CONTENTS

ORIGINAL ARTICLES 6

ABBREVIATIONS 7

1. INTRODUCTION 8

1.1. The complement system 8

1.1.1. Functions of the complement system 8

1.1.2. Activation pathways of complement 10

1.1.3. Soluble regulators 11

1.1.4. Membrane regulators 12

1.1.5. Removal of membrane attack complexes by vesiculation and

endocytosis 16

1.2. Complement activation in myocardial infarction 16

1.2.1. Complement activation products in infarcted myocardium 16

1.2.2. Triggering of complement activation 17

1.2.3. Inhibition of complement-mediated injury 17

1.3. Complement and atherosclerosis 18

2. AIMS OF THE STUDY 19

3. MATERIALS AND METHODS 20

3.1. Tissue samples and cell culture 20

3.1.1. Autopsy specimens 20

3.1.2. Blood samples 20

3.1.3. Experimental myocardial infarction 20

3.1.4. Cell culture 21

3.2. Microscopical and immunohistochemical methods 21

3.2.1. Microscopical and histochemical diagnosis of myocardial infarction 21

3.2.2. Immunofluorescence staining 21

3.2.3. Immunoperoxidase staining 22

3.2.4. Electron microscopy 22

3.3. Sodium dodecyl sulphate -polyacrylamide gel electrophoresis and

immunoblotting analysis 24

3.4. Isolation of protectin from normal human heart tissue, erythrocytes and

urine 24

3.5. Protein and lipoprotein analysis 25

3.5.1. Purification of lipoproteins 25

3.5.2. Quantification of proteins, apolipoproteins and lipids 25

3.6. Phospholipase C and neuraminidase treatment of tissues 26

3.7. Sucrose density gradient ultracentrifugation 26

3.8. Gel filtration and anti-apolipoprotein A-I affinity chromatography 26

3.9. Incorporation of protectin into cell membranes 27

3.10. Transfer of protectin between cells and lipoproteins 27

4. RESULTS 28

4.1. Immunohistochemical detection of complement components in human

myocardium (I, II) 28

4.2. Demonstration of MAC in sarcolemmal membranes of infarcted

myocardium by electron microscopy (II) 28

4.3. Expression and deposition of complement regulators in normal and

infarcted human myocardium 29

4.3.1. Expression of protectin (I) 29

4.3.2. Expression of other membrane regulators of complement in

myocardium (II) 30

4.3.3. C4bp, vitronectin and clusterin (II, III) 30

4.4. Structural and functional properties of myocardial protectin (I) 31

4.5. An experimental rat model for myocardial infarction (IV) 31

4.5.1. Complement components in normal and infarcted rat myocardium 31

4.5.2. Protectin in normal rat myocardium 32

4.5.3. Protectin in infarcted rat myocardium 32

4.6. Interactions between glycolipid-anchored protectin and plasma

lipoproteins (V) 33

4.6.1. Incorporation of phospholipid-tailed protectin into lipoprotein

particles 33

4.6.2. Transfer of protectin between cells and lipoprotein particles 33

4.6.3. Protectin in high density lipoprotein particles 34

5. DISCUSSION 35

5.1. Time course of complement activation in myocardial infarction 35

5.2. Specific features of complement activation in myocardial infarction 35

5.3. Complement activating factors in ischemic injury 36

5.4. Regulation of complement activation in human myocardium 36

5.4.1. Expression of complement membrane regulators 36

5.4.2. Molecular and functional properties of myocardial protectin 37

5.4.3. Protectin in myocardial infarction 37

5.4.4. Vitronectin and clusterin in myocardial infarction - clearance of cell

debris from injured myocardium? 39

5.5. Interaction between protectin and lipoproteins 39

5.6. A synopsis of the role of the complement system in the pathogenesis

of myocardial infarction 40

6. Summary and concluding remarks 43

7. ACKNOWLEDGMENTS 45

8. REFERENCES 47

ORIGINAL ARTICLES

This thesis is based on the following original publications:

I Väkevä A, Laurila P and Meri S. Loss of expression of protectin (CD59) is associated with complement membrane attack complex deposition in myocardial infarction. Lab Invest 1992, 67: 608-616

II Väkevä A, Laurila P and Meri S. Regulation of complement membrane attack complex formation in myocardial infarction. Am J Pathol 1993, 143: 65-75

III Väkevä A, Laurila P and Meri S. Co-deposition of clusterin with the complement membrane attack complex in myocardial infarction. Immunology 1993, 80: 177-182

IV Väkevä A, Morgan BP, Tikkanen I, Helin K, Laurila P and Meri S. Time course of complement activation and inhibitor expression after ischemic injury of rat myocardium. Am J Pathol 1994, 144: 1357-1368

V Väkevä A, Jauhiainen M, Ehnholm C, Lehto T and Meri S. High-density lipoproteins can act as carriers of glycophosphoinositol lipid-anchored CD59 in human plasma. Immunology 1994, 82: 28-33

ABBREVIATIONS

AP alternative pathway of complement

AMI acute myocardial infarction

apo A-I apolipoprotein A-I

apo B-100 apolipoprotein B-100

B factor B

BSA bovine serum albumin

C complement

C4bp C4b binding protein

C8bp C8 binding protein

CD59E erythrocyte CD59

CD59H myocardial CD59

CD59U urine CD59

CP classical pathway of complement

CR1 complement receptor type 1

CR3 complement receptor type 3

D factor D

DAF decay accelerating factor

FITC fluorescein isothiocyanate

GPI glycophosphoinositol

H factor H, ß1H-globulin

HDL high density lipoprotein

I factor I, C3b inactivator

IFL immunofluorescence microscopy

IP immunoperoxidase

kDa kilodalton

LDL low density lipoprotein

mAb monoclonal antibody

MAC membrane attack complex of complement

MCP membrane cofactor protein

Mr relative molecular weight

NBT nitroblue tetrazolium

NHS normal human serum

PAGE polyacrylamide gel electrophoresis

PBS phosphate buffered saline

PIPLC phosphatidylinositol-specific phospholipase C

PNH paroxysmal nocturnal hemoglobinuria

SDS sodium dodecyl sulphate

TRITC tetramethylrhodamine isothiocyanate

1. INTRODUCTION

1.1. The complement system

1.1.1. Functions of the complement system

The human complement (C) system (Fig. 1) is composed of about 20 components in blood plasma (Table 1) and 10 regulators or receptors on cell membranes. The main functions of C are to eliminate invading foreign cells and to mediate an inflammatory reaction. When activated C may also cause injury to host cells. The activation of C leads to several biological responses: chemotaxis of leukocytes, opsonization, increased vascular permeability and target cell lysis. The C anaphylatoxins, C3a, C4a and C5a, are C activation products that cause increased vascular permeability, smooth muscle contraction, release of histamine from basophils and mast cells, migration and activation of monocytes, macrophages and polymorphonuclear leukocytes (Hugli, 1984; Vogt, 1985a). The C3b activation product and its degradation fragment iC3b are important opsonins that mark target cells for elimination by phagocytosis (Ross and Medof, 1985) .

Fig. 1. A schematic outline of complement activation. Regulators of complement are boxed in rectangles. B = factor B. D = factor D. H = factor H. I = factor I (C3b inactivator). For explanation of other abbreviations see Table 3. Enzymatic activity is indicated by .

Table 1. Proteins involved in complement activation

Classical pathway

Protein	Molecular weight (kDa)	Serum concentration (µg/ml)
C1q	410	70
C1r	85	34
C1s	85	31
C4	206	600
C2	95	25
C3	195	1200

Alternative pathway

Protein	Molecular weight (kDa)	Serum concentration (µg/ml)
Properdin	153	25
Factor B	100	225
Factor D	25	1

Terminal pathway

Protein	Molecular weight (kDa)	Serum concentration (µg/ml)
C5	180	85
C6	128	60
C7	120	55
C8	150	55
C9	79	60

Activation of the terminal pathway of C leads to formation of the membrane attack complex (MAC; Müller-Eberhard, 1986). MAC consists of C5b, C6, C7, C8 and several C9 molecules that form a macromolecular aggregate on a target cell cell membrane (Müller-Eberhard, 1986; Esser, 1990; Fig. 2). The C5b-9 complex may contain up to 18 C9 molecules that have become polymerized into a barrel-like structure. Formation of the MAC causes changes in the permeability of the target cell membrane and finally osmotic cell lysis. It has been suggested that MAC-mediated cell injury is principally caused by the influx of calcium through the MAC channels (Campbell et al., 1981) . Although the C5b-8 complex is capable of causing lysis of certain cells, the lytic capacity of the C5b-9 complex is much greater (Morgan et al., 1986; Martin et al., 1987) . Sublytic amounts of MAC can also activate target cells (e.g. neutrophils, endothelial and epithelial cells) to secrete inflammatory mediators, such as reactive oxygen metabolites, prostaglandins and thromboxanes (Morgan, 1989a) .

The C system is involved in the pathogenesis of many diseases. Complement has been shown to have a direct role in the development of tissue injury in experimental models of autoimmune diseases like myasthenia gravis, experimental allergic alveolitis, Heymann's nephritis, immune complex-induced vasculitis and collagen-induced arthritis (Morgan, 1991). In addition, C-mediated tissue injury is also involved in disorders caused by different types of "nonspecific" tissue injury, such as burns or infarction (Morgan, 1991).

Fig. 2. Sequential assembly of the complement membrane attack complex (MAC). Formation of MAC is initiated by binding of C6 to C5b generated by one of the C5 convertase enzymes (A). After binding of C7 to C5b6 the C5b67 complex can insert into a nearby cell membrane (B). Thereafter one C8 and one or more C9 molecules can bind to the complex and insert into the cell membrane (C-E). Formation of the C5b-8 complex leads to permeability changes on the cell membrane, but formation of poly-C9 markedly increases the lytic capacity of the MAC by creating a transmembrane pore.

1.1.2. Activation pathways of complement

Activation of the complement system occurs via the classical and/or the alternative pathway in a sequential manner by proteolytic cleavages and association of inert precursor molecules (Müller-Eberhard, 1988; Fig. 1). Activation of the classical pathway (CP) is triggered by antibodies, whereas the alternative pathway (AP) is activated directly by the target surface. The central event in both pathways is the generation of the C3 convertase. The CP and AP converge at the level of C3 and C5 and continue along a common terminal pathway (Fig. 2). Activation of the CP is initiated by binding of C1q, a part of the first component of C (C1), to the Fc portions of immunoglobulins. This leads to activation of the C1r and C1s esterases which are other subcomponents of C1. Activated C1s cleaves C4 and C2 to generate the CP C3/C5 convertase C4b2a. Activation of the CP can also be triggered without antibodies, e.g. by complexes of C-reactive protein with polycations (Kaplan and Volanakis, 1974) , bacterial lipopolysaccharides (Clas et al., 1985) , certain viruses (Cooper et al., 1976; Hirsch et al., 1980) and by intracellular components - like mitochondrial membranes (Pinckard et al., 1973) and cytoskeletal intermediate filaments (Linder et al., 1979) . Association of mannan bind-ing protein (MBP) with the cell surface of a microbe can also initiate activation of the CP independently of C1q (Lu et al., 1990; Ohta et al., 1990) . In this route, binding of the C1r2C1s2 complex or of a novel serine esterase called MASP (Matsushita and Fujita, 1992) to MBP mediates activation of the CP.

The alternative pathway of C activation is a phylogenetically older defence mechanism than the CP. The initiation of the AP is based on the spontaneous hydrolysis of the internal thioester bond of C3. C3 has two molecular forms: a native form that possesses an intact thioester bond and an altered form, iC3, with a hydrolyzed internal thioester bond. iC3 can form a fluid phase C3 convertase, that may initiate activation of the AP. C3b components generated thereafter can bind to target surfaces and form new C3bBb convertases, which results in amplification of AP activation (the amplification loop). The C3bBb convertases are stabilized by an AP component called properdin (P).

1.1.3. Soluble regulators

The function of soluble regulators of C (Table 2 and Fig. 1) is to inhibit excessive activation of the C system. C1 inhibitor (C1INH) inactivates the C1r and C1s serine esterases (Ziccardi, 1982) . C1INH is also an important regulator of the kinin, intrinsic clotting and fibrinolytic systems (Sundsmo and Fair, 1983) . Factor I inactivates C3b and C4b by cleaving their a-chains (Pangburn et al., 1977) . C4b binding protein (C4bp) dissociates the C4b2a convertase of the classical pathway (Scharfstein et al., 1978) and factor H dissociates the analogous C3bBb convertase of the alternative pathway (Whaley and Ruddy, 1976) . In addition, C4bp and factor H are cofactors for factor I in cleaving C4b and C3b, respectively (Whaley and Ruddy, 1976; Pangburn et al., 1977; Scharfstein et al., 1978).

Factor H has a central role in the discrimination between activators and nonactivators of the AP. The presence of negatively charged polyanions, sialic acid or acidic glycosaminoglycans, on cell membranes leads to the binding of factor H to C3b, which prevents formation of the C3bBb complex on the surface of a nonactivator, whereas the lack of these polyanions allows formation of the C3bBb convertase on an activator surface (Pangburn and Müller-Eberhard, 1984; Meri and Pangburn, 1990a; Meri and Pangburn, 1994).

Two plasma proteins, vitronectin (Podack and Müller-Eberhard, 1979) and clusterin (SP40,40/Apo-J; Jenne and Tschopp, 1989) are capable of binding to the C5b-7 complex. In this way, the C5b-7 complex remains soluble and the assembly of the MAC is inhibited (Podack and Müller-Eberhard, 1979; Jenne and Tschopp, 1989).

