Identification of amino acids in antiplasmin involved in its noncovalent ‘lysine-binding-site’-dependent interaction with plasmin Haiyao Wang, Anna Yu, Bjo¨ rn Wiman and Sarolta Pap Department of Clinical Chemistry and Blood Coagulation, Karolinska Hospital, Karolinska Institute, Stockholm, Sweden The lysine-binding-site-mediated interaction between plas- min and antiplasmin is of great importance for the fast rate of this reaction. It also plays an important part in regulat- ing the fibrinolytic enzyme system. To identify structures important for its noncovalent interaction with plasmin, we constructed seven single-site mutants of antiplasmin by modifying charged amino acids in the C-terminal part of the molecule. All the variants were expressed in the Drosophila S2 cell system, purified, and shown to form stable complexes with plasmin. A kinetic evaluation revealed that two mutants of the C-terminal lysine (K452E or K452T) did not dif- fer significantly from wild-type antiplasmin in their reac- tions with plasmin, in either the presence or absence of 6-aminohexanoic acid, suggesting that this C-terminal lysine is not important for this reaction. On the other hand, modification of Lys436 to Glu decreased the reaction rate about fivefold compared with wild-type. In addition, in the presence of 6-aminohexanoic acid, only a small decrease in the reaction rate was observed, suggesting that Lys436 is important for the lysine-binding-site-mediated interaction between plasmin and antiplasmin. Results from computer- ized molecular modelling of the C-terminal 40 amino acids support our experimental data. Keywords: antiplasmin; fibrinolysis; lysine-binding site; muta- genesis; plasmin. The reaction between plasmin and its natural inhibitor in blood plasma, antiplasmin, normally occurs in several sequential steps [1–4]. The first step, which is rate limiting, takes place between one of the so called ‘lysine-binding sites’ in the plasmin molecule and a complementary site in the antiplasmin molecule. The second step is a noncovalent interaction between the substrate-binding pocket in the plasmin active site and the scissile peptide bond in the reactive centre loop of antiplasmin. Subsequently, peptide bond cleavage occurs, and, after formation of an ester bond between the carboxy group of the arginine in the newly cleaved peptide bond in antiplasmin and the hydroxy group of the active site serine in plasmin [4,5], major conforma- tional changes probably occur in both the enzyme and inhibitor, in a similar manner to that described for other serine proteinase–serpin reactions [6]. The interaction between a lysine-binding site in plasmin and a complement- ary site in antiplasmin is also of importance in regulating the fibrinolytic system. The same sites in the plasmin molecules are used in the interaction between plasmin and fibrin [7]. Fibrin-bound plasmin molecules therefore react much more slowly with antiplasmin compared to free plasmin, thereby keeping the fibrinolytic process localized [8]. Antiplasmininplasmaisslowlyconvertedintoa nonplasminogen-binding form, by proteolytic cleavage and removal of a C-terminal peptide [9,10]. This suggests that the lysine-binding-site-dependent interaction between antiplasmin and plasmin occurs in the C-terminal part of antiplasmin. It is also known that antiplasmin has a lysine as the C-terminal amino acid [11,12] and that proteins with an exposed C-terminal lysine typically interact with the lysine- binding sites in plasmin(ogen) [13,14]. Therefore it is usually accepted that the C-terminal lysine in antiplasmin is responsible for its interaction with the plasmin(ogen) lysine-binding sites [15]. From kinetic experiments using different natural or synthesized peptides that mimic the C-terminal part of antiplasmin as inhibitors of the plasmin– antiplasmin reaction, it was not possible to clearly determine the specific amino acids involved in this interaction [16–18]. However, it has been suggested that Lys436 in antiplasmin may be, at least partly, involved in the lysine-binding-site- mediated interaction with plasmin [17]. We investigated this in more detail, using site-directed mutagenesis of charged amino acids in the C-terminal portion of antiplasmin. Our first goal was to produce mutants in which the C-terminal lysine was replaced by amino acids