Lipid oxidation enhances the function of activated protein C.

Although lipid oxidation products are usually associated with tissue injury, it is now recognized that they can also contribute to cell activation and elicit anti-inflammatory lipid mediators. In this study, we report that membrane phospholipid oxidation can modulate the hemostatic balance. Oxidation of natural phospholipids results in an increased ability of the membrane surface to support the function of the natural anticoagulant, activated protein C (APC), without significantly altering the ability to support thrombin generation. Lipid oxidation also potentiated the ability of protein S to enhance APC-mediated factor Va inactivation. Phosphatidylethanolamine, phosphatidylserine, and polyunsaturation of the fatty acids were all required for the oxidation-dependent enhancement of APC function. A subgroup of thrombotic patients with anti-phospholipid antibodies specifically blocked the oxidation-dependent enhancement of APC function. Since leukocytes are recruited and activated at the thrombus or sites of vessel injury, our findings suggest that after the initial thrombus formation, lipid oxidation can remodel the membrane surface resulting in increased anticoagulant function, thereby reducing the thrombogenicity of the thrombus or injured vessel surface. Anti-phospholipid antibodies that block this process would therefore be expected to contribute to thrombus growth and disease.

The central role of the membrane surface in the reactions of the coagulation cascade has been well known for many years (1,2). A negative charge on the membrane is recognized as necessary for the binding of the vitamin K-dependent enzymes and substrates through their N-terminal Gla 1 domains. Through this binding, the local concentrations of the proteins are markedly increased, augmenting activation rates. Binding may also induce conformational changes in the proteins, aligning substrate cleavage sites with the active site of the enzyme (3). The nature of the phospholipid head group was known to play a role, and phosphatidylserine (PS) has been considered the most important (4,5). For many years, it was believed the reactions of the coagulation cascade shared similar requirements for the membrane surface. However, most experiments were performed with the prothrombin-activating complex, prothrombinase, with a tendency to generalize to the other complexes. More recently, it has become apparent that the presence of phosphatidylethanolamine (PE) potently enhanced the rate of inactivation of factor Va by the activated protein C (APC) complex (6) and the anticoagulant activity in plasma (7) while having little to no effect on the prothrombinase reaction when the PS concentration was optimal. Polyunsaturation of the phospholipids further enhanced the activity of the APC complex (8). Subsequently, roles for PE in factor VIII binding (9,10), tissue factor-factor VIIa activation of factor X (11), and prothrombin activation (12,13) were reported. In the latter studies, the incorporation of PE into the vesicles decreased the amount of PS required for optimal prothrombin activation (12) much as we observed (8). Unlike the APC system, the PE stimulation could be overcome by increased PS.
The role of oxidation in disease has gained considerable attention in the last several years. Oxidation is believed to play a key role in the pathogenesis of many inflammatory diseases including atherosclerosis, reperfusion injury, and autoimmune diseases (14 -17). Oxidized LDL plays a major role in the initiation and propagation of the atherosclerotic plaque. It can also lead to the activation of endothelium and platelets, possibly through lysophosphatidic acid, a product of oxidation (18). Hydrogen peroxide, a substance elaborated by activated leukocytes, can also act as a direct cell agonist (19,20). Endothelial cell activation can then lead to the elaboration of other inflammatory mediators which themselves exacerbate coagulation (21,22) and induce the expression of adhesion molecules, allowing the attachment of monocytes and neutrophils (23). Endothelial cells and leukocytes themselves can elaborate reactive oxygen species leading to oxidation of additional surfaces (16,24,25). Oxidation of red blood cells results in increased exposure of both PS and PE on the membrane surface, making these cells potentially procoagulant (15).
Oxidized LDL can also be immunogenic and lead to the development of antibodies that cross-react with cardiolipin or the surface of endothelial cells (26,27). In addition, apoptotic cells with negatively charged surfaces can be immunogenic and lead to the development of anti-phospholipid antibodies (APAs) with procoagulant activity (28). These observations led Horrko et al. (29,30) to determine that many APAs are directed toward epitopes of oxidized phospholipids.
