dEGR-IXaγ/dEGR-IXaδ mixture. The fibrinolysis observed with NPP is also presented. B, tPA mediated plasma clot lysis using FXIa as the clotting initiator. Concentration of FXIa used was 64 ng/mL and tPA was 0.5 μg/mL. NPP, 50% NPP:50% FIX-depleted plasma, and 20% NPP:80% FIX-depleted plasma were used. Control experiments consisted of no tPA.

Next, we examined if FIX/FIXa contributes to fibrinolysis when clotting is initiated by FXIa, a physiologic activator of FIX. Here, we used 64 ng/mL FXIa and 0.5 μg/mL tPA and 100% NPP, 50% NPP:50% FIX-depleted plasma, and 20% NPP:80% FIX-depleted plasma. These data are presented in Fig 7B. The mid-point of clot lysis was ~18 min for NPP, ~21 min for 50% NPP:50% FIX-depleted plasma, and ~25 min for 20% NPP:80% FIX-depleted plasma. These data clearly indicate that FIX/FIXa does contribute to fibrinolysis when a physiologic activator of FIX is used.

Discussion

The data presented here reveals that in the presence of Ca²⁺/PL, plasmin cleavage in FIXaβ at Lys316↓Gly317 (FIXaγ) leads to loss of catalytic activity. Notably, the cleaved fragment likely stays associated with FIXa as is the case in γ-thrombin [38]. This is due to the interaction of the cleaved C-terminal fragment with the NH₂-terminal heavy chain of FIXa involving several salt-bridges (~10) and hydrogen bonds (~32) [39,40]. The loss of catalytic activity (~10-fold for synthetic substrate and ~30-fold for FX) could stem from the impairment of catalytic triad and FVIIIa binding (~80-fold) [41–43] as observed for γ-thrombin activation of fibrinogen and FXIII [44] as well as impairment of GpIbα binding to exosite-II in γ-thrombin [45,46].

The mass spec data revealed that plasmin cleaves FIXaγ at Lys413-Leu414 generating a second C-terminal lysine in FIXaδ. The ligand blots demonstrate that FIXaγ has one while FIXaδ has two binding sites for plasminogen/DIP-plasmin. This difference has important biologic implication in that FIXaδ by simultaneous binding to tPA and plasminogen enhances tPA-mediated plasminogen activation and plasma clot lysis. Since C-terminal Lys413 is missing in FIXaγ, it can only bind either plasminogen or tPA that is insufficient to bring them in close proximity for plasminogen activation and plasma clot lysis. The second cleavage in FIXaγ at Lys413↓Leu414 is consistent with plasmin cleavages in FXaα at Lys330↓Gly331 and Lys435↓Ser436 for converting it to a fibrinolytic enhancer. Notably, the second cleavage at Lys435↓Ser436 in FXaα is not obligatory, since FXaα already has C-terminal Lys448 [13,20]. However, plasmin cleavage at Lys413↓Leu414 in FIXaγ is essential for its conversion to a fibrinolytic enhancer [13].

The biochemical properties of FIXaγ/FIXaδ are in agreement with previous studies describing a pathway between coagulation and fibrinolysis involving plasmin modulation of FVa, FXa, and FVIIIa [9–13,15–18]. In these studies, plasmin was shown to obliterate procoagulant functions of these proteins, while converting them into tPA and plasminogen/plasmin receptors to facilitate fibrinolysis. In previous studies, plasmin cleaved FIX at multiple sites (introduction and Fig. 2B) in the absence of PL membrane surface [14]. Interestingly, in the absence of PL, plasmin also cleaved the light chain of FX/FXa at residue Lys43 [8,10,11]. In the presence of Ca²⁺/PL, we also did not observe cleavage at Arg318-Ser319 in FIXaβ. The reason(s) for this is unknown; however, cleavage at this site does not generate C-terminal lysine and cannot participate in tPA or plasminogen/plasmin binding. Because of the methodology used, Samis et al. [14] also did not detect proteolysis in FIX at Lys413-Leu414, which in all probability occurred. If so, although not tested, their product could enhance fibrinolytic activity as we have observed in Fig 7B. Although the data presented in Fig 2A are non-physiological in nature, the results demonstrate that cleavages at Lys316-Gly317 and Lys413-Leu414 are most likely needed for generation of fibrinolytic activity. While the mechanism for two binding sites (plasminogen and tPA) on FIXaδ is plausible, the data are equally consistent with FIXaδ in conjunction with appropriately situated fibrin, serving as an accelerating scaffold.

The present study highlights the intricate balance between coagulation and fibrinolysis. Although many coagulation factors are viewed as only procoagulants, they can also contribute to fibrinolysis. These findings are critical from a clinical perspective especially in the setting of trauma. Many patients presenting with major trauma have thromboelastography (TEG) performed upon presentation to the emergency department. Often, the TEG shows a shortened R time (reaction time) indicating activation of the clotting cascade, and increased LY30 (percentage decrease in 30 min post-Maximum Amplitude) indicating increased fibrinolytic activity. When patients present with increased LY30, tranexamic acid is administered based upon the CRASH-2 study [47,48]. Tranexamic acid, a lysine analog competes with plasmingeon/plasmin and t-PA which utilize C-terminal lysine in binding to the Kringle domains. Tranexamic acid administration decreases fibrinolysis and helps restore hemostasis. Thus, this study and others serve to further our understanding of the biochemical process during the period of clotting activation and increased fibrinolysis as occurs in trauma patients.

Addendum

Amy E. Schmidt, Kanagasabai Vadivel and Julian Whitelegge performed the experiments. Amy Schmidt, Kanagasabai Vadivel and S. Paul Bajaj conceived, designed, analyzed and interpreted the data, and wrote the manuscript.