Journal of the Royal Society of Medicine Volume 85 July 1992
Recombinant natural anticoagulants:
a
407
review
P L Harper MRcp MRCPath J M Hermans BA R W Carrell FRCP FRCPath Department of Haematology, University of Cambridge, MRC Centre, Hills Road, Cambridge CB2 2QH Keywords: serpins; septic shock; recombinant antitrypsin; antithrombin; disseminated intravascular coagulation
Introduction A pressing need in haematology is an appropriate therapy for disseminated intravascular coagulation (DIC) and the shock syndromes. This WlIl most desirably require intervention at two levels; the inhibition of the contact' proteases involved in the onset of the process and the halting of the consumptive coagulation process that follows. To this end a good deal of effort is being made to design and produce appropriate engineered recombinant agents. One approach is the use of engineered- serpin inhibitors, discussed in this paper. The 'shock syndromes' and DIC The term 'shock syndrome' is' used to describe a number of pathological conditions associated with the activation of the proteolytic cascades of the blood; namely the activation of coagulation, fibrinolysis,' the kinin pathway and complement. Septic shock, secondary to gram negative septicaemia, is a classical example of this syndrome. This occurs in a 'wide variety of patients from preterm neonates with necrotizing enterocolitis to the elderly, often as a complication of a trivial problem such as a postoperative urinary tract infection. In the USA it effects 30 000 patients in every one million admissions and approximately 50% of patients diel. This is in spite of intensive care monitoring and appropriate antibiotic therapy. Many ofthe features ofseptic shock arise as a result ofeither contact activation ofthe coagulation pathway or via alterations in endothelial cell fimction. In addition to hypotension and increased vessel permeability, septic shock is almost always associated with DIC leading to microvascular thrombosis, platelet consumption and haemorrhage. A similar, but often less dramatic, clinical picture can be seein following trauma, burns, major surgery and some
inflammatory conditions, for example pancreatitis2. The primary event' in these conditions is brought about by endothelial cell damage and exposure of the subendothelial matrix. In sepsis bacterial endotoXin is likely to contribute to the endothelial cell'lysis. This has clearly been detnonstrated in bovine endothelI'ial cell culture where endotoxin leads to structural and metabolic changes causing increased permeability and complement independent cytotoxicity3. Human endothelial cells appear to be less susceptible to endotoxin damage and cell death is only seen in cell' cultures if neutrophils4 or low density lipoproteins6
Correspondence to: Dr P L Harper, Department of Haematology, West Suffolk Hospital, Hardwick Lane, Bury St Edmunds IP33 2QZ
Bradykinin release )ljo Kaltikrein
Pleklhkein
Xli
-
-c>aXI
110-
Complt acaivsion Fibrinolysis
a
Xl
..
Paper presented to Symposium on Application of
Biotechnology in Therapeutics
Xla
and Preventive IX
-*
Medicine,
IXa
4-6 December 1989
x
>X -)re
xa
C.
Pfodwombin
-*
Thombin
Fabrusogen -
IO Fibrin
Figure 1. Contact activation and the intrinsic pathway of coagulation (HMWK, high molecular weight kininogen)
are present in the culture medium. Alterations in endothelial cell function arise through the release of cytokines, in particular interleukin 1, interleukin 66 and tumour necrosis factor (INF)7 and via platelet activation8. These enhance granulocyte adhesion to the endothelium which causes the release of elastase and free radicals which in turn increases cell destruction. In addition interleukin 1 and TNF down regulate the expression of thrombomodulin and protein S receptors on the endothelium and cause endothelial cells to express tissue thromboplastin on the cell membrane, these enhance the activation of the coagulation pathway both by inhibiting protein C and by activating the extrinsic pathway of cagulation via factor VIP. The contact system involves interactions between four plasma proteins; factor XII, factor XI, prekallikrein and high-molecular-kininogen (Figure 1). The first step in this pathway is the conversion of factor XII to an activated form by the action of kallikrein. This action is greatly enhanced when factor XII is bound to a negatively charged surface. The physiological surface operating in vivo has not yet been identified, however collagen and activated platelets10 have been shown to provide suitable support for factor XII activation in in vitro models. It is likely therefore that exposed collagen following en,dothelial cell lysis initiates the contact system in man. Once activatedfactor XIa cleaves two ; namely prekallikrein to kallikrein and factor XI to factor XIa. Factor XIa triggers off the -intrinsic coagulation pathway which ultimately leads to the conversion of prothrombin to thrombin, which in turn converts fibrinogen to fibrin to form a stable clot 0141-0768/92/ (Figure 1). 070407-05/02.00/0 Kallikrein produces several of the features of the © 1992 shock syndromes in particular hypotension. This is The Royal brought about by the release of bradykinin following Society of -the proteolytic cleavage ofboth high-molecular-weight Medicine
408
Journal of the Royal Society of Medicine Volume 85 July 1992
Table 1. The major inhibitors of the intrinsic pathway of coagulation
Coagulation factor
Major inhibitor
aXIIa Kallikrein
C1 inhibitor C1 inhibitor antitrypsin antithrombin activated protein C activated protein C antithrombin
XIa Xa VIIa Va Thrombin (IIa)
and low-molecular weight kininogens. Bradykinin is and is thought to be responsible for early hypotension in sepsis, the severity and duration of which has a dramatic effect on the prognosis of septic patients. An additional action of kallikrein which exacerbates the bleeding associated with DIC is the direct activation of fibrinolysis by converting pro-urokinase to urokinase". In the normal physiological state the coagulation system is a dynamic process maintained in a homeostatic balance. On the one hand there is constant low grade activation of the clotting factors leading to thrombin formation, whilst on the other hand there is an anticoagulant system constantly inhibiting coagulation at several points (Table 1). The importance of these inhibitors is emphasized by the severity of the clinical condition associated with deficiency states, in particular congenital deficiency of antithrombin or protein C. a potent vasodilator
Antithrombin deficiency The clinical expression of either antithrombin or protein C deficiency is similar. The initial presenting event is usually pulmonary embolus or venous thrombosis of the lower limb. About 50% of affected individuals suffered their first thrombotic event before the age of 30 years and the majority of cases have had at least one thrombosis before the age of 50. The first event may be idiopathic, but in as many as 50% of cases there is an underlying precipitating factor such as surgery, pregnancy, the use of oestrogen contraceptives, immobility or trauma. A family history is common but not invariable12. The prevalence of antithrombin deficiency is in the order of 5% of all cases who present with venous thrombosis before the age of 45 years'3. Acquired antithrombin deficiency is seen far more frequently. A fall in antithrombin activity, due to consumption, is seen in most patients following surgical procedures and minor trauma, however the deficiency is most marked in DIC, particularly due to sepsis and major trauma2. In these cases there is dramatic activation of coagulation, which overwhelms the anticoagulant pathway leading to rapid consumption of C1 inhibitor, antithrombin and protein C. These deficiencies have severe consequences as the coagulation pathway is still being driven by the initial insult without any inhibitors to hold the process in check. The end result is uncontrolled fibrin formation leading to widespread microthrombi. In addition, the consumption of clotting factors, thrombocytopenia and activation of fibrinolysis leads to bleeding. The importance of antithrombin deficiency in sepsis and trauma is emphasized by several studies which have shown that antithrombin activity has prognostic
significance. In our study of the critically ill we have shown that intensive care patients with an antithrombin activity of less than 40% have a 50% mortality. Replacement antithrombin seems an appropriate first approach to stopping the uncontrolled consumptive coagulopathy seen in DIC.