Table 2. Soluble regulators of complement

Designation	Molecular weight (kDa)	Complement components recognized	Function
C1 inhibitor	105	C1s, C1r	Inhibitor of C1s and C1r
Factor H	150	C3b	Cofactor for C3b cleavage, decay of C3bBb
Factor I	88	C3b, C4b	C3b and C4b cleavage
C4b binding protein, C4bp	540	C4b	Cofactor for C4b cleavage
Vitronectin, S-protein	80	C5b-7	Keeps C5b-7 complex soluble
Clusterin, SP40,40, Apo-J	70	C5b-7	Keeps C5b-7 complex soluble

1.1.4. Membrane regulators

Complement regulators on cell membranes (Table 3) protect cells against C-mediated injury and participate in immune clearance. At the level of C3/C5 convertases C activation is regulated by complement receptor type 1 (CR1, C3b receptor; Fearon, 1979), decay accelerating factor (DAF; Nicholson-Weller et al., 1982) and membrane cofactor protein (MCP; Cole et al., 1986; Fig. 3). CR1 and MCP are integral membrane proteins, whereas DAF is bound to the cell membrane phospholipids via a glycophosphatidylinositol (GPI)-anchor. Erythrocytes and phagocytes bind C3b-coated immune complexes and microbes via their CR1 receptors. CR1 also acts as a cofactor for factor I in the cleavage of C3b and C4b thereby augmenting the release of immune complexes from erythrocytes within the sinusoids of spleen and liver. Thus, CR1 has an important role in both opsonisation and the clearance of immune complexes. Because of its length, CR1, unlike DAF and MCP, does not seem to act as an intrinsic inhibitor of C on those cells where it resides. DAF accelerates the decay of the C3bBb convertases by dissociating Bb from the complex (Nicholson-Weller et al., 1982) . MCP acts as a cofactor for factor I-mediated cleavage of C3b (Cole et al., 1986).

Table 3. Membrane regulators of complement

Designation	Molecular weight (kDa)	Complement components recognized	Function and distribution
Complement receptor type 1 (CR1, CD35)	190, 220	C3b, C4b	C3b/C4b receptor Recognition of opsonized particles, immune complex transport, cofactor for C3b/C4b inactivation Phagocytes, B-lymphocytes, erythrocytes, eosinophils, glomerular podocytes, Langerhans cells
Decay accelerating factor (DAF, CD55)	70	C3bBb, C4b2a	Decay of C3/C5 convertases Blood cells, endothelia, various epithelial cells, spermatozoa
Membrane cofactor protein (MCP, CD46)	48-56, 58-68	C3b	Cofactor for C3b inactivation Blood cells, endothelia, epidermis, various epithelial cells, spermatozoa
Protectin (CD59, P-18, MACIF, HRF-20, MIRL, H19)	18-25	C8, C9	Inhibition of MAC Blood cells, endothelia, epidermis, glomerular podocytes and other epithelial cells, spermatozoa
C8 binding protein (C8bp, HRF, MIP)	65	C8, C9	Inhibition of MAC Blood cells, endothelia, epidermis, glomerular podocytes, amniotic epithelial cells

Fig. 3. Functions of decay accelerating factor (DAF), membrane cofactor protein (MCP) and complement receptor type 1 (CR1). DAF dissociates C3/C5 convertases. MCP is a cofactor for inactivation of C3b by factor I. CR1 has both functions, but because of its length its activity is directed primarily to the surface of neighbouring cells or to immune complexes. CCP, complement control protein repeat units (CCP).

Cells can resist the lytic effect of MAC by inhibiting its assembly on their cell membranes. Until now, two membrane-bound inhibitors of the MAC have been found. The first inhibitor, C8 binding protein (HRF/MIP) is a 65 kDa protein, that has been found on blood cells and endothelial cells (Table 3; Schönermark et al., 1986; Zalman et al., 1986). It has been suggested that C8 binding protein inhibits the polymerization of C9 in the MAC (Schönermark et al., 1986; Zalman et al., 1986) . The amino acid sequence of C8 binding protein is not yet known. The other inhibitor of the MAC, protectin (CD59-antigen/MIRL/HRF-20; Sugita et al., 1988; Davies et al., 1989; Okada et al., 1989; Holguin et al., 1990; Meri et al., 1990b; Meri, 1994) is a 20-kDa glycoprotein with a tissue distribution similar to C8bp. The amino acid sequence of protectin has a 26% homology with the sequence of Ly-6, a GPI-anchored murine T cell activating protein (LeClair et al., 1986; Davies et al., 1989) . The human urokinase-type plasminogen activator (uPAR; Roldan et al., 1990) also shows homology with protectin in that it has three protectin-like domains (Ploug and Ellis, 1994; Meri, 1994). Protectin, like uPAR, is bound to cell membranes via a GPI-anchor (Fig. 4; Sugita et al., 1988; Davies et al., 1989; Okada et al., 1989; Holguin et al., 1990). In paroxysmal nocturnal hemoglobinuria (PNH), a defect in the biosynthesis of GPI-anchors causes deficiency of GPI-anchored proteins on the cell membranes of erythrocytes, leukocytes and platelets. Because of the lack of GPI-anchored C regulators PNH erythrocytes are vulnerable to C lysis. Protectin inhibits incorporation of C9 to the C5b-8 complex thereby preventing the polymerization of C9 in the MAC (Fig. 5; Meri et al., 1990b; Rollins and Sims, 1990; Lehto and Meri, 1993). The existence of several C inhibitors on human cell membranes is a clear indication of the need for cells to protect themselves against C attack.

Fig. 4. Structure of the GPI-anchor. The enzymatic cleavage sites in GPI-anchor are indicated by . PI-PLD = phosphatidylinositol-specific phospholipase D. PI-PLC = phosphatidylinositol-specific phospholipase C. PLA2 = phospholipase A2.

Fig. 5. Mechanism of function of protectin (CD59). By binding to the C5b-8 complex CD59 limits the number of C9 molecules interacting with C5b-8 and the insertion of C9 molecules into the cell membrane (A). CD59 may prevent the formation of poly-C9 also by binding directly to C9 (B). Adapted from Morgan and Meri, 1994.

1.1.5. Removal of membrane attack complexes by vesiculation and endocytosis

Previous studies have shown that several types of nucleated cells - e.g. neutrophils (Campbell and Morgan, 1985; Morgan et al., 1987) , glomerular epithelial cells (Camussi et al., 1987) , oligodendrocytes (Scolding et al., 1989) and platelets (Sims and Wiedmer, 1986) - can resist C-mediated attack by vesiculation (ectocytosis) or endocytosis of MAC-containing cell membrane particles. However, it is unknown, how well injured cells, e.g. in ischemic cardiomyocytes, can resist the formation of MAC by energy-demanding processes, like vesiculation.

1.2. Complement activation in myocardial infarction

1.2.1. Complement activation products in infarcted myocardium

Several experimental and clinical studies have suggested that C activation is involved in the pathogenesis of myocardial infarction. Immunohistochemical analyses of human autopsy specimens have shown that C3d, C5, C8, C9 components and C5b-9 neoantigens of C are selectively deposited in infarcted areas (Schäfer et al., 1986; Rus et al., 1987) . Approximately a five-fold higher concentration of complement C5b-9 neoantigen complexes has been observed in lesions of acute myocardial infarction (AMI) than in normal myocardium by ELISA (Rus et al., 1987; Hugo et al., 1990) . It has been suggested that the detection of C5b-9 complexes is the most sensitive tool for diagnosing early ischemic changes in infarcted myocardium since these complexes have been observed in the earliest histologically detectable (approximately 7 hours old) myocardial infarction lesions (Schäfer et al., 1986) . C activation has also been observed in canine (Maroko et al., 1978; McManus et al., 1983; Rossen et al., 1985) , baboon (Pinckard et al., 1980; Crawford et al., 1988) and rat (Weisman et al., 1990) models of AMI.

Evidence for systemic activation of C in AMI has been observed in some studies. Elevated serum C3d (Earis and Bernstein, 1985) and SC5b-9 (Langlois and Gawryl, 1988; Yasuda et al., 1990) levels have been reported after AMI. Furthermore, there was a good correlation between the plasma C5b-9 complex and creatine phosphokinase levels (Yasuda et al., 1990) . However, another study failed to detect increased levels of C3 activation products or SC5b-9 complexes in the plasmas of AMI patients (Mollnes et al., 1988) . The observed differences may be due to variable sensitivities of the assays employed. Low levels of C activation products in plasma may be explained by the fact that C activation is primarily a local event in the ischemic myocardium (Mollnes et al., 1988) .

1.2.2. Triggering of complement activation

Although there are some experimental data that, in part, have elucidated the possible mechanisms behind C activation in AMI, the reasons and factors that trigger C activation in AMI are not well understood. Activation of the C system appears to occur principally during reperfusion of the ischemic tissue. As a consequence of ischemic tissue injury intracellular components may become released from cardiomyocytes. Previous studies based on in vitro experiments and animal models of AMI have shown that myocardial mitochondrial structures are able to activate the C cascade by directly binding C1q (Pinckard et al., 1975; Peitsch et al., 1988; Rossen et al., 1988; Kagiyama et al., 1989) . In addition, Linder et al. have shown that cytoskeletal intermediate filaments can activate the classical pathway of C in vitro (Linder et al., 1979) .

Plasmin can activate the C system by cleaving C1 (Ratnoff and Naff, 1967; Taubman and Lepow, 1971) , C3 (Ward, 1967) or C5 (Arroyave and Müller-Eberhard, 1973) . The treatment of AMI patients with recombinant tissue plasminogen activator has also led to C activation (Bennett et al., 1987) . This may attenuate the efficacy of the thrombolytic therapy. Recently, it has been shown that the treatment of AMI patients with streptokinase caused systemic C activation (Franci et al., 1994) . In this study an approximately 10-fold increase in the plasma levels of C3a, C4a and SC5b-9 was measured in AMI patients with streptokinase treatment, but not in patients without the treatment. It is conceivable that streptokinase activates the C system via plasmin (Sundsmo and Fair, 1983) .

Oxygen-derived free radicals can activate the terminal C cascade by converting C5 to a functionally active C5b-like metabolite (Vogt et al., 1986) . Similarly, the release of reactive oxygen species from activated neutrophils could trigger C activation (Shingu and Nobunaga, 1984) .

Some AMI patients develop C-fixing antibodies that are directed against heart structures. These antibodies could be involved in the pathogenesis of complications such as in the postmyocardial infarction syndrome or cause additive tissue injury in recurrent myocardial infarction (Kaplan and Frengley, 1969; DeScheerer, 1984; Langlois and Gawryl, 1988) .

1.2.3. Inhibition of complement-mediated injury

Experimental studies have shown that inhibition of the complement system reduces the size of myocardial infarction. Cobra venom factor (CVF) forms a stable C3 convertase with factor B and results in C3 depletion. In a dog myocardial infarction model the treatment of animals with CVF significantly reduced the size of an experimental infarction (Maroko et al., 1978) . In recent studies the treatment of rats with recombinant soluble human CR1 receptor (sCR1) was found to reduce myocardial infarction size by 38-44 % (Smith et al., 1992; Weisman et al., 1990) . Analogous reduction in tissue injury by sCR1 has been observed in experimental reperfusion injury of rabbit isolated heart (Homeister et al., 1992; Homeister et al., 1993), intestines (Hill et al., 1992) or skeletal muscle (Lindsay et al., 1992). sCR1 acts as an inhibitor of both the classical and the alternative pathway of C. The ability of sCR1 to inhibit both rat and human C makes it a useful tool for experimental studies and, possibly, for therapeutic trials in human patients (Weisman et al., 1990) .

1.3. Complement and atherosclerosis

The most common underlying disease of myocardial infarction is coronary atherosclerosis. Endothelial cell damage in an atherosclerotic lesion may trigger the formation of a coronary artery occluding thrombus and is therefore a crucial factor in the initiation of myocardial infarction. Previous studies have suggested, that the C system could be involved in the pathogenesis of atherosclerosis. Atherosclerotic lesions have been shown to contain deposits of C1q, C3c, C4 and complexes of C5b-9 (Vlaicu et al., 1985) , whereas blood vessels of newborns do not have C deposits (Schäfer et al., 1986) . The intensity of C5b-9 deposits in the atherosclerotic lesions in an immunofluorescence study (Vlaicu et al., 1985) , as well as the amount of C5b-9 quantitated in an ELISA (Niculescu et al., 1987a) , correlated with the severity of atherosclerosis. Animal models have shown a temporal and spatial colocalization of C5b-9 deposits with accumulation of lipid in the aortic tunica intima of cholesterol-fed rabbits (Seifert et al., 1989) . In C6-deficient rabbits fewer and less severe aortic atherosclerotic lesions have been detected than in control animals (Geertinger and Sørensen, 1975) .

It has been shown that crystalline cholesterol alone (Cosio et al., 1985; Seifert and Kazatchkine, 1987) and lipids derived from atheromas (Vogt et al., 1985b; Seifert et al., 1990) can activate the C system in vitro. Mitochondrial membranes (Pinckard et al., 1975) and intracellular intermediate filaments of endothelial cells (Linder et al., 1979) can also activate the C system in vitro. Thus, an endothelial cell injury in an atherosclerotic vessel may lead to a contact of intracellular structures with the C system and activate the cascade.

2. AIMS OF THE STUDY

1. to investigate activation of the complement system in acute myocardial infarction in autopsy specimens of infarcted human hearts and in a rat model

2. to examine mechanisms by which human and rat heart are protected against C-mediated injury and reasons why this protection fails in acute myocardial infarction

3. to analyze the behaviour of soluble (clusterin, vitronectin) and membrane-associated (protectin, C8bp) regulators of the membrane attack complex of complement in the post-injury clearance process

4. to examine the fate of glycolipid-tailed protectin in human plasma after release from cell membranes

3. MATERIALS AND METHODS

3.1. Tissue samples and cell culture

3.1.1. Autopsy specimens

Tissue samples of normal and infarcted human myocardium were used for the immunohistochemical analysis of C components and C regulators and for the immunochemical analysis of protectin and clusterin (I-IV). Myocardial tissue specimens were obtained at autopsy (Department of Pathology, University of Helsinki) from patients (n=15; female/male ratio 7/8), who had died from an acute myocardial infarction. The ages of AMI (= time interval between the initiation of chest pain and the death of patient) varied from 2.5 hours to 14 days. Normal hearts (n=5) were obtained from patients with neither anamnestic nor histopathological evidence of infarction. The post mortem time interval between death and autopsy was 1-8 days in the myocardial infarction group and 2-7 days in the control group.