Oxidation has long been recognized as having an effect in the in vitro coagulation assays (31), usually observed as a change in clotting time with the age of the phospholipid preparation. In general, the response to this observation has been to minimize oxidation by using only freshly prepared lipid vesicles. Very recently, Weinstein et al. (32) have reported enhanced prothrombinase activity in response to oxidation using defined systems. Whether the hemostatic balance is affected as a consequence of the oxidation was not directly addressed.
Clinical studies have shown that decreases of as little as 50% in the plasma protein C concentrations increase the risk of thrombosis. There is an even more profound effect when other relatively common risk factors, such as factor V Leiden, are also present (33). Clinical studies have shown that the levels of activated protein C in the circulation correlate directly with the levels of plasma protein C (34), most likely because the physiological protein C concentration is below the K m value for the physiological activation complex. Given the relationship between protein C levels and the risk of thrombosis, it follows that mechanisms that modulate APC anticoagulant function even 2-3-fold will have a substantial impact on the probability of thrombus formation. In our studies of the phospholipid requirements of the APC complex in contrast to those of the prothrombinase complex, we noticed a dramatically different effect of aging of the phospholipids on the two reactions. Here we report the differential effect of lipid oxidation on these reactions and the identification of anti-phospholipid antibody populations that selectively inhibit the influence of oxidation.
Liposome Preparation-Extruded vesicles were prepared as described (6) using a 100-nm Nucleopore membrane. PS:PC vesicles were 20% PS, 80% PC and PE:PS:PC vesicles were 40% PE, 20% PS, 40% PC. Lipids were mixed in the weight proportions indicated, dried under argon, and lyophilized 4 h to remove organic solvents. [ 14 C]PC (Amersham Pharmacia Biotech) was included as tracer for the determination of lipid concentrations. The lipids were then reconstituted at 5 mg/ml under argon in 150 mM NaCl, 20 mM HEPES, pH 7.5, which had been Chelex-treated, extensively degassed, and bubbled with argon.
Phospholipid Oxidation-To 1 ml of liposomes (200 g/ml) was added 1 l of 10 mM CuSO 4 in a glass tube. The suspension was vortexed to introduce air into the solution, and the liposomes were incubated at 37°C for the appropriate period (39). Standard oxidized liposomes were composed of brain-derived phospholipids oxidized for 20 h.
Analysis of Oxidation Products-The extent of oxidation was determined spectrophotometrically (40) or by 2-thiobarbituric acid reactivity (41). 130 nmol of phospholipid as liposomes were added to 1 ml of absolute ethanol. The amount of conjugated dienes or trienes present was determined by absorbance at 233 or 270 nm, respectively, in a Hewlett-Packard diode array 400N spectrophotometer. An "oxidation index" was defined as the ratio of absorbance at 233 nm to that at 215 nm, with a ratio of 0.02 corresponding to 0.1% oxidation (40). Malondialdehyde (MDA), an end product of lipid peroxidation, was measured by reactivity with 2-thiobarbituric acid by standard techniques. A standard curve was constructed by acid treatment of 1,1,3,3-tetraethoxypropane (Sigma) as described (41).
Lipoxygenase Oxidation of Phospholipids-Liposomes (2 mg/ml) were dispersed in 100 mM n-octyl-␤-D-glucopyranoside, 200 mM sodium borate, pH 10, with or without soybean lipoxygenase type I-B (Sigma) at 5000 units/ml (42). The suspensions were vortexed to introduce air into the solutions. The mixtures were then incubated at room temperature for 1 h. Each sample was then dialyzed versus 150 mM NaCl, 20 mM HEPES, pH 7.5, with three buffer changes over 2 days. Chelex 100 resin was present in the dialysis vessels. Immunoglobulin Purification-Plasma or serum was precipitated with 50% NH 4 SO 4 . The precipitate was resuspended and dialyzed versus 150 mM NaCl, 20 mM Tris-HCl, pH 7.5. The material was applied to a protein G column, washed, and eluted with 100 mM glycine, pH 2.5. The eluate was immediately neutralized with 1 M Tris-HCl, pH 9, and dialyzed versus 150 mM NaCl, 20 mM Tris-HCl, pH 7.5. Total IgG fractions were concentrated with Millipore Ultrafree30 centrifuge concentrators. IgG fractions were aliquoted and stored at Ϫ80°C.
