Author + information
- Received October 26, 2011
- Revision received December 16, 2011
- Accepted December 22, 2011
- Published online April 17, 2012.
- Gregor Leibundgut, MD⁎,†,
- Kiyohito Arai, MD†,‡,
- Alexina Orsoni, PhD§,
- Huiyong Yin, PhD∥,
- Corey Scipione, BS¶,
- Elizabeth R. Miller, BS†,
- Marlys L. Koschinsky, PhD¶,
- M. John Chapman, PhD§,
- Joseph L. Witztum, MD† and
- Sotirios Tsimikas, MD†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Sotirios Tsimikas, Vascular Medicine Program, University of California San Diego, 9500 Gilman Drive, La Jolla, California 92993-0682
Objectives This study sought to assess whether plasminogen, which is homologous to lipoprotein (a) [Lp(a)], contains proinflammatory oxidized phospholipids (OxPL) and whether this has clinical relevance.
Background OxPL measured on apolipoprotein B-100 (OxPL/apoB), primarily reflecting OxPL on Lp(a), independently predict cardiovascular disease (CVD) events.
Methods The authors examined plasminogen from commercially available preparations and plasma from chimpanzees; gorillas; bonobos; cynomolgus monkeys; wild-type, apoE−/−, LDLR−/−, and Lp(a)-transgenic mice; healthy humans; and patients with familial hypercholesterolemia, stable CVD, and acute myocardial infarction (AMI). Phosphocholine (PC)-containing OxPL (OxPC) present on plasminogen were detected directly with liquid chromatography–mass spectrometry (LC-MS/MS) and immunologically with monoclonal antibody E06. In vitro clot lysis assays were performed to assess the effect of the OxPL on plasminogen on fibrinolysis.
Results LC-MS/MS revealed that OxPC fragments were covalently bound to mouse plasminogen. Immunoblot, immunoprecipitation, density gradient ultracentrifugation, and enzyme-linked immunosorbent assay analyses demonstrated that all human and animal plasma samples tested contained OxPL covalently bound to plasminogen. In plasma samples subjected to density gradient fractionation, OxPL were present on plasminogen (OxPL/plasminogen) in non-lipoprotein fractions but on Lp(a) in lipoprotein fractions. Plasma levels of OxPL/apoB and OxPL/apo(a) varied significantly (>25×) among subjects and also strongly correlated with Lp(a) levels. In contrast, OxPL/plasminogen levels were distributed across a relatively narrow range and did not correlate with Lp(a). Enzymatic removal of OxPL from plasminogen resulted in a longer lysis time for fibrin clots (16.25 vs. 11.96 min, p = 0.007). In serial measurements over 7 months, OxPL/plasminogen levels did not vary in normal subjects or in patients with stable CVD, but increased acutely over the first month and then slowly decreased to baseline in patients following AMI.
Conclusions These data demonstrate that plasminogen contains covalently bound OxPL that influence fibrinolysis. OxPL/plasminogen represent a second major plasma pool of OxPL, in addition to those present on Lp(a). OxPL present on plasminogen may have pathophysiological implications in AMI and atherothrombosis.
Plasminogen plays a key role in the fibrinolytic system and has also been implicated in several other pathophysiological properties, including tissue remodeling, angiogenesis, embryogenesis, tumor metastasis, infections, wound healing, and leukocyte migration (1). Plasminogen consists of 5 tandem kringle domains and a protease domain. It is activated to plasmin by physiological activators, such as tissue-type plasminogen activator (tPA), and in turn, plasmin degrades fibrin-rich thrombi through a functional serine protease domain. Kringles are common motifs in coagulation and growth factors and in apolipoprotein (a) (apo(a)).
Lipoprotein (a) (Lp(a)) is composed of apo(a) covalently bound via a single disulfide bond on kringle IV type 9 to apolipoprotein B-100 (apoB) of low-density lipoprotein (LDL). Apo(a) is highly homologous to plasminogen and is believed to have evolved from duplication of parts of the plasminogen gene present on the long arm of chromosome 6. The apo(a) gene sits opposite the plasminogen gene on chromosome 6q26–27 and consists of multiple KIV repeats, of which subtype KIV-2 is present in multiple and variable numbers, KV, and an inactive protease domain due to a Ser561-Ile562 substitution for Arg561-Val562 present on plasminogen (2). Unlike plasminogen, which is present widely across species, apo(a) appeared late during evolution, approximately 40 to 60 million years ago, and is present only in humans, non-human primates, and old world monkeys, as well as an unrelated version in European hedgehogs. Lp(a) is now generally recognized as a causal, independent, genetic cardiovascular risk factor for coronary artery disease (CAD) and myocardial infarction (3–5).
