Enhanced plasmin inhibition by a reactive center lysine mutant of the Kunitz-type protease inhibitor domain of the amyloid beta-protein precursor.

The Alzheimer's disease related protein, amyloid beta-protein precursor (A beta PP), contains a domain homologous to Kunitz-type serine protease inhibitors (KPI). The recombinant KPI domain of A beta PP is a potent inhibitor of coagulation factors XIa and IXa and functions as an anticoagulant in vitro. Here we report the expression, purification, and characterization of a reactive center lysine mutant of the KPI domain of A beta PP (KPI-Lys17). An expression plasmid for the KPI-Lys17 domain of A beta PP encoded amino acids 285-345 of the A beta PP cDNA containing a lysine substitution at arginine 17 in the KPI domain. The secreted 61-amino acid product was purified to homogeneity and functionally characterized. The protease inhibitory properties of the KPI-Lys17 domain were compared to those of the native KPI domain of A beta PP. Both KPI domains equally inhibited trypsin, chymotrypsin, and coagulation factors IXa and Xa. However, the KPI-Lys17 domain was an approximately 25-fold less effective inhibitor of coagulation factor XIa resulting in markedly less prolongation of the activated partial thromboplastin time compared to the native KPI domain of A beta PP. On the other hand, the KPI-Lys17 domain was an approximately 10- and 5-fold better inhibitor of plasmin in a chromogenic substrate assay and in a fibrinolytic assay, respectively, than the native KPI domain of A beta PP. Together, these studies suggest that the KPI-Lys17 domain has enhanced anti-fibrinolytic and diminished factor XIa inhibitory properties compared to the native KPI domain of A beta PP.

The amyloid ␤-protein precursor (A␤PP) 1 is the parent molecule to the Alzheimer's disease amyloid ␤-protein (1)(2)(3)(4). A␤PP can be translated from predominantly three alternatively spliced mRNA species to yield polypeptides of 695, 751, and 770 amino acids (5)(6)(7). The latter two species contain an additional insert which codes for a domain that is homologous to Kunitz-type serine protease inhibitors (KPI) (5)(6)(7). The KPI-containing isoforms of A␤PP are identical to the previously described cell-secreted protease inhibitor, protease nexin-2 (PN-2) (8,9). Messenger RNA encoding the KPI-lacking A␤PP 695 isoform is found primarily in brain. However, mRNA encoding the KPIcontaining A␤PP 751/770 isoforms are also abundant in brain and found in most peripheral tissues, suggesting that these isoforms of A␤PP may have a common function throughout the body.
Several studies have suggested a potential physiologic function for the KPI domain of PN-2/A␤PP in hemostasis. For example, measurements of protease inhibition equilibrium constants revealed that PN-2/A␤PP and its recombinant KPI domain are very potent inhibitors of intrinsic blood coagulation factors XIa and IXa (10 -14). In this regard it is noteworthy that PN-2/A␤PP is an abundant platelet ␣ granule protein and is secreted in high concentrations by platelets that are activated by physiologic agonists (11,(15)(16)(17). Together, these findings suggest that secreted platelet PN-2/A␤PP may play a role in regulating the intrinsic blood coagulation cascade at sites of vascular injury upon release by activated platelets (10,11,15,17). Fibrinolysis is another regulated proteolytic process that occurs at sites of vascular injury. Previous studies have shown that PN-2/A␤PP and its recombinant KPI domain also inhibit plasmin but to a lesser extent than coagulation factors XIa and IXa (10,(12)(13)(14).
Here we report the high level expression, purification, and biochemical characterization of a reactive center lysine mutant of the KPI domain of A␤PP (KPI-Lys 17 ). The protease inhibitory, anticoagulant and anti-fibrinolytic properties of the KPI-Lys 17 domain and the native KPI domain of A␤PP were compared. The results of these studies indicate that the KPI-Lys 17 domain has enhanced anti-plasmin and diminished factor XIa inhibitory properties compared to the native KPI domain of A␤PP. These properties suggest that KPI-Lys 17 may have utility as an anti-fibrinolytic agent.

