Autoproteolysis or plasmin-mediated cleavage of factor Xaalpha exposes a plasminogen binding site and inhibits coagulation.

Blood coagulation factor Xa (FXa) has recently been shown to function as a plasminogen receptor in the presence of procoagulant phospholipid (phosphatidylserine; PS) and Ca2+. In the current work, the possible effect of autoproteolytic and plasmin-mediated cleavage of FXa on complex formation with plasminogen was investigated. 125I-plasminogen binding to derivatives of FXa electrotransferred to polyvinylidene difluoride revealed that the autoproteolytic conversion of FXaα to FXaβ was required for the expression of a plasminogen binding site. In the presence of PS and Ca2+, plasmin was shown to convert FXaα to a FXaβ-like species at least 3 orders of magnitude faster than the autoproteolytic mechanism. This also resulted in the exposure of a plasminogen binding site. Further processing by plasmin generated a fragment (33 kDa) due to cleavage at Gly331 in the FXa heavy chain. Production of this species enhanced apparent plasminogen binding compared with FXaβ and resulted in the loss of FXa amidolytic and clotting activity. In the absence of either PS or Ca2+, the plasmin-mediated fragmentation of FXaα was altered to include a FXaβ-like molecule and a species (40 kDa) with intact β-heavy chain disulfide linked to a COOH-terminal fragment of the light chain starting at Tyr44. Neither of these products was observed to interact with plasminogen. The 40-kDa species had amidolytic activity comparable with FXaα but inhibited clotting activity. Cumulatively the data provide the first evidence for a functional difference between the FXa subforms and suggest a mechanism where autoproteolysis and plasmin-mediated cleavage modulate the function of FXaα from a procoagulant enzyme to a profibrinolytic plasminogen receptor.

Blood coagulation factor Xa (FXa) has recently been shown to function as a plasminogen receptor in the presence of procoagulant phospholipid (phosphatidylserine; PS) and Ca 2؉ . In the current work, the possible effect of autoproteolytic and plasmin-mediated cleavage of FXa on complex formation with plasminogen was investigated. 125 I-plasminogen binding to derivatives of FXa electrotransferred to polyvinylidene difluoride revealed that the autoproteolytic conversion of FXa␣ to FXa␤ was required for the expression of a plasminogen binding site. In the presence of PS and Ca 2؉ , plasmin was shown to convert FXa␣ to a FXa␤-like species at least 3 orders of magnitude faster than the autoproteolytic mechanism. This also resulted in the exposure of a plasminogen binding site. Further processing by plasmin generated a fragment (33 kDa) due to cleavage at Gly 331 in the FXa heavy chain. Production of this species enhanced apparent plasminogen binding compared with FXa␤ and resulted in the loss of FXa amidolytic and clotting activity. In the absence of either PS or Ca 2؉ , the plasmin-mediated fragmentation of FXa␣ was altered to include a FXa␤-like molecule and a species (40 kDa) with intact ␤-heavy chain disulfide linked to a COOH-terminal fragment of the light chain starting at Tyr 44 . Neither of these products was observed to interact with plasminogen. The 40-kDa species had amidolytic activity comparable with FXa␣ but inhibited clotting activity. Cumulatively the data provide the first evidence for a functional difference between the FXa subforms and suggest a mechanism where autoproteolysis and plasmin-mediated cleavage modulate the function of FXa␣ from a procoagulant enzyme to a profibrinolytic plasminogen receptor.
The serine protease factor Xa (FXa) 1 functions to generate the principal biological effector of blood coagulation, thrombin (1). To avoid thrombosis, the expression of FXa activity is strictly limited. One level of regulation is the requirement that FXa associate with the nonenzymatic cofactor Va (FVa) and procoagulant phospholipid (e.g. phosphatidylserine (PS)) in the presence of Ca 2ϩ to form prothrombinase. Within this complex, the catalytic rate of FXa toward the thrombin precursor, pro-thrombin, is accelerated 5 orders of magnitude and becomes biologically significant (2). The need to assemble into prothrombinase furthermore confines FXa activity to sites of vascular damage, where PS is accessible (3)(4)(5)(6).