3.1.2. Blood samples

Samples of human blood were taken by venipuncture from healthy donors of our laboratory and allowed to clot for 30 min at 22 °C. The sera were separated from the clot by centrifugation and stored in aliquots at -70 °C. Erythrocytes of rats, guinea pigs and rabbits were obtained from the animal facility of our department.

3.1.3. Experimental myocardial infarction

Male Wistar rats obtained from the Unit of Clinical Physiology, Minerva Foundation Institute for Medical Research, Helsinki, Finland, were used in the animal myocardial infarction model (IV). The animals were 4-6 months old and weighed 400-600 g. The study protocol was approved of by the ethical committee of the Meilahti Theoretical Institutes and by the local health authority (STO 2534).

Myocardial infarctions were produced in rats as described earlier (Selye et al., 1960; Tikkanen et al., 1987) . Animals were anesthetized with ketamine (Ketalar, Parke-Davis, Barcelona, Italy), 150 mg/kg, given as an intraperitoneal injection and xylazine (Rombun, Bayer Corp., Leverkusen, Germany), 6 mg/kg, given as an intramuscular injection. Tracheostomized animals were ventilated by a positive pressure respirator (Medith Corp., Helsinki, Finland). After thoracotomy the left coronary artery was ligated between the pulmonary artery outflow tract and the left atrium. The rats were sacrificed by carbon dioxide inhalation and decapitation.

Rats were divided into 8 groups each containing 4 rats. In groups I-VI rats were sacrificed 1, 2, 3, 6, 24 or 72 hours after the coronary artery ligation. Group VII included sham-operated rats, which were killed 72 hours after the thoracotomy. In group VIII (post mortem autolysis controls) 2 rats were, and 2 rats were not, subjected to coronary artery ligation. Rats were killed at 6 hours, whereafter they were kept at 4°C for 3 days. The rats without signs of AMI in nitroblue tetrazolium or hematoxylin-eosin stainings were excluded from groups III-VI. No reliable signs for AMI could be observed by nitroblue tetrazolium or hematoxylin-eosin staining in AMI lesions that were less than 3 hours of age.

3.1.4. Cell culture

The human endothelial cell line EA.hy 926 (Edgell et al., 1983) was obtained from Dr. R. Renkonen at our department. This cell line is positive for factor VIII expression. EA.hy 926 cells were cultured in RPMI 1640 medium (Gibco, Paisley, UK) supplemented with 10 % heat-inactivated fetal calf serum (Gibco), 2 mM l-glutamine and 10 U/ml penicillin and 100 mg/ml streptomycin.

3.2. Microscopical and immunohistochemical methods

3.2.1. Microscopical and histochemical diagnosis of myocardial infarction

Frozen or paraffin sections (4 µm) of human and rat heart samples were stained with hematoxylin-eosin. The AMI was diagnosed by the following histopathological signs: stretching and waviness of myocardial fibers, cytoplasmic eosinophilic coagulative changes, condensation of the nuclear chromatin, shrinkage of the cells, intercellular oedema, loss of cross striation, neutrophil infiltration, pyknosis or loss of nuclei, and complete cellular necrosis. In the rat myocardial infarction model nitroblue tetrazolium (NBT) vital staining was used for macroscopic detection of AMI and hematoxylin-eosin staining was used for histopathological examination (Nachlas and Shnitka, 1963) .

3.2.2. Immunofluorescence staining

Samples for indirect immunofluorescence (IFL) microscopy were quickly frozen to -65°C using isopentane cooled with liquid nitrogen. Frozen sections (4 µm) were fixed for 10 min in cold (-20°C) acetone. In some experiments nonfixed sections were used. Paraffin sections were also used for immunohistochemical analysis. After deparaffination the sections were treated with pepsin (1% w/v, pH 1.8; 30 min at 37°C) and washed with phosphate buffered saline (PBS), pH 7.4.

For IFL analysis sequential sections of the tissues were incubated for 30 min at +22°C with the primary antibodies (see Tables 4 and 5). After washing with PBS the sections were treated by fluorescein isothiocyanate- (FITC-) or tetramethylrhodamine isothiocyanate- (TRITC-) conjugated secondary antibodies.

In the IFL analysis control incubations were done either by omitting the primary antibody, by using nonimmune sera or by using antisera with known specificities other than those of the first antibodies. The slides were mounted with Mowiol (Heimer and Taylor, 1974) and examined with a Zeiss Standard microscope equipped with filters specific for FITC- and TRITC-fluorescence.

3.2.3. Immunoperoxidase staining

Cryostat sections were incubated with the primary antibody for 16 hours at +22°C in a moist chamber. After washing, bound primary antibody was detected by a biotinylated secondary antibody against mouse or rabbit immunoglobulins complexed with avidin-peroxidase (Vectastain ABC kit, Vector Laboratories, Burlingame, CA). The color was developed by using 3-amino 9-ethylcarbazole, 0.2 mg/l in 0.05 M sodium acetate buffer, pH 5.0, as substrate during a 20 min incubation at room temperature. To detect cell nuclei the sections were counterstained with the Meyer stain (MSD, Darmstadt, Germany). Control stainings were done similarly to the IFL analysis.

3.2.4. Electron microscopy

Sarcolemmal membrane samples from normal and infarcted myocardium were disrupted by freezing and thawing the preparates twice (II). The membrane fractions were collected by centrifugation (40,000g, 15 min) and the pellets were suspended into H20 to further disrupt the cells. The membrane fragments were washed with PBS (40,000g, 15 min x3) and resuspended in PBS to a total volume of 50 µl. The specimens were negative contrast stained with 2.5 % (w/v) uranyl acetate, dried and examined in a JEOL (JEM-100 CX II, Tokyo, Japan) transmission electron microscope (accelerating voltage of 80 kV).

Complement treated sheep erythrocyte ghosts were used as positive controls of MAC lesions and sheep erythrocyte membranes without C treatment were used as negative controls. Sheep erythrocytes (3x108 cells/100 µl PBS) were incubated first with a rabbit anti-sheep erythrocyte antibody (final dilution 1/250) and then with normal human serum (NHS; final dilution 1/10) for 30 min at 37°C. Thereafter, erythrocyte ghosts were prepared by adding 700 µl H20 to the cell suspensions and washing three times with PBS and finally resuspended in PBS to a total volume of 100 µl.

Table 4. Antibodies against human proteins used in the study

Specificity	Type	Source or reference
C1q	rabbit pAb	Behringwerke AG, Marburg, Germany
C3c	-"-	-"-
C3d	-"-	-"-
C4	-"-	-"-
C5	goat pAb	Quidel Corp., San Diego, CA
C6	-"-	-"-
C7	-"-	-"-
C8	-"-	-"-
C9	rabbit pAb	-"-
C9	-"-	Behringwerke AG
C9 neoantigens	-"-	-"-
C9 neoantigen	mouse mAb	Quidel Corp.
Properdin	rabbit pAb	A. Sjöholm, University of Lund, Lund, Sweden
Properdin	-"-	Atlantic Antibodies Corp., Stillwater, MN
CR1	mouse mAb	Becton Dickinson Corp., Mountain View, CA
CR3	-"-	-"-
DAF (BRIC 110)	-"-	Bio-Products Laboratory, Elstree, England
DAF (BRIC 216)	-"-	-"-
DAF (IA10)	-"-	V. Nussenzweig, New York University School of Medicine, New York, NY
MCP (GB24)	-"-	J.P. Atkinson, Washington University School of Medicine, St. Louis, MO
C8bp	rabbit pAb	B.P. Morgan, University of Wales College of Medicine, Cardiff, Wales
C4bp	-"-	Behringwerke AG
Vitronectin	-"-	-"-
Clusterin	-"-	Ehnholm et al., 1991
Clusterin	mouse mAb	Quidel Corp.
Protectin (BRIC 229)	-"-	Bio-Products Laboratory
Protectin (YTH53.1)	rat mAb	Davies et al., 1989; Meri et al., 1990b
Protectin (R476)	rabbit pAb	-"-
HLA class I Cedarlane	rabbit pAb	Laboratories, Hornby, ON, Canada
Transferrin	-"-	Behringwerke AG
CRP	-"-	-"-
IgA	-"-	-"-
IgG	-"-	-"-
IgM	-"-	-"-
Albumin	-"-	-"-
Haptoglobin	-"-	-"-
Glycophorin A (YTH 89.1)	rat mAb	Davies et al., 1989
LFA 1	mouse mAb	Immunotech, Marseilles, France

pAb, polyclonal antibody; mAb, monoclonal antibody

Table 5. Antibodies against rat proteins

Specificity	Type	Source or reference
C1	rabbit pAb	Jones et al., 1990
C3	-"-	Cappel, Malvern, PA
C6	-"-	Jones et al., 1990
C8	-"-	-"-
C9	-"-	-"-
Rat protectin	mouse mAb	Hughes et al., 1992
IgG	rabbit pAb	Dakopatts, Glostrup, Denmark
Albumin	-"-	-"-

3.3. Sodium dodecyl sulphate -polyacrylamide gel electrophoresis and immunoblotting analysis

Sodium dodecyl sulphate -polyacrylamide gel electrophoresis (SDS-PAGE) gels were run according to the method of Laemmli (Laemmli, 1970) by using a mini gel system (Bio-Rad Laboratories, Richmond, CA). SDS-PAGE analysis of clusterin or protectin containing samples was performed using 8% or 12% homogenous gels, respectively. Molecular weight standards were from Bio-Rad Laboratories. In the immunoblotting analysis proteins were electrotransferred from SDS-PAGE gel to a nitrocellulose filter and nonspecific binding was blocked by incubating the filter with 3% bovine serum albumin (BSA) in PBS for 16 hours. The filter was incubated with a primary antibody, and bound immunoglobulin was detected with a peroxidase-conjugated antibody against the primary antibody (Dakopatts, Glostrup, Denmark) or by an alkaline phosphatase-conjugated rabbit anti-mouse immunoglobulin antibody (Orion-Diagnostica, Espoo, Finland) diluted in 3% BSA/PBS.

3.4. Isolation of protectin from normal human heart tissue, erythrocytes and urine

Protectin was purified from human heart by YTH53.1-Sepharose affinity column chromatography, as described earlier (Davies et al., 1989; Meri et al., 1990b) . In brief, human heart tissue obtained at autopsy (n=5) and from a patient undergoing heart transplantation was homogenized and solubilized in 2% NP40 detergent in PBS containing protease inhibitors. Insoluble material was removed by centrifugation (at 100,000g, 60 min at 4°C) and the supernatant was subjected to YTH53.1-Sepharose affinity column chromatography. Bound protectin was eluted from the column by 0.1 M glycine-HCl in 0.15 M NaCl and 0.1% NP40, pH 2.8, and collected into tubes containing 0.1% NP40 in 1 M Tris-HCl buffer, pH 9.0. Finally, the elution fractions were pooled, concentrated and dialyzed against 0.01% NP40 in PBS. Remaining contaminants were removed by gel filtration on a Superose 6 column (Pharmacia, Uppsala, Sweden). Erythrocyte protectin was isolated similarly, except that erythrocytes were first lysed by water to obtain erythrocyte ghosts. Soluble protectin was isolated from human urine. Urine samples were neutralized with 0.1 M NaOH, filtered through 0.22 µm filters and then subjected to the YTH53.1-Sepharose affinity column chromatography, as described above. The purity of protectin was examined by SDS-PAGE and silver staining.

Protectin samples purified from human heart, erythrocytes and urine were radiolabeled with Na-125I to specific activities of 106-107 cpm/µg using the Iodogen method (Pierce Chemical Co., Rockford, IL). Labeled proteins were examined by SDS-PAGE and autoradiography using a 14C-methylated protein mixture (Amersham, UK) as a standard.

N-terminal amino acid sequence of myocardial protectin was determined by automated Edman-degradation in an Applied Biosystems gas phase sequencer (Foster City, CA) at the Department of Medical Chemistry, University of Helsinki.

3.5. Protein and lipoprotein analysis

3.5.1. Purification of lipoproteins

Low density lipoproteins (LDL) and high density lipoproteins (HDL) were isolated from human plasma by sequential ultracentrifugation using KBr for density adjustments, washed by reflotation at d = 1.21 g/ml, gel filtered on a high resolution Superose 6B column and dialyzed against PBS, pH 7.4 (Havel et al., 1955; Jauhiainen et al., 1993) .

3.5.2. Quantification of proteins, apolipoproteins and lipids

A modified Lowry method was used for the determination of protein concentration (Markwell et al., 1981) , an immunoturbidometric method for the determination of apo A-I and apo B-100 concentrations (Riepponen et al., 1987) and an enzymatic method for the determination of cholesterol concentration (Röschlau et al., 1974) . Gradient gel electrophoresis was used for the measurement of average particle sizes of HDL2 (10 nm), HDL3 (8.5 nm) and LDL (25 nm; Blanche et al., 1981; Jauhiainen et al., 1993).

3.6. Phospholipase C and neuraminidase treatment of tissues

Cryostat sections of normal human heart specimens were treated with B. cereus phosphatidylinositol-specific phospholipase C (PIPLC; Sigma, St. Louis, MO) at 0.1-2 IU/ml; in PBS or with PBS alone for 30 min in a moist chamber at +37°C. The effect of PIPLC on sarcolemmal protectin was examined by IFL microscopy. Frozen sections of normal myocardial samples were incubated with C. perfringens neuraminidase (1 IU/ml in PBS; Sigma) for 30 min in a moist chamber at +37°C. The loss of cell membrane sialic acid in the infarcted heart specimens was analyzed by the appearance of binding sites for FITC-conjugated peanut agglutinin (PNA) lectin (Holthöfer et al., 1981) . PNA, specific for ß-D-galactose was obtained from Vector Laboratories, Burlingame, CA. FITC-conjugated WGA lectin (EY LABS Inc., Biochemical Division, San Mateo, CA), specific for N-acetyl neuraminic acid and N-acetyl glucosaminyl residues, was also used for the detection of glycosylated molecules in myocardial cell membranes.