EXPERIMENTAL PROCEDURES
Materials-The oligonucleotides used were synthesized by Genosys. Human factors XIa, IXa, and Xa were obtained from Enzyme Research Laboratories (South Bend, IN). Bovine trypsin, bovine chymotrypsin, fibrinogen, p-nitrophenyl guanidinobenzoate, tosyl-Gly-Pro-Arg-p-nitroanilide, N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide, and reagents for the activated partial thromboplastin time (APTT) were from Sigma. Pyro-Glu-Pro-Arg-p-nitroanilide was purchased from Helena Laboratories (Beaumont, TX). Human plasmin and carbobenzoxy-Val-Gly-Arg-4-nitroanilide were obtained from Boehringer Mannheim. Human thrombin was purchased from Calbiochem (San Diego, CA). Human urokinase was generously provided by Dr. J. Henkin of Abbott Labs (Abbott Park, IL). The recombinant native KPI domain of PN-2/A␤PP was prepared and purified as described (12). Construction of the KPI-Lys 17 Expression Vector-The construction of Pichia pastoris vector, pKPI200, encoding the wild type KPI sequence has been described (12). To replace the reactive center arginine at position 17 with lysine, site-specific mutations were introduced by a two-step PCR mutagenesis procedure (18). The first PCR utilized a 5Ј primer which includes the mutations (5Ј-CTGAGACTGGTCCATGTA-AGGCCATGATTTCTAGATGGTACT-3Ј). The mutation (underlined) not only replaces an arginine codon (AGA) with a lysine codon (AAG) but also incorporated a diagnostic HaeIII restriction site (GGCC). The 3Ј primer (5Ј-GATTAGAATCTAGCAAGACC-3Ј) in the reaction is located in the AOX1 terminator region and downstream of a unique AgeI site in the pKPI200 plasmid. The pKPI200 plasmid served as the template in the reaction. The second PCR used the same 3Ј primer as the first reaction and a 5Ј primer (5Ј-CATAATTGCGACTGGTTCC-3Ј) located in the AOX1 promoter region upstream from a unique HindIII site. An aliquot of the first PCR containing a small amount of the minor long extension product was the template for the second reaction (18).
The principal product of the second PCR, a Ϸ630-base pair fragment, was digested with AgeI and HindIII, gel purified, ligated into the large fragment of AgeI and HindIII digested pKPI200, and transformed into bacteria. Plasmids from individual bacterial transformants were tested initially for the presence of the diagnostic HaeIII restriction site and the DNA sequenced to confirm the mutations and the integrity of the rest of the KPI sequence. The resulting plasmid, designated pKPI-R17K ( Fig.  1), contains the promoter (5Ј AOX1) and terminator (3Ј AOX1) sequences of the P. pastoris methanol-inducible alcohol oxidase gene and the mutated KPI sequence in frame with the Saccharomyces cerevisiae ␣-mating factor prepro signal sequence. The transformation of the KPI-R17K plasmid into P. pastoris was accomplished as described previously (19). The plasmid was linearized at the unique StuI restriction site within the HIS4 gene prior to transformation into the HIS4-negative strain of P. pastoris, GS115. Successful integration of the plasmid was assessed by growth on minimal medium indicating the reconstitution of the HIS4-positive phenotype. Expression and purification of the KPI-Lys 17 domain were conducted as described previously for the KPI domain of PN-2/A␤PP (12). Amino acid sequence analysis of the purified protein confirmed the presence of the lysine at position 17.
Protease Inhibition Measurements-The active sites of trypsin, chymotrypsin, and plasmin were titrated by the method of Chase and Shaw (20) using the burst titrant p-nitrophenyl guanidinobenzoate. This trypsin was used to titrate the protease inhibitory activity of the purified KPI domain and KPI-Lys domain. Briefly, 10 nM trypsin was incubated with increasing concentrations of the purified KPI domains in 100 l of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM CaCl 2 , containing 0.1% bovine serum albumin in 96-well microtiter plates for 60 min at room temperature. The remaining trypsin activity was then measured by the addition of 50 l of 0.5 mM carbobenzoxy-Val-Gly-Arg-4-nitroanilide and by following the change in absorbance at 405 nm in a microtiter plate reader (Molecular Devices).
