Purification of high and low molecular weight plasminogen activator inhibitor 1 from fibrosarcoma cell-line HT 1080 conditioned medium

Functional stability of plasminogen activator inhibitor-1

Plasminogen activator inhibitor-1 (PAI-1) is the main inhibitor of plasminogen activators, such as tissue-type plasminogen activator (t-PA) and urokinase-type plasminogen activator (u-PA), and a major regulator of the fibrinolytic system. PAI-1 plays a pivotal role in acute thrombotic events such as deep vein thrombosis (DVT) and myocardial infarction (MI). The biological effects of PAI-1 extend far beyond thrombosis including its critical role in fibrotic disorders, atherosclerosis, renal and pulmonary fibrosis, type-2 diabetes, and cancer. The conversion of PAI-1 from the active to the latent conformation appears to be unique among serpins in that it occurs spontaneously at a relatively rapid rate. Latency transition is believed to represent a regulatory mechanism, reducing the risk of thrombosis from a prolonged antifibrinolytic action of PAI-1. Thus, relying solely on plasma concentrations of PAI-1 without assessing its function may be misleading in interpreting the role of PAI-1 in many complex diseases. Environmental conditions, interaction with other proteins, mutations, and glycosylation are the main factors that have a significant impact on the stability of the PAI-1 structure. This review provides an overview on the current knowledge on PAI-1 especially importance of PAI-1 level and stability and highlights the potential use of PAI-1 inhibitors for treating cardiovascular disease.

Structures of importance for the stability of antiplasmin as studied by site-directed mutagenesis

Human antiplasmin, a fast-acting inhibitor of plasmin in plasma, belongs to the serpin super-family of proteins. Like other members of this family, antiplasmin has a scissile peptide bond exposed within a reactive centre loop, typically present at the surface of the molecule. Antiplasmin is stable at neutral pH, but at acidic pH or at elevated temperatures it rapidly becomes inactivated. Data regarding "native" antiplasmin have demonstrated that both polymerization processes and formation of latent molecules are important in this respect. In this work we used site-directed mutagenesis to produce 11 single-site mutants (mainly within Abeta-sheet, Bbeta-sheet and reactive centre loop), which were expressed in Drosophila S2 cells, purified and characterized. Five of the 11 mutants were found to have a deviating stability at decreased pH. Glu346Thr was the only mutant with a lesser stability as compared to wt-antiplasmin, but the other 4 were more stable. The most stable mutant, His341Thr, was 7-fold more stable at pH 4.9 as compared to wt-antiplasmin. The wt-antiplasmin had a much more pronounced tendency to polymerize at decreased pH, as compared to "native" antiplasmin. However, many of the mutants clearly rather formed latent molecules, as judged both from PAGE-analysis at non-denaturing condition and reactivation experiments.

Inactivation of antiplasmin at low pH: evidence for the formation of latent molecules

Several serine proteinase inhibitors (serpins) are metastable proteins which under certain conditions may undergo conformational changes resulting in the insertion of the reactive centre loop into the so-called Abeta-sheet and hence forming latent molecules. Here we have studied the inactivation of antiplasmin as a function of pH and temperature with time. At decreased pH (4.9-5.8) and at room temperature, antiplasmin activity decreased following first-order kinetics. Analysis by polyacrylamide gel electrophoresis under non-denaturing conditions demonstrated that only minor amounts of polymerized material formed after extensive incubation (4 days) at room temperature. However, on incubation at elevated temperatures (45 or 55 degrees C), a rapid formation of polymerized material was observed. We also demonstrated that antiplasmin inactivated by treatment at pH approximately 5 at room temperature spontaneously slowly regained some activity if incubated in a buffer of neutral pH. Furthermore, by treatment with 4 M guanidinium chloride for about 30 min, followed by dialysis against a neutral phosphate buffer, considerable activity (almost 40%) was regained. Thus, we conclude that antiplasmin, at least partially, at lower temperatures is transformed into a latent form, which could be reactivated, in a similar manner as PAI-1. At increased temperature, however, polymerization seems to be the predominant reason for inactivation.