3.7. Sucrose density gradient ultracentrifugation

Binding of heart protectin to the terminal C complex intermediate C5b-8 was examined by mixing radiolabeled protectin (105 cpm/0.1 µg) in 0.1% NP40 with C9-deficient human serum (a kind gift from Dr. B.P. Morgan, University of Wales College of Medicine, Cardiff, United Kingdom). Serum was incubated with or without 4% inulin for 60 min at 37 °C. Samples were layered on top of linear 10-50% sucrose gradients containing 0.1% NP40/PBS. After ultracentrifugation (16 hours at 200,000g) 200 µl fractions were collected and counted for radioactivity. Migration of serum IgG (7S) and IgM (19S) were used as standards for sedimentation. The sedimentation of C5b-8 was determined in separate experiments using 125I-C8 in C5b-8 as a marker.

3.8. Gel filtration and anti-apolipoprotein A-I affinity chromatography

Binding of heart, erythrocyte or urine protectin to serum lipoproteins was examined by incubating (30 min at 37 °C) radiolabeled preparations of protectin (3x105 cpm/5 µl) with 500 ml of normal human serum (1/2 dilution in Tris buffered saline, pH 7.4 [TBS]), HDL (protein concentration, 1.76 mg/ml), LDL (1.76 mg/ml) or TBS alone. After applying the samples on a Superose 6B high resolution gel filtration column (1x30 cm, Pharmacia), the column was eluted with TBS. Protein-containing fractions were detected by absorbance at 280 nm. Collected fractions (0.5 ml) were subjected to radioactivity counting and quantitation of apo A-I, apo B-100 and cholesterol.

In one experiment fractions containing radiolabeled CD59E and HDL were mixed, applied on an anti-apo A-I affinity chromatography column and the column-bound material was eluted with 0.1 M glycine-HCl, pH 2.8.

3.9. Incorporation of protectin into cell membranes

125I-labeled protectins (105 cpm) purified from heart and urine were incubated with a suspension of rabbit erythrocytes (109 cells) or EA.hy 926 endothelial cells (107 cells) for 30 min at 37 °C. After washing the cell-bound radioactivity was measured. The incorporation of protectin into rabbit erythrocytes was also analyzed by IFL microscopy using FITC-conjugated anti-protectin YTH53.1 F(ab')2.

Inhibition of protectin incorporation into cell membranes by lipoproteins was examined by incubating rabbit erythrocytes (109 cells) with 125I-CD59H (105 cpm) in the presence of 0.1 or 1 mg/ml HDL2, HDL3, LDL, apo A-I or BSA. Cell-bound and free 125I-CD59 were separated by layering the mixtures on top of 20% sucrose (250 µl) in test tubes which were centrifuged at 5000g for 1 min.

3.10. Transfer of protectin between cells and lipoproteins

125I-CD59H was incorporated into rabbit erythrocytes (109 cells) as described above. After incubating these cells with HDL2, HDL3, LDL, apo A-I, BSA (0.1 mg/ml or 1 mg/ml) or PBS, cell-bound and free 125I-CD59H were separated by centrifuging through 20% sucrose.

The ability of HDL to transport protectin between two cell populations was tested by first incubating HDL2 (0.1 or 1 mg/ml) or PBS with rabbit erythrocytes (109 cells) or EA.hy 926 endothelial cells (107 cells) containing incorporated 125I-CD59H. After centrifugation (2000g, 5 min) the supernatants containing 125I-CD59H were incubated with unlabeled rabbit erythrocytes (109 cells) or EA.hy 926 endothelial cells (107 cells), respectively. After washing the amount of reincorporated 125I-CD59H was counted.

4. RESULTS

4.1. Immunohistochemical detection of complement components in human myocardium (I, II)

Human autopsy specimens were used for studies of C activation in myocardial infarction lesions. The deposition of C components was analyzed by immunofluorescence and immunoperoxidase microscopy in 15 cases with and in 5 cases without acute myocardial infarction. Both cryostat and paraffin embedded samples of the heart tissues were used for the studies. The estimated ages of the infarctions (time difference between the beginning of the clinical episode and death) ranged from 2.5 hours to 14 days.

Activation of the C system was observed in human infarction lesions, that were >8 hours old. In these samples components of both the early (C1q, C4, C3c, C3d) and late (C5, C6, C8, C9 and C5b-9) pathways of C were deposited selectively in the AMI lesions. In more recent AMI lesions (< 3 days) the C deposits were seen throughout the lesions, while in older (>3 days) lesions the necrotic center appeared free of C components and C deposits were seen in the periphery of the lesions. Deposits of C1q, C3c, C4, C5 and C6 were usually less intense than those of C3d, C8, C9 and C5b-9 neoantigens. In the scars of old infarction lesions no deposits of C were found. In control heart specimens no lesional C deposits were observed, but basement membranes of blood vessels appeared occasionally positive for C3c, C3d or C9 irrespective of the time interval between death and autopsy (range 2-7 days).

Antisera against several human plasma proteins (transferrin, haptoglobin, IgA, IgG, IgM) did not stain the infarcted myocardial lesions. Neither was staining of myocardial cells seen in control experiments when the first antibody was omitted or replaced by nonimmune rabbit serum.

4.2. Demonstration of MAC in sarcolemmal membranes of infarcted myocardium by electron microscopy (II)

The sarcolemmal membrane fragments from infarcted and non-infarcted myocardium were processed and negatively stained for transmission electron microscopy. Transmission electron microscopy analysis showed numerous regularly shaped lesions with diameters of 9-12 nm in the infarcted myocardium. The highest density of myocardial MACs was estimated to be 1600 lesions/µm2. In comparison, these lesions were similar in size and shape as MACs generated on the surface of antibody-sensitized sheep erythrocytes, on which the density of MACs was approximately 1100 lesions /µm2. No MAC-like lesions were seen in the negatively stained membrane preparation of normal myocardium.

4.3. Expression and deposition of complement regulators in normal and infarcted human myocardium

4.3.1. Expression of protectin (I)

The immunofluorescence (IFL) and immunoperoxidase (IP) techniques were used for the detection of protectin in frozen and paraffin sections of human myocardium. A polyclonal antibody (R476) reacted with both types of sections whereas the monoclonal antibodies (YTH53.1 and BRIC 229) showed specific reactivity only in frozen sections.

Protectin was observed in myocardial fibers and vascular endothelium of normal human heart tissue. The expression of protectin appeared as a linear, band-like fluorescence on the cell membranes of cardiomyocytes and endothelial cells of cardiac blood vessels. Protectin was absent at the center of infarction lesions aged 8 hours or more and its expression gradually increased towards the border areas between lesions and normal tissue. Analysis of C9 showed that it colocalized with the protectin-negative infarction lesions. In older AMI-lesions (> 3 days of age) deposits of C components were primarily localized in the periphery of lesions, whereas the loss of protectin was observed throughout the entire lesion. In contrast with a clearly diminished expression of protectin in the infarcted cardiomyocytes protectin was found to persist in the endothelial cells of coronary blood vessels in the infarction lesions.

When cryostat sections of the heart tissue specimens were treated with the PIPLC enzyme the staining of protectin in sarcolemmal cell membranes was abolished. This indicated that myocardial protectin was tethered to the cell membranes via a PIPLC-sensitive glycophosphoinositol anchor.

As controls for protectin staining, rat anti-glycophorin A mAb (YTH89.1) with the same isotype as YTH53.1 and a preimmune rabbit serum were used. YTH89.1 stained neither cardiomyocytes nor endothelial cells, but reacted with human erythrocytes within blood vessels. The preimmune rabbit serum showed no reactivity with the human heart tissue. As another control a rabbit polyclonal antibody against HLA class I antigens was used. This antibody showed a linear membrane staining throughout normal and infarcted myocardium.

Expression of glycosylated proteins in the heart tissue was examined by direct staining of frozen tissue specimens with the WGA lectin and by looking for the appearance of binding sites for the ß-D-galactose reactive PNA lectin after neuraminidase treatment. These studies indicated that glycoproteins with terminal sialic acid residues were present throughout normal and infarcted areas of human heart. The results suggest that most sialylated glycoproteins remain in the AMI lesions and that the loss of protectin is not simply the result of a general loss of all membrane glycoproteins due to ischemia.

4.3.2. Expression of other membrane regulators of complement in myocardium (II)

The expression of the C3/C5 convertase regulators MCP, CR1 and DAF was studied by IFL and IP techniques in normal and infarcted myocardial specimens. Neither a mAb against MCP (GB24) nor CR1 stained normal or infarcted muscle cells. Also, no staining for CR3, the receptor for iC3b (CD11b/CD18), was observed in cardiomyocytes of the myocardial tissue. For the IFL and IP microscopical analysis of DAF three different mAbs (BRIC 110, BRIC 216 and IA10) were used. All three antibodies stained the endothelial lining of blood vessels, but no staining or only very weak linear staining of sarcolemmal membranes in both normal or infarcted cardiomyocytes could be detected. Frozen sections of human placenta were used as positive controls. All three anti-DAF mAbs reacted well with the human placental syncytiotrophoblast cell membranes. IFL and IP microscopy with antibodies against C8 binding protein (C8bp), a putative 65 kDa GPI-anchored regulator of MAC, showed that C8bp was expressed in the sarcolemmal membranes of both normal and infarcted specimens. Occasionally, C8bp appeared to form similar deposits as the terminal C components within the infarction lesions.

4.3.3. C4bp, vitronectin and clusterin (II, III)

The expression and deposition of C regulators in both normal and infarcted human myocardium were examined by IFL on frozen and paraffin sections and by IP microscopy on paraffin sections. No deposits of soluble C control proteins, C4bp, vitronectin or clusterin were seen in the cardiomyocytes of normal myocardium, although basement membranes of coronary vessels sometimes appeared positive for these regulators. However, in the infarcted myocardium C4bp, vitronectin and clusterin were observed to colocalize with components of the MAC. The staining intensity of these regulators was strong in all infarcted lesions regardless of the age of the lesion.

Immunoblotting analysis of extracts from infarcted myocardial samples was performed for a more specific molecular characterization of clusterin deposition in AMI lesions. For this procedure AMI specimens that had shown strong deposits of clusterin in the IFL analysis were chosen. Using a mAb against clusterin, a band of 80 kDa after SDS-PAGE under nonreducing conditions was detected. A similar band was observed in samples of normal human serum and seminal plasma. Only a faintly staining band with an Mr of 80 kDa was found in extracts from normal heart specimens.

4.4. Structural and functional properties of myocardial protectin (I)

Western blot analysis of purified human heart protectin, using the YTH53.1 anti-CD59 mAb, detected a major band with an apparent Mr of 18-22 kDa. The purified protectin from human erythrocyte membranes gave a main band of 18 kDa in both SDS-PAGE and immunoblotting analysis. In contrast, the urinary protectin had an approximate Mr of 24 kDa in SDS-PAGE analysis.

The N-terminal amino acid sequence of heart protectin showed a sequence: LQXYNX PNPT (X is for residues not identifiable during sequencing) which is identical to the N-terminal sequence of erythrocyte and urine protectin and to the amino acid sequence derived from the CD59 cDNA sequence (LQCYNCPNPT; Davies et al., 1989) . In sucrose density gradient analysis the purified radiolabeled, myocardial protectin bound to the nascent SC5b-8 complex.

4.5. An experimental rat model for myocardial infarction (IV)

Myocardial infarctions in rats were produced surgically by permanent ligation of the left main coronary artery. Rats were sacrificed at 1, 2, 3, 6, 24 or 72 hours after the ligation. Macroscopically, infarction lesions aged 3 hours or more were detectable by NBT staining. Microscopically, the earliest signs of AMI could be detected 3-6 hours after coronary artery ligation by routine hematoxylin-eosin staining.

4.5.1. Complement components in normal and infarcted rat myocardium

Complement activation in rat heart was examined by specific antibodies against rat C components. In both normal hearts and in noninfarcted areas no apparent cardiomyocyte-associated C deposits were observed. Two of four rats in the 2-hour infarction group had C3 deposition in myocardial areas supplied by the ligated coronary artery, but no deposition of C1, C8 or C9. The C3 deposits appeared as distinct foci within the infarcted heart.

In the 3-hour AMI lesions activation of the entire C cascade was detected: C1, C3, C8 and C9 deposits were selectively accumulated in the ischemic myocardial lesions. In older AMI-lesions (6 hours - 3 days) the C deposits were also detected selectively throughout the myocardial infarcted areas. The infarcted areas appeared as clearly demarcated from the normal areas. Ischemic cells could be distinguished from viable cells by reactivity with the anti-C antibodies.

Control stainings by omitting the primary antibody, by using nonimmune rabbit serum, or antibodies against rat albumin or rat IgG showed no reactivity with the infarcted myocardial areas.

4.5.2. Protectin in normal rat myocardium

Rat protectin, as examined with the TH9 mouse mAb, was found to be expressed in the sarcolemmal membranes of normal cardiomyocytes throughout the myocardium as well as in endocardium and in the endothelial cells of blood vessels. Immunoblotting analysis of homogenates of normal rat myocardium, rat aorta and erythrocyte ghost preparations with the TH9 mAb showed a major band with an apparent Mr of 21 kDa. Protectin in rat urine had a slightly higher apparent Mr (24 kDa) than the cell membrane-associated forms of protectin. Monoclonal antibody against rat protectin (TH9) did not react with human heart protectin in IFL or immunoblotting analyses (IV). Similarly, a mAb against human protectin (YTH53.1) showed no reactivity with rat protectin.

