Congenital alpha(2)-plasmin inhibitor deficiencies: a review.

Abstract

Plasmin is generated from its precursor plasminogen by plasminogen activators. To avoid excessive proteolysis and tissue damage, a precise, co-ordinated regulation of the plasminogen activators and plasmin is required. Alpha2-plasmin inhibitor (α2-PI) was isolated from human plasma successively by two groups (Moroi & Aoki, 1976; Wiman & Collen, 1977). As shown in Fig 1, it was rapidly apparent that α2-PI (also called α2-anti-plasmin) was the most important inhibitor of plasmin (Collen, 1976; Aoki et al, 1977). The human gene was sequenced in 1988 (Hirosawa et al, 1988) and the murine gene in 1997 (Okada et al, 1997). Knock-out mice have recently been described (Lijnen et al, 1999, 2000). The discovery of several genetic abnormalities with the recent elucidation of some biochemical consequences prompted us to review the congenital deficiencies of α2-PI deficiencies. Place of α2-plasmin inhibitor in the fibrinolytic pathway. t-PA, tissue-type plasminogen activator; PAI-1, plasminogen activator inhibitor 1; α2-PI, alpha2-plasmin inhibitor; α2-M, α2-macroglobulin; α1-AT, α1-anti-trypsin. In this paper we will focus on (a) the α2-PI gene and protein, (b) the main molecular abnormalities found in congenital deficiencies, and (c) some clinical aspects of α2-PI congenital deficiencies. The human α2-PI gene contains 10 exons and nine introns spanning approximately 16 kilobases of DNA. There is a polymorphism caused by the presence of a minor allele with a deletion of about 720 base pairs in intron 8 of the gene (Miura et al, 1989a). Other polymorphisms have been described (Lind & Thorsen, 1999). All introns are located in the 5′-half of the corresponding mRNA. The 5′-untranslated region and the leader sequence are interrupted by two introns. A TATA box sequence is located 17 nucleotides upstream from the transcription initiation site. The NH2 terminal region is encoded by the 4th exon. The reactive site and the plasminogen-binding site are encoded by the 10th exon (Fig 2). Homologies to α1-anti-trypsin, anti-thrombin and angiotensinogen suggest that the α2-PI gene originated from a common ancestor of serine proteinase inhibitors (serpin). Gene, mRNA and protein of alpha2-plasmin inhibitor. Using polymerase chain reaction (PCR) analysis of rodent human hybrid panels, Welch & Francke (1992) assigned α2-PI to human chromosome 17, more precisely in 17pter–p12. Kato et al (1993) mapped the α2-PI gene to 17p13 using Southern blot analysis of somatic cell hybrids and fluorescence in situ hybridization studies using a genomic DNA and cDNA probes. α2-PI was shown to be present in human liver parenchymal cells using the immunofluorescence technique and plasma α2-PI levels are reduced to as low as 8% of normal in severe liver diseases (Aoki & Yamanaka, 1978), suggesting that α2-PI is mainly produced in the liver. Saito et al (1982) observed synthesis and secretion of α2-PI by established human liver cell lines. Murine α2-PI is also synthesized in the liver (Okada et al, 1997), but other potential sites of production have been found. Murine mRNA has been detected by in situ hybridization in epithelial cells lining the convoluted portion of proximal tubules (Menoud et al, 1996). This kidney production is modulated by testosterone. Human kidneys (particularly the cortical region) also contain high levels of this transcript. Moderate amounts of α2-PI mRNA are present in other murine tissues such as muscle, intestine, central nervous system and placenta. The functional relevance of this production remains unknown. Recently, mice in which the α2-PI gene has been specifically inactivated have been obtained (Lijnen et al, 1999). After amputation of tail or toe tips, bleeding stopped spontaneously in the wild-type mice and also in the two mutated genotypes. Spontaneous lysis of [125I]-fibrin-labelled plasma clots injected intravenously was significantly higher in α2-PI−/– homozygotes than in wild-type mice after 4 h. In another set of experiments, fibrin deposition in kidneys was reduced in α2-PI−/– mice compared with normal mice 4–8 h after endotoxin injection. Mutant mice had higher fibrinolytic activity, as measured by zymography, 1 week after a femoral arterial injury (Lijnen et al, 2000). Nevertheless, after 2 and 3 weeks, it was comparable in all genotypes. Thus, the α2-PI−/– mice have a transient enhanced endogenous fibrinolytic activity. In spite of this enhanced fibrinolytic activity, the α2-PI−/– mice do not have an overt bleeding tendency, in