PAI-1 stability: the role of histidine residues

The role of the 13 histidine residues in plasminogen activator inhibitor 1 (PAI-1) for the stability of the molecule was studied by replacing these residues by threonine, using site-directed mutagenesis. The generated mutants were expressed in Escherichia coli, purified and characterized. All variants had a normal activity and formed stable complexes with tissue-type plasminogen activator. Most of these PAI-1 variants displayed a similar pH-dependency in stability as wild-type PAI-1, with increased half-lives at lower pH. However, the variant His364Thr had a half-life of about 50 min at 37 degrees C and had almost completely lost its pH-dependency. Therefore, our data suggest that His(364), in the COOH-terminal end of the molecule might be responsible for the pH-dependent stability of PAI-1.

Tissue plasminogen activator (tPA) antigen in plasma: correlation with different tPA/inhibitor complexes

In many studies, tPA antigen has been a strong predictor of myocardial infarction. However, only a few percent of the total tPA antigen present in plasma samples taken at rest constitutes active tPA. The rest is enzymatically inactive and consists of a heterogeneous mixture of tPA in complex with inhibitors such as PAI-1, antiplasmin and C1 inhibitor. In the present study we developed specific two-site ELISA methods for determining the individual protease/inhibitor complexes constituting tPA antigen. We subsequently measured the concentrations of the different complexes in plasma samples taken from 30 healthy individuals. The results show that the concentration of the complex between tPA and PAI-1 in plasma correlated strongly with that of tPA antigen in plasma, as measured with a commercially available kit. Also the correlation between tPA/PAI-1 complex levels in plasma and the PAI-1 activity concentration was significant. However, no significant correlations were found between tPA antigen concentration and tPA/C1 inhibitor or tPA/antiplasmin concentrations in plasma.

Functional effects of single amino acid substitutions in the region of Phe113 to Asp138 in the plasminogen activator inhibitor 1 molecule

Thirteen amino acid substitutions have been introduced within the stretch Phe113 to Asp138 in the plasminogen activator inhibitor 1 (PAI-1) molecule by site-directed mutagenesis. The different proteins and wild-type (wt) PAI-1 have been overexpressed in Escherichia coli and purified by chromatography on heparin-Sepharose and on anhydrotrypsin-agarose. The PAI-1 variants have been characterized by their reactivity with tissue plasminogen activator (tPA), interactions with vitronectin or heparin, and stability. Most PAI-1 variants, except for Asp125-->Lys, Phe126-->Ser and Arg133-->Asp, displayed a high spontaneous inhibitory activity towards tPA, which did not change greatly on reactivation with 4 M guanidinium chloride, followed by dialysis at pH 5.5. The variants Asp125-->Lys and Arg133-->Asp became much more active after reactivation and they were also more rapidly transformed to inactive forms (t12 22-31 min) at physiological pH and temperature than the other variants. However, in the presence of vitronectin they were both almost equally stable (t12 2.3 h) as wtPAI-1 (t12 3.0 h). The mutant Glu130-->Lys showed an increased stability, both in the absence and in the presence of vitronectin compared with wtPAI-1. Nevertheless a similar affinity between all the active PAI-1 variants and vitronectin was observed. Further, all mutants, including the three mutants with low activity, were to a large extent adsorbed on anhydrotrypsin-agarose and were eluted in a similar fashion. In accordance with these data, the three variants with a low activity were all to a large extent cleaved as a result of their reaction with tPA, suggesting that they occurred predominantly in the substrate conformation. Our results do not support the presence of a binding site for vitronectin in this part of the molecule, but rather that it might be involved in controlling the active PAI-1 to substrate transition. Partly, this region of the PAI-1 molecule (Arg115 to Arg118) seems also to be involved in the binding of heparin to PAI-1.

