A novel plasma proteinase potentiates α₂-antiplasmin inhibition of fibrin digestion

Human α₂-antiplasmin (α₂AP), also known as α₂-plasmin inhibitor, is the main inhibitor of plasmin.¹  Plasmin plays a critical role in fibrin proteolysis and tissue remodeling. The physiologic relevance of plasmin inhibition by α₂AP to blood clotting and fibrinolytic homeostasis is supported by the following observations: (1) the rate of free plasmin inactivation by circulating α₂AP is much faster than fibrin(ogen) digestion by plasmin,²  thereby eliminating the possibility of a systemic lytic state and consequent bleeding; (2) α₂AP is cross-linked to forming fibrin by activated blood clotting factor XIII (FXIIIa) and inhibits plasmin-mediated lysis in direct proportion to the amount incorporated^3-5 ; and (3) patients with homozygous α₂AP deficiency manifest serious hemorrhagic tendencies, while heterozygotes tend to bleed only after major trauma or surgery.⁶  Human α₂AP is synthesized primarily in the liver, and during circulation in plasma, the secreted precursive Met-α₂AP form, a 464-residue protein having Met as the N-terminus, undergoes proteolytic cleavage between Pro12 and Asn13 to yield Asn-α₂AP, a 452-residue version with Asn as the N-terminus.⁷  Met-α₂AP accounts for approximately 30% of circulating α₂AP, and Asn-α₂AP, approximately 70%.^7,8  While 3-fold more Asn-α₂AP than recombinant Met-α₂AP was shown to cross-link to fibrin,⁹  no data have been reported for native circulating Met-α₂AP. Moreover, the effect of different ratios of the 2 α₂AP forms on clot lysis has not been reported. Finally, the enzyme responsible for converting Met-α₂AP to Asn-α₂AP has not been identified, although it is presumed to be in plasma.⁷ 

We report here the purification and partial characterization of antiplasmin-cleaving enzyme (APCE), a serine proteinase from human plasma that cleaves the Pro12-Asn13 bond of Met-α₂AP to yield Asn-α₂AP, and the comparison of these 2 circulating forms of α₂AP for the ability to cross-link with fibrin and to inhibit fibrinolysis.

A mixture of Met-α₂AP and Asn-α₂AP was isolated using a modification of a published procedure.¹⁰  Briefly, human plasma was precipitated using polyethylene glycol 6000 (final concentration 5%), and the supernatant was applied to a lysine–Sepharose 4B (Amersham Biosciences, Piscataway, NJ) affinity column to remove plasminogen. The flow-through volume was applied to an affinity column consisting of plasminogen kringles 1 to 3 bound to Sepharose 4B. After washing with 40 mM sodium phosphate–0.5 M NaCl, pH 7.0, α₂AP was eluted from this column using a 0- to 50-mM linear gradient of ϵ-aminocaproic acid in the same buffer. To separate Met-α₂AP and Asn-α₂AP, a goat antibody to the N-terminal 12–amino acid sequence unique to Met-α₂AP was prepared, using as the immunogen a multiple antigenic peptide (MAP) that contained 8 copies of the N-terminal peptide linked via their C-termini to a core peptide of 7 lysines.¹¹  The α₂AP mixture was applied to an immunoaffinity column made by conjugating the purified goat antibody to Affigel-10 (Bio-Rad, Richmond, CA). After washing with 40 mM Tris (tris(hydroxymethyl)aminomethane) · HCl–0.5 M NaCl, pH 7.5, Met-α₂AP was eluted with 0.1 M glycine, pH 2.5. Fractions were immediately neutralized by collection into tubes containing 1.0 M Tris · HCl, pH 9.0. Asn-α₂AP eluted in the flow-through fractions from the immunoaffinity column. N-terminal amino acid sequence analyses confirmed that unbound and bound fractions were Asn-α₂AP and Met-α₂AP, respectively. FXIII was purified from fresh human plasma according to published methods.¹² 

Factor XIIIa–catalyzed cross-linking reactions were performed, and free α₂AP and α₂AP-fibrin cross-linked complexes were detected by immunostaining with an antibody to α₂AP as previously described.¹³  The rate of formation of α₂AP-fibrin cross-linked complexes (Figure 1A-B) was determined by measuring the appearance of bands of α₂AP-fibrin complexes at approximately 140 kDa and approximately 210 kDa and the concomitant disappearance of the 70-kDa α₂AP band from gel scans using a Personal Densitometer SI (Molecular Dynamics, Piscataway, NJ) and image analysis software (ImageQuant, version 5.2; Molecular Dynamics, Piscataway, NJ). The percentage of α₂AP-fibrin cross-linked complexes was expressed as [(densitometric values for the ∼ 140 kDa and ∼ 210 kDa bands)/(densitometric values for the ∼ 70 kDa, ∼ 140 kDa, and ∼ 210 kDa bands)] × 100 (Figure 1B).