We made the initial observation that Lp(a) is a preferential lipoprotein carrier of oxidized phospholipids (OxPL) in humans (6–10). We developed an enzyme-linked immunosorbent assay (ELISA) that quantitates OxPL on human apoB particles (OxPL/apoB), which primarily reflects the presence of OxPL on Lp(a). We have demonstrated that OxPL/apoB levels reflect changes in vascular and endothelial function and coronary calcium, identified the presence and progression of carotid and femoral atherosclerosis, and angiographically determined CAD (8,11). Elevated OxPL/apoB levels occur following acute coronary syndromes (ACS) (6) and percutaneous coronary intervention (12) and predict the occurrence of new cardiovascular disease (CVD) events in previously healthy subjects (3,13). Thus, OxPL/apoB appears to reflect the adverse consequences of Lp(a) on cardiovascular outcomes, but are also independently associated with CVD risk above and beyond Lp(a) levels (reviewed in Taleb et al. ).
In this study, we examine the hypothesis that plasminogen, due to its high homology to apo(a), may also contain OxPL and assess whether OxPL levels on plasminogen vary with differences in plasma Lp(a) levels in patients with familial hypercholesterolemia (FH), affect fibrinolysis, and vary temporally in patients with stable CAD and following ACS.
A detailed description of the Methods section is available in the Online Appendix.
Human plasma was obtained from healthy volunteers, patients with heterozygous FH with highly elevated LDL cholesterol levels but varying Lp(a) levels, and serial samples from patients with stable CAD and acute myocardial infarction (AMI) (6).
Source of animal plasma samples
Plasma samples from bonobos, chimpanzees, gorillas, cynomolgus monkeys, and wild-type, transgenic Lp(a) (15), apoE−/−, and LDLR−/− mice were obtained for measurement of OxPL on plasminogen.
Plasminogen and antibodies
Murine and human plasminogen were purchased commercially. A rabbit polyclonal anti-human plasminogen antibody (Santa Cruz Biotechnology, Santa Cruz, California) raised against amino acids 16 to 105 at the N-terminal end of human plasminogen, a sequence not present in apo(a), was used for immunoblot analysis to avoid cross-reactivity with apo(a). A murine monoclonal anti-human plasminogen antibody not cross-reacting with apo(a) and a polyclonal, biotinylated, guinea pig anti-plasminogen antibody were used as capture and detection antibodies, respectively. Monoclonal antibodies MB47, binding human apoB, LPA4 binding apo(a), and E06 binding the phosphocholine (PC) headgroup of OxPL were previously described (12).
Tandem liquid chromatography–mass spectrometry analysis of OxPL on mouse plasminogen
Liquid chromatography–mass spectrometry (LC-MS/MS) was utilized to assess the covalent binding of PC-containing OxPL on mouse plasminogen (Haematologic Technologies, Essex Junction, Vermont) using a triple quadruple instrument Thermal TSQ Vantage mass spectrometer (Thermo Fisher Scientific, Waltham, Massachusetts) coupled with a Waters NanoAcquity autosampler/UPLC system (Waters Corporation, Milford, Massachusetts), as previously described (16). Full-scan monitoring was carried out to scan a mass range of m/z 350 to 1,500. Precursor ion scanning monitoring was carried out to identify a product ion of m/z 184, which is characteristic for the PC headgroup.
Non-reducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis was carried out using precast gradient gels with 4% to 12% polyacrylamide concentrations.
Density gradient ultracentrifugation and OxPL on lipoproteins
Isopycnic density gradient ultracentrifugation was used to fractionate plasma, providing 24 fractions plus the non-lipoprotein plasma “bottom” fraction as previously described (10). OxPL on apoB (OxPL/apoB) (14) and OxPL on apo(a) (OxPL/apo(a)) (17) were measured as described.
ELISA to measure plasminogen levels and OxPL on plasminogen
To measure plasminogen levels, microtiter well plates (Dynex Technologies, Chantilly, Virginia) were incubated with a mouse monoclonal anti-human plasminogen antibody (Meridian Life Science, Saco, Maine) at 5 μg/ml overnight at 4°C, the plates washed, human plasma added (1:32,000 dilution), and plasminogen detected with biotinylated guinea pig anti-human plasminogen antibody using chemiluminescence ELISA.