The expression of rat protectin in cardiomyocytes, endothelial layers of cardiac blood vessels and endocardium disappeared after a 30 min treatment of sections of normal rat heart with PIPLC; 1 IU/ml. This indicated that myocardial rat protectin also had a PIPLC-sensitive glycophosphoinositol-type anchorage.

4.5.3. Protectin in infarcted rat myocardium

Expression of rat protectin diminished gradually from the 6-hour to the 72-hour AMI lesions. The expression of rat protectin was nearly absent in the AMI lesions aged 24 hours or more in most cases. On the other hand, rat protectin was still present in the earlier AMI lesions, but it was often seen in condensed patches. In the border areas of lesions aged 24 hours or more, vesicle-like formations positive for rat protectin were occasionally seen. Deposits of C components (C1, C3, C8, C9) were often found within the protectin-negative infarction lesions.

No loss of rat protectin was detected in the hearts of sham-operated rats or rats that had undergone a 3-day post mortem autolysis period. The majority of cell membrane glycoproteins remained present throughout the normal and infarcted cardiac muscle, as indicated by staining with the FITC-conjugated WGA-lectin.

4.6. Interactions between glycolipid-anchored protectin and plasma lipoproteins (V)

4.6.1. Incorporation of phospholipid-tailed protectin into lipoprotein particles

The loss of myocardial protectin from infarcted heart tissue raised the question of the fate of lipid-tailed protectin when entering human plasma. To examine this, radiolabeled human erythrocyte protectin (CD59E) was mixed with NHS. Subsequent gel filtration analysis demonstrated that CD59E comigrated with serum lipoprotein particles (Fig. 2 in V). Up to 85 % of 125I-CD59E comigrated with HDL and 15 % with LDL-containing fractions. Radiolabeled soluble urinary protectin which lacks the phospholipid of the GPI-anchor lipid did not change its gel filtration pattern in human plasma.

In an additional experiment, fractions containing both radiolabeled CD59E and HDL were pooled and applied to an anti-apo A-I affinity chromatography column. Fifty eight percent of the radioactivity bound to the column. This suggests that erythrocyte protectin had become incorporated into the HDL particles. In another gel filtration analysis, purified HDL was mixed with 125I-CD59E and the two were found to comigrate (Fig. 3 in V). In similar gel filtration experiments radiolabeled GPI-anchored protectin isolated from human heart (CD59H) partially comigrated with HDL and a proportion of CD59H comigrated with LDL (Fig. 3 in V).

4.6.2. Transfer of protectin between cells and lipoprotein particles

The ability of lipoproteins to incorporate 125I-CD59H that had become spontaneously shed from cells and to deliver protectin further to other cells was tested in vitro. Approximately 15% of 125I-CD59H could be incorporated into rabbit erythrocytes, whereas only 0.2% of soluble urine 125I-protectin (CD59U) bound to these cells under identical conditions. When rabbit erythrocytes containing 125I-CD59H were incubated with HDL2 (1 mg/ml), 43% of the radiolabel was released into the HDL-containing supernatant. The corresponding figure was 25% for HDL3, 20% for LDL, 17% for BSA and 7% for PBS buffer (spontaneous release; Fig. 5 in V). Similar experiments using smaller amounts of lipoproteins indicated the release of 125I-CD59H from cell surfaces was dependent on the concentration of lipoproteins in the extracellular milieu. Purified lipoproteins were also found to be efficient inhibitors of protectin incorporation into rabbit erythrocytes. The maximum inhibition was 85% by HDL2 (at 1 mg/ml), 64% by HDL3, 45% by LDL and 61% by apo A-I (Fig. 5 in V).

The ability of HDL to transport protectin from one cell type to another was tested using rabbit erythrocytes and human EA.hy 926 endothelial cells with incorporated 125I-CD59H. In this experiment approximately 42% of incorporated protectin was released from each cell type into the HDL2 (protein concentration 1 mg/ml) containing supernatant. Furthermore, after incubating this supernatant with unlabeled EA.hy 926 endothelial cells or rabbit erythrocytes, 7% of 125I-CD59H released from rabbit erythrocytes by HDL2 became transferred to EA.hy 926 endothelial cells and 14% from EA.hy 926 cells to rabbit erythrocytes.

4.6.3. Protectin in high density lipoprotein particles

Because phospholipid-tailed protectin was found to incorporate into HDL particles, the protectin content of HDL particles in the plasma was next examined. Isolated HDL particles were subjected to an anti-human CD59 (YTH53.1) immunoaffinity column in the presence of 0.1% NP40 detergent. Immunoblotting analysis of concentrated material eluted from the column showed a diffuse band with an apparent Mr of approximately 20 kDa, similar to that of erythrocyte or heart protectin (Fig. 6 in V). In contrast, no positive immunoblotting reaction was observed with an analogously processed LDL or with an unenriched HDL preparation. Immunoblotting analysis of whole human plasma appeared too insensitive for the detection of protectin.

5. DISCUSSION

5.1. Time course of complement activation in myocardial infarction

In the human AMI specimens, C deposition was observed in the ischemic lesions that were about 8 hours old or more but not in the 2.5-hour-old lesions. This finding is in accordance with a previous study based on autopsy specimens of AMI patients showing that the MAC deposits could be detected in 7-hour-old AMI-lesions (Schäfer et al., 1986) . These results suggest that C is activated during the early hours of AMI. However, studies based on autopsy specimens (I-III) do not allow determination of the exact time-course of C activation, because the age of early AMI lesions (< 6 hours of age) cannot be reliably estimated by histopathological methods. Also, estimation of the duration of AMI by clinical data alone is unreliable and autolysis of the heart tissue post mortem could affect the results. For these reasons, the C activation process was studied also in a well characterized rat model of AMI (Selye et al., 1960; Tikkanen et al., 1987) . In this study (IV) deposition of C3, an initial sign of C activation, could be observed in two of four rats at 2 hours after ligation of the coronary artery, whereas full activation of C was detected in all rats with a 3-hour AMI. From experimental studies it is known, that the majority of the ischemic damage in heart will develop during the first 6 hours of AMI (Reimer and Jennings, 1979) . Thus, analysis of the time course of complement activation suggests the C system has a pathogenetic role in extending the injury of ischemic myocardium. The timing of C activation apparently coincidences with reperfusion of the ischemic areas by blood (or plasma) from collateral blood vessels or from a partially reopened coronary artery. It is notable that in our experimental model the infarction was induced by a permanent ligation of the coronary artery. It is likely that a more rapid C activation response would follow, if reperfusion is allowed to occur earlier by opening the ligation. This is in accordance with a study on an experimental rabbit AMI model showing that C5b-9 complexes were accumulated in the ischemic myocardium 5-6 hours after a permanent coronary occlusion, whereas C5b-9 deposits were seen in the reperfused heart already after a 30 min period of coronary occlusion (Mathey et al., 1994).

5.2. Specific features of complement activation in myocardial infarction

In studies on human AMI (I-III), components of the early (C1q, C4, C3c, C3d) and late C pathway (C5, C6, C8, C9 and C5b-9) were present in the AMI lesions. The most specific component of the alternative pathway, properdin, was not seen in the human AMI lesions. This might indicate that C activation had occurred via the CP in human myocardial infarction. However, the possible role of the AP in the C activation process could not be excluded, because specific components of the AP are difficult to demonstrate by immunohistochemical methods.

The fact that selective deposition of C3, but not of C1, was observed in the 2-hour AMI lesions in the rat model (IV) suggests that the C system could be initiated via spontaneous activation of the AP. Deposition of C1, C8 and C9 in the AMI lesions aged 3 hours or more indicates that the CP is also recruited and that the C system becomes fully activated within 3 hours after the beginning of AMI. The results agree with a previous study of experimental baboon infarction, which showed that C3 deposits were present in cardiomyocytes 4 hours after coronary artery ligation and gradually increased during the next 3 hours (Pinckard et al., 1980) . Similarly, in an experimental canine model of myocardial infarction, radiolabeled CP component C1q was shown to be selectively accumulated in the ischemic reperfused myocardial lesions after a 45-minute coronary artery occlusion. The accumulation of C1q in the myocardial infarction lesions was reciprocally related to regional blood flow (Rossen et al., 1985) . Taken together, the current data suggest that the CP and the AP become activated in AMI. Early deposition of C components into infarcted myocardium indicates that tissue ischemia induces a profound change in the C activating potential of affected tissue.

5.3. Complement activating factors in ischemic injury

Lack of immunoglobulin deposits in both rat and human AMI lesions (I-IV) indicates that C activation had occurred in an antibody-independent manner, perhaps primarily via spontaneous activation of the AP of C and subsequently via a direct interaction of C1q with the cell constituents, like mitochondrial membranes (Pinckard et al., 1975; Ciglas et al., 1979; Peitsch et al., 1988) and intermediate filaments (Linder et al., 1979) . It is possible that the metabolic changes in the infarcted tissue (low pH, production of oxygen-derived free radicals) trigger spontaneous activation of C from the level of C3, C4 or C5 (Shingu and Nobunaga, 1984; Vogt et al., 1986) . It appears, however, that the mere presence of an activator of C is not sufficient to cause a C attack against autologous tissue. It was therefore postulated that ischemia may cause changes in the ability of the myocardial tissue to resist C activation.

5.4. Regulation of complement activation in human myocardium

5.4.1. Expression of complement membrane regulators

Human heart seems to be better protected against the late than the early C cascade, because regulators of the C3/C5 step, CR1, DAF and MCP, were absent or expressed at low levels, while regulators of MAC, protectin and C8bp, were strongly expressed in normal cardiomyocytes (I, II). Thus, the contact of plasma with insufficiently protected ischemic cardiomyocytes may lead to generation of the alternative and classical pathway C3/C5 convertases, which in turn would initiate formation of the terminal C complexes.

In myocardial infarction lesions the expression of C8bp was strong and in some lesions it was slightly increased (II). The latter could be due to the accumulation of a soluble plasma form of C8bp (Watts et al., 1990) in the lesions or an increased expression of cell membrane-bound C8bp. The inability of C8bp to prevent MAC deposition may be due to the low activity of C8bp in preventing C-mediated cell lysis. In functional assays C8bp (Watts et al., 1990) has at least a 100-fold weaker functional activity than protectin (CD59), which is now considered to be the main inhibitor of MAC (Meri et al., 1990b; Meri, 1994).

5.4.2. Molecular and functional properties of myocardial protectin

Immunoblotting, N-terminal amino acid sequence analysis and the ability to bind to the SC5b-8 complex in sucrose density ultracentrifugation analysis verified that the YTH53.1 immunoaffinity-purified myocardial molecule is myocardial protectin (I). IFL analysis showed that both human and rat protectin were strongly expressed in the sarcolemmal membranes of normal cardiomyocytes and in the endothelia of blood vessels (I, IV). In immunoblotting analysis using mAb against human and rat protectin, bands with an Mr of approximately 20 kDa were detected in extracts of both human and rat myocardium. The release of both human and rat myocardial protectin from sarcolemmal cell membranes by the PIPLC-enzyme suggested that both molecules have a GPI-an chor (I, IV). Similarities in the distribution, type of anchor and molecular weight between rat and human heart protectin indicated that the two represent species analogues of the same molecule. Despite similarities the human and rat protectins are antigenically distinct. The finding that urinary protectin has a higher apparent Mr in SDS-PAGE analysis than heart protectin could be explained by the lack of the phospholipid-tail from urinary protectin (I). Lack of the phospholipid would alter the mobility of the soluble form in electrophoresis.

5.4.3. Protectin in myocardial infarction

Protectin was found to be strongly expressed in normal human heart. In contrast with the expression of C8bp, the expression of protectin was totally lost or clearly diminished in the infarcted AMI lesions (I, II). This result suggested that the disappearance of protectin could make the human heart tissue sensitive to the membranolytic effects of the MAC. In the study based on autopsy specimens (I) it was not, however, possible to investigate a cause-effect-relationship between the loss of protectin and C activation.

In the study of human autopsy material, protectin was observed to be lost from sarcolemmal membranes in human AMI lesions in patients whose infarctions were aged 8 hours or more (I). The fact that protectin-negative lesions were accompanied by concomitant deposits of C5b-9 complexes, suggested that the loss of protectin could allow a full assembly of MAC in the ischemic lesions. The loss of protectin could also be detected by IFL analysis in the study of rat AMI (IV). The loss of rat protectin occurred gradually between 6 hours and 72 hours after the onset of AMI. A nearly total loss of rat protectin was observed by 24 hours and a total loss was evident in 72-hour-old AMI lesions. Thus, in the animal model the loss of rat protectin appeared to occur subsequently to C activation.

In the AMI model (IV) based on permanent ligation of the left coronary artery, the additional damaging effect of reflow of blood into the ischemic area (reperfusion injury) may not have been fully achieved (Engler et al., 1986; Ito et al., 1990; Entman et al., 1991; Dreyer et al., 1992) . It is likely, that in human AMI (with or without thrombolysis therapy) a more extensive reperfusion will occur. In human AMI more C activation products may be deposited in the infarcted lesions and augment local inflammatory responses. If the loss of rat protectin from the sarcolemmal membranes is a consequence of C activation, disappearance of protectin might be detected more clearly and at an earlier phase of AMI in those AMI models, where the ischemic phase is followed by reperfusion of blood.

Protectin is strongly expressed on vascular endothelial cells (Meri et al., 1991b). Unlike sarcolemmal membranes protectin expression did not appear to diminish on myocardial endothelial cells. This may be due to the high baseline expression of protectin on blood vessels or a less severe ischemic attack on endothelial cells. In addition, vascular endothelial cells are capable of upregulating protectin expression as a response to various inflammatory stimuli (Meri et al., 1993).