Inhibition equilibrium constants (K i ) for the purified KPI domains and factor XIa, trypsin, chymotrypsin, and plasmin were determined by the method of Bieth (21) as described previously (10,12). The K i values for the purified KPI domains and coagulation factor IXa were determined using a polylysine and polyethylene glycol based factor X activation assay (13,14,22,23). Factor Xa (1 nM) inhibitory activity was measured by incubation with the purified KPI domains (5-10 nM) in 0.1 M triethanolamine, pH 8.0, 0.1 M NaCl, containing 0.1% polyethylene glycol (M r ϭ 8000) and 100 g/ml bovine serum albumin in triplicate in 96-well microtiter plates. Incubations were conducted at 25°C for at least 40 min to allow for equilibrium between factor Xa and the KPI domains. At the end of the incubation 0.3 M tosyl-Gly-Pro-Arg-p-nitroanilide was added and hydrolysis proceeded for 60 min at 25°C. Graphical analysis yielded an apparent K i (K i,app ). Since the inhibition is reversible a correction must be made for the K m of the protease-substrate reaction. The K m for each protease and its corresponding chromogenic substrate was determined independently and the true K i was calculated using the following equation: APTT Assay-Microtiter plate activated partial thromboplastin time (APTT) coagulation assays were conducted as described previously (24). Briefly, 30 l of pooled normal, citrated-plasma, 30 l of APTT reagent, and 30 l of Tris-buffered saline were incubated in triplicate microtiter plate wells in the absence or presence of increasing concentrations of the purified KPI domains for 10 min at 22°C. The APTT assay was initiated by the addition of 30 l of 25 mM CaCl 2 and the time to clot formation was monitored by the change in absorbance at 405 nm in a V max kinetic microtiter plate reader (Molecular Devices). The absorbance readings were recorded every 5 s for 3-5 min.
Fibrinolysis Assay-Fibrin microtiter plates were prepared by incubating 5.4 M fibrinogen and 1.5 nM thrombin in a final volume of 100 l of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM CaCl 2 , for 30 min at room temperature (25). Triplicate fibrin wells were prepared in the absence or presence of increasing concentrations of the KPI-Lys 17 domain or the native KPI domain of A␤PP. Control experiments showed that the concentrations of the KPI-Lys 17 domain or the native KPI domain of A␤PP used in these studies had no effect on the ability of thrombin to generate fibrin or cleave its chromogenic substrate. Five l of 0.6 M plasmin were then added to the fibrin wells and incubated shaking for 16 h at 37°C in a humidified environment. The extent of fibrinolysis was determined as a loss in absorbance at 405 nm in a V max microtiter plate reader.

RESULTS AND DISCUSSION
A DNA encoding the mutant KPI-Lys 17 domain was prepared by site-directed mutagenesis and using PCR and the plasmid KPI200 that encodes the native KPI domain of A␤PP (Fig. 1). DNA encoding the yeast-specific secretion signal of S. cerevisiae ␣-mating factor prepro signal sequence was attached to the DNA encoding the KPI-Lys 17 domain and incorporated into a P. pastoris expression vector to generate pKPI-R17K. The recombinant plasmid pKPI-R17K ( Fig. 1) was integrated into and expressed in the methylotrophic yeast P. pastoris. Transcription from the P. pastoris AOX1 promoter included in pKPI-R17K occurs at very high levels in P. pastoris cells grown in methanol providing an inducible expression system particularly designed for foreign gene expression (26). Transformed P. pastoris cells were first grown in a fermenter in the presence of glycerol and then switched to methanol as the carbon source to induce expression of the KPI-Lys 17 domain. The secreted KPI-Lys 17 domain (Ͼ1.0 g/liter) comprised Ͼ80% of the protein in the fermentation medium.
The expressed KPI-Lys 17 domain was purified from the fermentation medium by the same procedures previously described for purification of the native KPI domain of A␤PP (12). As shown in Fig. 2, the purified KPI-Lys 17 domain and native KPI domain of A␤PP had molecular masses of Ϸ6.5 kDa. Amino acid sequencing of the purified KPI-Lys 17 domain revealed the integrity of the expressed product and showed that the amino terminus of the protein was properly processed upon secretion (data not shown). Titration experiments with the purified KPI-Lys 17 domain revealed a 1:1 stoichiometry for trypsin inhibition, essentially the same for parallel titration experiments with the purified native KPI domain of A␤PP (Fig. 3). Importantly, this result demonstrated that the KPI-Lys 17 domain, as the native KPI domain of A␤PP, was fully active, properly folded, and the disulfide bonds were in the correct orientation when expressed and secreted by the Pichia cells. This assessment was further supported by the finding that treatment of either purified KPI domain with the reducing agent dithiothreitol completely abolished inhibitory activity toward the target proteases (data not shown).