contrast to α2-PI-deficient patients, who have a severe haemorrhagic tendency as a result of late bleeding. Regulatory mechanisms of thrombolysis in the mouse may be somewhat different from those in humans. α2-PI is a single-chain glycoprotein which contains about 13% of carbohydrate (Moroi & Aoki, 1976). Depending on the methods used, its molecular weight varies between 60 and 70 kDa. The mature human protein secreted from cells is composed of 464 amino acids with the NH2-terminal Met (Met-form) (Bangert et al, 1993). The mature protein loses its amino terminal 12 residues in blood plasma during circulation and is converted to the protein with the NH2-terminal Asn (Asn-form) (Koyama et al, 1994). The Asn-form is present in plasma as 60–70% of the total α2-PI (Bangert et al, 1993). α2-PI contains four cysteine residues but only one S–S bridge is present (Christensen et al, 1997). Reduction of S–S bridge, however, was not associated, with any change of inhibitory activity of plasmin (Moroi & Aoki, 1976), indicating that the S–S bridge may have no functional role. The reactive site, reacting with the active centre of plasmin, consists of Arg376–Met377 (according to amino acid numbering of Met-form) (Holmes et al, 1987a). In addition to inhibition of plasmin proteolytic activity, there are two characteristic functions: inhibition of plasmin(ogen) binding to fibrin and covalent binding (cross-linking) to fibrin. α2-PI has a strong affinity for plamin(ogen) and non-covalently binds to the site called the lysine-binding site (LBS) of plasminogen (Wiman et al, 1979). LBSs are the sites at which fibrin is also non-covalently bound. Hence, α2-PI competitively inhibits the binding of plasminogen to fibrin (Moroi & Aoki, 1977). Kluft & Los (1981) found a non-plasminogen-binding form, lacking a peptide in the carboxyl-terminal region that contains the plasminogen binding site (Sasaki et al, 1986). This non-plasminogen-binding form, converted from the native form and present as a minor component in plasma (Kluft et al, 1986), exhibits less potent inhibitory activity on fibrinolysis (Clemmensen et al, 1981). When blood clots, part of α2-PI present in plasma is rapidly covalently bound (cross-linked) to a fibrin α-chain by activated factor XIII (Sakata & Aoki, 1980). Cross-linking of α2-PI to fibrin endows fibrin with resistance to fibrinolysis (Sakata & Aoki, 1982; Jansen et al, 1987; Lee et al, 1999). The cross-linking site is located at Gln 14, the second residue from the amino terminus of Asn-form (Tamaki & Aoki, 1982). The Asn-form has a physiologically more potent inhibitory activity on fibrinolysis because of its increased capacity of cross-linking to fibrin compared with the Met-form (Sumi et al, 1989). α2-PI plasma concentration is 0·7 ± 0·06 mg/l. Its half-life is 2·6 d, whereas the half-life is 0·5 d for the plasmin–α2-PI complex (Collen & Wiman, 1979). α2-PI is present at low concentrations in the α-platelet granules (Mui et al, 1975; Plow & Collen, 1981), constituting only 0·05% of α2-PI in the whole blood. Figure 2 illustrates the three functional sites of α2-PI. Two are located in the carboxyl terminal part (the reactive site and the plasminogen-binding site), the third one (the cross-linking site) is found in the amino-terminal part. Plasmin inhibition by α2-PI is a very fast reaction. First, there is a rapid formation of a reversible complex between the LBS of plasmin and the plasminogen-binding site in α2-PI. Next, plasmin cleaves the reactive site (Arg 376–Met 377) of α2-PI, resulting in a covalent plasmin–α2-PI complex and the release of a peptide. Plasmin is also inhibited at the clot level by α2-PI when factor XIII catalyses covalent binding (cross-linking) of α2-PI with the alpha chains of fibrin. In human plasma, plasmin is inhibited with a half-life of approximately 0·1 s by α2-PI. In contrast, the rate of inhibition of fibrin-bound plasmin by free α2-PI is two orders of magnitude slower (Wiman & Collen, 1977). This is one way in which the activity of plasmin generated by thrombolytic therapy is targeted towards fibrin deposits, although fibrin-bound plasmin is initially inhibited by fibrin-bound α2-PI. In addition to inhibition of plasmin, α2-PI inhibits plasminogen activators (Moroi & Aoki, 1976; Rijken et al, 1983). The inhibition of fibrin-bound tPA by fibrin-bound α2-PI may significantly contribute to the inhibition of thrombolysis (Robbie et al, 2000). Congenital deficiencies of α2-PI are rare, the real prevalence being unknown. The transmission is autosomal recessive. The five