As indicated by control stainings, the loss of protectin expression from sarcolemmal membranes in AMI was not due to a non-specific loss of all membrane constituents. At present, the mechanism behind the selective loss of protectin from infarcted myocardium is unknown. The fact that myocardial protectin was sensitive to the PIPLC enzyme (I, IV) suggests the loss of protectin could be caused by a phospholipase (or proteolytic enzymes) released from the injured cardiomyocytes. Candidate enzymes include phospholipase C inside the cells (Low and Saltiel, 1988) or phospholipase D in human plasma (Davitz, et al., 1987). Both enzymes are capable of cleaving the GPI-anchor of protectin (Fig. 4). Alternatively, protectin may be "used up" in trying to prevent the assembly of MAC. It has been previously shown that shedding of MAC-containing cell membrane vesicles is an important mechanism by which nucleated cells are protected against C-mediated cell lysis (Campbell and Morgan, 1985; Morgan et al., 1987; Morgan, 1989b; Morgan, 1991) . Our observations suggest that vesiculation or exocytosis could also be involved in the removal of GPI-anchored molecules from the ischemic cardiomyocytes (I, IV). Protectin-containing vesicles were frequently observed in the infarcted areas. Hypothetically, an association of MAC with cell-membrane-bound protectin could trigger a cellular signaling machinery that leads to the vesiculation of GPI-anchored molecules. This suggestion is in accordance with a recent study showing that GPI-anchor-deficient erythrocytes have an impaired ability to vesiculate (Whitlow et al., 1993) . Similarly, it has been demonstrated that GPI-linked molecules may associate with intracellular tyrosine kinases with potential signaling functions (Stefanova et al., 1991) . The physiological function of removing GPI-anchored complement regulators could be to mark injured cells for clearance by the C system and phagocytes.

5.4.4. Vitronectin and clusterin in myocardial infarction - clearance of cell debris from injured myocardium?

Soluble regulators of C, C4bp, vitronectin and clusterin, were codeposited with the MAC in the AMI lesions (II, III). Previous studies have shown that vitronectin associates with MAC in AMI (Rus et al., 1987; Hugo et al., 1990) . However, the deposition of C4bp or clusterin in the AMI lesions has not previously been reported. The association of vitronectin and clusterin with a cytolytically active MAC suggests that these regulators had not inhibited the assembly of MAC on the cardiomyocytes during acute myocardial infarction. This is in accordance with a previous finding that at least vitronectin can associate with a cytolytically active MAC (Bhakdi et al., 1988) .

Vitronectin and clusterin are multifunctional plasma proteins. In addition to the C regulating activity vitronectin promotes cell attachment and binds to glycosaminoglycans and proteoglycans (Jenne and Stanley, 1985; Suzuki et al., 1985) , whereas clusterin has been suggested to be involved in tissue repair and remodeling after tissue damage or during development (Blaschuk et al., 1983; Cheng et al., 1988; Buttyan et al., 1989; Jenne and Tschopp, 1992) . Clearance of the deposited MAC complexes and membrane fragments could be mediated via binding of vitronectin and clusterin to phagocyte (Savill et al., 1990) or apolipoprotein (Jenne et al., 1991) receptors, respectively. A possible physiological role of vitronectin and clusterin in myocardial infarction areas would be to recruit clearance mechanisms for removal of injured cells and thereby participate in the healing process.

5.5. Interaction between protectin and lipoproteins

The studies of this thesis showed that protectin is lost from cardiomyocytes following ischemia-associated cell injury (I, IV). However, the fate of myocardial protectin released from cell membranes remained unknown. The possibility that lipid-tailed protectin could incorporate into lipoprotein particles in human plasma was considered because GPI-anchored molecules have the ability to incorporate into cell membranes (Meri et al., 1990b; Zhang et al., 1992). It was observed (V) that high density lipoprotein particles in human plasma were capable of serving both as an acceptor and as a donor of GPI-anchored protectin in vitro. Furthermore, anti-protectin affinity chromatography and immunoblotting analysis indicated that purified high density lipoprotein contained small quantities of protectin. Under normal conditions protectin is present in human serum at relatively low concentrations (Meri et al., 1991a; Ratnoff et al., 1993; Lehto et al., unpublished observations). Although the ratio between lipid-tailed and soluble protectin in plasma is unknown, it is highly likely that the HDL-associated form represents the lipid-tailed form. At present, the concentrations of protectin in the sera of AMI patients have not yet been determined. It remains to be examined whether AMI or other forms of tissue injury lead to increased serum concentrations of protectin. According to the results of the present study, lipid-tailed protectin released from in jured tissues may be incorporated into HDL-particles and be recycled into other tissues (V). HDL-particles thus resemble "prostasomes", cholesterol- and protectin-rich particles in human seminal plasma (Rooney et al., 1993) .

High density lipoprotein particles and apo A-I have previously been shown to be capable of inhibiting C-mediated cell lysis (Rosenfeld et al., 1983; Packman et al., 1985) . The transfer of protectin from one cell type to another by lipoproteins may be one mechanism whereby HDL provides protection against C attack. In part, the inhibitory effect of HDL could be due to clusterin, which is associated with apo A-I in HDL particles (Jenne et al., 1991; Choi-Miura et al., 1993) . Some of the antiatherogenic activity of HDL (Tall, 1990) could, in fact, be due to the C inhibiting activities of high density lipoproteins (Rosenfeld et al., 1983; Packman et al., 1985) . Protection against excessive C attack is probably important, since C deposits have been detected in human atherosclerotic lesions (Niculescu et al., 1987a and b) , the amount of MAC correlates with the degree of experimental atherosclerosis (Seifert et al., 1989) and C depletion inhibits the development of experimental atherosclerosis (Geertinger et al., 1970; Geertinger et al., 1975). However, further studies are required to assess the specific role of the C system in the pathogenesis of atherosclerosis.

5.6. A synopsis of the role of the complement system in the pathogenesis of myocardial infarction

Activation of the C system appears to be a direct consequence of changes in the myocardium. Several factors probably contribute to the initiation and propagation of C activation in AMI. Spontaneous activation of the AP might be accelerated by acidosis. Ischemia leads to alterations on the endothelial and cardiomyocyte cell membranes, which could transform the cell surfaces to activators of C. Formation of free oxygen radicals might activate the C system directly or by damaging the cell membrane. Contact of C components, notably of C1q, with intracellular structures, like intermediate filaments or mitochondrial membranes, might also activate the C system. Because of the relative deficiency of C3/C5 convertase regulators (II) and acquired loss of protectin from the cardiomyocyte cell membranes (I, IV), the ischemic myocardium is unduly sensitive to the inflammation-promoting and cytotoxic activities of complement.

The C system is capable of causing injury to the ischemic lesions by both direct and indirect mechanisms. The direct effects are caused by the MAC complex, whereas the indirect damage is mediated by the anaphylatoxins, C3a and C5a, and to some extent also by MAC via its cell stimulatory activities. C5a has a potent and C3a a weak chemotactic activity for neutrophils and both anaphylatoxins can also activate neutrophils (Hugli, 1984; Vogt, 1985a) . Activated neutrophils produce tissue damaging oxygen free radicals and lysosomal enzymes (Hugli, 1984; McCord, 1985; Vogt, 1986) and adhere to capillaries of ischemic myocardial areas thereby impairing regional perfusion (Hugli, 1984; Vogt, 1985a; Schmid-Schoenleben and Engler, 1987; Stahl et al., 1990; Kilgore and Lucchesi, 1993). Activated neutrophils trigger the lipo- and cyclo-oxygenase pathways, that further enhance tissue injury by their chemotactic, leukocyte- and platelet-activating products (Mehta et al., 1988; Ito et al., 1990) . C3a can also aggregate and activate platelets to secrete serotonin and thromboxane A2 (Polley and Nachman, 1983) .

Deposition of MAC on cell membranes will lead to the loss of cell integrity by the formation of transmembrane channels. Sublytic concentrations of MAC have also been shown to lead to calcium influx into the cells, membrane depolarization, arachidonate conversion to thromboxane A2 and activation of cellular protein kinases (Mehta et al., 1988; Morgan, 1989b) . A further increase in the intracellular calcium concentration is presumed to lead to an irreversible cell injury (Farber, 1982) . Thromboxane A2 is a central mediator of coronary vasoconstriction (Willerson et al., 1986) . The formation of MAC complexes on the membranes of endothelial cells can augment the clotting cascade by stimulating endothelial cells to secrete the von Willebrand factor (Hattori et al., 1989; Hamilton et al., 1990) . As this effect is enhanced by the treatment of endothelial cells with anti-protectin antibodies (Hamilton et al., 1990), a similar phenomenon would be expected to occur after membrane shedding of protectin. MAC deposition on platelets will activate them, and this activation is also increased by treatment with anti-protectin antibody prior to exposing the platelets to the terminal C cascade (Sims and Wiedmer, 1991) .

In the present studies we have suggested that the physiological purpose of C activation in the ischemic areas is the postischemic clearance of injured cellular material. It is believed that during the secondary phase of C attack the irreversibly injured infarcted cardiomyocytes are demarcated from viable cells by the MAC. Deposited C factors principally iC3b and vitronectin, are thought to mark the cells for clearance by phagocytes. Phagocyte receptors possibly involved in the clearance process include CR3 (CD11b/CD18), CR1 (CD35), vitronectin receptor (CD51) and a hypothetical "clusterin receptor". Formation of MAC may trigger vesiculation of protectin-containing particles thereby further sensitizing the tissue to the cytolytic activity of MAC. On the other hand, release of protectin to circulation and possible association with HDL particles may allow reuse of this important molecule elsewhere in the human body (V). Unfortunately, all too often the consequences of AMI prove fatal to the patients. Future clinical studies will show whether therapeutic intervention of AMI by C suppressing agents, including analogues of CR1 and protectin, could alleviate the suffering of the patients.

6. Summary and concluding remarks

Within recent years, studies on experimental myocardial infarction have shown that complement (C) activation has a significant pathogenetic role in the development of ischemic tissue injury. It is, however, not understood why the C system becomes triggered during AMI and why the autologous heart tissue loses its tolerance against the cytotoxic C system. The present study has examined factors involved in the regulation of C activation and focused on the role of terminal C pathway regulators in protecting normal and infarcted tissue against C-mediated cell damage.

Deposition of C components, activation products and soluble regulators (vitronectin, clusterin, C4bp) as well as expression of membrane regulators of C (protectin, CR1, DAF, MCP, C8bp) were analyzed by immunohistological and immunochemical methods. Heart specimens were obtained at autopsies from patients who had died because of AMI. Additional studies were performed in rats that were subjected to an experimental AMI. Deposits of C components and activation products (C1q, C3c, C3d, C4, C5, C6, C8, C9, C5b-9 neoantigens) were detected in human AMI lesions aged 8 hours or more. Formation of the membrane attack complex (MAC) within the infarction lesions was associated with deposits of vitronectin, clusterin and C4bp. The MAC inhibitor protectin (CD59) was strongly expressed in normal human cardiomyocytes, but it was lost from AMI lesions in association with deposits of MAC. The C3/C5 convertase regulators, CR1, DAF or MCP, were not, or only weakly, expressed in human cardiomyocytes of normal and infarcted areas. In the rat AMI model deposits of C1, C3, C8 and C9 were observed in AMI lesions aged 3 hours or more. C3 deposits, but not those of C1, C8 or C9, were seen in 2-hour-old AMI lesions suggesting that C activation is initiated via the alternative pathway. Rat protectin was observed to become gradually lost from AMI lesions at day 1 and thereafter. Protectin-positive vesicle-like structures were often seen in both human and rat AMI lesions. This suggests that shedding of cell membrane vesicles is a potential elimination mechanism of protectin from cell membranes in areas of tissue injury. In vitro experiments showed that GPI-anchored lipid-tailed protectin incorporated readily into high density lipoprotein (HDL) particles in human plasma. In addition, HDL particles were found to be able to transfer protectin between e.g. erythrocytes and endothelial cells.

Results of the present study show that C activation is a relatively early event in AMI. After the initial occlusion of a coronary artery and development of ischemia the C system is involved in the pathogenesis of AMI during the reperfusion stage. The complement anaphylatoxins, C3a and C5a, and a direct MAC-mediated membranolytic effect induce an inflammatory reaction and extend AMI associated tissue damage. Tolerance of cardiomyocytes against a MAC attack appears to become lost because of shedding of the MAC inhibitor protectin from the cell membranes. During the late phase of AMI the terminal C complexes in association with the multifunctional molecules, vitronectin and clusterin, may participate in the clearance of irreversibly injured cardiomyocytes. The incorporation of protectin into HDL suggests that the lipoprotein particles could be involved in the recycling of protectin, and possibly of other GPI-anchored proteins, from injured cells to viable tissues.

It remains unknown what the initial changes in ischemic cardiomyocytes and endothelial cells are that transform these cells into activators of C. Regardless of the mechanisms behind C activation in AMI, it will be of scientific and clinical interest to investigate whether some form of protectin or its analogue could be useful in the therapy of AMI. One potential form of therapy in reperfusion injury is based on results presented in paper V and on other studies showing that the GPI-lipid-anchored protectin is able to reincorporate into cell membranes. The lipid-tailed protectin might be transferred e.g. on the surface of liposomes to the site(s) of tissue injury. However, it is possible that inhibition of the C system at the C3/C5 level - e.g. by soluble CR1 (Weisman et al., 1990) - will be therapeutically more efficient than the putative treatment with protectin or its recombinant analogue. In conclusion, once the mechanisms behind the activation of C in AMI are entirely understood, more specific therapy to inhibit C-mediated injury in AMI can be developed.

7. ACKNOWLEDGMENTS

This study was carried out at the Department of Bacteriology and Immunology, University of Helsinki during the years 1990-1994. I thank Professor Olli Mäkelä for placing the excellent facilities of the Department at my disposal. Professor Eero Saksela, Department of Pathology, University of Helsinki is gratefully acknowledged for providing the opportunity to obtain human autopsy specimens. Professor Frej Fyhrquist, Head of the Unit of Clinical Physiology, the Minerva Foundation Institution for Medical Research is especially acknowledged for placing the unique facilities for the animal studies at my disposal.

I am most grateful to my supervisor Docent Seppo Meri, who introduced me to the field of immunology. I could only admire the unfailing energy and enthusiasm expressed in his scientific activity during my work.