Although FXa biochemistry is well studied, a question that still remains concerns the functional significance of an autoproteolytic step that excises a 4-kDa glycopeptide (␤-peptide) from the heavy subunit COOH terminus (7)(8)(9). The intact species, FXa␣, is converted to FXa␤. This process is greatly accelerated by Ca 2ϩ -dependent binding to vesicles containing PS (7, 9 -11). FXa␤ has been shown to undergo a second autoproteolytic step when phospholipid binding is facilitated (i.e. PS and Ca 2ϩ ) that produces a species referred to as FXa␥ (10). Since this cleavage is at a position NH 2 -terminal to the active site Ser, irreversible inhibition of enzymatic function results. In the absence of PS, the second autoproteolytic cleavage appears to occur at a position on the light chain that inhibits clotting but not amidolytic activity (12).
In a recent study, FXa was observed to accelerate the generation of the fibrinolytic enzyme plasmin by tissue plasminogen activator (tPA) (13). This was found to involve an interaction with the precursor, plasminogen (13), which is known from other work to require Lys at the COOH terminus of the receptor (14,15). The acceleration of tPA by FXa was found to be dependent on the presence of both PS and Ca 2ϩ and was inhibited by the inactivation of FXa␣ by Glu-Gly-Arg-chloromethylketone. One explanation for these observations may be the inhibition of FXa autoproteolysis, which could be necessary for complex formation with plasminogen. Evidence for an involvement of the FXa COOH terminus was given by the finding that carboxypeptidase pretreatment of FXa inhibited the acceleration of tPA (13). Since autoproteolysis creates a new COOH terminus, FXa␣ subform conversion may be required to provide an accessible Lys for plasminogen binding.
In addition to the possible involvement of autoproteolysis in the tPA cofactor function of FXa, the involvement of plasminmediated proteolysis was suggested by the observation that plasmin treatment of FXa decreased the profibrinolytic activity by approximately 50% (13). The proteolysis of FXa by plasmin has not been previously reported, although plasmin is known to have a relatively broad specificity (16). As a pertinent example, cleavage of the FXa cofactor, FVa, by plasmin has been observed and was reported to alter function through inhibition of coagulation (17,18) and acceleration of tPA (13). Evidence to suggest an effect of plasmin on FXa function in coagulation is not available.
The goal of the current study was to clarify the roles of autoproteolysis and plasmin-mediated cleavage in the plasminogen binding function of FXa. We now report that plasmin rapidly converts FXa␣ into a FXa␤-and FXa␥-like species and that either autoproteolytic or plasmin-mediated processing of FXa␣ is required for expression of a plasminogen binding site.
Proteins-Human coagulation factor X (FX) was purified from fresh frozen plasma (20) or from prothrombin complex concentrate diluted to 1 unit/ml factor X clotting activity (obtained as generous gifts from the Canadian Red Cross Society, Ottawa Collection Centre, or from Bayer, Inc., respectively). For comparison to in-house preparations, FX was also purchased (Haematologic Technologies, Inc.). FXa␣ was generated from FX (21,22) by treatment with the purified activator from Russell's viper venom (RVV, Haematologic Technologies) (23) followed by affinity chromatography using benzamidine-Sepharose (Pharmacia Biotech Inc.) to remove the FX activation peptide, RVV, and residual inactivated FX (22). Autoproteolysis during the course of FX activation was inhibited by conducting the reaction in a slurry of benzamidine-Sepharose (ϳ0.3 ml of packed resin per ml). Human Glu-plasminogen was purified from fresh frozen plasma as described (24) and passed over benzamidine-Sepharose (Pierce) to remove any trace of serine protease activity. Plasminogen was radioiodinated (50,000 -150,000 dpm/g) using Iodogen (Pierce) and chromatographically desalted (Excellulose 5, Pierce) or dialyzed to remove unincorporated 125 I. The homogeneity of proteins was assessed by SDS-polyacrylamide electrophoresis (SDS-PAGE) (25) and where applicable by autoradiography using X-Omat AR film (Kodak) and Quanta III intensifying screens (Dupont). Purified human plasmin was prepared commercially (Haematologic Technologies).