Coagulation Assays-Clotting times were determined by a modified factor Xa one-stage clotting assay using the factor X-activating enzyme from Russell's viper venom (X-CP) to activate factor X and an ST4 coagulometer (Diagnostica Stago). The assays were adapted to allow evaluation of the phospholipid or APC dependence or the effect of IgG on the reactions. The buffer used was 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, 0.1% gelatin. In the standard assay, normal pooled plasma (50 l) was mixed with X-CP sufficient to obtain a 30-s clotting time in the absence of APC, phospholipid (10 g/ml), in the presence or absence of APC (0.2 g/ml) in a total volume of 150 l. When present, IgG was at the concentrations indicated in the figure. After 1 min of incubation at 37°C, 50 l of 25 mM CaCl 2 was added to initiate clotting.
Determination of Thrombin Generation in Defibrinated Plasma-Plasma was defibrinated by the addition of Ancrod to 0.1 unit/ml citrated plasma for 10 min at 37°C. The fibrin was removed by two centrifugations at 8000 ϫ g for 10 min. 240 l of defibrinated plasma was mixed with 40 l of 200 g/ml PE:PS:PC vesicles, 50 l of buffer, and 30 l of X-CP (2.3 nM). The buffer in all cases was 100 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.5% bovine serum albumin. The mixtures were incubated for 3 min at 37°C. At regular intervals, thrombin generation was started by the addition of 10 l of the above mixture to 10 l of 16 mM CaCl 2 with or without 2 nM APC in a "reverse" time course. This resulted in the final concentrations of 10 g/ml phospholipid, 100 pM X-CP, 1 nM APC (when present), and 8 mM CaCl 2 in the assay. At the end of the time course, 80 l of 100 mM NaCl, 50 mM Tris-HCl, pH 8, 10 mM EDTA, and 100 mM benzamidine HCl was added. After further dilution of 1:10 in 100 mM NaCl, 50 mM Tris-HCl, pH 8, 10 mM EDTA, Spectrozyme-TH was added, and the rate of hydrolysis measured in a V max microplate reader (Molecular Devices). Thrombin concentration was determined by comparison to a standard curve of known concentration.
Measurement of APC and Prothrombinase Activity in Purified Systems-Factor Va (50 nM) was reacted with 2.5 pM APC in the presence or absence of 20 nM protein S and 20 g/ml oxidized or nonoxidized PE:PS:PC vesicles at 37°C. At various times, the APC was inhibited by the addition of benzamidine HCl to 10 mM. After dilution, the remaining factor Va was determined by standard one-stage clotting assays using factor V-deficient plasma (43) or by its activity in the prothrombinase complex as described previously (6,37).
To determine the effect of oxidation on the prothrombinase reaction directly, reaction mixtures contained 0.2 nM factor Va, 2 nM factor Xa, 1-120 g/ml phospholipid, and 5 mM CaCl 2 . After 5 min of preincubation at 37°C, the reactions were started with the addition of prothrombin to 1.4 M final. After 3 min the reaction was stopped with the addition of 10 l 50 mM EDTA, 0.2 M HEPES, pH 7.5, in a final volume of 100 l. Aliquots were further diluted in 0.15 M NaCl, 20 mM HEPES, 5 mM EDTA, pH 7.5, and the resultant thrombin was measured using a chromogenic assay with Spectrozyme TH and a V max microplate reader (Molecular Devices). All buffers contained 1 mg/ml gelatin and 10 mg/ml bovine serum albumin.