without a positive charge. We also produced variants in which other charged amino acids in this portion of the molecule were changed to either uncharged residues or residues of opposite charge. All the antiplasmin variants were expressed in insect cells (Drosophila S2 cells), purified, and characterized with regard to their reactions with plasmin. Materials and methods Chemicals and reagents The vector pMT/BiP/V5 (Invitrogen, Stockholm, Sweden) was used to express antiplasmin, using the efficient Correspondence to B. Wiman, Department of Clinical Chemistry, Karolinska Hospital, SE-17176 Stockholm, Sweden. Fax: + 46 851776150, Tel.: + 46 851773124, E-mail: bjorn.wiman@ks.se Enzyme: Human plasmin (EC 3.4.21.7). (Received 12 February 2003, accepted 17 March 2003) Eur. J. Biochem. 270, 2023–2029 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03578.x Drosophila metallothionein (MT) promoter. pMT/BiP/V5 also contains the Drosophila Bip secretion signal, which efficiently targets high levels of BiP to the endoplasmic reticulum in the S2 cell line. The Drosophila Schneider S2 cell (DES system, Invitrogen) was cultured in Schneider medium with L -glutamine, heat-inactivated fetal bovine serum at a final concentration of 10% (v/v), and penicillin– streptomycin at final concentrations of 50 UÆmL )1 penicil- lin G and 50 lgÆmL )1 streptomycin sulfate. For selection, the pCoHygro vector (Invitrogen) was used in medium containing Hygromycin-B (Roche Diagnostics Scandinavia AB, Stockholm, Sweden). Goat antiplasmin polyclonal antibody (Biopool AB, Umea ˚ , Sweden) was conjugated with horseradish peroxidase as described elsewhere [19]. DEAE-Sepharose CL6B and anhydrotrypsin–agarose were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden) and Sigma-Aldrich Sweden AB (Stockholm, Sweden), respectively. Human plasmin was prepared from purified human plasminogen by activation with strepto- kinase as described [20]. Native antiplasmin was purified from human plasma as described [21]. The plasmin substrate Flavigen Pli was purchased from Biopool AB, and the inhibitor 6-aminohexanoic acid was obtained from Sigma-Aldrich Sweden AB. Construction and expression of wild-type antiplasmin (wt-antiplasmin) Plasmid cDNA of wt-antiplasmin (a gift from Roger Lijnen, Center for Vascular Research, University of Leuven, Belgium) was subcloned into the expression vector pMT/ Bip/V5 using the BglII and XhoI sites. The nucleotide sequence of the insert was confirmed by DNA sequencing and the recombinant protein expressed by transient trans- fection. The wt-antiplasmin plasmid (19 lg) was transfected into Drosophila Schneider S2 cells, and CuSO 4 was added to the medium (final concentration 500 l M ) to induce expression. After 24 h, the supernatant was harvested and the presence of antiplasmin was detected by ELISA. To obtain a stable cell line, the wt-antiplasmin plasmid and the selection plasmid pCoHygro were cotransfected into Droso- phila Schneider S2 cells (3 · 10 6 mL )1 , in a total volume of 3 mL), using a concentration ratio between expression plasmid and selection plasmid of 19 : 1 (19 lg expression plasmid and 1 lg selection plasmid). The cells were then grown in the presence of Hygromycin-B at a final concen- tration of 250 lgÆmL )1 . The medium was changed every fourth day, and after 3 weeks a stable cell line that could express wt-antiplasmin was established. Mutagenesis of antiplasmin The cDNA of wt-antiplasmin, subcloned into the expression vector pMT/Bip/V5 at the BglII and XhoI sites, was used to produce the selected mutants of antiplasmin. Site-directed mutagenesis was performed using modified primers result- ing in the desired modification (Quick Change Mutagenesis Method; Stratagene, Stockholm, Sweden). The primers, each complementary to opposite strands of the template cDNA, were extended during temperature cycling using PfuTurbo DNA polymerase. The PCR product was then treated with DpnI endonuclease, which is specific for methylatedandhemimethylatedDNA.Itisusedtodigest the parental DNA template, but not newly synthesized DNA copies. The nicked plasmid DNA containing the desired mutations was transformed into XL1-Blue super competent cells. Seven constructs were made to obtain the following mutants of antiplasmin: K429E; K436E; E442G; E443G; D444G; K452E; K452T (Table 1, Fig. 1). The nucleotide