After adding Met-α₂AP, Asn-α₂AP, or selected mixtures of the 2 forms (0.5 μM) to α₂AP-depleted plasma (Enzyme Research, South Bend, IN), a mixture of 1 U/mL thrombin, 16 mM CaCl₂, and 45 IU/mL urokinase (uPA) (Abbott, Chicago, IL) was added to catalyze almost instant fibrin clot formation and to initiate fibrinolysis. The rate of plasma clot lysis was determined by a turbidimetric microtiter plate method.^14-16 

APCE was isolated from 2.0 L of human citrated plasma (purchased from Sylvan Goldman Blood Institute, Oklahoma City, OK) by sequential chromatographic steps that followed ammonium sulfate fractionation (15%-40%). The pellet was dissolved in 25 mM sodium phosphate–1 mM EDTA (ethylenediaminetetraacetic acid) buffer, pH 7.5, containing 10% ammonium sulfate and applied to a phenyl-Toyopearl hydrophobic interaction chromatography column (Tosohaas, Montgomeryville, PA). After washing with 10% ammonium sulfate–25 mM sodium phosphate–1 mM EDTA, pH 7.5, bound proteins were eluted by a decreasing ammonium sulfate gradient (linear, 5%-0%) in the same buffer. To monitor APCE activity in chromatographic fractions, hydrolysis of the fluorogenic substrate Z-Gly-Pro-AMC was measured at excitation/emission wavelengths of 360/460 nm. Active fractions were pooled and further purified by elution from a Q-Sepharose anion exchange column (Amersham Biosciences) using a 0- to 0.5-M NaCl linear gradient in the buffer described above. The activity peak was applied to a T-gel thiophilic column (Pierce, Rockford, IL) equilibrated in 0.5 M ammonium sulfate in the same phosphate buffer, and APCE activity was eluted by a decreasing ammonium sulfate gradient (linear, 0.5-0 M). N-terminal sequence of the dominant protein in the active fractions, as judged by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), was determined by Edman sequence analysis. Based upon the N-terminal sequence (IVLRPSRVHNSEENT) obtained, a multiple antigenic peptide (MAP) was prepared for immunizing a goat. The linear form of the peptide was used to make an affinity column from which the antibody was purified from goat serum. An immunoaffinity column made with this antibody was used to purify the cleaving activity of Z-Gly-Pro-AMC from the T-gel column. Following application and washing, the immunoaffinity column was eluted with 3 M sodium thiocyanate, pH 7.5. The active fractions were pooled, dialyzed against 25 mM sodium phosphate–1 mM EDTA–20% glycerol, pH 7.5, and stored at –80° C. On the basis of the Z-Gly-Pro-AMC assay, selected fractions were pooled from each chromatographic step, and each pooled sample was also assayed for Met-α₂AP cleaving activity (Figure 4A). Met-α₂AP (5 μg) and each chromatographic sample (54 μg phenyl-Toyopearl chromatography fraction, 9 μg Q-Sepharose chromatography fraction, 0.6 μg T-gel thiophilic chromatography fraction, or 0.2 μg immunoaffinity chromatography fraction) was incubated at 22° C. At selected times, an aliquot of each reaction mixture was subjected to SDS-PAGE and transferred to a nitrocellulose membrane. Met-α₂AP was treated with antibody specific for its N-terminal sequence and then detected by the horseradish-peroxidase system. Met-α₂AP and Asn-α₂AP were detected by a monoclonal antibody (American Diagnostica, Greenwich, CT), which was reactive to the C-terminal region of either form of α₂AP (Figure 4B, right panel). Also, Met-α₂AP (5 μg) was incubated with selected amounts of APCE (0.02-0.16 μg) for 6 hours, and Western blot analysis was performed using Met-α₂AP–specific antibody (Figure 4C).