Antithrombin replacement in shock Experimentally antithrombin concentrates have been studied in several animal shock models where shock was induced by either endotoxin or E. coli infusion'4 15. In all cases antithrombin infusion provided considerable protection against the lethal effects of shock. However, these animal trials have limited value for comparison with the pathological shock syndromes seen in man. Firstly, in the animal models antithrombin supplementation was administered prior to the endotoxin infusion and secondly, high doses of antithrombin, at 4-10 times the normal plasma concentration, were needed to achieve protection. Several clinical studies which demonstrate the beneficial effects of antithrombin supplementation have been reported in cases of septicaemia'6, peritonitis'7, trauma16, burns18 and pancreatic disease'7. However, these studies involved small groups of patients, many had multiple aetiologies contributing to DIC and most studies were neither controlled, nor randomized. We have recently completed a randomized study examining the effect of antithrombin concentrate in intensive care patients'9. Our results showed that antithrombin replacement had no effect on the coagulation parameters, clinical outcome or length of stay on the intensive care unit, but marginally reduced the incidence of renal failure, the benefit, however, was not sufficient to warrant the widespread use of antithrombin replacement in intensive care management. Limitations of antithrombin in the management of shock Antithrombin may prove to have a role in the management of shock; however this particular inhibitor has a number of limitations. In general all plasma proteins prepared from human plasma carry the risk oftransmitting viral infections, of which the most important are clearly HIV and hepatitis B. Although precautions are taken to screen donor samples and the concentrates themselves are heattreated it is impossible to guarantee a totally safe product. This fact alone is a major incentive to prepare recombinant proteins. A second problem common to the preparation of all plasma products is that the amount of protein produced is dependent on the supply of plasma. If a clear therapeutic indication was identified for a protease inhibitor, such as antithrombin, it is unlikely that the transfusion service would be able to cope with the increased demand. In addition to these problems antithrombin has specific characteristics which limit its value in septic shock namely; (1) it has a narrow spectrum of inhibition, (2) it requires heparin for activation and (3) it undergoes proteolytic degradation in the presence of neutrophil elastase. Narrow spectrum of inhibition The clinical features of septic shock are caused by the activation of a number of proteolytic pathways with contact activation playing a central role in this process. Antithrombin inhibits only the coagulation pathway and is a poor inhibitor of contact activation
Journal of the Royal Society of Medicine Volume 85 July 1992
and has no effect on fibrinolysis. Theoretically Cl-inhibitor would be a more logical therapeutic agent in shock as it should inhibit contact activation, reduce the release of kinins and therefore prevent hypotension. In practice, however, Cl-inhibitor used in an animal shock model did not prevent the features of shock except for a small reduction in platelet consumption20. The likely explanation for this poor therapeutic response is that, in shock there is not only activation of the intrinsic pathway via contact activation, but also activation of the extrinsic pathway via factor VII. Cl-inhibitor does not inhibit the extrinsic system. Heparin activation Antithrombin alone is a relatively poor inhibitor of thrombin, but its inhibitory effect is increased 10 000 fold in the presence of heparin. In vivo the antithrombin heparin interaction probably takes place at the endothelial cell surface. Antithrombin becomes bound to a heparin-like glycosaminoglycan, heparan sulphate, and inhibits thrombin and factor Xa locally. In septicaemia, where there is widespread endothelial damage, the action of antithrombin may be impaired. In addition under conditions of a fulminant inflammatory response, as- occurs during E. coli sepsis, the expression of heparin like receptor on the vascular endothelium may be down regulated in the same manner as the thrombomodulin and protein S receptors. This may in part explain why high concentrations of antithrombin are necessary to prevent shock in animal models. Theoretically combined antithrombin and heparin therapy should be more effective than antithrombin alone in the management of shock, but unfortunately this form of treatment did not improve the outcome in shocked patients and was associated with an increased risk of bleeding17. Proteolytic cleavage of antithrombin by neutrophil elastase During septic shock activated neutrophils release large amounts of the proteolytic enzyme elastase. This causes proteolytic degradation of antithrombin, antiplasmin and Cl-inhibitor (Figure 2).Elastase is normally inhibited by a-1-antitrypsin and therefore in shock, the proteolytic inactivation of antithrombin may be reduced by using combined antithrombin and antitrypsin replacement. This combination therapy
H202
elastase
|al-antitr003
*o
yVal, Leu
an| ArgC LysC
anti vbin h,oniibm
C1
-
1/iitor st/ \
alk.e,n
anti fin plasm,n
fibrinolysis
coagulation
complement activation
kinin release
Figure 2. The effect of neutrophil elastase on the protease inhibitors ofcoagulation. a--antifrypsin is the prime inhibitor of neutrophil elastase but is subject to inactivation by oxidation. This allows elastase to ceave and inactivate the other inhibitors shown
Table 2. The reactive centre of natural and recombinant serpins Reactive centre P2-Pl-Pl' Major targets
Antitrypsin Antithrombin Cl inhibitor
Antiplasiin
Pro-Met-Ser Gly-Arg-Ser Ala-Arg-Ser Ser-Arg-Met
Recombinants of antrysm Pittsburgh Pro-Arg-Ser Ala-Pittsburgh Ala-Arg-Ser Lys-recombinant Pro-Lys-Ser Val-recombinant Pro-Val-Ser Ala-recombinant Pro-Ala-Ser
Elastase Thrombin Kallikrein, Cl esterase Plasmin
Thrombin
Plasmin Elastase Elastase
has been used in a sheep shock model, where it was shown to reduce significantly pulmonary lymph flow, an important factor in the development ofrespiratory distress. Neither antithrombin nor a-1-antitrypsin alone had the same effect2l. From the study of the mechanisms involved in the genesis of DIC and from the results of clinical trials there is a good theoretical basis to propose that an inhibitor with the following characteristics would be superior to antithrombin in the management ofshock; namely an inhibitor that (1) has a broad spectrum of inhibitory activity (eg inhibits thrombin, kallikrein, factor XIa etc), (2) does not require heparin activation and (3) is resistant to elastase proteolysis. No single naturally occurring inhibitor has all of these properties and at first sight it appears a tall order to engineer an appropriate inhibitor. However, studies on the structure and function of the serpins have revealed that this family of proteins is eminently suitable for molecular manipulation to engineer agents with these characteristics.