Effect of Lipid Oxidation on APC Anticoagulant
Function-In our recent studies we noticed that the ability of the liposomes to support APC anticoagulant activity increased as the liposomes aged. Little effect was seen on the ability to support clotting initiated at the level of factor X activation using the Russell's viper venom factor X activator, X-CP. PE is known to increase the sensitivity of phospholipid membranes to oxidation (44). Because our liposomes now routinely incorporate PE, we theorized that variability in our coagulation assays with time might represent oxidation. We therefore investigated the impact of phospholipid oxidation on the coagulation reactions (Fig. 1). Whether or not the liposomes contained PE, the clotting time of plasma in the absence of APC was not affected significantly by oxidation. Oxidation of vesicles containing only PS and PC or PE and PC (see below) did not enhance the ability of the vesicles to support APC anticoagulant activity. Oxidation of PE:PS:PC vesicles dramatically enhanced the anticoagulant activity of APC, at least doubling the APC prolongation of the clotting time.
Effect of Oxidation on the Coagulation Reactions in Purified Systems-To be sure that membrane oxidation was affecting the APC complex directly and not acting through another plasma component, the effect on factor Va inactivation by APC was investigated in a purified system (Fig. 2). In addition, the effect of oxidation on prothrombinase in the purified system was studied (Fig. 3). Phospholipid oxidation enhanced factor Va inactivation in the presence or absence of protein S. Protein S was not required for oxidation-dependent enhancement of APC inactivation of factor Va. However, interestingly, the ability of protein S to enhance APC-dependent factor Va inactivation was much greater on oxidized lipids. Thus, oxidation may have a multifaceted effect on the anticoagulant pathway. Equivalent results were obtained whether remaining factor Va was assessed by clotting or prothrombinase assay. In contrast to a recent report (32), prothrombinase was not affected at any concentration of phospholipid studied under the conditions of our study (see "Discussion").
A reagent previously found to be useful in dissecting the membrane requirements of the APC complex is a chimeric form of the protein in which the Gla domain of APC is replaced with that of prothrombin, referred to as APC-PtGla. This molecule shows enhanced anticoagulant activity in plasma relative to the native protein. However, in addition to protein S not being required for activity, the presence of PE does not affect its activity. We therefore investigated whether it was also unaffected by oxidation. Oxidation of the vesicles had no effect on the activity of the protein either in clotting assays (Fig. 4) or the purified system described above.
The natural brain-derived phospholipids are composed of different fatty acids of varying length and degree of unsaturation. Increasing unsaturation increases the potential for oxidation (14,28). In addition, according to the supplier, the brainderived PE contains ϳ50% plasmalogen forms of PE. To control these variables better, we investigated whether synthetic phos-pholipids could support oxidation-enhanced APC activity to the extent of the brain-derived material (Fig. 5). Direct comparisons of clotting times are not straightforward, as changes in lipid composition affect the base-line clotting times, especially in the presence of APC, when the same clotting stimulus (concentration of X-CP) is used (8). However, it is apparent that APC activity is not affected by oxidation of minimally unsaturated phospholipid (all components 16:0 -18:1) (Fig. 5A). Oxidation of lipids containing two diunsaturated fatty acids (18:2) only marginally enhanced the APC anticoagulant response. Inclusion of arachidonate (20:4) increased the clotting time by 50% (Fig. 5B). When oxidation was allowed to continue in these highly polyunsaturated liposomes, clotting times increased even in the absence of APC. Light scattering was performed to determine whether this was due to the loss of liposome structure (Fig. 6). After 20 h of copper-catalyzed oxidation, the liposomes made with synthetic phospholipids no longer scattered light, indicating that the liposomes were no longer intact. Similar light-scattering experiments indicated that liposomes made with brain-derived components remained intact. Plasmalogens, present in the brain phospholipids and the plasma membranes of most cells, are known to stabilize membranes against oxidation (45). Thus, it is possible that liposomes made with the highly unsaturated synthetic phospholipids were not as effective as the brain-derived vesicles because the liposome concentration was decreasing with increasing oxidation, thereby offsetting any continued enhancement of APC function by continued oxidation.