To assess APCE activity in human plasma and its recovery in each purification step, we designed and synthesized a fluorescence resonance energy transfer (FRET) peptide substrate that contained the APCE-sensitive Pro12-Asn13 bond within the Thr9-Gln16 sequence of Met-α₂AP as follows: Arg-Lys(DABCYL)-Thr-Ser-Gly-Pro-Asn-Gln-Glu-Gln-Glu(EDANS)-Arg. Hydrolysis of the Pro-Asn bond separates the fluorophore, EDANS (5-[(2-aminoethyl)amino]-naphthalene-1-sulfonic acid), from the quenching group, DABCYL (4-(4-dimethylaminophenylazo)benzoyl), to give an increase in fluorescence. APCE cleavage of the Pro-Asn bond in the FRET peptide substrate was confirmed by analysis of products by liquid chromatography–mass spectrometry (LC/MS; data not shown). APCE was selectively removed from normal human plasma by an antibody specific for the N-terminal region of APCE (see above for purified goat antibody as made for APCE purification). A nonspecific goat antibody was used as a control. Each antibody was coupled to POROS EP20 beads (Applied Biosystems, Framingham, MA), and the coupled beads were incubated separately with plasma diluted 1:5 with 25 mM sodium phosphate–1 mM EDTA, pH 7.5, on a rocker for 15 hours at 4° C. Any APCE activity remaining in the supernatant, and therefore not bound to the antibody-POROS beads, was measured as a function of its ability to cleave the APCE-specific FRET peptide (0.1 mM, 10 μL), which was added to 200 mL of the supernatant and incubated at 22° C. Fluorescence was monitored with time at excitation and emission wavelengths of 360 and 460 nm, using a BIO-TEK FL600 fluorescence plate reader (Winooski, VT).

We determined the N-terminal sequence of purified α₂AP from 52 normal plasma samples and found that Met-α₂AP is present as 33% ± 10% (mean ± SD) of the total α₂AP, which is in accord with previous values from 4 pools of human plasma.⁸  We isolated native forms of both Met-α₂AP and Asn-α₂AP from human plasma and studied these instead of recombinant forms.⁹  To compare cross-linking rates of Met-α₂AP versus Asn-α₂AP to fibrin, mixtures of fibrinogen, Ca²⁺, FXIII, and Met-α₂AP or Asn-α₂AP were clotted by thrombin and allowed to cross-link for selected times; then solubilized in urea–sodium dodecyl sulfate (SDS)–dithiothreitol (DTT); subjected to polyacrylamide gel electrophoresis (PAGE); analyzed by Western blots; and quantified by densitometry. The content of α₂AP immunoreactivity in high-molecular-weight cross-linked complexes of α₂AP-fibrin increased with time, and simultaneously immunoreactivity decreased in the lower molecular-weight α₂AP band (Figure 1A). Using essentially physiologic concentrations, we found that the extent of purified native Asn-α₂AP cross-linking to fibrin was approximately 13 times greater than purified native Met-α₂AP during the initial 10 minutes as fibrin gel formation achieved completion (Figure 1B). Our finding that native Asn-α₂AP is far more rapidly incorporated into fibrin than its native precursive form, Met-α₂AP, clarifies and extends a prior observation that recombinant Met-α₂AP cross-linked to fibrin at one-third of the amount of purified plasma Asn-α₂AP.⁹  Holmes et al¹⁷  reported a similar finding in that recombinant Asn-α₂AP having an extension of 3 additional N-terminal amino acids could no longer be cross-linked to fibrin. Our results can be explained by comparing the 2 purified plasma forms, rather than plasma α₂AP with recombinant Met-α₂AP as in the previous report. Recombinant Met-α₂AP has been shown to have a bit slower initial rate of free plasmin inhibition than plasma α₂AP,⁹  raising the question of whether slight structural differences might exist between circulating α₂AP and recombinant Met-α₂AP that could also impact cross-linking. We could detect no differences between the 2 purified circulating forms in this regard (Figure 1C). Important to note is that the purified plasma α₂AP used by Sumi et al⁹  was not characterized for content of the 2 α₂AP forms, and no separation method was described. Hence, if their plasma α₂AP preparation were in fact a mixture of Met-α₂AP and Asn-α₂AP, this could account for the much narrower difference between the 2 forms for becoming cross-linked to fibrin than we report here.⁹  Finally, the molar ratio of fibrinogen to α₂AP used by Sumi et al⁹  was around 70:1, which is significantly greater than the physiologic ratio of 7:1 used in our work, and for that matter, even higher than ratios in pathologic settings where fibrinogen concentration is markedly elevated.