Structure and function of serpins The serine protease inhibitors play an important part in the control of coagulation, fibrinolysis and complement activation. As their name suggests, they inactivate and remove enzymes capable of cleaving tissue proteins. This inhibition is brought about by the formation of a 1: 1 complex between the inhibitor and its specific substrate. The specificity of each serpin depends on the structure of the reactive centre of the molecule. The reactive centre functions by providing an ideal substrate for the target enzyme, primarily based on a single peptide bond, that between P1 and P1' amino acid residues22 (Table 2). Thus in oe-1-antitrypsin the reactive centre is a Met-Ser as neutrophil elastase characteristically cleaves at methionine residues; and for antithrombin the reactive centre is an Arg-Ser as thrombin characteristically cleaves at arginine. The demonstration that a single residue at the reactive centre defines the primary specificity of inhibition opens the possibility of engineering reactive centre mutants ofthese inhibitor with predictable activity. This potential was first demonstrated by a unique accident of nature; the formation of the Pittsburgh mutant of antitrypsin. Pittsburgh mutant of antitryp In 1978 Dr Jessica Lewis and colleages in Pittsburgh published the findings in the case of a 10-year-old boy
409
410
Journal of the Royal Society of Medicine Volume 85 July 1992
Table 3. Protease inhibitor association constants (Units are reciprocal seconds)
Elastase Thrombin Plasmin FXIIa Kallikrein
Antitrypsin
Antithrombin Antitrypsin (with heparin) Pittsburgh
6.5x107 4.8x101
cleaved
1.9x102 nil 4.2 x 101
1.0x104
2.8x107
1.6x103 2.0x105 1.3x105 3.5x 104 6.9x 104
who had a life-long history of bleeding problems. They noted that his plasma contained a protein with anticoagulant properties acting as an antithrombin. However, on antigenic studies this protein was shown to be a fraction of antitrypsin with altered electrophoretic mobility. It was proposed that this child had a methionine to arginine mutation at the reactive centre of a-1-antitrypsin accounting for the clinical findings. This prediction was confirmed by isolation and sequence studies of the abnormal electrophoretic component from the patient23. Functional studies showed that the mutation had, in effect, converted the antitrypsin to an antithrombin. There was complete loss of inhibitory activity to pancreatic elastase and a 104-fold decrease in the inhibition of neutrophil elastase. This was accompanied by a 105-fold increase in activity as a thrombin inhibitor. Later studies have shown a-i-antitrypsin Pittsburgh to be a highly effective inhibitor of plasmin, factor XIIa, and kallikrein24 (Table 3). Solving the Pittsburgh mutation confirmed that altering a single amino acid changed the inhibitory activity of the protein. The therapeutic potential ofthis mutant was also immediately recognized as this inhibitor has many of the properties we proposed for an improved agent in the management of shock.
Engineered variants Trial quantities of several antitrypsin mutants have been produced using techniques based on human cDNA clones of antitrypsin expressed outside the endoplasmic pathway to give the non-glycosylated protein25. Using these techniques it has been possible to produce a good yield of a range of ai--antitrypsin variants in both yeast and E. coli. Large scale production of the Pittsburgh mutant has been achieved and trial quantities are available. Preliminary studies using recombinant antitrypsin Pittsburgh in an animal model looked promising. Some ofthe features of septic shock were alleviated by this inhibitor, in particular there was a less dramatic fall in antithrombin, fibrinogen and factor XI in the treated animals than in the controls, however antitrypsin Pittsburgh had no effect on the haemodynamic parameters of shock or on the overall survival. At the time of this study only small quantitiesof antitrypsin Pittsburgh were available and therefore only a low dose regimen of treatment was given. It was proposed that a larger dose may improve survival. Surprisingly, however, when antitrypsin Pittsburgh was administered at a higher dose in a baboon shock model, quite the opposite effect was seen. Animals treated with antitrypsin Pittsburgh died within 6 hours of the onset of shock whereas control animals lived for atleast twice as long. The cause of these unexpected results remain unclear, but two possible contributing
factors have been identified. Firstly, during the induced shock in these animals a significant proportion of the engineered protein underwent proteolytic cleavage with the release of a chemotactic peptide which could itself enhance the inflammatory process. Secondly, antitrypsin Pittsburgh, as well as being a good inhibitor of contact factors and coagulation, also inhibits activated protein C. The inhibition of this natural anticoagulant may have a major detrimental effect in septic shock. In support ofthis is the finding that the administration of protein C antibodies exacerbate shock in animal models, whereas supplementation with activated protein C protects them.
The characterization of engineered mutants At first sight antitrypsin Pittsburgh appeared the ideal agent for the management of shock because of its broad spectrum of activity and its resistance to proteolytic cleavage by elastase. The animal studies, however, clearly demonstrate that other factors have to be considered in designing inhibitors and modifications will be necessary before these agents can have a therapeutic role. It is now clear that inhibitory specificity does not depend entirely on the single amino acid at the reactive centre but other residues around the Pl-Pi' site are involved, in particular the P2 amino acid. Alterations on the P' side of the reactive centre do not alter the main inhibitory function of the molecule but alter the kinetic characteristics of inhibition. This potentially allows fine tuning of the inhibitory features of engineered variants, by these means it should be possible to engineer a good inhibitor of contact factors and thrombin which has little effect on protein C activity. The formation of the protease-protease inhibitor complex depends on several factors, one of which appears to be the electrostatic interactions between the two molecules. Therefore, charged residues near contact points between the protease and its inhibitor are likely to be involved in the formation of a stable complex. This has recently been demonstrated using mutants of tissue plasminogen activator (tPA)N. The major inhibitor of tPA is the serpin plasminogen activator inhibitor 1 (PAM-1). A model for the interaction of tPA with PAI-1 has been proposed from studies of other protease inhibitor complexes, in particular the reaction between trypsin and bovine pancreatic trypsin inhibitor. It is predicted from this model that tPA at Arg 304 forms a salt bridge with the negatively charged glutamic acid at 350 on PAI-1. On the tPA molecule adjacent to this contact point, there is a stretch of 7 amino acids, four of which are positively charged. The predicted region of PAI-I (P4'-P9') in contact with tPA contains three negatively charged residues. It is proposed that these opposing charges help to stabilize the complex. Site directed mutants of tPA with reduced positive charge in this region are less rapidly inhibited than wild type tPA. It is likely therefore that modifications of charged residues on PAI-1 and other serpins will alter the inhibitory profile of these molecules. Future prospects There is little doubt that engineered protease inhibitors will have a place in the management of many conditions in clinical medicine and certainly the full potential of these agents has not yet been realized. Clearly in this review the emphasis has been centred on the management of shock syndromes.