sequences of all the constructs were confirmed. All mutants were then transfected as described for wt-antiplasmin. Expression of antiplasmin variants To express the antiplasmin variants, transfected Drosophila Schneider S2 cells were cultured and extended in Schneider medium containing L -glutamine, heat-inactivated fetal bovine serum (final concentration 10%, v/v), 50 UÆmL )1 penicillin G and 50 lgÆmL )1 streptomycin sulfate. After the volume had been increased to 500 mL with a cell concen- tration of about 3 · 10 6 mL )1 , the cells were transferred to thesamevolumeofDrosophila serum-free medium con- taining L -glutamine and penicillin/streptomycin at the same concentrations as above. Pluronic F68 (final concentration 0.05%) and CuSO 4 at a final concentration of 500 l M were also added. The cells were cultured at room temperature in darkness for 3 days with gentle stirring. The cells were removed by centrifugation at 2000 g for 30 min, and the supernatant containing antiplasmin was stored at )70 °C. Table 1. Identification of oligonucleotides used to introduce mutations in antiplasmin cDNA and sequencing antiplasmin variants. Amino acid exchanged Nucleotides exchanged Primer length K429E A1285G G1270–C1303 K436E A1306G, A1308G T1291–G1324 E442G A1324G, G1325C C1309–G1342 E443G A1327G, G1328C G1312–T1345 D444G A1330G, T1331C C1315–C1348 K452E A1354G C1339–G1372 K452T A1354C C1339–G1372 Fig. 1. Schematic presentation of the antiplasmin structure and sites where mutations were introduced. 2024 H. Wang et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Determination of antiplasmin activity and antigen concentration Antiplasmin activity was determined by a titration method against plasmin of known concentration, essentially as described [20]. Antiplasmin antigen concentration was determined by an ELISA method. For this purpose, Maxisorp microtiter plates (96 wells; Nunc, Labdesign AB, Stockholm, Sweden) were coated for 2 h at room temperature with goat anti-antiplasmin IgG, diluted in 0.1 M NaHCO 3 buffer, pH 9.6. The plates were then incubated at room temperature for 30 min with 0.04 M sodium phosphate buffer, pH 7.3, containing 0.1 M NaCl and 1 mgÆmL )1 BSA and washed three times with the phosphate/NaCl buffer without BSA. The samples (200 lL) to be analysed were compared with an antiplasmin standard (purified native antiplasmin from human plasma at the following concentrations: 40, 20, 10, 5, 2.5 and 0 lgÆL )1 ). The microtiter plates were incubated for 2 h at room temperature and then washed four times with the phos- phate/NaCl buffer. Then, the anti-antiplasmin IgG conju- gated with horseradish peroxidase was added to the samples. After incubation for 1 h and washing the plates four times with phosphate/NaCl buffer, the horseradish peroxidase substrate o-phenylenediamine (Sigma) in the presence of H 2 O 2 was added. After another incubation for 10 min, 50 lL stop solution (3 M H 2 SO 4 ) was added to each well, and A 492 recorded by a microtiter plate reader. Antiplasmin antigen concentration in the samples was then calculated from the standard curve obtained. Purification of antiplasmin variants Three steps were used for purification of the antiplasmin variants from the S2 cell cultures. First, the supernatants (about 500 mL) were incubated with DEAE-Sepharose CL6B in a batch procedure (about 70 mL, equilibrated with 0.05 M Tris/HCl buffer, pH 8.0) with slow stirring for 2 h at 4 °C. Each suspension was then filtered through a Bu ¨ chner funnel and washed with about 1 L 0.05 M Tris/HCl buffer, pH 8.0 (until the A 280 was less than 0.1), and then with the equilibration buffer, containing 0.2 M NaCl. Anti- plasmin was then eluted from the Bu ¨ chner funnel using about 100 mL equilibration buffer containing 0.4 M NaCl. The fractions containing antiplasmin (detected by ELISA) were dialysed overnight at 4 °C against 0.05 M Tris/HCl buffer, pH 8.0. The dialysate was then applied to a DEAE- Sepharose CL6B column (2.5 cm diameter · 5 cm), equili- brated with 0.05 M Tris/HCl buffer, pH 8.0. After a wash with this buffer, the column was eluted with a linear gradient from 0 to 0.4 M NaCl in the Tris/HCl buffer [5]. The antiplasmin concentration was determined by ELISA (see above), and the fractions containing antiplasmin were pooled and dialysed overnight against 0.04 M sodium