OxPL on plasminogen (OxPL/plasminogen) was determined in a similar manner except the plasma dilution was 1:400 and biotinylated EO6 was the detection antibody. This assay normalized all wells to the same amount of plasminogen and is therefore independent of plasma plasminogen levels.
To assess whether OxPL are physically associated with plasminogen, increasing amounts of the murine monoclonal anti-human plasminogen antibody (Meridian), not cross-reacting with apo(a), were added to human plasma to preferentially precipitate plasminogen. Plasminogen, OxPL/plasminogen, Lp(a), and OxPL/apo(a) were then measured in the supernatant.
Phospholipase A2 treatment of plasminogen
Plasminogen, free of Lp(a), was purified from fresh-frozen plasma using lysine-Sepharose affinity chromatography and incubated with or without 35 U/ml of phospholipase A2 (PLA2) at 37°C for 90 min and then PLA2 quenched by the addition of phenylmethanesulfonylfluoride. The treated plasminogen was then isolated by lysine-Sepharose and subjected to sodium dodecyl sulfate–polyacrylamide gel followed by silver staining to verify the absence of degradation. OxPL removal was confirmed by Western blot analysis using EO6.
In vitro clot lysis assay
Assays were performed in a manner similar to that previously published (18). Fibrin clots were formed by the addition of a solution containing 1 mg/ml purified fibrinogen and 0.66 μmol/l plasminogen (with or without PLA2 treatment) to small, separated aliquots of CaCl2, thrombin, and tPA at final concentrations of 10 mmol/l, 6 nmol/l, and 100 pmol/l, respectively. Clot lysis at 37°C was monitored by measurement of turbidity at 405 nm and defined as the time required to reach the midpoint between the maximum and minimum turbidity excursions.
Determination of OxPL on coagulation factors with kringle-like structures
To determine whether kringle domains of other coagulation factors also contain OxPL detectable by EO6, we used commercially available antibodies to specifically capture prothrombin, urokinase, and tPA in microtiter well plates from plasma of 6 healthy human individuals, and OxPL was determined with E06.
Baseline characteristics of the human study population
The baseline characteristics of the human subjects from whom plasma was derived for the various studies are depicted in Table 1. The characteristics are typical of their underlying diagnoses.
Determination of OxPL on mouse plasminogen using LC-MS/MS
We studied mouse plasminogen (with no possibility of Lp(a) contamination as mice do not have Lp(a)) to determine whether covalently bound OxPL-containing phosphocholine (OxPC) were present. The presence of OxPC on trypsin digests of plasminogen was assessed by precursor ion scanning for m/z 184, which is the signature of PC. Since we postulated that the OxPL were covalently bound by Schiff base adducts between oxidized sn-2 side chains of the OxPL and epsilon amino groups of lysine, we also examined for the presence of OxPC on the plasminogen digests before and after saponification with NH4OH.
As shown in Figure 1A, a number of prominent peptide peaks containing OxPC were present in the LC-chromatogram when examined by precursor ion scanning mass spectrometry, indicating OxPC-modified peptides. More importantly, these peaks disappeared when the protein was first saponified by NH4OH (Fig. 1B). Full-scan experiments demonstrated the presence of similar patterns of peptides in samples with or without NH4OH pretreatment (data not shown). These results suggest that there are multiple, but limited, sites on plasminogen that were covalently modified by OxPC.