It has not been established whether Asn-α₂AP protects fibrin from fibrinolysis more effectively than Met-α₂AP. Therefore we compared the 2 forms by urokinase (uPA)–induced plasma clot lysis. After addition of selected ratios of purified Met-α₂AP and/or Asn-α₂AP to give a total α₂AP concentration of 0.50 μM in α₂AP-depleted plasma, a mixture of thrombin, Ca²⁺, and uPA was added to catalyze cross-linked fibrin formation and to initiate fibrinolysis. As shown in Figure 2, turbidity was monitored at 405-nm wavelength, and plasma clot lysis times (PCLTs) were defined as the midpoint between the highest and lowest absorbances. A control clot prepared from α₂AP-depleted plasma was rapidly digested by plasmin (PCLT = 20 minutes). When only Met-α₂AP was added, clot lysis was slightly prolonged (PCLT = 23 minutes). As the proportion of Asn-α₂AP increased, clot lysis was progressively delayed. When 100% Asn-α₂AP was added, clot lysis was maximally delayed (PCLT = 42 minutes), becoming about 2-fold longer than when only Met-α₂AP was present. These results indicated a dose-response effect for ratios of the 2 forms of circulating α₂AP on fibrinolysis. It should be noted that normal human plasma that had been depleted of α₂AP using immobilized polyclonal antibodies to α₂AP was used in these studies, thereby negating any competitive impact that C-terminal shortened, non–plasminogen-binding forms of α₂AP may exhibit for becoming cross-linked to fibrin.^18,19  The Met-α₂AP and Asn-α₂AP added to α₂AP-depleted plasma were purified using their C-terminal affinities for immobilized plasminogen kringles 1-3; hence, C-terminal degraded forms of α₂AP would not bind. These studies demonstrate clearly that Asn-α₂AP inhibits plasmin-mediated fibrin digestion more effectively than Met-α₂AP.

Given the greater efficiency of Asn-α₂AP incorporation into fibrin, we reasoned that the unidentified plasma proteinase responsible for converting Met-α₂AP to Asn-α₂AP might be important in modulating the susceptibility of fibrin to plasmin; we therefore attempted to isolate the proteinase. The proteinase was purified from human plasma using an initial ammonium sulfate fractionation (A), followed by phenyl-Toyopearl hydrophobic interaction (P), Q-Sepharose anion exchange (Q), T-gel thiophilic (T), and immunoaffinity (I) chromatographies. Through the successive purification steps prior to the immunoaffinity chromatography, the enzyme activity increased concomitantly with the increase of the dominant 97-kDa protein band as detected by SDS-PAGE (Figure 3). This band was subjected to protein sequencing following transfer to a polyvinylidene difluoride membrane, and the identification of the first 15 amino acids (IVLRPSRVHNSEENT) allowed the synthesis of an identical peptide, which was used as an immunogen to prepare a goat antibody. After purification, the antibody was linked to a gel support for immunoaffinity chromatography (I). Throughout the purification processes, proteolytic activity was monitored by hydrolysis of either Z-Gly-Pro-AMC²⁰  or the FRET peptide. Active fractions in each purification step were pooled and analyzed by SDS-PAGE (Figure 3). Cleavage of Met-α₂AP to Asn-α₂AP was assessed by immunoblot analyses; the rate of Met-α₂AP conversion by the proteinase clearly increased as its purification progressed (Figure 4A). We termed this proteinase “antiplasmin-cleaving enzyme” (APCE). Table 1 summarizes a typical purification of APCE from human plasma.

Recently it was reported that phenyl hydrophobic interaction chromatography could be used to separate 2 distinct Z-Gly-Pro-AMC–hydrolyzing activities from bovine serum: one bound to the phenyl column, while the other did not.²¹  We obtained a similar result when ammonium sulfate–fractionated human plasma was subjected to phenyl-column chromatography, with each fraction being assayed using Z-Gly-Pro-AMC. Importantly, however, only the phenyl-bound fraction possessed proteinase activity that cleaved the FRET peptide substrate and Met-α₂AP (Figure 4A, P). Table 1 shows a dramatic increase in purification with a high yield of activity for this phenyl-Toyopearl step. To determine APCE activity in human plasma, commercially available Z-Gly-Pro-AMC cannot be used, since more than 1 Pro-Xaa cleaving enzyme is present in human plasma. Therefore, taking advantage of the FRET peptide as a specific substrate, total APCE activity was assessed in 3 different pooled normal plasma samples, using the total activity in each plasma sample and specific activity of purified APCE from each to estimate the concentration of APCE in each plasma sample as 560, 561, and 493 μg/L.