phosphate buffer, pH 7.5, containing 0.1 M NaCl. The dialysate was applied to an anhydrotrypsin–agarose column (column volume  2 mL), equilibrated with the phosphate/NaCl buffer. The column was washed with the same buffer until the A 280 was less than 0.1. Elution was performed with the equilibration buffer, containing 0.3 M arginine. Fractions containing antiplasmin were dialysed against 0.04 M sodium phosphate buffer, pH 7.3, and then stored at )70 °C. SDS/PAGE SDS/PAGE was performed in a Mini-protean II electro- phoresis apparatus (Bio-Rad, Stockholm, Sweden) as described by Laemmli [22]. Proteins were separated in 10% (w/v) polyacrylamide gels and stained with Coomassie Brilliant Blue R-250. Determination of rate constants in the reaction between plasmin and the antiplasmin variants The two reactants were mixed at low concentrations, in either 0.1 M sodium phosphate buffer, pH 7.3, or the same buffer containing 1.0 m M 6-aminohexanoic acid. The final plasmin concentration (active site titrated) used in these experiments was 0.6 n M , whereas the antiplasmin concen- tration varied between 1 and 5 n M . After specified times of reaction (0–300 s), samples were withdrawn into tubes containing high concentration of a plasmin substrate (0.6 m M Flavigen Pli, final concentration), 20 m M 6-aminohexanoic acid and polyclonal rabbit anti-(human antiplasmin) IgG (1 mgÆmL )1 ). By this procedure further inhibition of plasmin was rapidly and efficiently decreased, allowing long incubation times with the plasmin substrate, which is necessary to accurately measure the low plasmin concentrations. After incubation for 1.5 h, plasmin cleavage of the chromogenic substrate was stopped by addition of acetic acid (final concentration 1%, v/v) and the A 405 recorded. A 405 is thus a reliable measure of the residual plasmin concentration at the time of sampling. Then, the reaction rate constants were calculated from the results using the classic formula for second-order reactions [1], using data obtained before 50% of the plasmin was inhibited. In the experiments performed in the presence of 6-aminohexanoic acid, in which the antiplasmin concentra- tion was almost 10-fold higher than the plasmin concentra- tion, pseudo-first-order conditions were assumed and the reaction rate constants were calculated from the half-lives of plasmin in these experiments (also before 50% of the plasmin activity was inhibited). Computer model of the C-terminal 40 amino acids in antiplasmin A computer model of the C-terminal 40 amino acids in antiplasmin [11] was constructed by CS CHEM 3 DULTRA , version 7.0 (Cambridge Soft, Cambridge, MA, USA), followed by energy minimization with the MM2 protocol. Modelling was performed on different lengths of the C-terminal portion of the antiplasmin molecule, ranging from 30 to 50 residues from Lys452. However, energy minimization did not work well on structures with more than 40 amino acids. Results Generation of antiplasmin mutants Using the QuickChange mutagenesis method, seven anti- plasmin mutants (K429E, K436E, E442G, E443G, D444G, K452E and K452T) were constructed. The cDNA structure was confirmed by nucleotide sequencing. The mutant Ó FEBS 2003 Antiplasmin–plasmin interaction (Eur. J. Biochem. 270) 2025 K452T contained an unwanted mutation in position 180, changing the expected Phe to a Leu. As the functional behaviour of this mutant was found to be almost identical with the other mutant of this specific residue, K452E, and with wt-antiplasmin, we did not correct this mistake. The conservative mutation from one hydrophobic to another hydrophobic amino acid distant from the C-terminal portion of antiplasmin seems therefore to be of little importance. Expression and purification of antiplasmin variants wt-antiplasmin and seven antiplasmin mutants were expressed in Drosophila S2 cells. With the expression plasmid used (see Material and methods), the antiplasmin variants were exported via the endoplasmic reticulum to the conditioned medium, where they are found in soluble forms. The concentrations of the different antiplasmin variants in the conditioned medium were typically quite high, from 5 to 70 lgÆmL )1 (Table 2). After purification by DEAE-Sepharose CL6B and anhydrotrypsin–agarose chro- matography, the typical yield from the harvested condi- tioned medium was  20%. All