Phospholipid Concentration Dependence of APC Activity-A possible function of oxidation is to alter the phospholipid concentration dependence of the coagulation reactions. Therefore, the effect of oxidation on phospholipid concentration dependence was determined for PS:PC, PE:PC, and PE:PS:PC vesicles (Fig. 7). In the absence of APC, all curves showed an initial drop in clotting time, consistent with the phospholipid concentration dependence of the prothrombinase reaction. With PS:PC vesicles, there was some increased anticoagulant function in the presence of APC at relatively high phospholipid concentrations, but this was not affected by oxidation. PE:PC vesicles mirrored the prothrombinase curves, and oxidation had no effect. With PE:PS:PC vesicles in the absence of oxidation, the rise in anticoagulant function after the initial drop in clotting time was dramatic. Lipid oxidation increased the APC anticoagulant response at all phospholipid concentrations and also abrogated the initial decrease in clotting time. Thus, although oxidation of these vesicles altered the concentration dependence at low lipid concentrations, the overall phospholipid concentration dependence of APC anticoagulant activity was not altered in that enhanced activity was observed at all lipid concentrations. It should be noted that the PE:PC vesicles made using brain-derived material do not show enhanced activity with oxidation. This implies that PE or PE plasmalogens per se are not sufficient to result in enhanced APC anticoagulant activity.
Effect of Oxidation on Total Thrombin Generation-A prolonged clotting time may indicate only a lengthening of the lag period before feedback loops lead to the explosive generation of thrombin, rather than a decrease in the total amount of throm-bin eventually formed (46,47). These two possibilities may not have the same ultimate physiological effect. To determine which was the case here, thrombin generation rather than clotting time was determined (Fig. 8). APC slightly prolonged the lag time for thrombin generation on lipids with or without oxidation. The additional effect of oxidation was to decrease total thrombin generation. The decrease in thrombin with time represents inhibition by antithrombin and ␣ 2 -macroglobulin (48). The residual amidolytic activity is due to the fact that ␣ 2 -macroglobulin does not completely inhibit thrombin amidolytic activity.
Preliminary Attempts to Determine the Active Species in Oxidized Phospholipids-The formation of conjugated dienes and trienes and MDA was monitored as a function of time and correlated to the anticoagulant activity of the resulting liposomes (Fig. 9). When PE was present in the vesicles, the formation of diene and triene correlated with increased anticoagulant function. "Unoxidized" liposomes were Ͻ1% oxidized based on absorbance at 233 nm. Liposomes were never more than 4% oxidized based on conjugated diene formation for optimal activity. If PE was not present, even though increases in these parameters were observed spectrophotometrically, no increase in function was observed. Formation of MDA did not correlate with increased anticoagulant function.
We used soybean lipoxygenase, which produces conjugated dienes only (38), to determine whether formation of these structures alone was sufficient to produce enhanced APC anticoagulant activity. At no time was enhanced anticoagulant activity observed, even when the concentration of conjugated dienes present was equivalent to or greater than that observed in the oxidized phospholipids as monitored by absorbance at 233 nm (data not shown). Lysophospholipids represent a possible product of lipid peroxidation. The lyso forms of oleoyl-PE, palmitoyl-PE, oleoyl-PC, or brain plasmalogen PE were incorporated into vesicles from 1 to 25% of the total phospholipid. No concentration of these molecules reproduced the effect observed with copper/air oxidation of the vesicles. If anything, incorporation of lyso-plasmalogen PE or lyso-PC slightly inhibited APC anticoagulant activity (data not shown).