cases published to date are summarized in Table I. Four allelic variants concern homozygous patients. To be certain that the mutations were responsible for the deficiencies and to elucidate their mechanisms, expression vectors containing the mutated sequence were transfected into Cos 7 or Chinese hamster ovary (CHO) cells for transient expression in four of the five variants: α2-PI Enschede (Holmes et al, 1987b), α2-PI Okinawa (Miura et al, 1989b), α2-PI Nara (Miura et al, 1989c; Miura & Aoki, 1990) and α2-PI Paris Trousseau (Yoshinaga et al, 2000). For α2-PI Nara and α2-PI Okinawa, the mutant proteins were found to be retained in the rough endoplasmic reticulum (RER) and degraded within the cells, with only a small proportion of proteins secreted. Normally, the proteins synthesized in the RER are glycosylated by the addition of a high mannose carbohydrate chain. During its transit to the Golgi apparatus, the CHO side chain is converted to a complex type of oligosaccharides. Recently, Chung et al (2000) have studied the degradation of these mutants and found that these two mutated proteins, instead of going to the Golgi, are degradated by proteosomes. Modification of mannose residues in the misfolded proteins probably impairs an efficient binding to calnexin, a RER resident protein that allows the transfer to the Golgi. In the mutant α2-PI Paris Trousseau, a short protein is secreted. This impaired secretion is as a result of a splicing donor mutation. For the last quantitative variant, the authors (Lind & Thorsen, 1999) had strong but indirect evidence for the implication of the mutation because its biochemical consequence has not yet been elucidated. Finally, the qualitative variant α2-PI Enschede (Holmes et al, 1987b) consists of an alanine insertion near the active site region of the molecule. This abnormal α2-PI is converted from an inhibitor of plasmin to a substrate. There is no simple coagulation assay that raises suspicion of the diagnosis of PI deficiency. Therefore, a specific PI assay should be performed when a patient has a bleeding diathesis that the usual screening tests do not identify. Based on the results of functional or immunological assays, two types of biological deficiencies can be recognized: type I (quantitative), defined by a similar decrease of α2-PI by both types of assays, and type II (qualitative), in which there is a discrepancy between the activity which is low and the antigen concentration which is normal. Koie et al (1978) and Aoki et al (1979) reported the first case of homozygous deficiency. The patient was a 25-year-old man who had suffered from haemorrhages (haemothorax, haemarthrosis, gingival bleeding, ecchymoses, as well as prolonged bleeding after minor trauma) since childhood. Familial study revealed many consanguineous marriages. Analysis of fibrinolytic state of the patient revealed that his bleeding tendency was caused by premature dissolution of haemostatic plugs before the restoration of injured vessels (Aoki et al, 1980, 1983). Thrombi formed in vivo including haemostatic plugs are subject to physiologically occurring endogenous fibrinolysis caused by fibrin-associated plasminogen activation. This fibrinolytic process is normally hampered by the presence of α2-PI. Since then, at least 13 cases have been reported (Table II). Most bleeds are severe, appear during childhood and, in a few cases, umbilical bleeding is the first manifestation. Nevertheless, some homozygous patients seem to present only moderate bleeding. A very unusual localization of bleeding, intramedullary haematoma in the diaphyses of long bones, has been described in four cases (Takahashi et al, 1991; Devaussuzenet et al, 1998). Radiography indicates homogeneous hyperlucent lesions with regular limits and without marginal sclerosis, which can be difficult to distinguish from cystic fibrous dysphasia, Langerhans cell histocytosis or metastasis of neuroblastoma. Accurate diagnosis of these intramedullary haematomas can be performed with magnetic resonance imaging which shows a homogeneous hyperintense signal in the medulla and a hypointense signal surrounding the lesion. Similar bone haematomas have been also described in afibrinogenaemia (Lagier et al, 1980). The real importance of bleeding in heterozygous cases is a matter of controversy. The majority of heterozygous subjects discovered in familial studies of homozygous patients and isolated reported cases (Stormorken et al, 1983; Knot et al, 1986) have no bleeds. However, a few cases of congenital α2-PI deficiency seem to have bleeding