antiplasmin variants could be purified using this procedure. Activity determination by titration against plasmin of known concentration and measuring free plasmin with a chromogenic plasmin sub- strate suggested that all antiplasmin variants were fully or close to fully active (data not shown). Formation of SDS-stable complexes between antiplasmin variants and plasmin The ability of the different variants to form stable complexes with plasmin was studied by SDS/PAGE. As shown in Fig. 2, the most important antiplasmin variants (wt, K436E, K452E and K452T) were  80% pure after the described purification procedure and could almost quanti- tatively form stable complexes with plasmin. Influence of the antiplasmin variants on the plasmin–antiplasmin reaction To study the reaction between the antiplasmin variants and plasmin in the absence of 6-aminohexanoic acid, plasmin (final concentration 0.6 n M ) was mixed with antiplasmin (final concentration 1.3 n M ). Samples were taken from 0 to 60 s and added to a mixture of plasmin substrate, 6-aminohexanoic acid and anti-antiplasmin IgG as des- cribed in Material and methods. After incubation for 90 min at room temperature and addition of acetic acid, A 405 was recorded. The absorbance value at a certain time, compared with the absorbance value at zero time, was used to calculate residual plasmin activity at that time. The prerequisite was that less than 50% of the added plasmin had been inhibited. The rate constants were calculated from the classical formula of second-order reactions [1]. Similar experiments were performed in the presence of 1.0 m M 6-aminohexanoic acid. The plasmin concentration was the same, but the concentrations of the antiplasmin variants were higher (5.0 n M ) and the incubation time prolonged (0–300 s). Residual plasmin activity was measured, and rate constants were calculated assuming pseudo-first-order kine- tics. The rate constants for the reactions between plasmin and the antiplasmin variants are shown in Table 3 (in both the presence and absence of 6-aminohexanoic acid). The reactions between ‘native’ human antiplasmin and plasmin in the presence or absence of 6-aminohexanoic acid were also studied for comparison. All variants of antiplasmin except for K436E had a rate constant higher than 10 7 M )1 Æs )1 . This is not far from the rate constant obtained with native antiplasmin. In addition, the method used here gave very similar results to those reported in earlier studies [1,2]. Interestingly, the two mutants K452E and K452T did not differ in activity from wt-antiplasmin, suggesting that the C-terminal lysine is of little importance in the lysine- Table 2. Concentration of antiplasmin in the conditioned media from the various S2 cells expressing the different variants of antiplasmin. About 500 mL was harvested from each cell line. Antiplasmin variant Concentration (lgÆmL )1 ) wt-antiplasmin 12.8 K429E 13.6 K436E 72.0 E442G 4.5 E443G 7.7 D444G 6.8 K452E 42.0 K452T 23.3 Fig. 2. SDS/PAGE of some of the antiplasmin variants in the presence (lanes 1–4) or absence (lanes 5–8) of plasmin. The antiplasmin vari- ants shown are: wt (lanes 4 and 8); K436E (lanes 3 and 7); K452E (lanes 2 and 6); K452T (lanes 1 and 5). In addition pure plasmin is showninlane9. 2026 H. Wang et al.(Eur. J. Biochem. 270) Ó FEBS 2003 binding-site-mediated interaction between plasmin and antiplasmin. On the other hand, the variant K436E reacts much more slowly (about fivefold) than the other variants, suggesting that this residue is important in this interaction. In the presence of 6-aminohexanoic acid, the reaction rate decreased 10-fold or more for most variants. Also in this case the results with wt-antiplasmin and the two mutants of the C-terminal lysine did not differ. Only K436E was less affected by 6-aminohexanoic acid (2.5-fold decrease in reaction rate), again suggesting that this residue is involved in the lysine-binding-site-mediated interaction between plasmin and antiplasmin. Molecular modelling of the C-terminal portion of antiplasmin The amino-acid sequence of the C-terminal 40 residues in antiplasmin is GNKDFLQSLKGFPRGDKLFGPDLKL VPPMEEDYPQFGSPK-OH [11]. The C-terminal lysine is residue 452 in the antiplasmin molecule. Molecular model- ling resulted in the structure shown in Fig. 3. We construc- ted a number of models with different