Immunoblot and ELISA analysis demonstrating OxPL on human and murine plasminogen
We next examined for the presence of OxPL on plasminogen from a variety of sources by Western blotting techniques using monoclonal E06 specific for PC-containing OxPL. Polyacrylamide gel electrophoresis of commercially available, lysine-Sepharose column purified, human and murine plasminogen, as well as freshly procured plasma samples from wild-type, LDLR−/−, apoE−/−, and apo(a) and Lp(a) transgenic mice (all on standard mouse chow) was performed and the gels incubated with: 1) biotinylated species-appropriate anti-plasminogen antibodies; 2) monoclonal antibody LPA4 to detect apo(a); and 3) monoclonal antibody EO6 to detect OxPL. Figure 2A (left panel) demonstrates that plasminogen is present in all lanes at the appropriate molecular weight (∼88 to 92 kDa), as expected. Figure 2A (middle panel) demonstrates that OxPL is present in all samples corresponding to the molecular weight of plasminogen and also at the appropriate molecular weight for apo(a) in lanes containing plasma from apo(a) and Lp(a) transgenic mice. There are no other E06-positive bands throughout the gel, suggesting that plasminogen and apo(a)/Lp(a) are the major protein/lipoprotein carriers of OxPL in plasma. Figure 2A (right panel) confirms the presence of apo(a) immunoreactivity, detected by antibody LPA4, in the apo(a) and Lp(a) transgenic mouse plasma. Interestingly, Lp(a) is also present in the commercial preparation of human plasminogen, which undoubtedly coelutes with plasminogen on the lysine-Sepharose columns used to purify plasminogen, since they share similar lysine binding sites, particularly on KIV-10. The size of the human Lp(a) is larger than in the Lp(a) transgenic mouse, as the Lp(a)-transgenic mice express a mini apo(a) construct (19). An OxPL band corresponding to the Lp(a) contaminant in the human purified plasminogen is not visible, likely due to much higher (∼50×) sensitivity of LPA4 versus E06 on immunoblots, as previously shown (7).
Figure 2B displays the plasma plasminogen and OxPL/plasminogen in 6 healthy human subjects using a sandwich ELISA format. It is evident that OxPL are strongly present on plasminogen captured on the microtiter well plate. In contrast, there is no evidence that apo(a) or apoB are present on the captured plasminogen, ruling out nonspecific physical interactions of apo(a) or apoB as potential contributions of OxPL on plasminogen. Importantly, like Lp(a), plasminogen is not “oxidized” per se, as supported by the observation that murine monoclonal antibody MDA2 (10), which recognizes malondialdehyde (MDA)-lysine epitopes and which are commonly present during generalized lipid peroxidation, does not show immunoreactivity with plasminogen.
Assessment of the presence of OxPL on plasminogen using immunoprecipitation of human plasma
Incubating increasing amounts of a murine monoclonal anti-human plasminogen antibody with human plasma demonstrates that at a molar ratio of ∼15:1 (anti-plasminogen antibody/plasminogen), nearly all of the plasminogen was precipitated (Fig. 3A). In parallel, a similar decrease in OxPL/plasminogen (Fig. 3B) was noted. In contrast, Lp(a) (Fig. 3C) and apoB (Fig. 3D) remained in the supernatant, suggesting that OxPL were physically associated with plasminogen. In a prior study, we demonstrated that immunoprecipitation of Lp(a) with LPA4 immunoprecipitated the OxPL associated with Lp(a) (10).
Coagulation factors with kringle-like structures do not contain OxPL
Using specific antibodies to capture plasminogen (5 kringles), prothrombin (factor II) (2 kringles), tPA (2 kringles), and urokinase (1 kringle) on microtiter well plates and adding biotinylated E06, it was demonstrated that OxPL were only present on plasminogen (Fig. 4).
OxPL on plasminogen and lipoproteins isolated by density gradient ultracentrifugation
Plasma from 2 patients with an Lp(a) level of ∼90 mg/dl were subjected to density gradient ultracentrifugation. Direct plating of the density fractions on microtiter well plates was used to assess the presence of apo(a) and apoB in each fraction. A set of capture assays was also performed as described in the previous text, where specific antibodies for plasminogen, Lp(a), and apoB were used to capture these particles, respectively. Specific antibodies were then used to detect the presence of plasminogen, Lp(a), apoB, and OxPL on the captured proteins or lipoproteins. An example from 1 patient shows that plasminogen was present in the heaviest density fractions and Lp(a) present in the modestly dense fractions (density 1.050 to 1.090 g/ml, corresponding to fractions 11 to 15), as expected. These aliquots also directly correspond to the presence of OxPL on plasminogen or Lp(a). ApoB is widely distributed across the density range, but OxPL/apoB is primarily present in the Lp(a) density range, consistent with the fact that these apoB particles are associated with Lp(a) and not LDL (Fig. 5). The OxPL/apo(a) and OxPL/apoB and Lp(a) data are also consistent with prior observations from a similar analysis (10).