Two-dimensional Electrophoresis-The disulfide-linked subunit compositions of plasmin-digested FXa␣ end products were investigated by nonconventional two-dimensional electrophoresis (two-dimensional SDS-PAGE), as described previously (26). The protocol involved excising the appropriate lane from a 0.75-mm-thick nonreduced SDS-PAGE (12% acrylamide) and incubating it with 1 ϫ Laemmli sample buffer (25) containing 300 mM dithiothreitol for 20 min at 55°C. The excised lane was rinsed once with reservoir buffer and slipped into a 1.5-mmthick 14% acrylamide gel. A stacking gel with two wells for reference proteins and a single well large enough to accept the reduced gel slice was precast for the second dimension of SDS-PAGE. To identify proteolytic cleavage sites, the two-dimensional SDS-PAGE or one-dimensional SDS-PAGE under reducing conditions was transferred to PVDF, and bands were excised for amino acid sequence analysis (Dr. A. Kurosky, Protein Chemistry Laboratory, University of Texas at Galveston, Medical Branch) (27,28).
Ligand Blots-To identify FX-or FXa␣-derived species that interact with plasminogen, ligand blotting experiments were performed according to methods published by several laboratories (29 -31). In our studies, FX or FXa␣ fragments were electrotransferred to PVDF after SDS-PAGE (27,28). The PVDF was blocked overnight at 4°C in bovine serum albumin (Sigma, 10 mg/ml) and then incubated with 125 I-plasminogen (0.1 M) for 1 h at 22°C in the presence of the protease inhibitors, 2-guanidinoethylmercaptosuccinic acid (50 nM) and aprotonin (Calbiochem, 50 kallikrein-inactivating units/ml). Following extensive washing with HBS, the location of bound 125 I-plasminogen was determined by autoradiography and compared with the electrophoretic patterns made visible by staining the PVDF with Coomassie Brilliant Blue R-250.
FXa Activity Assays-The effect of plasmin on FXa␣ amidolytic and procoagulant activity over time was evaluated. Identical time courses were initiated as described above. Amidolytic activity was followed at 22°C in HBS using S2337 (200 M) as a chromogenic substrate. The rate of color development due to the activity associated with 10 nM FXa␣ at various times of incubation was monitored in a kinetic multiwell plate reader (V max , Molecular Devices). Procoagulant activity was measured in one-stage tilt clotting assays using FX-deficient plasma (Sigma) as substrate in the presence of 50 M PCPS. At each incubation time point, the plasmin activity was inhibited by adding a small volume of aprotonin to a final concentration of 50 kallikrein-inactivating units/ ml. The FXa activity was determined using a standard curve in which known concentrations of purified FXa were titrated in the presence of plasmin and aprotonin to duplicate the conditions of the time course samples.

Comparison of Plasminogen Binding to FXa␣ and FXa␤-
The binding of 125 I-plasminogen to proteins electrotransferred to PVDF (or nitrocellulose) has been used extensively in the past for the identification of plasminogen receptors (29 -31). To determine the effect of FXa␣ autoproteolysis on the interaction with plasminogen, similar ligand blotting experiments were conducted. In this system ligand blotting is advantageous because the relative amount of plasminogen associated with discrete proteolytic products of FXa␣ in a mixture can be directly analyzed due to the resolution offered by electrophoresis. Fig. 1 shows the autoproteolysis of FXa␣ in the absence (panel A) and presence (panel C) of conditions that facilitate PCPS binding. The resulting autoradiograms (panels B and D, respectively), revealed that expression of new plasminogen binding sites paralleled FXa␤ generation. The amount of plasminogen-FXa␣ complex detected was small compared with that observed with FXa␤. In the absence of PCPS binding (panels A and B), subform conversion was insignificant, and as a result expression of new plasminogen binding sites was not evident. This demonstrated that structural changes in FXa␣ over the course of the experiment that are indiscernible by SDS-PAGE do not account for the observed plasminogen association. The amount of intrinsic plasminogen binding to FXa␣ varied somewhat between preparations but was always low compared with FXa␤. The batch of FXa␣ used to conduct the experiment in Fig. 1 had the highest amount of endogenous plasminogen binding that we observed. All other preparations of FXa␣ (or FX) bound even less plasminogen relative to FXa␤ and were used in the remaining experiments presented here (e.g. Figs. 2, 5, and 6).
Pretreatment of the FXa␤ with carboxypeptidase B prior to conducting the ligand blotting experiment, as described before (13), or inclusion of ⑀-aminocaproic acid (10 mM) with the 125 Iplasminogen inhibited the interaction (data not shown). These observations indicated an involvement of a COOH-terminal Lys. The specificity of 125 I-plasminogen binding was demonstrated by inhibition due to excess unlabeled plasminogen and the lack of binding to human prothrombin transferred to PVDF (data not shown). Furthermore, no binding was detected when the 125 I-plasminogen was substituted for 125 I-FVa or 125 I-FXa.