Journal of the Royal Society of Medicine Volume 85 July 1992
These conditions are not common, but are associated with a high mortality and morbidity. Although the initial studies with antitrypsin Pittsburgh have proved unsuccessful, the serpins are particularly suitable for further modification and this should allow the production of a range of inhibitors, some with a broad spectrum of inhibition and others with more specific inhibitory capacity. In addition, it is likely that alterations to glycosylation will be achieved which will allow modification of half-life and tissue distribution. A particularly interesting area for the future study involves directing inhibitors to specific target organs. This can be achieved via natural binding sites on the serpins themselves, such as the fibrin binding site on antiplasmin, or by complexing the inhibitor with a monoclonal antibody. Using these techniques it should be possible to localize the action of a specific serpin. Preliminary results using antitrypsin Pittsburgh linked to a monoclonal antibody which recognizes activated platelets (GMP140) look promising. In an in vitro model the conjugate acts as a potent anticoagulant which is more efficient than antitypsin Pittsburgh alone. These proposed developments hopefully emphasize that these engineered products are more than mere replacement therapy, and will have potentially wide applications in many spheres of medicine. Acknowledgments: We wish to acknowledge the British Heart Foundation, Wellcome Trust and Medical Research
10 Walsh PN, Griffin JH. Contribution of human platelets to the proteolytic activation ofblood coagulation factors XII and XI. Blood 1981;57:106-18 11 Ichinose A, Fujikawa K, Suyama T. The activation of prourokinase by plasma kallikrein and its inactivation by thrombin. J Biol Chem 1986;261:3486-9 12 Winter JH, Fenech A, Ridley W. Familial antithrombin m deficiency. Q J Med 1982;51:29-33 13 Harper PL, Luddington RJ, Daly M, et aL The incidence of dysfunctional antithrombin variants: four cas in 210 patients with thromboembolic disease. Br J Haematol 1991;77:360-4 14 Emerson TE, Fournel TE, Leach WJ, Redens TB. Protection against intravascular coagulation and death by antithrombin m in the E. coli.ndotoxaemic rat. Circ Shock 1987;21:1-13 15 Taylor FB, Emerson TE, Chang JR, Blick KE. Antithrombin m prevents lethbal effects of E. coli infusion in baboons. Circ Shock 1988;26:227-35 16 Hellgren M, Javelin L, Hagnevik K. Antitbrombin m nntrate as an adjuvant in d med intravasula coagulation patients: a pilot study in 9 severely ill patients. Thromb Res 1984;356459-66 17 Blauhut B, Necek S, Vinazzer H, Bergman H. Substitution therapy with antitbrombin mU concentrate in shock and DIC. Thromb Ree 1982;27:271-8 18 Ono I, Ohura T, Hamamoto J. Clinical observation of co-administration of antithrombin replacement in intensive care management: the effects on mortality, coagulation and renal function. Trans Med 1991;1: 112-18 19 Harper PL, Williamson L, Luddington RJ, Park GR, Smith JK, Carrell RW. A pilot study of antithrombin replacement in intensive care management. The effects
Council for financial support. References 1 Hinshaw LB, Peduzzi P, Wilson M. The veterans administration Systemic sepsis cooperative group: effect of high dose glucocorticoid therapy on mortality in patients with clinical signs of systemic sepsis. N Engl J Med 1987;817:659 2 Bick RL. Disseminated intravascular coagulation and related syndromes: a clinical review. Semin Thromb Haemostas 1988;14:299-338 3 Harlan JM, Harker LA, Reidy MA, et aL Lipopolysaccharide mediated bovine endothelial cell injury in vitro. Lab Invest 1983;48:269-74 4 Harlan JM, Harker LA, Striker GE, Weaver LJ. Neutrophil mediated endothelial injury in vitro. Mechanisms of cell detachment. J Clin Invest 1983;68:1394-403 5 Harlan JM, Harker LA, Striker GE, Weaver LJ. Effects of lipopolysaccharide on human endothelial cells in culture. Thromb Res 1983;29:15-26 6 Bevcilacqua MP, Pober JS, Majeau GR. Interleukin 1 (IL-1) induces biosynthesis and cell surfaceexpression of procoagulant activity in human vascular endothelial cells. J Exp Med 1984;160:618 7 Cerami A, Beutler B. The role of cachetinwTNF in endotoxin shock and cachexia. Immunology Today
1988;9:28 8 Gimbrone MA, Buchanan MR. Interactions of platelets and leukocytes with vascular endothelium, in vitro studies. Ann N Y Acad Sci 1982;401:171 9 Nawroth PP, Stern DM. Modulation of endothelial cell
haemostatic properties by tumour necrosis factor. JExp Med 1986;163:740-5
20
21
22 23
24
25
26
onmortality,coagulationandr function ransfusion Med 1991;1:121-8 Triantapbyllopoulos DG, Cho MS. Effects of injection of Cl inactivator on the platelet count and coagulation in rabbits infused with endotoxin. Thromb Haemostas 1986;55:293-7 Redens TB, Leach WJ, Bogdanoff DA, Emerson TE. Synergistic protection from lung damage by combining antithrombin m and alpha-1-protease inhibitor in the E coli endotoxaemic sheep pulmonary dysfimction model. Circ Shock 1988;26:15-26 Carrell RW, Christey PB, Boswell DR. In Verstraete M, Vermylen J, Lijnen R, Arnout J, eds. Thrombosis and haematosis1987.Leuven, Blgium:Pergamo,1987:1-15 Owen MC, Brennan SO, Lewis JH, Carrell RW. Mutation of antitrypsin to antithrombin, alpha-1antitrypsin Pittsburgh (358Met-Arg), a fatal bleeding disorder. N Engi J Med 1983;309:694-8 Scott CF, Carrell RW, Glaser CB, et aL Alpha-iantitrypsin Pittsburgh-. A potent inhibitor of human plasma factor XIa, kallikrein and factor XII. J Clin Invest 1985;77:631-4 Courney M, Jallat S, Tessier LIH, et aL Synthesis in E. coli of alpha-l-antitrypsin variants of therapeutic potential for emphysema and thrombosis. Nature 1985;313:149-51 Madison EL, Goldsmith EJ, Gerard RD, Gething MH, Sambrook JF. Serpin Resistant mutants of human tissue-type plasminogen activator. Nature
1989;389:721-4 (Accepted 30 Dlecember 1991)
411
Recombinant natural anticoagulants:
a
407
review
P L Harper MRcp MRCPath J M Hermans BA R W Carrell FRCP FRCPath Department of Haematology, University of Cambridge, MRC Centre, Hills Road, Cambridge CB2 2QH Keywords: serpins; septic shock; recombinant antitrypsin; antithrombin; disseminated intravascular coagulation
Introduction A pressing need in haematology is an appropriate therapy for disseminated intravascular coagulation (DIC) and the shock syndromes. This WlIl most desirably require intervention at two levels; the inhibition of the contact' proteases involved in the onset of the process and the halting of the consumptive coagulation process that follows. To this end a good deal of effort is being made to design and produce appropriate engineered recombinant agents. One approach is the use of engineered- serpin inhibitors, discussed in this paper. The 'shock syndromes' and DIC The term 'shock syndrome' is' used to describe a number of pathological conditions associated with the activation of the proteolytic cascades of the blood; namely the activation of coagulation, fibrinolysis,' the kinin pathway and complement. Septic shock, secondary to gram negative septicaemia, is a classical example of this syndrome. This occurs in a 'wide variety of patients from preterm neonates with necrotizing enterocolitis to the elderly, often as a complication of a trivial problem such as a postoperative urinary tract infection. In the USA it effects 30 000 patients in every one million admissions and approximately 50% of patients diel. This is in spite of intensive care monitoring and appropriate antibiotic therapy. Many ofthe features ofseptic shock arise as a result ofeither contact activation ofthe coagulation pathway or via alterations in endothelial cell fimction. In addition to hypotension and increased vessel permeability, septic shock is almost always associated with DIC leading to microvascular thrombosis, platelet consumption and haemorrhage. A similar, but often less dramatic, clinical picture can be seein following trauma, burns, major surgery and some