When Met-α₂AP was incubated with APCE activity from the final immunoaffinity step (Figure 4B), it was completely and specifically cleaved at the Pro12-Asn13 bond as shown by the following: (1) The 70-kDa Met-α₂AP band, which was easily demonstrable at 0 time by antibody specific for Met-α₂AP, had completely disappeared by 6 hours (Figure 4B, left panel). When immunostained with an antibody to both α₂AP forms, however, the intensities of the 70-kDa band at 0 hours and 6 hours did not change (Figure 4B, right panel), but the Met-α₂AP band disappeared in direct proportion to the concentration of APCE when analyzed with an antibody that detected only Met-α₂AP (Figure 4C). (2) N-terminal analyses of the 0-hour sample (Figure 4B) showed only the Met-α₂AP sequence, MEPLGRQLTSGP, while the 6-hour sample had only the Asn-α₂AP N-terminal sequence, NQEQVSPLTLLK. The proteinase isolated by the last purification step showed a single 97-kDa band (Figure 3, I), with an N-terminal sequence identical to that of the major protein band in sample T. Interestingly, the sequence matched that deduced from the cDNA sequence for Ile24-Thr38 of fibroblast activation protein (FAP),²²  a proline-specific serine proteinase also termed seprase by another group.²³ 

To determine additional sequence information from the protein, the 97-kDa protein band from the SDS-PAGE gel was reduced, alkylated, and digested with trypsin.²⁴  The tryptic digest was subjected to liquid chromatography–mass spectrometry–mass spectrometry (LC/MS/MS) to obtain molecular weights and MS/MS fragment ion spectra and to use the MASCOT MS/MS ion search engine to query the OWL nonidentical protein database (http://www.matrixscience.com/cgi/search_form.pl?FORMVER=2&SEARCH=MIS). From our search results, 5 peptides had sequences identical with FAP/seprase^22,23 : residues 210-219 (YALWWSPNGK); residues 247-254 (TINIPYPK); residues 487-499 (ILEENKELENALK); residues 500-509 (NIQLPKEEIK); and residues 522-530 (MILPPQFDR). This degree of homology suggested that the 97-kDa APCE proteinase had similarity to FAP/seprase; however, significant differences remained. FAP/seprase is reported to be expressed on the cell surface of reactive stromal fibroblasts of healing wounds and epithelial cancers, but not on normal tissue or benign epithelial tumors.^25,26  It is believed to be particularly associated with an invasive phenotype of human melanoma and carcinoma cells.²³  Despite the plausibility that FAP/seprase may have a role in tissue remodeling, a physiologic substrate has not been demonstrated for the native enzyme, and hence a clearly defined biologic function remains obscure. Although gelatin and collagen I appeared to serve as substrates for recombinant FAP expressed on a human embryonic kidney epithelial cell line (293 cells), products of fragmentation or actual cleavage sites have not been identified.²⁷  Importantly, a previous report indicated that neither soluble FAP nor its cleavage products exist in the circulation.²⁸  The APCE we describe here was isolated from normal human plasma.

The possibility that human plasma APCE could be proteolytically derived from cell membrane–bound FAP/seprase is raised by the following observations: (1) The N-terminal 15 amino acids of APCE are identical to residues 24 to 38 of FAP/seprase, the latter estimated to have its first 6 N-terminal amino acids in the fibroblast cytoplasm, with another predicted 20-residue transmembrane domain, and then a 734-residue extracellular C-terminal catalytic domain.^22,23  (2) Internal sequences of APCE obtained from 5 tryptic peptides are identical to those deduced from the cDNA sequence of a fairly long segment of FAP/seprase (amino acids 210 through 530). (3) Nondenaturing gel filtration chromatography of APCE indicates a native mass of approximately 180 kDa, and, whether or not reduced, its denatured molecular weight by SDS-PAGE is 97 kDa (Figure 3, I). Both values are consistent with the reported dimeric structure of 190 kDa for FAP/seprase.²⁷  (4) APCE is inhibited by diisopropyl fluorophosphate (DFP), phenylmethyl-sulfonyl fluoride (PMSF), and AEBSF (4-(2-aminoethyl)-benzene-sulfonyl fluoride), but not by iodoacetate, E-64, or EDTA, just as reported for FAP/seprase.²⁷  (5) Analogously, a closely related type II transmembrane serine proteinase with prolyl specificity apparently undergoes proteolytic cleavage of its N-terminal 38-residue cytoplasmic/transmembrane domain to give rise to the circulating human dipeptidyl peptidase IV (DPP IV, aka T-cell activation CD26, with 50% homology to FAP/seprase)²⁹ ; however, DPP IV is strictly a dipeptidyl aminopeptidase,²⁰  and we found that it does not cleave Pro12-Asn13 of Met-α₂AP as does APCE. If APCE indeed results from proteolysis of the transmembrane domain of FAP/seprase, a proteinase that might cleave the Cys23-Ile24 bond has yet to be identified. The hydrophobic residue Ile24 as the P1′ position of the bond could make this a preferred cleavage site for an extracellular enzyme such as a matrix metalloproteinase.³⁰ 