Stability of plasminogen activator inhibitor-1: role of tyrosine221

Using site-directed mutagenesis, changes of Tyr221 in plasminogen activator inhibitor-1 (PAI-1) have provided mutants with normal activity, but with increased stability. At physiological conditions, the transition of the PAI-1 mutants Tyr221His and Tyr221Ser to the latent form was significantly prolonged (half-lives 14.8 and 4.1 h, respectively) as compared to wild-type PAI-1 (2.0 h). Their half-lives, especially for the Tyr221Ser mutant, were even more prolonged in the presence of vitronectin (23.8 and 53.7 h, respectively). While wild-type PAI-1 was more stable at lower pH, the PAI-1 mutants Tyr221His and Tyr221Ser had stability optima at about pH 6.5, but displayed shorter half-lives at pH 5.5.

The cluster of basic amino acids in vitronectin contributes to its binding of plasminogen activator inhibitor-1: evidence from thrombin-, elastase- and plasmin-cleaved vitronectins and anti-peptide antibodies

Derivatives of vitronectin obtained by specific cleavage at its cluster of basic amino acids with thrombin, elastase and plasmin are shown to have a decreased ability to bind plasminogen activator inhibitor-1 (PAI-1). The identification and localization of the segment involved in the binding of PAI-1 (Lys348-Arg379) were carried out by purification of these cleaved vitronectins and their subsequent structural characterization (sequence analysis, phosphorylation of Ser378 with cAMP-dependent protein kinase and immunostaining with peptide-specific antibodies), then measurement of the vitronectin-PAI-1 interaction by (a) a two-phase system (ELISA); (b) co-precipitation of the vitronectin-PAI-1 complex out of solution, and (c) analysis of the stereospecific interaction between the active conformation of PAI-1 and a peptide derived from the above-mentioned cluster; this interaction occurs when the peptide is composed of all-l-amino acids but not when it is composed of all-d-amino acids. Our results explain why workers who have used immobilized vitronectin to study this interaction could not have observed the involvement of the cluster of basic amino acids in PAI-1 binding, since the immobilization of vitronectin is shown to render this cluster inaccessible for interaction. We propose that vitronectin binds active PAI-1 by interaction via amino acid residues that originate from distal locations in the N- and C-termini of vitronectin.

Phosphorylation of vitronectin on Ser362 by protein kinase C attenuates its cleavage by plasmin

Vitronectin, found in the extracellular matrix and in circulating blood, has an important role in the control of plasminogen activation. It was shown to be the major protein substrate in human blood fluid for a protein kinase A (PKA) released from platelets upon their physiological stimulation with thrombin. Since vitronectin was shown to have only one PKA phosphorylation site, but to contain 2-3 mol covalently bound phosphate, it was reasonable to assume that other protein kinases might phosphorylate vitronectin at other sites in the protein. We have reported earlier that human serum contains at least three protein kinases, one of which was found to be cAMP independent and to phosphorylate a repertoire of plasma proteins that was very similar to that obtained upon phosphorylation of human plasma with protein kinase C (PKC). Since there are now several examples of proteins with extracellular functions that are phosphorylated by PKC, we undertook to study the phosphorylation of vitronectin by PKC. Here, we show that vitronectin is a substrate for PKC, and characterize the kinetic parameters of this phosphorylation (Km approximately tenfold lower than the concentration of vitronectin in blood), indicating that, from the biochemical point of view, this phosphorylation can occur at the locus of a hemostatic event. We also identify Ser362 as the major PKC phosphorylation site in vitronectin, and confirm this localization by means of synthetic peptides derived from the cluster of basic amino acids in vitronectin surrounding Ser362. We show that the PKC phosphorylation at Ser362 alters the functional properties of vitronectin, attenuating its cleavage by plasmin at Arg361-Ser362. This phosphorylation has the potential to regulate plasmin production from plasminogen by a feedback mechanism involving the above-mentioned plasmin cleavage, a loosening of the vitronectin grip on inhibitor 1 of plasminogen activators, and a subsequent latency of this regulatory inhibitor.