The Effect of Patient Immunoglobulin on APC Activity-Previously we showed that plasma or immunoglobulin derived from at least a subset of lupus patients could inhibit APC function selectively in a PE-dependent manner (37,49). Screening of IgG from several patients with thrombosis and APA or lupus anticoagulant indicated the anticoagulant activity of APC was inhibited more effectively on oxidized phospholipid than on nonoxidized phospholipid (data not shown). A typical IgG titration curve for one patient is shown in Fig. 10.

DISCUSSION
The results from the present studies begin to address several important and unresolved clinical issues. First, the mechanism by which new thrombi or newly injured vessels which are very thrombogenic become less so with time, a process known as "passivation" (50,51). The ability of phospholipid oxidation, a time-dependent process, to enhance APC anticoagulant function would be expected to contribute to this process. Second, although protein S is clinically established as a key component of the anticoagulant pathway with total deficiencies resulting in massive thrombosis, protein S effects on factor Va inactivation observed in purified systems have been relatively modest. As seen in Fig. 2, the ability of protein S to accelerate APC inactivation of factor Va is markedly enhanced by phospholipid oxidation. If oxidation is indeed involved in thrombus passivation, this may demonstrate a potentially important function for protein S in that process. Third, patients with APAs have an increased propensity to thrombose. Oxidized lipids are a target for at least some of these APAs. APAs that can block the ability of the oxidized lipid to enhance anticoagulant function may contribute to a prothrombotic state.
Our previous studies indicated the importance of PE and polyunsaturation in the expression of APC activity (6,8,49). The studies reported here indicate yet another feature that distinguishes procoagulant from anticoagulant phospholipid surfaces. We have found that oxidization of the phospholipid membrane further enhances the activity of APC in the factor Va inactivation complex with little or no effect on the prothrombinase reaction. The enhanced factor Va inactivation results in increased anticoagulant activity. That the effect is greater in the presence of protein S in the purified system indicates oxidation may be impacting more than one interaction. The present results demonstrate that lipid oxidation has a significant effect on the ability of APC to block total thrombin generation without significantly influencing thrombin generation in the absence of APC. These findings exclude the possibility that the increased anticoagulant activity in plasma is due to an increase in the lag time before thrombin generation is initiated. It is of interest that studies of deficiency states and inhibitors have demonstrated that the impact on total thrombin generation is usually greater than the effect on the clotting time. Indeed, studies of total thrombin generation were able to demonstrate the presence of an acquired inhibitor of APC in pregnant women and women on birth control pills that is thought to contribute to thrombotic risk (52). Standard coagulation tests failed to detect this inhibitor.
For the oxidation to be effective, both PE and PS must be present in addition to a significant degree of polyunsaturation. Based on the studies with APC-PtGla, it is apparent that the oxidation effect is mediated by features of the Gla domain of protein C that differ from those of prothrombin. We demonstrated previously that the PE head group specificity was also resident in the Gla domain of protein C (37). The requirements for the assembly of the anticoagulant complex include interaction of the enzyme, APC, and the substrate, factor Va, with the membrane surface. Optimal factor Va binding requires the presence of the PS head group, probably explaining the need for PS. APC is essentially inactive in the absence of PE. PE, polyunsaturation, and lipid oxidation all favor formation of hexagonal phase lipids (28,53). Based on the observations that PE, polyunsaturation, and oxidation all favor APC function, we would presume that the oxidation is favoring hexagonal phase lipid formation. This would be consistent with the ability of the anti-PL antibodies to block the oxidation effect as other investigators (54,55) have demonstrated these antibodies bind preferentially to hexagonal phase phospholipids and that oxidation can enhance their binding.