Alternatively, APCE may be a product of intracellular processing. Conceivably, 2 pathways could exist in the same cell: one that normally synthesizes, processes, and secretes APCE without a transmembrane segment; and another that produces a membrane-bound enzyme (FAP/seprase) when fibroblasts become activated and proliferate in inflammatory states, embryogenesis, or metastases of certain epithelial-derived malignancies.^25,26  Finally, the possibility remains that APCE is a product of a different gene or that it results from alternative splicing. Our data indicate that APCE is produced under normal physiologic states, and unlike the case of FAP/seprase, pathologic disorders are not required to stimulate gene expression. The fact that APCE is found in normal plasma and cleaves Met-α₂AP suggests that one of its physiologic functions is the regulation of Asn-α₂AP availability for plasmin inhibition within cross-linked fibrin.

To determine if whole-plasma APCE is specifically responsible for cleaving the Met-α₂AP sequence that generates Asn-α₂AP, antibody to the N-terminal 15 residues of APCE was coupled to beads and compared with a control antibody for the ability to remove significant amounts of APCE activity from human plasma as described in “Materials and methods.” Control antibody did not remove FRET peptide cleaving activity, whereas antibody specific for APCE removed approximately 50%. Since the FRET peptide substrate is specific for APCE activity, the remaining FRET peptide cleaving activity in this experiment may be due to N-terminally degraded forms of APCE that react poorly or not at all with the antibody, particularly given that only the linear epitopes among the first 15 residues of APCE would be recognized. Although it must still be considered that human plasma may contain other proteinases that cleave the N-terminal 12-residue peptide of Met-α₂AP, our data indicate that this is unlikely because (1) only the 15% to 40% ammonium sulfate fraction was found to contain Met-α₂AP cleaving activity; (2) when ammonium sulfate–fractionated human plasma was subjected to phenyl-column chromatography, phenyl-bound and unbound cleaving activities toward Z-Gly-Pro-AMC were indeed observed, but only the bound activity cleaved the FRET peptide and Met-α₂AP; (3) only one peak of proteolytic activity for the FRET peptide and Met-α₂AP was observed in elution profiles from Q-Sepharose and T-gel chromatographies (Figure 4A); and (4) on reduced SDS-PAGE, the peak from the immunoaffinity chromatography contained only one band and it had a single N-terminal sequence (Figure 3, I).

We conclude that human Met-α₂AP is a physiologically important substrate of APCE, since proteolytic cleavage between Pro12-Asn13 yields the derivative Asn-α₂AP, which becomes cross-linked significantly faster to fibrin by FXIIIa, and, as a consequence, enhances the resistance of fibrin to digestion by plasmin. Since human APCE augments the inhibition of fibrinolysis by increasing the availability of the faster cross-linking form (ie, Asn-α₂AP), potent and selective inhibitors of APCE might allow titration of Asn-α₂AP production to lower in vivo levels, and thereby enhance both endogenous and exogenous fibrin removal. Bleeding complications would be unlikely, based on the observation that persons heterozygous for α₂AP deficiency have minimal hemorrhagic risk.⁶  Hence, a window of safety may exist for lowering α₂AP function in healthy persons. Particularly in clinical situations where fibrin formation is likely, the development of an agent that inhibits APCE might result in decreased amount of Asn-α₂AP available for cross-linking to fibrin as thrombi develop or inflammation progresses. Then endogenous levels of generated plasmin, or plasmin produced by administering small amounts of a plasminogen activator, might be sufficient to effect fibrin removal so that vessel patency and organ function are maintained and bleeding risk is minimized. We believe that this hitherto unknown circulating enzyme, APCE, may provide a valuable target for pharmacologic modulation of the fibrinolytic system.

We thank C. S. Lee and J. G. Chun for technical assistance.