Identification of a PAI-1 binding site in vitronectin

Active PAI-1 (plasminogen activator inhibitor 1) is bound to vitronectin in plasma and in the extracellular matrix. In this study we aimed at identifying the PAI-1 binding site in vitronectin, which at present is a matter of dispute. Vitronectin was cleaved with trypsin and the fragments were tested for inhibitory effect on the PAI-1/vitronectin interaction using vitronectin-coated microtiter plates. Intact vitronectin and the tryptic digest of vitronectin both caused a 50% reduction in PAI-1 binding at a concentration of about 2 nmol/I. Gel-filtration on Sephadex G-50 superfine of the tryptic peptides resulted in one main peak of inhibitory activity. The elution volume, Kav, was 0.55 indicating (a) medium-size peptide(s). The peptide was further purified by reverse-phase HPLC. Structural analysis revealed that it constituted the 45 NH2-terminal amino-acid residues in vitronectin. The NH2-terminal vitronectin peptide caused a 50% decrease in PAI-1 binding to the vitronectin-coated microtiter plates at a concentration of about 13 nmol/l. Thus, the peptide is a little less effective in this respect than intact vitronectin. Reduced and S-carboxymethylated peptide had no effect on the interaction. The NH2-terminal vitronectin fragment increased the stability of active PAI-1 by about 60%, which is a little less than with intact vitronectin. The peptide also prevented PAI-1 from oxidation with chloramine T. The half-life was prolonged about 4-fold as compared to about 30-fold with intact vitronectin.

Attenuation of thrombolysis by release of plasminogen activator inhibitor type-1 from platelets

Although platelets contain approximately 90% of the total amount of plasminogen activator inhibitor type-1 (PAI-1) present in blood, the functional significance of PAI-1 in platelets has been controversial. Most assessments of platelet PAI-1 have been performed with platelet lysates in which the PAI-1 derived from platelets may have been inactivated during the course of lysis. This study was performed to determine whether elaboration of PAI-1 from platelets activated physiologically by thrombolysis of pre-formed clots inhibits activation of plasminogen by tissue-type plasminogen activator (t-PA). Human whole blood clots were formed in Chandler tubes, and release of PAI-1 from platelets was quantified during and after clot formation. Subsequently, clots were placed in different Chandler tubes, and the effects of platelet PAI-1 on lysis induced by t-PA were characterized. Both the activity and concentrations of PAI-1 elaborated from platelets peaked approximately 15 min after induction of clotting. Induction of clot lysis with t-PA, 1,000 to 5,000 ng/ml, was inhibited by platelet-rich compared with platelet-poor plasma. Platelets inhibited lysis of preformed clots by t-PA and plasminogen in buffer solutions as well. Both the inhibition of clot lysis and accumulation of PAI-1 released from platelets were prevented by attenuation of thrombin-mediated activation of platelets with hirudin. Furthermore, the PAI-1 mediated inhibition was obviated by blockade of PAI-1 activity with a neutralizing monoclonal antibody to PAI-1. These results indicate that platelets inhibit clot lysis induced by t-PA by releasing functionally active PAI-1.

Serum-derived vitronectin influences the pericellular distribution of type 1 plasminogen activator inhibitor

Bovine aortic endothelial cells (BAEs) were used as a model system to study the nature and origin of protein(s) in the extracellular matrix that bind to type 1 plasminogen activator inhibitor (PAI-1). Matrix samples were fractionated by SDS-PAGE and analyzed by PAI-1 ligand binding and by immunoblotting using antibodies to vitronectin (Vn). PAI-1 bound primarily to two Vn-related polypeptides of Mr 63,000 and 57,000, and both of these partially degraded polypeptides were present in the culture serum. Radiolabeling experiments failed to detect significant Vn biosynthesis by BAEs (less than 0.03% of total), or by human umbilical vein endothelial cells and HT 1080 cells. The binding of PAI-1 to Vn was relatively specific since direct binding studies failed to demonstrate significant interactions between PAI-1 and other matrix proteins (e.g., fibronectin, type IV collagen, laminin, or matrigel). Kinetic studies indicate that PAI-1 rapidly accumulates in the matrix when BAEs are plated on Vn, appearing in the conditioned medium only after a significant lag period (1-2 h). However, no PAI-1 was detected in the matrix when the cells were plated on fibronectin-coated dishes, and there was no lag period for PAI-1 accumulation in the medium. These results indicate that PAI-1 binds specifically to serum-derived Vn in the matrix, and suggest that the composition of both the matrix and serum itself may influence the pericellular distribution of this important inhibitor.