We have not yet obtained a liposome composition using synthetically produced phospholipids that is as effective as the naturally occurring brain-derived phospholipids. This may be due to the progressive loss of liposome structure during the oxidation process of highly unsaturated phospholipids. This is quite possibly a function of the difficulty to control the rate of oxidation in these preparations. When these liposomes are oxidized in the absence of copper, greater APC activity is observed after an extended period, but the unpredictability of this reaction makes it impractical for detailed study.
One constituent present in the brain-derived material that is absent in the synthetic mixtures is various plasmalogen forms of PE that may serve to stabilize the membranes (45). Although possibly serving this role, plasmalogens themselves are not sufficient for the effect, as liposomes composed solely of brainderived PC and PE, which contain the plasmalogens, lack any enhancing activity after oxidation. Lysophospholipid or conjugated diene derivatives of the fatty acid side chains are also not sufficient to support the enhanced activity of these membranes. Conjugated diene formation was a useful monitor of oxidation. However, experiments using soybean lipoxygenase, which produces only conjugated dienes (42), failed to produce enhanced activity, even though the concentration of conjugated dienes formed was equivalent to or greater than that observed in the copper-oxidized phospholipids (data not shown). Although conjugated triene formation appears to correlate with activity in the data presented, this was not always the case (data not shown). Attempts to produce trienes specifically to directly test whether their formation is sufficient have not been successful. Only a small degree of oxidation of the phospholipids (Ͻ4%) is required for the observed effects as monitored by conjugated diene/triene formation. Thus, isolation and identification of the putative moieties responsible for the activity will require sensitive and extensive investigation. Such studies are ongoing.
Very recently, Weinstein et al. (32) reported on enhanced prothrombinase activity in the presence of oxidized 1-stearoyl-2-arachidonoyl-PC (18:0, 20:4-PC) (SAPC). However, the conditions used by these authors are very different from those used in the studies reported here, making comparisons difficult. In the cited reference, other than the SAPC, the vesicles are composed of dimyristoyl (14:0, 14:0)-PC and/or -PS. These are very sterically hindered nonphysiological forms of these molecules. Extremely strong oxidation conditions (2 mM H 2 O 2 and 200 M CuSO 4 ) are used relative to the mild conditions used here (10 M CuSO 4 and air), which will most likely yield different products. The use of EDTA at concentrations as high as 75 mM may also result in unexpected effects. Butylated hydroxytoluene is added to all mixtures to stop oxidation. However, it may also be incorporated into the vesicles used for the prothrombinase reactions and have direct effects on observed activity. The tocopherols (and ascorbic acid to some extent), used in some experiments to block oxidation, would be coextracted with the SAPC and presumably reincorporated into the vesicles used for the prothrombinase reactions. Incorporation of tocopherols into membranes has been found to decrease membrane fluidity (56,57), further complicating interpretation of these experiments. Indeed, the reversal of the known effects of unsaturated fatty acids (in this case, SAPC whether or not it is oxidized) on prothrombinase activity (58, 59) would argue for a direct effect of the lipophilic antioxidants on the membrane structure, further complicating interpretation of these experiments.
In addition to cell or platelet surfaces, plasma lipoproteins have been found to be active in the coagulation reactions. Both Rota et al. (60) and Moyer et al. (61) observed very low density lipoproteins supported prothrombin generation. Rota et al. (60) also found that copper oxidation of the low density (LDL) population led to a marked increase in prothrombin activation. However, oxidation of the other lipoprotein classes did not lead to enhanced activation. Since the authors showed that the relative concentration and composition of phospholipids did not vary significantly between the lipoprotein populations, oxidation of the phospholipid per se was unlikely to be the major mechanism of the enhanced anticoagulant activity. More likely, oxidation of the protein moieties in the LDL permitted improved display of the phospholipid present in these particles. Griffin et al. (62) observed APC anticoagulant function on high density lipoprotein particles but not LDL. Neither very low density lipoproteins nor the effect of oxidation on any of the lipoproteins was investigated.