Plasminogen Binding to FX Activation Products-The activation of FX requires proteolysis of only the NH 2 terminus of the heavy subunit. Therefore FX and FXa␣ should possess the same COOH-terminal amino acid residue, which is predicted from the cDNA sequence to be Lys (32,33). To test whether our relative inability to detect plasminogen binding to FXa␣ was acquired during zymogen activation, FX was activated with RVV and allowed to subsequently autoproteolyze. The time course of this reaction (panel A) and corollary plasminogen blot (panel B) are presented in Fig. 2. These data showed that neither FX nor FXa␣ interact appreciably with plasminogen in comparison with FXa␤.
Cleavage of FXa␣ by Plasmin-To directly establish if FXa␣ is a substrate for plasmin, as suggested by previous functional data (13), digestion time courses were carried out and are shown in Fig. 3. The data in panel B demonstrated that FXa␣ is indeed cleaved by plasmin. For comparison, the autoproteolysis of FXa␣ under identical conditions except in the absence of plasmin is shown in panel A. In the presence of PCPS and Ca 2ϩ , plasmin generated molecules corresponding in electrophoretic mobility under nonreducing conditions to FXa␤ and FXa␥ (33 kDa). The appearance of a faint band was routinely observed just above the dye front (ϳ13 kDa) and correlated to the appearance of the 33-kDa species. Although 50-fold less plasmin was present in the reaction mixtures than FXa␣, the conversion of FXa␤ generation after 1 min in the presence of plasmin (panel B) is comparable with that after 30 min by autoproteolysis (panel A). This indicated that on a molar basis FXa␣ subform conversion is mediated more than 1500 times faster by plasmin than by FXa.
A dependence on PCPS and Ca 2ϩ was previously observed for the participation of FXa in plasminogen activation by tPA (13). To determine whether the Ca 2ϩ -dependent binding to PCPS was necessary for fragmentation of FXa␣ by plasmin, an identical experiment in the presence of the chelator EDTA was conducted. The fragmentation time-course shown in panel C demonstrated that inhibition of the interaction of FXa␣ with PCPS by EDTA altered the recognition of cleavage sites by plasmin. In the presence of EDTA, production of a FXa␤-like species was also observed as the initial fragment. However, this migrated as an electrophoretic doublet, suggesting an additional or alternate cleavage(s). Subsequently a product with an apparent molecular mass of 40 kDa was generated. The same SDS-PAGE pattern was observed when the reaction was conducted in the presence of Ca 2ϩ but without PCPS. This showed that binding to PCPS influenced the plasmin-dependent FXa␣ cleavage pattern and that conformational changes in FXa that are induced by Ca 2ϩ (34 -37) were not responsible.
To obtain information regarding the location of the plasminmediated cleavage sites on FXa␣, 30-min digestions were conducted and subjected to two-dimensional SDS-PAGE (the first dimension was nonreduced with a reduced second dimension). As shown in Fig. 4A, the 33-kDa species that was produced when FXa was bound to PCPS resolved into subunits corresponding to the intact light chain of FXa and a 15-kDa fragment of the FXa heavy chain. Amino acid sequence analysis of the 15-kDa species demonstrated identity to the NH 2 terminus of the FXa heavy subunit ( Table I). The location of the heavy chain cleavage site was obtained by sequence analysis of the 13-kDa fragment not disulfide-linked to the light chain, which revealed a new NH 2 terminus corresponding to Gly 331 (Table  I). 2 A similar proteolytic pattern was observed when Ca 2ϩ was present but PCPS was omitted (data not shown). The same 33-kDa species was produced if FXa␣ was first allowed to convert to FXa␤ by autoproteolysis, followed by the addition of plasmin (data not shown).