lengths, ranging from 30 to 50 residues from the C-terminal Lys452. All models were similar around the two sites Lys452 and Lys436. In these models, the side chain of the C-terminal lysine residue (K452) seems to be in the near vicinity of the side chain of Phe448. Lys436, on the other hand, is found at the surface of the molecule with a protruding side chain. Discussion Antiplasmin belongs to the serpin superfamily of proteins [6], from which many individual members have been crystallised. The general structures of these proteins are therefore well established [6]. However, the C-terminal portion of antiplas- min is unique to this inhibitor [11,12] with no known similarities to the other members of this protein family. The lysine-binding-site-mediated interaction between plas- min and antiplasmin is of great importance in regulating the fibrinolytic process and keeping the plasmin active in the right place and at the correct time, as fibrin and antiplasmin compete for active plasmin molecules [7,8]. In addition, it is known that the lysine-binding sites in plasminogen are important for binding to other proteins, such as receptor proteins at the cellular surface, in both mammalian cells [13,23] and bacteria [14,24,25]. It has been reported that plasminogen bound to such receptor proteins is more readily activated to plasmin [26,27], providing the cells with a proteolytic shield, and thereby enhancing processes such as invasive growth and cell migration. Increasing our knowledge about structures involved in the interaction with the lysine- binding sites in plasmin(ogen) may be important for finding new agents that can effectively interfere with these processes. We have here studied the lysine-binding-site-dependent interaction between plasmin and antiplasmin in detail by constructing seven single-site mutants of antiplasmin. Fig. 3. Computer model of the C-terminal 40 amino acids in antiplasmin. Some of the residues are labelled to facilitate viewing. Table 3. Rate constants (in 10 6 M –1 Æs –1 ) in the reactions between plasmin and the different antiplasmin variants in the absence (No 6-AHA) or presence (6-AHA) of 1.0 m M 6-aminohexanoic acid. Antiplasmin variant No 6-AHA 6-AHA Native antiplasmin 25.3 ± 1.7 2.5 wt-antiplasmin 10.9 ± 0.3 1.1 K429E 27.3 ± 2.5 2.7 K436E 2.1 ± 0.3 0.8 E442G 19.5 ± 1.0 1.5 E443G 24.3 ± 1.2 1.6 D444G 21.6 ± 0.9 1.6 K452E 11.5 ± 0.7 0.8 K452T 12.7 ± 1.0 0.9 Ó FEBS 2003 Antiplasmin–plasmin interaction (Eur. J. Biochem. 270) 2027 All mutations were performed in the C-terminal 23 residues. After expression and purification of the different antiplas- minvariants,theywerecharacterizedwithregardtotheir reactions with plasmin, both structurally (SDS/PAGE) and kinetically. The results were compared with results obtained for ‘native’ human antiplasmin. All antiplasmin variants were found to be very active, forming stable complexes with plasmin (Fig. 2) in a comparable way to ‘native’ human antiplasmin [1,2,4]. The complexes formed were completely stable during analysis by SDS/PAGE. Also, the rate constant determined for the reaction between plasmin and wt-antiplasmin was only slightly lower than that found for ‘native’ antiplasmin using an identical experimental system. The latter constant is also very similar to that pre- viously reported [1,2]. Furthermore, the reaction between wt-antiplasmin and plasmin is decreased by about one order of magnitude in the presence of 1 m M 6-aminohexanoic acid, which is almost identical with the results with ‘native’ antiplasmin using the same experimental set up (Table 3). This clearly demonstrates that the overall structure and main functions of wt-antiplasmin were not altered by the expression in S2 cells or during purification. As already pointed out, all the constructed antiplasmin variants formed SDS-stable complexes with plasmin. In addition, the reaction rate with plasmin for most of these variants was comparable to that of the ‘native’ antiplasmin, again demonstrating the reliability of our techniques. A major finding in this report is that replacement of the C-terminal amino acid in antiplasmin, Lys452, with an acidic residue (Glu) or a neutral hydrophilic residue (Thr) did not significantly change the activity or kinetic properties. This is interesting, as it has been generally accepted that this residue was responsible