We also examined whether plasminogen levels and OxPL/plasminogen varies among patients with different Lp(a) levels. We analyzed 15 subjects with FH segregated evenly into 3 groups of 5 subjects each, according to low (10 ± 3 mg/dl), intermediate (42 ± 5 mg/dl), and high (103 ± 8 mg/dl) Lp(a) levels (p < 0.0001 by analysis of variance [ANOVA]) (Fig. 6). We also assessed the non–lipoprotein-containing fraction (“bottom” fraction) for plasminogen and OxPL/plasminogen. It is noted that the plasma OxPL/apo(a) levels (p < 0.0001 by ANOVA) track nearly identically with the plasma Lp(a) levels (p < 0.0001 by ANOVA) (Figs. 6A and 6B), but vary significantly (>25-fold) among groups. However, by contrast, plasminogen levels (p = 0.08 by ANOVA) and OxPL/plasminogen (p = 0.14 by ANOVA) are not significantly different among the groups (Figs. 6C and 6D) and vary minimally (<2-fold). As a confirmation of the presence of plasminogen (p = 0.28 by ANOVA) and OxPL/plasminogen (p = 0.71 by ANOVA) in non-lipoprotein fractions, we also evaluated the “bottom” density fraction which contains all the non-lipoprotein plasma proteins, and the findings are similar to the plasma data for plasminogen (Figs. 6E and 6F).
OxPL/plasminogen of monkeys and apes
Plasminogen and OxPL/plasminogen were measured in plasma from 6 cynomolgus monkeys, 14 gorillas, 5 chimpanzees, and 4 bonobos using the same assays used for the human plasma. OxPL was detected on the plasminogen of each species, and the OxPL/plasminogen levels corresponded to the level of plasminogen (Online Fig. 1). Comparison of the relative differences between groups cannot be made because differences in the affinities of the capture and detection antibodies among the species may exist, which were not directly tested due to the difficulty in obtaining purified plasminogen from apes.
In vitro clot lysis assay with native human plasminogen
To enzymatically remove the covalently bound OxPL, purified human glu-plasminogen was treated with PLA2, and the OxPL was successfully removed as documented by immunoblotting with E06 (Fig. 7, inset). The ability of plasminogen species with and without OxPL to degrade fibrin clots was then tested with a well-validated in vitro clot assay (18). This demonstrates that the enzymatic removal of the OxPL component (without degrading the plasminogen protein integrity) results in a 36% longer clot lysis time (975 ± 41.2 s vs. 718 ± 6.6 s, p = 0.007, or 16.25 min vs. 11.96 min) compared with the intact plasminogen containing OxPL (Fig. 7). Clot lysis time is defined as the transition midpoint that is halfway between the minimum and maximum excursions. This suggests that the presence of OxPL on plasminogen facilitates clot lysis.
Temporal trends in plasminogen, OxPL/plasminogen in normal human subjects, patients with CAD and ACS
To assess changes with time, we measured plasminogen and OxPL/plasminogen levels in serial time points over 7 months in 18 healthy volunteers, 17 patients with stable CAD, and 8 patients with AMI, 6 of which had an ST-segment elevation myocardial infarction (Fig. 8). Interestingly, the baseline levels of plasminogen and OxPL/plasminogen were lower in the AMI patients compared with the healthy subjects and patients with stable CAD (44,432 ± 5,184 relative light units [RLU], 64,649 ± 9,043 RLU, 67,283 ± 19,821 RLU, respectively, p = 0.001 by ANOVA) (Fig. 8A). These RLU values correspond to plasminogen levels of approximately 15 to 20 mg/dl based on the standard curve of the plasminogen ELISA. Baseline levels of OxPL/plasminogen levels were also lower in the AMI patients compared with patients with stable CAD but not compared with healthy subjects (56,369 ± 19,290 RLU, 85,809 ± 30,475 RLU, 70,795 ± 15,172 RLU, respectively, p = 0.015 by ANOVA) (Fig. 8B).
Evaluating the data as a mean percent change over time across each group by ANOVA, the plasminogen levels were significantly elevated in the AMI group at discharge (p = 0.01) and after 30 days (p = 0.01), but not after 120 days and 7 months (Fig. 8B). The OxPL/plasminogen levels were also elevated at discharge (p = 0.01) and after 30 days (p = 0.05), but not after 120 days and 7 months (Fig. 8B). In contrast, there were no significant changes in plasminogen levels and OxPL/plasminogen in normal individuals (p = 0.86 and p = 0.98 by ANOVA, respectively) and patients with stable CAD (p = 0.46 and p = 0.31 by ANOVA, respectively) over time. For comparison between groups, significant differences were noted at the 30-day time point for both plasminogen and OxPL/plasminogen, but not at the other time points. Plasminogen and OxPL/plasminogen levels did not correlate with Lp(a), OxPL/apoB, or OxPL/apo(a) levels (data not shown).