The 40-kDa plasmin-mediated fragment formed when FXa␣ was not bound to PCPS was demonstrated by two-dimensional SDS-PAGE (Fig. 4, panel B) to be composed of the intact FXa␤ heavy subunit and a 13-kDa fragment of the FXa light chain. The species that was detected under nonreducing conditions that migrated slightly below the FXa␤-like species to form a doublet (Fig. 3, panel C) is shown faintly in Fig. 4 to be composed of an 18-kDa light chain fragment disulfide linked to the FXa␤ heavy chain. Since this species did not accumulate, it was presumed to be an intermediate leading to the production of the 40-kDa fragment. Sequence analysis of the band corresponding to the 13-kDa fragment observed on one-dimensional SDS-PAGE under reducing conditions revealed an NH 2 terminus beginning at Tyr 44 (Table I). A lesser species that comigrated with the 13-kDa light subunit fragment was also identified by sequence analysis and aligned with the NH 2 terminus of the FXa heavy subunit (data not shown). This may be due to production of a FXa derivative composed of the 13-kDa light subunit piece disulfide-linked to a 13-kDa fragment of the heavy chain, which we have yet to resolve.
Both panels A and B showed lower apparent molecular 2 Numbering based on the amino acid sequence predicted from FX cDNA (33). weight bands that were not detected in Fig. 3 due to less protein loaded onto the gel. Their identities are currently being investigated. A small amount of the 40-kDa fragment was observed in panel A, which indicated that a fraction of the FXa␣ was not bound to phospholipid under the conditions of this digest.
Plasminogen Binding to Plasmin-mediated FXa␣ Products-To determine whether plasminogen binding sites are exposed when FXa␣ is cleaved by plasmin, the electrophoretic time courses shown in Fig. 3 were transferred to PVDF and blotted with 125 I-plasminogen. The resulting autoradiographs are presented in Fig. 5. When conditions were established to promote the association of FXa␣ and PCPS (panel A), plasminogen binding initially correlated with the production of the FXa␤-like species. The second cleavage induced by plasmin (giving rise to the 33-kDa species) was observed to substantially enhance plasminogen binding. Under reducing conditions (data not shown), the association of plasminogen was found to be predominantly to the 15-kDa heavy subunit-derived fragment.
In sharp contrast to the plasmin-mediated products of FXa␣ that are formed when the association of PCPS is favored (panel A), plasminogen binding was observed to be insignificant to the fragments of FXa␣ cleaved by plasmin in EDTA (panel B). Although similar in electrophoretic mobility, these data suggested that the first species formed by plasmin in the absence of phospholipid binding differs from FXa␤.
Comparison of Autoproteolytic and Plasmin-mediated FXa␣ Fragments-Having demonstrated that plasmin generates a FXa␣ fragment in the presence of PCPS and Ca 2ϩ with similar apparent molecular weight as authentic FXa␥, we compared the affinity of the two species for plasminogen. The stained gel in Fig. 6A shows the relative amount of protein blotted to PVDF, which was subsequently probed with 125 I-plasminogen to produce the autoradiograph in panel B. Although the electrophoretic migration of FXa␥ formed by autoproteolysis is the same as that produced by plasmin, the amount of plasminogen binding is clearly disproportionate and favors the latter. This experiment also confirms the inability of FX, FXa␣, and plasmin-mediated fragments formed in the presence of EDTA to bind plasminogen.
Effect of Plasmin on FXa␣ Enzymatic Activity-Having established that FXa can be cleaved by plasmin, we next determined the enzymatic activity of the plasmin-processed FXa. Autoproteolytic or plasmin digestion time courses that were identical to those shown in Fig. 3 were assayed for FXa activity. The data presented in Fig. 7A showed that the cleavage of    (panel B). The slower loss of clotting activity in the presence of EDTA correlated with the relative rates of 33-versus 40-kDa fragment production (Fig. 3).

DISCUSSION
The studies described in this paper are an extension of a previous report that identified FXa as a plasminogen receptor and accelerator of the tPA-mediated generation of plasmin (13). Evidence is presented in the current study revealing that the expression of a plasminogen binding site on FXa requires prior conversion of FXa␣ to the ␤-subform. This conclusion provides a possible explanation for our earlier observation that inhibition of the FXa cofactor effect on tPA by chloromethylketone (13) may have been due to blocking of the autoproteolytic production of FXa␤. To our knowledge, these are the first data that demonstrate a functional difference between FXa␣ and FXa␤ in a well defined system.