Complex formation between plasminogen activator inhibitor 1 and vitronectin in purified systems and in plasma

Functionally active PAI-1 is bound to a discrete binding or carrier protein in plasma, which was recently identified as vitronectin. In the present study, the interaction between PAI-1 and vitronectin has been studied in purified systems and in plasma by agarose gel electrophoresis using non-denaturing conditions and by crossed immunoelectrophoresis using an antiserum produced towards purified PAI-1/vitronectin complex. Both methods revealed a clearly distinguishable complex with electrophoretic mobility in between the parent molecules. Virtually all of the purified vitronectin, which did not contain any appreciable amounts of polymerized material, and almost all of the vitronectin in plasma, had the capacity to form a complex with PAI-1. The results suggested a stoichiometry of 1:1 as the most likely ratio between the two molecules in the complex. In contrast to functionally active PAI-1, latent or chloramine T-inactivated PAI-1 did not form such a complex with vitronectin.

Purification and characterization of human plasminogen activator inhibitor type I expressed in Saccharomyces cerevisiae

The rapidly acting inhibitor of plasminogen activators, PAI-1, was produced intracellularly in Saccharomyces cerevisiae by using the ADH2 promoter to drive the expression of the human PAI-1 cDNA. Approximately 8 mg of human PAI-1 was produced per liter of confluent yeast culture. A purification scheme which resulted in 20% recovery of isolated PAI-1 from the broken yeast cell homogenate was devised. Yeast-derived human PAI-1 differs from endothelial-type PAI-1 isolated from HT1080 fibrosarcoma cells in that the recombinant inhibitor does not contain carbohydrate side chains. Nevertheless, the activity and other functional attributes of yeast-derived PAI-1 are similar to those exhibited by HT1080 fibrosarcoma cell-derived PAI-1. Hence, this study demonstrates that expression of human PAI-1 in yeast is a viable strategy for the production of ample quantities of this key modulator of plasminogen activator-mediated proteolysis.

The mechanism of the reaction between human plasminogen-activator inhibitor 1 and tissue plasminogen activator

The structural events taking place during the reaction between PAI-1 (plasminogen-activator inhibitor 1) and the plasminogen activators sc-tPA (single-chain tissue plasminogen activator) and tc-tPA (two-chain tissue plasminogen activator) were studied. Complexes were formed by mixing sc-tPA or tc-tPA with PAI-1 in slight excess (on an activity basis). The complexes were purified from excess PAI-1 by affinity chromatography on fibrin-Sepharose. Examination of the purified complexes by SDS/polyacrylamide-gel electrophoresis (SDS/PAGE) and N-terminal amino acid sequence analysis demonstrated that a stoichiometric 1:1 complex is formed between PAI-1 and both forms of tPA. Data obtained from both complexes revealed the amino acid sequences of the parent molecules and, in addition, a new sequence: Met-Ala-Pro-Glu-Glu-. This sequence is found in the C-terminal portion of the intact PAI-1 molecule and thus locates the reactive centre of PAI-1 to Arg346-Met347. The proteolytic activity of sc-tPA is demonstrated by its capacity to cleave the 'bait' peptide bond in PAI-1. The complexes were inactive and dissociated slowly at physiological pH and ionic strength, but rapidly in aq. NH3 (0.1 mol/l). Amidolytic tPA activity was generated on dissociation of the complexes, corresponding to 0.4 mol of tPA/mol of complex. SDS/PAGE of the dissociated complexes indicated a small decrease in the molecular mass of PAI-1, in agreement with proteolytic cleavage of the 'bait' peptide bond during complex-formation.