I would like to thank Docent Sirpa Jalkanen and Docent Timo Paavonen for their constructive criticism of the manuscript.

I express my warmest thanks to Docent Pekka Laurila for sharing his vast knowledge of pathology. I wish to thank Professor Christian Ehnholm and Docent Matti Jauhiainen, Docent Ilkka Tikkanen, Karri Helin, Paul Morgan and Marc Baumann for pleasant collaboration.

All members of the "Complement Club" are acknowledged for their friendship and contribution to this work. The stimulating discussions with Timo Lehto, Juha Hakulinen, Sami Junnikkala and Sakari Jokiranta are greatly appreciated.

Special thanks are due to my colleagues at the Theoretical Institutes of Meilahti, National Public Health Institute, Helsinki, and Minerva Institute for Medical Research for their positive attitude and support of this work. I am particularly grateful to Aaro Miettinen, Harry Holthöfer, Ilkka Seppälä, Martti Vaara, Ilmo Leivo, Perttu Lindsberg, Risto Renkonen, Mikko Hurme, Pentti Kuusela, Jorma Tissari, Jukka Reivinen and Sirkka Kontiainen for many helpful and inspiring discussions. Professor Carl-Henrik von Bonsdorff, Department of Virology, University of Helsinki, is gratefully acknowledged for help in electron microscopic analysis. I sincerely thank Don Haut for revising the language of this thesis.

I am greatly indebted to Kirsti Tuominen, Ulla Ikävalko, Riitta Väisänen, Ritva Majuri and Marita Siren for their excellent technical assistance.

Finally, the everlasting patience and love of my wife Liisa deserves a recognition that can be fully appreciated only by those who have faced a similar challenge.

This work has been financially supported by the Ilmari Ahvenainen Foundation, the Ida Montin Foundation, the Finnish Medical Foundation Duodecim, the Sigrid Jusélius Foundation and the Academy of Finland.

Helsinki, December 1994

Antti Väkevä 8. REFERENCES

Arroyave, C and Müller-Eberhard, HJ: Interactions between human C5, C6 and C7 and their functional significance in complement-dependent cytolysis. J Immunol. 3: 536-545, 1973.

Bennett, W, Yawn, D, Migliore, J, Young, J, Pratt, C, Raizner, A, Roberts, R and Bolli, R: Activation of the complement system by recombinant tissue plasminogen activator. J Am Coll Card. 10: 627-632, 1987.

Bhakdi, S, Kaflein, R, Halstensen, TS, Hugo, F, Preissner, KT and Mollnes, TE: Complement S-protein (vitronectin) is associated with cytolytic membrane bound C5b-9 complexes. Clin Exp Immunol. 74: 459-464, 1988.

Blanche, PJ, Gong, EL, Forte, TM and Nichols, AV: Characterization of human high density lipoproteins by gradient gel electrophoresis. Biochim Biophys Acta. 665: 408-419, 1981.

Blaschuk, OW, Burdzy, K and Fritz, IB: Purification and characterization of a cell-aggregating factor (clusterin), the major glycoprotein in ram rete testis fluid. J Biol Chem. 12: 7714-7720, 1983.

Buttyan, R, Olsson, CA, Pintar, J, Chang, C, Bandyk, M, Ng, PY and Sawczuk, IS: Induction of the TRPM-2 gene in cells undergoing programmed death. Mol Cell Biol. 9: 3473-3481, 1989.

Campbell, AK, Daw, RA, Hallett, MB and Luzio, JP: Direct measurement of the increase in intracellular free calcium ion concentration in response to the action of complement. Biochem J. 194: 551-560, 1981.

Campbell, AK and Morgan, BP: Monoclonal antibodies demonstrate protection of polymorphonuclear leucocytes against complement attack. Nature 317: 164-166, 1985.

Camussi, G, Salvidio, G, Biesecker, G, Brentjens, J and Andres, G: Heymann antibodies induce complement-dependent injury of rat glomerular visceral epithelial cells. J Immunol. 139: 2906-2914, 1987.

Cheng, CY, Mathur, PP and Grima, J: Structural analysis of clusterin and its subunits in ram rete testis fluid. Biochemistry 27: 4079-4088, 1988.

Choi-Miura, NH, Sakamoto, T, Tobe, T, Nakano, Y and Tomita, M: The role of HDL consisting of SP-40,40, apo A-I, and lipids in the formation of SMAC of complement. J Biochem. 113: 484-487, 1993.

Ciglas, PC, Pinckard, RN and Olson, MS: In vitro activation of complement by isolated human heart subcellular membranes. J Immunol. 122: 146-151, 1979.

Clas, FG, Schmid, TG and Loos, M: The role of the classical pathway for the bactericidal effect of normal sera against gram-negative bacteria. Current topics in microbiology and immunology: Bacteria and complement. Berlin, Springer-Verlag. Editor M. Loos. 19-72, 1985.

Cole, J, Housley, GA, Dykman, TR, MacDermott, RP and Atkinson, JP: Identification of an additional class of C3-binding membrane proteins of human peripheral blood leukocytes and cell lines. Proc Natl Acad Sci USA. 82: 859-864, 1986.

Cooper, NR, Jensen, FC, Welsh, RM and Oldstone, MBA: Lysis of RNA tumor viruses by human serum: Direct antibody independent triggering of the classical complement pathway. J Exp Med. 144: 970-980, 1976.

Cosio, FG, Zager, RA and Sharma, HM: Atheroembolic renal disease causes hypocomplementaemia. Lancet 2: 118-122, 1985.

Crawford, MH, Grover, FL, Kolb, WP, McMahan, CA, O'Rourke, RA, McManus, LM and Pinckard, RN: Complement and neutrophil activation in the pathogenesis of ischemic myocardial injury. Circulation 78: 1449-1458, 1988.

Davies, A, Simmons, DL, Hale, G, Harrison, RA, Tighe, H, Lachmann, PJ and Waldmann, H: CD59, an Ly-6 like protein expressed in human lymphoid cells, regulates the action of the complement membrane attack complex on homologous cells. J Exp Med. 170: 637-654, 1989.

DeScheerer, I.: Association of anti-heart antibodies and circulating immune complexes in the post-pericardiotomy syndrome. Clin Exp Immunol. 57: 423-430, 1984.

Dreyer, WJ, Michael, LH, Nguyen, T, Smith, CW, Anderson, DC, Entman, ML and Rossen, RD: Kinetics of C5a release in cardiac lymph of dogs experiencing coronary artery ischemia-reperfusion injury. Circ Res. 71: 1518-1524, 1992.

Earis, JE and Bernstein, A: Complement activation after myocardial infarction. Chest 87: 186-190, 1985.

Edgell, CJS, McDonald, CG and Graham, JB: Permanent cell line expressing factor VIII-related antigen established by hybridization. Proc Natl Acad Sci USA. 80: 3734-3739, 1983.

Ehnholm, C, Bozas, SE, Tenkanen, H, Kirszbaum, L, Metso, J, Murphy, B and Walker, ID: The apolipoprotein A-I binding protein of placenta and the SP-40,40 protein of human blood are different proteins which both bind to apolipoprotein A-I. Biochim Biophys Acta. 1086: 255-260, 1991.

Engler, RL, Dahlgren, MD, Peterson, MA, Dobbs, A and Schmid-Schonbein, GW: Accumulation of polymorphonuclear leukocytes during 3h experimental myocardial ischemia. Am J Physiol. 251: H93-H100, 1986.

Entman, ML, Michael, L, Rossen, RD, Dreyer, WJ, Anderson, DC, Taylor, AA and Smith, CW: Inflammation in the course of early myocardial ischemia. FASEB J. 5: 2529-2537, 1991.

Esser, AF: The membrane attack pathway of complement. Year Immunol. 1989-90. 6: 229-244, 1990.

Farber, JL: Membrane injury and calcium homeostasis in the pathogenesis of coagulative necrosis. Lab Invest. 47: 114-118, 1982.

Fearon, DT: Regulation of the amplification C3 convertase of human complement by an inhibitory protein isolated from human erythrocyte membranes. Proc Natl Acad Sci USA. 76: 5867-5872, 1979.

Franci, D, Gardinali, M, Conciato, L, Cafaro, C, Pozzoni, L and Agostoni, A: Abrupt complement activation and transient neutropenia in patients with acute myocardial infarction treated with streptokinase. Circulation 89: 76-80, 1994.

Geertinger, P and Sørensen, H: Complement as a factor in arteriosclerosis. Acta Pathol Microbiol Scand Sect A. 78: 284-290, 1970.

Geertinger, P and Sørensen, H: On the reduced atherogenic effect of cholesterol feeding on rabbits with congenital complement (C6) deficiency. Artery 1: 177-185, 1975.

Hamilton, KK, Zhao, J, Rollins, S, Stewart, BH and Sims, PJ: Regulatory control of the terminal complement proteins at the surface of human endothelial cells - neutralization of a C5b-9 inhibitor by antibody to CD59. Blood 76: 2572-2577, 1990.

Hattori, R, Hamilton, KK, McEver, RP and Sims, PJ: Complement proteins C5b-9 induce secretion of high molecular weight multimers of endothelial von Willebrand factor and translocation of granule membrane protein GMP-140 to the cell surface. J Biol Chem. 264: 9053-9060, 1989.

Havel, RJ, Eder, HA and Bragdon, JR: The determination and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest. 34: 1345-1353, 1955.

Heimer, GV and Taylor, CED: Improved mountant for immunofluorescence preparations. J Clin Pathol. 27: 254-256, 1974.

Hill, J, Lindsay, TF, Ortiz, F, Yeh, CG, Hechtman, HB and Moore Jr, FD: Soluble complement receptor type 1 ameliorates the local and remote organ injury after intestinal ischemia-reperfusion in the rat. J Immunol. 149: 1723-1728, 1992.

Hirsch, RL, Winkelstein, JA and Griffin, DE: The role of complement in viral infections. III. Activation of the classical and alternative complement pathways by Sindbis virus. J Immunol. 124: 2507-2520, 1980.

Holguin, MH, Wilcox, LA, Bernshaw, NJ, Rosse, WF and Parker, CJ: Erythrocyte membrane inhibitor of reactive lysis: effects of phosphatidylinositol-specific phospholipase C on the isolated and cell-associated protein. Blood 75: 284-293, 1990.

Holthöfer, H, Virtanen, I, Pettersson, E, Törnroth, T, Alftan, O, Linder, E and Miettinen, A: Lectins as fluorescence microscopic markers for saccharides in the human kidney. Lab Invest. 45: 391-399, 1981.

Homeister, JW, Satoh, PS, Lucchesi, BR: Effects of complement activation in the isolated heart. Role of the terminal complement components. Circ Res. 71: 303-319, 1992.

Homeister, JW, Satoh, PS, Kilgore, KS, Lucchesi, BR: Soluble complement receptor type 1 prevents human complement-mediated damage of the rabbit isolated heart. J Immunol. 150: 1055-1064, 1993.

Hughes, TR, Piddlesden, SJ, Williams, JD, Harrison, RA and Morgan, BP: Isolation and characterization of a membrane protein from rat erythrocytes which inhibits lysis by the membrane attack complex of rat complement. Biochem J. 284: 169-176, 1992.

Hugli, TE: Structure and function of the anaphylatoxins. Springer Semin Immuno-pathol. 7: 193-219, 1984.

Hugo, F, Hamdoch, T, Mathey, D, Schäfer, H and Bhakdi, S: Quantitative measurement of SC5b-9 and C5b-9(m) in infarcted areas of human myocardium. Clin Exp Immunol. 81: 132-136, 1990.

Iida, K, Whitlow, M and Nussenzweig, V: Membrane vesiculation protects erythrocytes from destruction by complement. J Immunol. 147: 2638-2642, 1991.

Ito, B, Schmid-Schönleben, G and Engler, R: Effects of leukocyte activation on myocardial vascular resistance. Blood Cells 16: 145-166, 1990.

Ito, BR, Roth, DM and Engler, RL: Thromboxane A2 and peptidoleukotrienes contribute to the myocardial ischemia and contractile dysfunction in response to intracoronary infusion of complement C5a in pigs. Circ Res. 66: 596-607, 1990.

Jauhiainen, M, Metso, J, Pahlman, R, Blomqvist, S, van Tol, A and Ehnholm, C: Human plasma phospholipid transfer protein causes high density lipoprotein conversion. J Biol Chem. 6: 4032-4036, 1993.

Jenne, D and Stanley, KK: Molecular cloning of S-protein, a link between complement, coagulation and cell-surface adhesion. EMBO J. 4: 3153-3157, 1985.

Jenne, DE and Tschopp, J: Molecular structure and functional characterization of a human complement cytolysis inhibitor found in blood and seminal plasma: identity to sulphated glycoprotein 2, a constituent of rat testis fluid. Proc Natl Acad Sci USA. 86: 7123-7127, 1989.

Jenne, DE, Lowin, B, Peitsch, MC, Bottcher, A, Schmitz, G and Tschopp, J: Clusterin (complement lysis inhibitor) forms a high density lipoprotein complex with apolipoprotein A-I in human plasma. J Biol Chem. 266: 11030-11036, 1991.

Jenne, DE and Tschopp, J: Clusterin: the intriguing guises of a widely expressed glycoprotein. Trends Biochem Sci. 17: 154-159, 1992.

Jones, J, Laffafian, I and Morgan, BP: Purification of C8 and C9 from rat serum. Compl Inflamm. 7: 42-51, 1990.

Kagiyama, A, Savage, HE, Michael, LH, Hansson, G, Entman, ML and Rossen, RD: Molecular basis of complement activation in ischemic myocardium: identification of specific molecules of mitochondrial origin that bind human C1q and fix complement. Circ Res. 64: 607-615, 1989.

Kaplan, MH and Frengley, JD: Autoimmunity to the heart in cardiac disease. Am J Cardiol. 24: 458-473, 1969.