Since we found in the earlier study that plasmin pretreatment of FXa alters its tPA cofactor activity (13), we investigated the ability for plasmin to cleave FXa. The current work shows that FXa␣ and FXa␤ are previously unrecognized substrates for plasmin. A schematic summarizing the plasminmediated cleavage sites in FXa␣ is shown in Fig. 8. Under conditions that facilitate binding of FXa␣ to phospholipid vesicles, the fragmentation pattern closely resembled that of autoproteolysis, which included the sequential generation of species identical by reduced and nonreduced SDS-PAGE to FXa␤ and FXa␥ (FXa␤ Pn,Ca and FXa␥ Pn,Ca , respectively). A notable difference was that on a molar basis plasmin produced these species more than 3 orders of magnitude faster than the autoproteolytic process.
Combined with the conclusion that authentic FXa␤ but not FXa␣ binds plasminogen, the finding that plasmin rapidly generates a FXa␤-like molecule suggested a mechanism of regulation in which plasmin production could directly accelerate the conversion of FXa␣ into a plasminogen receptor. Indeed, similar to the autoproteolytic processing of FXa␣, proteolysis by plasmin was shown to liberate a plasminogen binding site upon production of FXa␤ Pn,Ca . Furthermore, plasmin was observed to expose a plasminogen binding site on FXa␥ Pn,Ca , which appeared to have an even higher affinity than the first cleavage product. Like authentic FXa␥ (10), the production of FXa␥ Pn,Ca involved cleavage at a site NH 2 -terminal to the active site Ser, which we identified as Gly 331 . Consequently, inhibition of amidolytic and clotting activity resulted due to disruption of the active site catalytic triad shown in Fig. 8 (H, D, and S).
When conditions were established to block the interaction between FXa␣ and phospholipid (e.g. chelation of Ca 2ϩ by EDTA), the plasmin-mediated fragmentation pattern was altered. The first cleavage product of FXa␣ by plasmin resulted in a species that resembled FXa␤ (FXa␤ Pn,EDTA ). Unlike the species generated by plasmin when FXa␣ was bound to PCPS, the final digestion product in the presence of EDTA did not resemble FXa␥. This species was 40 kDa as determined by nonreduced SDS-PAGE and consisted of the intact FXa␤-like heavy subunit disulfide linked to a 13-kDa fragment of the light chain (FXa␦ Pn,EDTA ). The single disulfide bond that bridges the subunits of FXa includes a Cys only eight amino acids from the COOH terminus of the light chain (38). Therefore, the 13-kDa subunit of FXa␦ Pn,EDTA must span a COOHterminal light chain region. To corroborate this, amino acid sequence analysis revealed a cleavage at Tyr 44 , which, as shown in Fig. 8, excises the ␥-carboxyglutamic acid-containing phospholipid binding domain. The production of FXa␦ Pn,EDTA did not effect amidolytic activity, because the catalytic domain is left intact by this cleavage pathway. However, clotting activity was inhibited and was likely due to the loss of phospholipid binding. The enzymatic and electrophoretic properties of FXa␦ Pn,EDTA are similar to those reported for the second autoproteolytic product that was previously identified in the absence of PS binding (12). Considerable evidence has accumulated that suggests a COOH-terminal Lys in a receptor may be the only structural characteristic required for plasminogen binding (14,15). Since the cDNA sequence for the FX and FXa␣ heavy subunit predicts the COOH terminus to be Lys 448 (32,33), it is interesting that association of plasminogen with either FX or FXa␣ was found to be insignificant compared with FXa␤. This observation can be explained by recent reports of constitutive (39) and inducible (40,41) carboxypeptidase B activity in plasma, which excises COOH-terminal Lys and consequently attenuates plasmin production. The mechanisms by which these carboxypeptidases are inhibited when clot lysis is desirable has been tied to the control of thrombin production by activated protein C (40). At least one of the plasma carboxypeptidases is activated (40,41) and subsequently inactivated (40) by thrombin. While direct COOH-terminal sequencing has not yet been conducted, the lack of plasminogen binding suggested that FX is a substrate for the carboxypeptidase(s). Evidence for this hypothesis was previously obtained by amino acid sequence analysis of the ␤-peptide excised from FX/FXa␣, which showed the predicted COOH-terminal Lys was missing (42). Thus, Fig. 8 shows processing of FX by a plasma carboxypeptidase, which would result in FXa␣ having Leu 447 as a COOH terminus.