for the lysine-binding-site-mediated interaction between antiplasmin and plasmin [15–18]. In fact, many other proteins with a C-terminal lysine seem to bind to plasmin(ogen) quite efficiently [13,14]. However, in antiplasmin the C-terminal lysine does not seem to be important in this respect. In the presence of 1.0 m M 6-aminohexanoic acid, the reaction rates between plasmin and the antiplasmin variants typically decreased about one order of magnitude. The only exception was the variant K436E, for which the reaction rate decreased by only a factor of 2–3. On the other hand, in the absence of 6-aminohexanoic acid, it reacted about fivefold more slowly than the wild-type with plasmin. Both these findings clearly suggest that Lys436 is important for the interaction of antiplasmin with the lysine-binding sites in plasmin. The relatively small (fivefold) difference in reaction rate suggests that other structures in the vicinity of Lys436 are probably involved, but the positive charge of Lys436 most certainly has a key function. It was previously shown that Lys436 may be involved in the lysine-binding-site- mediated interaction between plasmin and antiplasmin [17]. Replacement of several other charged residues in this portion of the molecule did not significantly affect the reaction rate with plasmin. To shed more light on possible mechanisms explaining the behaviour of our mutants, we constructed a computer model of the C-terminal part (40 residues) of antiplasmin (Fig. 3). We cannot claim that this model shows an absolutely correct picture of the structure of the C-terminal portion of antiplasmin. However, there are a large number of proline residues in this part (10 of the C-terminal 55 residues), increasing the possibility of obtaining a model that at least partly mimics the true structure. In fact, the computer model supports our experimental data. The side chain of Lys452 in this model is found in close vicinity to the side chain of Phe448. If this is true, it may explain possible restrictions in the interaction between this residue and the lysine-binding sites in the intact plasmin molecule. In addition, the side chain of Lys436 in our computer model is found at the surface of the molecule and may definitely be involved in interactions with a lysine-binding site in plasmin. Some of the mutants, especially K429E and D443G, reacted slightly more rapidly than wt-antiplasmin with plasmin. The reason for this is not known. Since submission of the original version of this paper, a study on the interaction of a recombinant C-terminal 55-residue peptide from antiplasmin, expressed in Escheri- chia coli, and isolated ‘kringle’ 1 or ‘kringle’ 4 structures from plasminogen has been published [28]. The authors concluded that Lys452 is important in these interactions, but that other structures may also be involved. In view of our data with complete molecules, it is indeed possible that Lys452 may be more involved in interactions with smaller molecules, such as isolated ‘kringles’, but to a much lesser extent with the complete plasmin(ogen) molecule. More work is certainly needed to resolve this question. Acknowledgements Skilful technical assistance by Anette Dahlin and Marie Haegerstrand- Bjo ¨ rkman is gratefully acknowledged. We thank Dr Roger Lijnen, Center for Vascular Research, University of Leuven, Belgium, for providing us with antiplasmin cDNA. Financial support was obtained from the Swedish Medical Research Council (project no. 05193), the Swedish Cancer Foundation, the Heart and Lung Foundation and funds from Karolinska Institute. References 1. Wiman, B. & Collen, D. (1978) On the kinetics of reaction between human antiplasmin and plasmin. Eur. J. Biochem. 84, 573–568. 2. Christensen, U., Bangert, K. & Thorsen, S. (1996) Reaction of human alpha2-antiplasmin and plasmin stopped flow fluorescence kinetics. FEBS Lett. 387, 58–62. 3. Wiman, B., Boman, L. & Collen, D. (1978) On the kinetics of the reaction between human antiplasmin and a low-molecular-weight form of plasmin. Eur. J. Biochem. 87, 143–146. 4. Wiman, B. & Collen, D. (1979) On the mechanism of the reaction between human a 2 -antiplasmin and plasmin. J. Biol. Chem. 254, 9291–9197. 5. Nilsson, T. & Wiman, B. (1982) On the structure of the stable complex between plasmin and a 2 -antiplasmin. FEBS Lett. 142, 111–114. 