inflammatory conditions, for example pancreatitis2. The primary event' in these conditions is brought about by endothelial cell damage and exposure of the subendothelial matrix. In sepsis bacterial endotoXin is likely to contribute to the endothelial cell'lysis. This has clearly been detnonstrated in bovine endothelI'ial cell culture where endotoxin leads to structural and metabolic changes causing increased permeability and complement independent cytotoxicity3. Human endothelial cells appear to be less susceptible to endotoxin damage and cell death is only seen in cell' cultures if neutrophils4 or low density lipoproteins6
Correspondence to: Dr P L Harper, Department of Haematology, West Suffolk Hospital, Hardwick Lane, Bury St Edmunds IP33 2QZ
Bradykinin release )ljo Kaltikrein
Pleklhkein
Xli
-
-c>aXI
110-
Complt acaivsion Fibrinolysis
a
Xl
..
Paper presented to Symposium on Application of
Biotechnology in Therapeutics
Xla
and Preventive IX
-*
Medicine,
IXa
4-6 December 1989
x
>X -)re
xa
C.
Pfodwombin
-*
Thombin
Fabrusogen -
IO Fibrin
Figure 1. Contact activation and the intrinsic pathway of coagulation (HMWK, high molecular weight kininogen)
are present in the culture medium. Alterations in endothelial cell function arise through the release of cytokines, in particular interleukin 1, interleukin 66 and tumour necrosis factor (INF)7 and via platelet activation8. These enhance granulocyte adhesion to the endothelium which causes the release of elastase and free radicals which in turn increases cell destruction. In addition interleukin 1 and TNF down regulate the expression of thrombomodulin and protein S receptors on the endothelium and cause endothelial cells to express tissue thromboplastin on the cell membrane, these enhance the activation of the coagulation pathway both by inhibiting protein C and by activating the extrinsic pathway of cagulation via factor VIP. The contact system involves interactions between four plasma proteins; factor XII, factor XI, prekallikrein and high-molecular-kininogen (Figure 1). The first step in this pathway is the conversion of factor XII to an activated form by the action of kallikrein. This action is greatly enhanced when factor XII is bound to a negatively charged surface. The physiological surface operating in vivo has not yet been identified, however collagen and activated platelets10 have been shown to provide suitable support for factor XII activation in in vitro models. It is likely therefore that exposed collagen following en,dothelial cell lysis initiates the contact system in man. Once activatedfactor XIa cleaves two ; namely prekallikrein to kallikrein and factor XI to factor XIa. Factor XIa triggers off the -intrinsic coagulation pathway which ultimately leads to the conversion of prothrombin to thrombin, which in turn converts fibrinogen to fibrin to form a stable clot 0141-0768/92/ (Figure 1). 070407-05/02.00/0 Kallikrein produces several of the features of the © 1992 shock syndromes in particular hypotension. This is The Royal brought about by the release of bradykinin following Society of -the proteolytic cleavage ofboth high-molecular-weight Medicine
408
Journal of the Royal Society of Medicine Volume 85 July 1992
Table 1. The major inhibitors of the intrinsic pathway of coagulation
Coagulation factor
Major inhibitor
aXIIa Kallikrein
C1 inhibitor C1 inhibitor antitrypsin antithrombin activated protein C activated protein C antithrombin
XIa Xa VIIa Va Thrombin (IIa)
and low-molecular weight kininogens. Bradykinin is and is thought to be responsible for early hypotension in sepsis, the severity and duration of which has a dramatic effect on the prognosis of septic patients. An additional action of kallikrein which exacerbates the bleeding associated with DIC is the direct activation of fibrinolysis by converting pro-urokinase to urokinase". In the normal physiological state the coagulation system is a dynamic process maintained in a homeostatic balance. On the one hand there is constant low grade activation of the clotting factors leading to thrombin formation, whilst on the other hand there is an anticoagulant system constantly inhibiting coagulation at several points (Table 1). The importance of these inhibitors is emphasized by the severity of the clinical condition associated with deficiency states, in particular congenital deficiency of antithrombin or protein C. a potent vasodilator
Antithrombin deficiency The clinical expression of either antithrombin or protein C deficiency is similar. The initial presenting event is usually pulmonary embolus or venous thrombosis of the lower limb. About 50% of affected individuals suffered their first thrombotic event before the age of 30 years and the majority of cases have had at least one thrombosis before the age of 50. The first event may be idiopathic, but in as many as 50% of cases there is an underlying precipitating factor such as surgery, pregnancy, the use of oestrogen contraceptives, immobility or trauma. A family history is common but not invariable12. The prevalence of antithrombin deficiency is in the order of 5% of all cases who present with venous thrombosis before the age of 45 years'3. Acquired antithrombin deficiency is seen far more frequently. A fall in antithrombin activity, due to consumption, is seen in most patients following surgical procedures and minor trauma, however the deficiency is most marked in DIC, particularly due to sepsis and major trauma2. In these cases there is dramatic activation of coagulation, which overwhelms the anticoagulant pathway leading to rapid consumption of C1 inhibitor, antithrombin and protein C. These deficiencies have severe consequences as the coagulation pathway is still being driven by the initial insult without any inhibitors to hold the process in check. The end result is uncontrolled fibrin formation leading to widespread microthrombi. In addition, the consumption of clotting factors, thrombocytopenia and activation of fibrinolysis leads to bleeding. The importance of antithrombin deficiency in sepsis and trauma is emphasized by several studies which have shown that antithrombin activity has prognostic
significance. In our study of the critically ill we have shown that intensive care patients with an antithrombin activity of less than 40% have a 50% mortality. Replacement antithrombin seems an appropriate first approach to stopping the uncontrolled consumptive coagulopathy seen in DIC.