It is well known that serine proteases (e.g. FXa and plasmin) preferentially cleave at basic amino acids. To explain our observation that several proteolytic derivatives of FXa␣ appear identical by SDS-PAGE but express different plasminogen binding activities, the FX sequence shows Lys and Arg situated several amino acids apart at plausible cleavage locations. By exposing COOH-terminal Lys, but not Arg, autoproteolysis and plasmin-mediated cleavage of FXa␣ may create new plasminogen binding sites. Therefore, authentic FXa␤ and FXa␤ Pn,Ca , which were observed by us to bind plasminogen, may have Lys 435 or Lys 433 as COOH termini. 2 Both cleavage sites have been implicated as new COOH termini due to autoproteolysis, with the latter being a minor product (42). To explain the lack of plasminogen binding to the FXa␤ Pn,EDTA electrophoretic doublet, it is proposed that Arg 429 may be the COOH terminus.
The exact autoproteolytic site(s) in the FXa heavy chain that produces FXa␥ has not been determined. However, a region corresponding to the ␣-thrombin autolysis loop has been identified in FX, which undergoes extensive cleavage (43) and would give rise to a species consistent with the electrophoretic properties of FXa␥. These cleavage sites could expose new COOH-terminal Lys at positions 330 and 338 or Arg at positions 326, 332, and 336 (43). By NH 2 -terminal sequence analysis of the fragment that is excised by plasmin, the exposure of COOH-terminal Lys 330 in FXa␥ Pn,Ca is suggested to account for the acquisition of plasminogen binding. A proportion of authentic FXa␥ having COOH-terminal Arg may explain why it appeared to associate with lower affinity to plasminogen than FXa␥ Pn,Ca . Based on NH 2 -terminal amino acid sequence and plasminogen binding data, the predicted COOH termini exposed on FXa␣ due to processing by plasmin are summarized in Fig. 8.
When FXa␣ was cleaved by plasmin under conditions that did not facilitate binding to PCPS, the exposure of binding sites for plasminogen was not observed by ligand blotting. This observation was inconsistent with the amino acid sequence data that predicted the creation of COOH-terminal Lys at position 43 upon production of FXa␦ Pn,EDTA , as shown in Fig. 8. One explanation for this discrepancy is that the light chain fragment containing the relevant COOH-terminal Lys 43 may be trapped poorly by PVDF and that consequently an association with plasminogen was not detectable by ligand blotting experiments.
In order to explain how phospholipid alters the observed plasmin-mediated cleavage pattern of FXa␣, two determining factors may be involved. The first is that the cleavage site in the Gla domain may become inaccessible to plasmin simply because it is masked when bound to PCPS. This cannot account for the absence of heavy chain cleavage when the FXa␣ is free in solution. Therefore, the association with PCPS may also promote a conformational change in FXa␣ that exposes the heavy chain cleavage site. Evidence for a conformational change in FXa due to PCPS binding has been previously reported (44) and may contribute to the prothrombinase cofactor effect of PCPS.
From previous reports a pathway of communication between coagulation and fibrinolysis has emerged that involves the modulation of prothrombinase by plasmin. In these studies plasmin was found to 1) inhibit the procoagulant activity of FVa (17,18), 2) convert the FVa into a plasminogen receptor and tPA accelerator (13), and 3) inhibit the attenuation of FXa as a tPA cofactor by native FVa (13). The current study adds to this model by providing evidence for a direct effect of plasmin on FXa function and a role for autoproteolytic conversion of FXa␣ to FXa␤. In summary, the new data show that neither FX nor FXa␣ can interact appreciably with plasminogen. Autoproteolysis results in the expression of a plasminogen binding site on FXa␤, presumably by creating a COOH-terminal Lys. This would facilitate the functioning of FXa␤ as a plasminogen receptor with consequential acceleration of plasmin generation (13). A feedback system is indicated when plasmin is present, which rapidly converts FXa␣ bound to phospholipid into a plasminogen receptor that resembles FXa␤ or FXa␥. In order for this pathway to be properly coordinated with sufficient fibrin production, it is assumed that autoproteolysis of FXa␣ relative to the generation of thrombin is slow. The data shown here also demonstrate that plasmin functions as an anticoagulant by proteolytic inhibition of FXa. Thus, plasmin is capable of directly modulating the activity of FXa from procoagulant to profibrinolytic.