6. Huntington, J.A. & Carrell, R.W. (2001) The serpins: nature’s molecular mousetraps. Sci. Prog. 84, 125–136. 7. Wiman, B., Lijnen, H.R. & Collen, D. (1979) On the specific interaction between the lysine-binding sites in plasmin and com- plementary sites in a 2 -antiplasmin and in fibrinogen. Biochim. Biophys. Acta 579, 142–154. 8. Wiman, B. & Collen, D. (1978) Molecular mechanisms of phy- siological fibrinolysis. Nature (London) 272, 549–550. 9. Clemmensen, I., Thorsen, S., Mu ¨ llertz, S. & Petersen, L.C. (1981) Properties of three different molecular forms of the a 2 -plasmin inhibitor. Eur. J. Biochem. 120, 105–112. 2028 H. Wang et al.(Eur. J. Biochem. 270) Ó FEBS 2003 10. Wiman, B., Nilsson, T. & Cedergren, B. (1982) Studies on a form of a 2 -antiplasmin in plasma which does not interact with the lysine-binding sites in plasminogen. Thromb. Res. 28, 193–199. 11. Holmes, W.E., Nelles, L., Lijnen, H.R. & Collen, D. (1987) Primary structure of human a 2 -antiplasmin, a serine protease inhibitor (Serpin). J.Biol.Chem.262, 1659–1664. 12. Lijnen, H.R., Holmes, W.E., Van Hoef, B., Wiman, B., Rodri- guez,H.&Collen,D.(1987)Amino-acidsequenceofhuman a 2 -antiplasmin. Eur. J. Biochem. 166, 565–574. 13. Miles, L.A., Dahlberg, C.M., Plescia, J., Felez, J., Kato, K. & Plow, E.F. (1991) Role of cell-surface lysines in plasminogen binding to cells: identification of a-enolase as a candidate plasminogen receptor. Biochemistry 30, 1682–1691. 14. Sjo ¨ stro ¨ m, I., Gro ¨ ndahl, H., Falk, G., Kronvall, G. & Ullberg, M. (1997) Purification and characterisation of a plasminogen-binding protein from Haemophilus influenzae. Sequence determination reveals identity with aspartase. Biochim. Biophys. Acta 1324, 182–190. 15. Hortin, G.L., Gibson, B.L. & Fok, K.F. (1988) a 2 -Antiplasmin’s carboxy-terminal lysine residue is a major site of interaction with plasmin. Biochem. Biophys. Res. Commun. 155, 591–596. 16. Sasaki, T., Morita, T. & Iwanaga, S. (1986) Identification of the plasminogen-binding site of human alpha 2-plasmin inhibitor. J. Biochem. (Tokyo) 99, 1699–1705. 17. Wiman, B., Almqvist, A ˚ .&Ra ˚ nby, M. (1979) The non-covalent interaction between human plasmin and a 2 -antiplasmin. Fibrino- lysis 3, 231–235. 18. Sugiyama, N., Sasaki, T., Iwamoto, M. & Abiko, Y. (1988) Binding site of a 2 -plasmin inhibitor to plasminogen. Biochim. Biophys. Acta 952, 1–7. 19. Tijssen, P. & Kurstak, E. (1984) Highly efficient and simple methods for the preparation of peroxidise and active peroxidase– antibody conjugates for enzyme immunoassays. Anal. Biochem. 136, 451–457. 20. Wiman, B. (1981) Methods in Enzymology: Human A 2 -Antiplas- min, pp. 395–408. Academic Press, New York. 21. Wiman, B. (1980) Affinity-chromatographic purification of human a 2 -antiplasmin. Biochem. J. 191, 229–232. 22. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriphage T4. Nature (London) 227, 680–685. 23. Hajjar, K.A., Harpel, P.C., Jaffe, E.A. & Nachman, R.L. (1986) Binding of plasminogen to cultured human endothelial cells. J.Biol.Chem.261, 11656–11662. 24. Ullberg, M., Kronvall, B. & Wiman, B. (1989) New receptor for human plasminogen on gram positive cocci. APMIS 97, 996–1002. 25. Ullberg, M., Karlsson, I., Wiman, B. & Kronvall, G. (1992) Two types of receptors for human plasminogen on group G strepto- cocci. APMIS 100, 21–28. 26. Eberhard, T., Ullberg, M., Sjo ¨ stro ¨ m, I., Kronvall, G. & Wiman, B. (1995) Enhancement of t-PA-mediated plasminogen activation by bacterial surface receptors. Fibrinolysis 9, 65–70. 27. Gong, Y., Kim, S.O., Felez, J., Grella, D.K., Castellino, F.J. & Miles, L.A. (2001) Conversion of Glu-plasminogen to Lys-plas- minogen is necessary for optimal stimulation of plasminogen activation on the endothelial cell surface. J. Biol. Chem. 276, 19078–19083. 28. Frank, P.S., Douglas, J.T., Locher, M., Llinas, M. & Schaller, J. (2003) Structural/functional characterization of the a2-plasmin inhibitor C-terminal peptide. Biochemistry 42, 1078–1085. Ó FEBS 2003 Antiplasmin–plasmin interaction (Eur. J. Biochem. 270) 2029 . Identification of amino acids in antiplasmin involved in its noncovalent ‘lysine-binding-site’-dependent interaction with plasmin Haiyao Wang,. lysine-binding-site-dependent interaction between antiplasmin and plasmin occurs in the C-terminal part of antiplasmin. It is also known that antiplasmin