The protease inhibitory properties of the purified KPI-Lys 17 domain were compared to those of purified native KPI domain of A␤PP (Table I). The protease inhibition equilibrium constants obtained for the inhibition of trypsin, chymotrypsin, and coagulation factors IXa and Xa by the purified KPI-Lys 17 domain were essentially the same as those obtained for the inhibition of these proteases by the purified native KPI domain of A␤PP (Table I). However, the KPI-Lys 17 domain was an Ϸ25fold less effective inhibitor of coagulation factor XIa than the native KPI domain of A␤PP. This finding indicates that the arginine residue at the reactive center of the KPI domain of A␤PP is important for optimal inhibition of factor XIa. On the other hand, the KPI-Lys 17 domain was found to be an Ϸ10-fold better inhibitor of plasmin compared to the native KPI domain of A␤PP (Table I). Neither KPI domain inhibited urokinase.
Since the KPI-Lys 17 domain inhibited factors XIa, IXa, and Xa, we performed studies to determine if it inhibited the coagulant activity of normal human plasma (Fig. 4). The native KPI domain of A␤PP at 1 M prolonged the APTT Ͼ2-fold over pooled normal plasma as previously reported (13). In contrast, only at concentrations Ϸ50 M or more was the APTT prolongation Ͼ2-fold by the KPI-Lys 17 domain. At 50 M, the native KPI domain of A␤PP prolonged the APTT Ϸ15-fold. These data suggest that the KPI-Lys 17 domain has diminished anticoagu-lant properties in in vitro clotting assays compared to the native KPI domain of A␤PP. Since clotting in the APTT assay involves the activity of numerous coagulation factors in the intrinsic pathway, the weaker anticoagulant activity of the KPI-Lys 17 domain observed is consistent with the above finding that it is a less effective inhibitor of coagulation factor XIa than the native KPI domain of A␤PP (Table I).
The KPI-Lys 17 domain was found to have enhanced inhibi- tion of plasmin in a chromogenic substrate assay compared to the native KPI domain of A␤PP (Table I). Therefore, we conducted studies to compare the abilities of the KPI-Lys 17 domain and the native KPI domain of A␤PP to inhibit plasmin degradation of its natural substrate, fibrin. The KPI-Lys 17 domain inhibited plasmin in the in vitro fibrinolytic assay starting at concentrations Ͼ0.1 M and with an IC 50 of Ϸ1 M (Fig. 5). In contrast, the native KPI domain of A␤PP inhibited plasmin in the same assay starting at concentrations Ͼ1 M and with an IC 50 of Ϸ5 M, consistent with its less effective inhibition of plasmin in the chromogenic substrate assay (Table I).
Together, these studies indicate that the KPI-Lys 17 domain possesses enhanced antiplasmin and diminished factor XIa inhibitory and anticoagulant properties compared to the native KPI domain of A␤PP. Although the present studies show that the native KPI domain of A␤PP is an inhibitor of plasmin and fibrinolysis in in vitro assays, its additional anticoagulant properties may interfere in the normal genesis of fibrin clots. On the other hand, KPI-Lys 17 domain exhibits increased anti-fibrinolytic activity yet it is a much less effective anticoagulant in in vitro assays. Since the inhibition of factors IXa and Xa by the KPI-Lys 17 domain is similar to that of the native KPI domain, the Arg 3 Lys mutation of KPI-Lys 17 may result in an inhibitor which overall is a better anti-fibrinolytic agent. This hypothesis needs to be examined in in vivo models. An anti-fibrinolytic inhibitor may be useful in cardiopulmonary bypass as an alternative to aprotinin administration. Not inhibited Not inhibited a The protease inhibition equilibrium constants (K i ) were determined as described under "Experimental Procedures." The data represent the mean Ϯ S.D. of Ն5 separate determinations. Significant differences from KPI-Lys 17 and KPI; b p Ͻ 0.0004; c p Ͻ 0.02 by unpaired group t test.