Antithrombin replacement in shock Experimentally antithrombin concentrates have been studied in several animal shock models where shock was induced by either endotoxin or E. coli infusion'4 15. In all cases antithrombin infusion provided considerable protection against the lethal effects of shock. However, these animal trials have limited value for comparison with the pathological shock syndromes seen in man. Firstly, in the animal models antithrombin supplementation was administered prior to the endotoxin infusion and secondly, high doses of antithrombin, at 4-10 times the normal plasma concentration, were needed to achieve protection. Several clinical studies which demonstrate the beneficial effects of antithrombin supplementation have been reported in cases of septicaemia'6, peritonitis'7, trauma16, burns18 and pancreatic disease'7. However, these studies involved small groups of patients, many had multiple aetiologies contributing to DIC and most studies were neither controlled, nor randomized. We have recently completed a randomized study examining the effect of antithrombin concentrate in intensive care patients'9. Our results showed that antithrombin replacement had no effect on the coagulation parameters, clinical outcome or length of stay on the intensive care unit, but marginally reduced the incidence of renal failure, the benefit, however, was not sufficient to warrant the widespread use of antithrombin replacement in intensive care management. Limitations of antithrombin in the management of shock Antithrombin may prove to have a role in the management of shock; however this particular inhibitor has a number of limitations. In general all plasma proteins prepared from human plasma carry the risk oftransmitting viral infections, of which the most important are clearly HIV and hepatitis B. Although precautions are taken to screen donor samples and the concentrates themselves are heattreated it is impossible to guarantee a totally safe product. This fact alone is a major incentive to prepare recombinant proteins. A second problem common to the preparation of all plasma products is that the amount of protein produced is dependent on the supply of plasma. If a clear therapeutic indication was identified for a protease inhibitor, such as antithrombin, it is unlikely that the transfusion service would be able to cope with the increased demand. In addition to these problems antithrombin has specific characteristics which limit its value in septic shock namely; (1) it has a narrow spectrum of inhibition, (2) it requires heparin for activation and (3) it undergoes proteolytic degradation in the presence of neutrophil elastase. Narrow spectrum of inhibition The clinical features of septic shock are caused by the activation of a number of proteolytic pathways with contact activation playing a central role in this process. Antithrombin inhibits only the coagulation pathway and is a poor inhibitor of contact activation
Journal of the Royal Society of Medicine Volume 85 July 1992
and has no effect on fibrinolysis. Theoretically Cl-inhibitor would be a more logical therapeutic agent in shock as it should inhibit contact activation, reduce the release of kinins and therefore prevent hypotension. In practice, however, Cl-inhibitor used in an animal shock model did not prevent the features of shock except for a small reduction in platelet consumption20. The likely explanation for this poor therapeutic response is that, in shock there is not only activation of the intrinsic pathway via contact activation, but also activation of the extrinsic pathway via factor VII. Cl-inhibitor does not inhibit the extrinsic system. Heparin activation Antithrombin alone is a relatively poor inhibitor of thrombin, but its inhibitory effect is increased 10 000 fold in the presence of heparin. In vivo the antithrombin heparin interaction probably takes place at the endothelial cell surface. Antithrombin becomes bound to a heparin-like glycosaminoglycan, heparan sulphate, and inhibits thrombin and factor Xa locally. In septicaemia, where there is widespread endothelial damage, the action of antithrombin may be impaired. In addition under conditions of a fulminant inflammatory response, as- occurs during E. coli sepsis, the expression of heparin like receptor on the vascular endothelium may be down regulated in the same manner as the thrombomodulin and protein S receptors. This may in part explain why high concentrations of antithrombin are necessary to prevent shock in animal models. Theoretically combined antithrombin and heparin therapy should be more effective than antithrombin alone in the management of shock, but unfortunately this form of treatment did not improve the outcome in shocked patients and was associated with an increased risk of bleeding17. Proteolytic cleavage of antithrombin by neutrophil elastase During septic shock activated neutrophils release large amounts of the proteolytic enzyme elastase. This causes proteolytic degradation of antithrombin, antiplasmin and Cl-inhibitor (Figure 2).Elastase is normally inhibited by a-1-antitrypsin and therefore in shock, the proteolytic inactivation of antithrombin may be reduced by using combined antithrombin and antitrypsin replacement. This combination therapy
H202
elastase
|al-antitr003
*o
yVal, Leu
an| ArgC LysC
anti vbin h,oniibm
C1
-
1/iitor st/ \
alk.e,n
anti fin plasm,n
fibrinolysis
coagulation
complement activation
kinin release
Figure 2. The effect of neutrophil elastase on the protease inhibitors ofcoagulation. a--antifrypsin is the prime inhibitor of neutrophil elastase but is subject to inactivation by oxidation. This allows elastase to ceave and inactivate the other inhibitors shown
Table 2. The reactive centre of natural and recombinant serpins Reactive centre P2-Pl-Pl' Major targets
Antitrypsin Antithrombin Cl inhibitor
Antiplasiin
Pro-Met-Ser Gly-Arg-Ser Ala-Arg-Ser Ser-Arg-Met
Recombinants of antrysm Pittsburgh Pro-Arg-Ser Ala-Pittsburgh Ala-Arg-Ser Lys-recombinant Pro-Lys-Ser Val-recombinant Pro-Val-Ser Ala-recombinant Pro-Ala-Ser
Elastase Thrombin Kallikrein, Cl esterase Plasmin
Thrombin
Plasmin Elastase Elastase
has been used in a sheep shock model, where it was shown to reduce significantly pulmonary lymph flow, an important factor in the development ofrespiratory distress. Neither antithrombin nor a-1-antitrypsin alone had the same effect2l. From the study of the mechanisms involved in the genesis of DIC and from the results of clinical trials there is a good theoretical basis to propose that an inhibitor with the following characteristics would be superior to antithrombin in the management ofshock; namely an inhibitor that (1) has a broad spectrum of inhibitory activity (eg inhibits thrombin, kallikrein, factor XIa etc), (2) does not require heparin activation and (3) is resistant to elastase proteolysis. No single naturally occurring inhibitor has all of these properties and at first sight it appears a tall order to engineer an appropriate inhibitor. However, studies on the structure and function of the serpins have revealed that this family of proteins is eminently suitable for molecular manipulation to engineer agents with these characteristics.