complications. They occur in response to conditions such as trauma, surgery and dental extraction (Leebeek et al, 1988). Two reports have described severe bleeding in heterozygous subjects (Kordich et al, 1985; Griffin et al, 1993). The first clinical events appeared in adults (33–45-years-old) but, in one case, bleeding after surgery was reported in a 19-month-old baby (Griffin et al, 1993). The bleeding tendency may increase with age. One heterozygous patient aged 83 years had haemorrhagic episodes repeatedly after the age of 79 years although he had not exhibited any bleeding tendency previously (Ikematsu et al, 1996). Intramedullary haematomas have never been described in heterozygotes. To explain why some homozygous subjects have moderate bleeds whereas some heterozygous subjects have severe bleeds, it can be hypothesized that other abnormalities such as a von Willebrand disease or mutations R506Q of factor V or G20210A of the prothrombin are associated, as it has been described, for example, to eventually explain the variable clinical expression in haemophilia A (Nichols et al, 1996). Anti-fibrinolytic agents (tranexamic acid or ε-aminocaproic acid) are used to treat patients who bleed or to avoid haemorrhagic complications in those who are undergoing surgical interventions. These agents prevent the binding of plasminogen to fibrin and thereby inhibit endogenous fibrinolysis and stabilize the haemostatic plug (Aoki et al, 1983) or increase the inhibitory activity of α2-macroglobulin (Takada & Takada, 1980). These agents can be administrated either orally or intravenously. Few side-effects (essentially, headaches) have been described. The opinion differs among authors about the best dosage and duration of their use. For a bleed (Ikematsu et al, 1996), intravenous infusion of tranexamic acid 1 g/d for 6 d has been administered followed by oral intake of 1·5 g/d. The pre- and perioperative treatment of intramedullary haematoma has been reported by two groups (Takahashi et al, 1991; Devaussuzenet et al, 1998). Tranexamic acid alone (250–500 mg/d) was effective in relieving symptoms. If the pain does not disappear, the intramedullary haematomas are curetted by surgery in conjunction with a perioperative intravenous administration of tranexamic acid (40 mg/kg/d) from the day before the operation until 2 weeks post operation. Fibrin glue and tranexamic acid or hydroxyapatite can be instilled into the lesion (Miyauchi et al, 1996). Fresh-frozen plasma (FFP) can be an alternative to anti-fibrinolytic agents. Yoshioka et al (1982) compared the efficacy of tranexamic acid with FFP for the preoperative management of a dental extraction in the same patient. Concentration and activity of α2-PI reached 15·6–19% immediately at the end of FFP infusion and diminished gradually afterwards. The half-life of concentration and activity was 35·5 and 21 h, respectively, for the infused α2-PI. Absence of α2-PI catabolism was shown by crossed immunoelectrophoresis. This result is different after tranexamic infusion but the clinical efficacy of the two procedures are identical. Nevertheless Kluft et al (1982) described an ineffectiveness of FFP in a homozygous subject owing to some differences in clinical condition, i.e. either a reduced factor XIII or an extensive consumption of α2-PI after transfusion. In a case of open heart surgery, some authors have successfully corrected the deficiency preoperatively using FFP (Shahian & Levine, 1990). The reason for this perioperative strategy was the unknown ability of anti-fibrinolytic agents to control the combined hyperfibrinolytic states (α2-PI deficiency and extra corporeal circulation). Anti-fibrinolytic agents seem an effective therapy for the prevention or the treatment of bleeding in contrast to FFP. It should be mentioned that the procedure of preparation of plasma can play a role because it has been recently demonstrated that, in solvent-/detergent-treated plasma, 100% of α2-PI are either in the latent or the polymerized conformation and therefore lack inhibitory activity (Mast et al, 1999). Therefore, if it is possible, tranexamic acid should be given instead of FFP, particularly as FFP carries a potential danger of disease transmission. In summary, the discovery of several genetic abnormalities with the recent elucidation of some biochemical consequences has renewed the interest in α2-PI and has enabled a more precise understanding of the mechanisms leading to some congenital deficiencies. This could lead to new therapeutic approaches, including gene therapy.