When a coagulant stimulus is generated, leukocytes are recruited into the forming clot (63,64). Platelet activation occurring during this time results in the expression of P-selectin on the platelet surface, serving to rosette the incoming neutrophils and monocytes around the activated platelet (65). Lipid oxidation as a result of the leukocyte-platelet interaction leads to the evolution of lipoxins, eicosanoids with anti-inflammatory effects including inhibition of neutrophil chemotaxis (66,67). Activation of the adherent leukocytes may also enhance the anticoagulant properties of the platelet through lipid oxidation.
Relative to phospholipid surfaces, the platelet expresses a relatively high prothrombinase activity compared with their ability to support factor Va inactivation (68). Despite this in FIG. 10. Titration of an oxidation-dependent inhibitory APA immunoglobulin. Normal or patient IgG was purified as described under "Materials and Methods" and incorporated in the clotting assays at the final concentrations indicated. Assays were performed as described in Fig. 1 vitro observation, it has been shown that APC (69), or thrombin infusion which elicits APC formation (70), can block platelet deposition on thrombogenic surfaces, a system that has been shown previously to involve thrombin formation. The latter observations are opposite to what would be predicted from the in vitro experiments. We would propose that the key difference between the in vitro and in vivo results may be leukocytemediated oxidation that helps passivise the developing thrombus. The mechanisms responsible for this process are unknown and undoubtedly complex. Lipid oxidation caused by the adherent leukocytes could contribute significantly to this phenomenon that is critical for successful vascular repair. The first response to a low or moderate inflammatory stimulus would be the development of a procoagulant surface. As leukocytes enter the clot, oxidation of cell surfaces would begin after a time lag. The anticoagulant response would then be promoted to prevent excessive thrombus growth. Membrane constituents such as plasmalogens (45,71) or sphingomyelin (72) may help stabilize minimally oxidized surfaces to maintain APC activity for a longer period. The red cell membrane could also be oxidized by the infiltrating leukocytes in addition to the platelets normally envisioned, possibly adding anticoagulant surface despite the externalization of PS and PE that can occur on these cells in response to oxidation (15).
The effect of oxidation on the APC complex may also further differentiate subclasses of autoimmune antibodies. Previously we showed that plasma or immunoglobulin derived from at least a subset of lupus patients could inhibit APC function selectively in a PE-dependent manner (37,49). Although normal precautions against oxidation were taken, including storage under argon and usage within 1 week of liposome preparation, the oxidation state of the phospholipids used in those studies was not specifically known. In a small sampling of patients, immunoglobulin populations have been identified that inhibit APC anticoagulant activity on oxidized phospholipids much more effectively than on nonoxidized liposomes. It is not known how common this reactivity is. The diagnostic utility of this distinction is also unknown at this time as many more patients will need to be screened for oxidation dependence of APC inhibition. Clinically, if lipid oxidation is involved in making clots and vascular injury sites less thrombogenic, then APAs with specificity for oxidized lipids may have a different physiological effect than those that are not oxidation-dependent. Oxidation-dependent antibodies would minimize the ability to passivise the clot or injured surface. Such antibodies may not be significantly procoagulant in the absence of an oxidized surface. Violi and co-workers (73,74) have reported that markers of clotting activation correlate with markers of in vivo lipid peroxidation in patients with APA but not in individuals devoid of APAs. Both markers decrease in response to antioxidants.
It might be argued that the 2-3-fold enhancement of APC anticoagulant function and factor Va inactivation is too small an effect to be physiologically relevant. However, decreases in protein C levels to 50% of normal are sufficient to increase thrombotic risk (75). Furthermore, small changes in the level or function of natural anticoagulants can have a major effect on potential thrombin generation. Studies of Mann and colleagues (46,47) indicate that a "threshold effect" occurs in coagulation. Thrombin must be generated to a level greater than that threshold for clotting to ensue. Doubling the rate of inhibition of the procoagulant reactions can be sufficient to keep thrombin generation below that required for threshold, changing the balance such that normal hemostasis is maintained.