Structure and function of serpins The serine protease inhibitors play an important part in the control of coagulation, fibrinolysis and complement activation. As their name suggests, they inactivate and remove enzymes capable of cleaving tissue proteins. This inhibition is brought about by the formation of a 1: 1 complex between the inhibitor and its specific substrate. The specificity of each serpin depends on the structure of the reactive centre of the molecule. The reactive centre functions by providing an ideal substrate for the target enzyme, primarily based on a single peptide bond, that between P1 and P1' amino acid residues22 (Table 2). Thus in oe-1-antitrypsin the reactive centre is a Met-Ser as neutrophil elastase characteristically cleaves at methionine residues; and for antithrombin the reactive centre is an Arg-Ser as thrombin characteristically cleaves at arginine. The demonstration that a single residue at the reactive centre defines the primary specificity of inhibition opens the possibility of engineering reactive centre mutants ofthese inhibitor with predictable activity. This potential was first demonstrated by a unique accident of nature; the formation of the Pittsburgh mutant of antitrypsin. Pittsburgh mutant of antitryp In 1978 Dr Jessica Lewis and colleages in Pittsburgh published the findings in the case of a 10-year-old boy
409
410
Journal of the Royal Society of Medicine Volume 85 July 1992
Table 3. Protease inhibitor association constants (Units are reciprocal seconds)
Elastase Thrombin Plasmin FXIIa Kallikrein
Antitrypsin
Antithrombin Antitrypsin (with heparin) Pittsburgh
6.5x107 4.8x101
cleaved
1.9x102 nil 4.2 x 101
1.0x104
2.8x107
1.6x103 2.0x105 1.3x105 3.5x 104 6.9x 104
who had a life-long history of bleeding problems. They noted that his plasma contained a protein with anticoagulant properties acting as an antithrombin. However, on antigenic studies this protein was shown to be a fraction of antitrypsin with altered electrophoretic mobility. It was proposed that this child had a methionine to arginine mutation at the reactive centre of a-1-antitrypsin accounting for the clinical findings. This prediction was confirmed by isolation and sequence studies of the abnormal electrophoretic component from the patient23. Functional studies showed that the mutation had, in effect, converted the antitrypsin to an antithrombin. There was complete loss of inhibitory activity to pancreatic elastase and a 104-fold decrease in the inhibition of neutrophil elastase. This was accompanied by a 105-fold increase in activity as a thrombin inhibitor. Later studies have shown a-i-antitrypsin Pittsburgh to be a highly effective inhibitor of plasmin, factor XIIa, and kallikrein24 (Table 3). Solving the Pittsburgh mutation confirmed that altering a single amino acid changed the inhibitory activity of the protein. The therapeutic potential ofthis mutant was also immediately recognized as this inhibitor has many of the properties we proposed for an improved agent in the management of shock.
Engineered variants Trial quantities of several antitrypsin mutants have been produced using techniques based on human cDNA clones of antitrypsin expressed outside the endoplasmic pathway to give the non-glycosylated protein25. Using these techniques it has been possible to produce a good yield of a range of ai--antitrypsin variants in both yeast and E. coli. Large scale production of the Pittsburgh mutant has been achieved and trial quantities are available. Preliminary studies using recombinant antitrypsin Pittsburgh in an animal model looked promising. Some ofthe features of septic shock were alleviated by this inhibitor, in particular there was a less dramatic fall in antithrombin, fibrinogen and factor XI in the treated animals than in the controls, however antitrypsin Pittsburgh had no effect on the haemodynamic parameters of shock or on the overall survival. At the time of this study only small quantitiesof antitrypsin Pittsburgh were available and therefore only a low dose regimen of treatment was given. It was proposed that a larger dose may improve survival. Surprisingly, however, when antitrypsin Pittsburgh was administered at a higher dose in a baboon shock model, quite the opposite effect was seen. Animals treated with antitrypsin Pittsburgh died within 6 hours of the onset of shock whereas control animals lived for atleast twice as long. The cause of these unexpected results remain unclear, but two possible contributing
factors have been identified. Firstly, during the induced shock in these animals a significant proportion of the engineered protein underwent proteolytic cleavage with the release of a chemotactic peptide which could itself enhance the inflammatory process. Secondly, antitrypsin Pittsburgh, as well as being a good inhibitor of contact factors and coagulation, also inhibits activated protein C. The inhibition of this natural anticoagulant may have a major detrimental effect in septic shock. In support ofthis is the finding that the administration of protein C antibodies exacerbate shock in animal models, whereas supplementation with activated protein C protects them.
The characterization of engineered mutants At first sight antitrypsin Pittsburgh appeared the ideal agent for the management of shock because of its broad spectrum of activity and its resistance to proteolytic cleavage by elastase. The animal studies, however, clearly demonstrate that other factors have to be considered in designing inhibitors and modifications will be necessary before these agents can have a therapeutic role. It is now clear that inhibitory specificity does not depend entirely on the single amino acid at the reactive centre but other residues around the Pl-Pi' site are involved, in particular the P2 amino acid. Alterations on the P' side of the reactive centre do not alter the main inhibitory function of the molecule but alter the kinetic characteristics of inhibition. This potentially allows fine tuning of the inhibitory features of engineered variants, by these means it should be possible to engineer a good inhibitor of contact factors and thrombin which has little effect on protein C activity. The formation of the protease-protease inhibitor complex depends on several factors, one of which appears to be the electrostatic interactions between the two molecules. Therefore, charged residues near contact points between the protease and its inhibitor are likely to be involved in the formation of a stable complex. This has recently been demonstrated using mutants of tissue plasminogen activator (tPA)N. The major inhibitor of tPA is the serpin plasminogen activator inhibitor 1 (PAM-1). A model for the interaction of tPA with PAI-1 has been proposed from studies of other protease inhibitor complexes, in particular the reaction between trypsin and bovine pancreatic trypsin inhibitor. It is predicted from this model that tPA at Arg 304 forms a salt bridge with the negatively charged glutamic acid at 350 on PAI-1. On the tPA molecule adjacent to this contact point, there is a stretch of 7 amino acids, four of which are positively charged. The predicted region of PAI-I (P4'-P9') in contact with tPA contains three negatively charged residues. It is proposed that these opposing charges help to stabilize the complex. Site directed mutants of tPA with reduced positive charge in this region are less rapidly inhibited than wild type tPA. It is likely therefore that modifications of charged residues on PAI-1 and other serpins will alter the inhibitory profile of these molecules. Future prospects There is little doubt that engineered protease inhibitors will have a place in the management of many conditions in clinical medicine and certainly the full potential of these agents has not yet been realized. Clearly in this review the emphasis has been centred on the management of shock syndromes.