Kaplan, MH and Volanakis, JE: Interaction of C-reactive protein complexes with the complement system. I. Consumption of human complement associated with the reaction of C-reactive protein with pneumococcal C-polysaccharide and the choline phosphatides lecithin and sphingomyelin. J Immunol. 112: 2135-2140, 1974.

Kilgore, KS and Lucchesi, BR: Reperfusion injury after myocardial infarction: the role of free radicals and the inflammatory response. Clin Biochem. 26: 359-370, 1993.

Laemmli, UK: Cleavage of structural proteins during the assembly of the head of the bacteriophage T4. Nature 227: 680-685, 1970.

Langlois, PF and Gawryl, MS: Detection of the terminal complement complex in patient plasma following acute myocardial infarction. Atherosclerosis 70: 95-105, 1988.

LeClair, KP, Palfree, RGE, Floor, PM, Hammerling, U and Bothwell, A: Isolation of a murine Ly6 cDNA reveals a new multigene family. EMBO J. 5: 3227-3334, 1986.

Lehto, T and Meri, S: Interaction of soluble CD59 with the terminal complement complexes: CD59 and C9 compete for a nascent epitope on C8. J Immunol. 151: 4941-4949, 1993.

Linder, E, Lehto, VP and Stenman, S: Activation of complement by cytoskeletal intermediate filaments. Nature 278: 176-178, 1979.

Lindsay, TF, Hill, J, Ortiz, F, Fudolph, A, Valeri, CCR, Hechtman, HB and Moore Jr., FD: Complement inhibition prevents skeletal muscle and pulmonary permeability following hindlimb ischemia-reperfusion. FASEB J. 6: A1049 (abstr.), 1992.

Low, MG and Saltiel, AR: Structural and functional roles of glycosyl-phosphatidylinositol in membranes. Science 239: 268-275, 1988.

Lu, J, Thiel, S, Wiedemann, H, Timpl, R and Reid, KBM: Binding of the pentamer/hexamer forms of mannan-binding protein to zymosan activates the proenzyme C1r2C1s2 complex, of the classical pathway of complement, without involvement of C1q. J Immunol. 144: 2287-2294, 1990.

Markwell, MA, Haas, SM, Tolbert, NE and Bieber, LL: Protein determination in membrane and lipoprotein samples: manual and automated procedures. Meth Enzymol. 72: 296-299, 1981.

Maroko, PR, Carpenter, CB, Chiariello, M, Fishbein, MC, Randvany, P, Kostman, JD and Hale, SL: Reduction by cobra venom factor of myocardial necrosis after coronary artery occlusion. J Clin Invest. 61: 661-670, 1978.

Martin, DE, Chiu, FJ, Gigli, I and Müller-Eberhard, HJ: Killing of human melanoma cells by the membrane attack complex of human complement as a function of its molecular composition. J Clin invest. 80: 226-233, 1987.

Mathey, D, Schofer, J, Schäfer, HJ, Hamdoch, T, Joachim, HC, Ritgen, A, Hugo, F, Bhakdi, S: Early accumulation of the terminal complement-complex in the ischemic myocardium after reperfusion. Eur Heart J. 15: 418-423, 1994.

Matsushita, M and Fujita, T: Activation of the classical complement pathway by mannose-binding protein in association with a novel C1s-like serine protease. J Exp Med. 176: 1497-1502, 1992.

McCord, JM: Oxygen-derived free radicals in post-ischemic tissue injury. N Engl J Med. 312: 886-890, 1985.

McManus, LM, Kolb, WP, Crawford, MH, O'Rourke, RA, Grover, FL and Pinckard, RN: Complement localization in ischemic baboon myocardium. Lab Invest. 66: 436-442, 1983.

Mehta, JL, Nichols, WW and Mehta, P: Neutrophils as potential participants in acute myocardial ischaemia: Relevance to reperfusion. J Am Coll Cardiol. 11: 20-28, 1988.

Meri, S and Pangburn, MK: Discrimination between activators and nonactivators of the alternative pathway of complement: regulation via a sialic acid/polyanion binding site on factor H. Proc Natl Acad Sci USA. 87: 3982-3986, 1990a.

Meri, S, Morgan, BP, Davies, A, Daniels, RH, Olavesen, MG, Waldmann, H and Lachmann, PJ: Human protectin (CD59), an 18,000-20,000 MW complement lysis restricting factor, inhibits C5b-8 catalysed insertion of C9 into lipid bilayers. Immunology 71: 1-9, 1990b.

Meri, S, Väkevä, A, Laari, T and Lachmann, PJ: Soluble forms of CD59-antigen: distribution in body fluids and functional activity. Compl Inflamm (abstr.). 8: 193, 1991a.

Meri, S., Waldmann, H, and Lachmann, PJ. Distribution of protectin (CD59), a complement membrane attack inhibitor, in normal human tissues. Lab Invest. 65: 532-537, 1991b.

Meri, S, Mattila, P and Renkonen, R: Regulation of CD59 expression on the human endothelial cell line EA.hy 926. Eur J Immunol. 23: 2511-2516, 1993.

Meri, S and Pangburn, MK: Regulation of alternative pathway complement activation by glycosaminoglycans: Specificity of the polyanion binding site on factor H. Biochem Biophys Res Comm. 198: 52-59, 1994.

Meri, S: Protectin (CD59). Complement lysis inhibitor and prototype domain in a new protein superfamily. Immunologist 2: 149-155, 1994.

Mollnes, TE, Tambs, KE, Myreng, Y and Engebretsen, LF: Acute phase reactants and complement activation in patients with acute myocardial infarction. Compl Inflamm. 5: 33-39, 1988.

Morgan, BP, Imagawa, DK, Dankert, JR and Ramm, LE: Complement lysis of U937, a nucleated mammalian cell line in the absence of C9: effect of C9 on C5b-8 mediated cell lysis. J Immunol. 136: 3402-3406, 1986.

Morgan, BP, Dankert, JR and Esser, AF: Recovery of human neutrophils from complement attack: removal of the membrane attack complex by endocytosis and exocytosis. J Immunol. 138: 246-253, 1987.

Morgan, BP: Mechanisms of tissue damage by the membrane attack complex of complement. Compl Inflamm. 6: 104-111, 1989a.

Morgan, BP: Complement membrane attack on nucleated cells: resistance, recovery and non-lethal effects. Biochem J. 264: 1-14, 1989b.

Morgan, BP: Complement. Clinical aspects and relevance to disease. London, Academic Press Ltd, 1991.

Morgan, BP and Meri, S: Membrane proteins that protect against complement lysis. Springer Semin Immunopathol. 15: 369-396, 1994.

Müller-Eberhard, HJ: The membrane attack complex of complement. Annu Rev Immunol. 4: 503-528, 1986.

Müller-Eberhard, HJ: Molecular organization and function of the complement system. Annu Rev Biochem. 57: 321-347, 1988.

Nachlas, MN and Shnitka, TK: Macroscopic identification of early myocardial infarcts by alterations in dehydrogenase activity. Am J Pathol. 42: 379-405, 1963.

Nicholson-Weller, A, Burge, J, Fearon, DT, Weller, PF and Austen, KF: Isolation of a human erythrocyte membrane glycoprotein with decay-accelerating activity for C3 convertase of the complement system. J Immunol. 129: 184-189, 1982.

Niculescu, F, Hugo, F, Rus, HG, Vlaicu, R and Bhakdi, S: Quantitative evaluation of the terminal C5b-9 complement complex by ELISA in human atherosclerotic arteries. Clin Exp Immunol. 69: 477-483, 1987a.

Niculescu, F, Rus, HG and Vlaicu, R: Immunohistochemical localization of C5b-9, S-protein, C3d and apolipoprotein B in human arterial tissues in atherosclerosis. Atherosclerosis 65: 1-11, 1987b.

Ohta, M, Okada, M, Yamashina, I and Kawasaki, T: The mechanism of carbohydrate-mediated complement activation by the serum mannan-binding protein. J Biol Chem. 265: 1980-1984, 1990.

Okada, N, Harada, R, Fujita, T and Okada, H: A novel membrane glycoprotein capable of inhibiting membrane attack by homologous complement. Int Immunol. 1: 205-208, 1989.

Packman, CH, Rosenfeld, S and Leddy, JP: High-density lipoprotein and its apolipoproteins inhibit cytolytic activity of complement. Studies on the nature of the inhibitory moiety. Biochim Biophys Acta. 812: 107-115, 1985.

Pangburn, MK, Schreiber, RD and Müller-Eberhard, HJ: Human complement C3b inactivator: Isolation, characterization, and demonstration of an absolute requirement for the serum protein ß1H for cleavage of C3b and C4b in solution. J Exp Med. 146: 257-270, 1977.

Pangburn, MK and Müller-Eberhard, HJ: The alternative pathway of complement. Spinger Semin Immunopathol. 7: 163-170, 1984.

Peitsch, M, Tschopp, J, Kress, A and Isliker, H: Antibody-independent activation of the complement system by mitochondria is mediated by cardiolipin. Biochem. J. 249: 495-500, 1988.

Pinckard, RN, Olson, MS, Kelley, RE, DeHeer, DH, Palmer, JD, O'Rourke, RA and Goldfein, S: Antibody-independent activation of human C1 after interaction with heart subcellular membranes. J Immunol. 110: 1376-1382, 1973.

Pinckard, RN, Olson, MS, Ciglas, PC, Terry, R, Boyer, JT and O'Rourke, RA: Consumption of classical complement components by heart subcellular membranes in vitro and in patients after acute myocardial infarction. J Clin Invest. 56: 740-750, 1975.

Pinckard, RN, O'Rourke, RA, Crawford, MH, Grover, FL, McManus, LM, Ghidoni, JJ, Storrs, SB and Olson, MS: Complement localization and mediation of ischemic injury in baboon myocardium. J Clin Invest. 66: 1050-1056, 1980.

Ploug, M and Ellis, V: Structure-function relationship in the receptor for urokinase-type plasminogen activator. Comparison to other members of the Ly-6 family and snake venom a-neurotoxins. FEBS Lett. 349: 163-168, 1994.

Podack, ER and Müller-Eberhard, HJ: Isolation of human S-protein, an inhibitor of the membrane attack complex of complement. J Biol Chem. 254: 9808-9814, 1979.

Polley, MJ and Nachman, RL: Human platelet activation by C3a and C3a desarg. J Exp Med. 150: 257-268, 1983.

Ratnoff, O and Naff, G: The conversion of C'1s to C'1 esterase by plasmin and trypsin. J Exp Med. 127: 337-358, 1967.

Ratnoff, WD, McCord, MC, Meri, S, Okada, H, Lachmann, PJ: Functional, immunoreactive, soluble CD59 is present in human plasma. Mol Immunol (Abstr.). 30: 46, 1993.

Reimer, KA and Jennings, RB: The "wavefront phenomenon" of myocardial ischemic cell death. II. Transmural progression of necrosis within the framework of ischemic bed size (myocardium at risk) and collateral blood flow. Lab Invest. 40: 633-640, 1979.

Riepponen, P, Marniemi, J and Rautaoja, T: Determination of apolipoproteins A-I and B in serum. Scand J Lab Invest. 47: 739-745, 1987.

Roldan, AL, Cubellis, MV and Masucci, MT: Cloning and expression of the receptor for human urokinase plasminogen activator, a central molecule in plasmin dependent proteolysis. EMBO J. 9: 8404-8410, 1990.

Rollins, SA and Sims, PJ: The complement inhibitory activity of CD59 resides in its capacity to block incorporation of C9 into membrane C5b-9. J Immunol. 144: 3478-3485, 1990.

Rooney, IA, Atkinson, JP, Krul, ES, Schonfeld, G, Polakoski, K, Saffitz, JE and Morgan, BP: Physiologic relevance of the membrane attack complex inhibitory protein CD59 in human seminal plasma: CD59 is present on extracellular organelles (prostasomes), binds cell membranes, and inhibits complement-mediated lysis. J Exp Med. 177: 1409-1420, 1993.

Röschlau, P, Bernt, E and Gruber, W: Enzymatische bestimmung des gesamtcholesterings in serum. Z Klin Biochem. 12: 226-230, 1974.

Rosenfeld, SI, Packman, CH and Leddy, JP: Inhibition of lytic action of cell-bound terminal complement components by human high density lipoproteins and apoproteins. J Clin Invest. 71: 795-808, 1983.

Ross, GD and Medof, ME: Membrane complement receptors specific for bound fragments of C3. Adv Immunol. 37: 217-267, 1985.

Rossen, RD, Michael, LH, Kagiyama, A, Savage, HE, Hanson, G, Reisberg, MA, Moake, JN, Kim, SH, Self, D, Weakley, S, Giannini, E and Entman, ML: Mechanism of complement activation after coronary artery occlusion: evidence that myocardial ischemia in dogs causes release of constituents of myocardial subcellular origin that complex with human C1q in vivo. Circ Res. 62: 572-584, 1988.

Rossen, RD, Swain, JL, Michael, LH, Weakley, S, Giannini, E and Entman, ML: Selective accumulation of the first component of complement and leukocytes in ischemic canine heart muscle. A possible initiator of an extra myocardial mechanism of ischemic injury. Circ Res. 57: 119-130, 1985.

Rus, HG, Niculescu, F and Vlaicu, R: Presence of C5b-9 complement complex and S-protein in human myocardial areas with necrosis and sclerosis. Immunol Lett. 16: 15-20, 1987.

Savill, J, Dransfield, I, Hogg, N and Haslett, C: Vitronectin receptor-mediated phagocytosis of cells undergoing apoptosis. Nature 343: 170-174, 1990.

Schäfer, H, Mathey, D, Hugo, F and Bhakdi, S: Deposition of the terminal C5b-9 complement complex in infarcted areas of human myocardium. J Immunol. 137: 1945-1949, 1986.

Scharfstein, J, Ferreira, A, Gigli, I and Nussenzweig, V: Human C4b-binding protein. Isolation and characterization. J Exp Med. 148: 207-220, 1978.

Schmid-Schoenleben, GW and Engler, RL: Granulocytes as active participants in acute myocardia