Journal of the Royal Society of Medicine Volume 85 July 1992
These conditions are not common, but are associated with a high mortality and morbidity. Although the initial studies with antitrypsin Pittsburgh have proved unsuccessful, the serpins are particularly suitable for further modification and this should allow the production of a range of inhibitors, some with a broad spectrum of inhibition and others with more specific inhibitory capacity. In addition, it is likely that alterations to glycosylation will be achieved which will allow modification of half-life and tissue distribution. A particularly interesting area for the future study involves directing inhibitors to specific target organs. This can be achieved via natural binding sites on the serpins themselves, such as the fibrin binding site on antiplasmin, or by complexing the inhibitor with a monoclonal antibody. Using these techniques it should be possible to localize the action of a specific serpin. Preliminary results using antitrypsin Pittsburgh linked to a monoclonal antibody which recognizes activated platelets (GMP140) look promising. In an in vitro model the conjugate acts as a potent anticoagulant which is more efficient than antitypsin Pittsburgh alone. These proposed developments hopefully emphasize that these engineered products are more than mere replacement therapy, and will have potentially wide applications in many spheres of medicine. Acknowledgments: We wish to acknowledge the British Heart Foundation, Wellcome Trust and Medical Research
10 Walsh PN, Griffin JH. Contribution of human platelets to the proteolytic activation ofblood coagulation factors XII and XI. Blood 1981;57:106-18 11 Ichinose A, Fujikawa K, Suyama T. The activation of prourokinase by plasma kallikrein and its inactivation by thrombin. J Biol Chem 1986;261:3486-9 12 Winter JH, Fenech A, Ridley W. Familial antithrombin m deficiency. Q J Med 1982;51:29-33 13 Harper PL, Luddington RJ, Daly M, et aL The incidence of dysfunctional antithrombin variants: four cas in 210 patients with thromboembolic disease. Br J Haematol 1991;77:360-4 14 Emerson TE, Fournel TE, Leach WJ, Redens TB. Protection against intravascular coagulation and death by antithrombin m in the E. coli.ndotoxaemic rat. Circ Shock 1987;21:1-13 15 Taylor FB, Emerson TE, Chang JR, Blick KE. Antithrombin m prevents lethbal effects of E. coli infusion in baboons. Circ Shock 1988;26:227-35 16 Hellgren M, Javelin L, Hagnevik K. Antitbrombin m nntrate as an adjuvant in d med intravasula coagulation patients: a pilot study in 9 severely ill patients. Thromb Res 1984;356459-66 17 Blauhut B, Necek S, Vinazzer H, Bergman H. Substitution therapy with antitbrombin mU concentrate in shock and DIC. Thromb Ree 1982;27:271-8 18 Ono I, Ohura T, Hamamoto J. Clinical observation of co-administration of antithrombin replacement in intensive care management: the effects on mortality, coagulation and renal function. Trans Med 1991;1: 112-18 19 Harper PL, Williamson L, Luddington RJ, Park GR, Smith JK, Carrell RW. A pilot study of antithrombin replacement in intensive care management. The effects
Council for financial support. References 1 Hinshaw LB, Peduzzi P, Wilson M. The veterans administration Systemic sepsis cooperative group: effect of high dose glucocorticoid therapy on mortality in patients with clinical signs of systemic sepsis. N Engl J Med 1987;817:659 2 Bick RL. Disseminated intravascular coagulation and related syndromes: a clinical review. Semin Thromb Haemostas 1988;14:299-338 3 Harlan JM, Harker LA, Reidy MA, et aL Lipopolysaccharide mediated bovine endothelial cell injury in vitro. Lab Invest 1983;48:269-74 4 Harlan JM, Harker LA, Striker GE, Weaver LJ. Neutrophil mediated endothelial injury in vitro. Mechanisms of cell detachment. J Clin Invest 1983;68:1394-403 5 Harlan JM, Harker LA, Striker GE, Weaver LJ. Effects of lipopolysaccharide on human endothelial cells in culture. Thromb Res 1983;29:15-26 6 Bevcilacqua MP, Pober JS, Majeau GR. Interleukin 1 (IL-1) induces biosynthesis and cell surfaceexpression of procoagulant activity in human vascular endothelial cells. J Exp Med 1984;160:618 7 Cerami A, Beutler B. The role of cachetinwTNF in endotoxin shock and cachexia. Immunology Today
1988;9:28 8 Gimbrone MA, Buchanan MR. Interactions of platelets and leukocytes with vascular endothelium, in vitro studies. Ann N Y Acad Sci 1982;401:171 9 Nawroth PP, Stern DM. Modulation of endothelial cell
haemostatic properties by tumour necrosis factor. JExp Med 1986;163:740-5
20
21
22 23
24
25
26
onmortality,coagulationandr function ransfusion Med 1991;1:121-8 Triantapbyllopoulos DG, Cho MS. Effects of injection of Cl inactivator on the platelet count and coagulation in rabbits infused with endotoxin. Thromb Haemostas 1986;55:293-7 Redens TB, Leach WJ, Bogdanoff DA, Emerson TE. Synergistic protection from lung damage by combining antithrombin m and alpha-1-protease inhibitor in the E coli endotoxaemic sheep pulmonary dysfimction model. Circ Shock 1988;26:15-26 Carrell RW, Christey PB, Boswell DR. In Verstraete M, Vermylen J, Lijnen R, Arnout J, eds. Thrombosis and haematosis1987.Leuven, Blgium:Pergamo,1987:1-15 Owen MC, Brennan SO, Lewis JH, Carrell RW. Mutation of antitrypsin to antithrombin, alpha-1antitrypsin Pittsburgh (358Met-Arg), a fatal bleeding disorder. N Engi J Med 1983;309:694-8 Scott CF, Carrell RW, Glaser CB, et aL Alpha-iantitrypsin Pittsburgh-. A potent inhibitor of human plasma factor XIa, kallikrein and factor XII. J Clin Invest 1985;77:631-4 Courney M, Jallat S, Tessier LIH, et aL Synthesis in E. coli of alpha-l-antitrypsin variants of therapeutic potential for emphysema and thrombosis. Nature 1985;313:149-51 Madison EL, Goldsmith EJ, Gerard RD, Gething MH, Sambrook JF. Serpin Resistant mutants of human tissue-type plasminogen activator. Nature
1989;389:721-4 (Accepted 30 Dlecember 1991)
411