This study demonstrates that plasminogen is a major carrier of OxPL in plasma of humans and animals, and OxPL appears to be important in facilitating fibrinolysis. OxPL on plasminogen are distinct from the OxPL present on Lp(a) and represent the second major pool of OxPL in plasma. Unlike OxPL/apoB and OxPL/apo(a) levels, which vary widely and which were previously shown to correlate with plasma Lp(a) levels (3,9), OxPL/plasminogen levels are distributed in a very narrow range and do not change over time among healthy subjects and patients with stable CAD. However, they rise acutely in patients following AMI. These findings may link atherogenic and thrombotic processes in defining the presence of OxPL as a unifying pathophysiological link among Lp(a) and plasminogen.
Glu-plasminogen is the zymogen that is converted to the active protease “plasmin” by cleavage of Arg561-Val562 resulting in an N-terminal heavy chain of 561 amino acids and a disulfide-linked C-terminal light chain of 230 amino acids. Plasmin can also activate glu-plasminogen to lys-plasminogen by removing the first N-terminal 77 amino acids in a positive feedback reaction. The catalytic triad of plasmin consists of His603, Asp646, and Ser741. Kringles I and IV participate in binding to lysine residues present on the surface of fibrin or cell membranes. We now present evidence in a variety of animal models, healthy humans, and patients with CAD and ACS that plasminogen is a major carrier of a distinct pool of OxPL within the non-lipoprotein fraction of plasma that is not associated with Lp(a). The LC-MS/MS results demonstrate that the OxPC covalently modify plasminogen. The precursor ion scanning experiments were carried out on peptides using a collision energy of 35 V that are indicative of the covalent nature of these modifications. One would expect that noncovalent association of PC species with plasminogen would not survive the stringent condition of denaturing, reduction, and digestion. Furthermore, disruption of noncovalent interaction in a collision-induced dissociation experiment requires much less energy. Finally, the demonstration that saponification removed the PC from the peptides is consistent with the postulated Schiff base formation between lysines of the protein and the reactive oxidized moieties present on the sn-2 side chains of the OxPL (20).
Although phospholipid surfaces are needed to activate clotting cascades, the role of OxPL in these pathways is not well defined. The few studies published in this area are not entirely consistent, but on the whole, suggest free OxPL species promote a pro-coagulant shift on the endothelium and mediate blood clotting (21). In addition, OxPL in concert with hyperlipidemia induce platelet aggregation through a CD36 scavenger receptor pathway in platelets (22). The current study suggests for the first time to our knowledge that OxPL on plasminogen are also important in facilitating fibrinolysis, as their removal results in a longer clot lysis time. This removal of OxPL may result in a decreased rate of plasminogen activation or a decreased enzymatic activity of plasmin. It may be postulated that the presence of OxPL on plasminogen may facilitate activation of plasminogen to plasmin by tPA and other plasminogen activators. Furthermore, the OxPL on plasminogen may bind to scavenger receptors on cells that are recruited at sites of inflammation, atherosclerosis, and thrombosis. It remains to be defined what percent of circulating plasminogen contains covalently bound OxPL and what their relative contributions to fibrinolysis are. In contrast, Lp(a) has been shown to inhibit fibrinolysis in vitro (18,23,24), but the specific role of the OxPL component on Lp(a) on fibrinolysis has not been investigated. The overall role of OxPL on plasminogen and Lp(a) in mediating all aspects of fibrinolysis awaits further dedicated mechanistic studies.
In vitro tissue culture studies have shown that the gene expression of plasminogen can increase significantly when cells are exposed to inflammatory mediators, such as IL-6 (25). In the current study, we demonstrate for the first time in serial time points that both plasminogen and OxPL/plasminogen levels were significantly lower at baseline in patients with AMI compared with stable CAD and normal subjects, but increase acutely following the proinflammatory milieu of AMI in subjects treated with percutaneous coronary intervention. Although this AMI study is small and hypothesis generating, it does imply that OxPL/plasminogen may vary before and following plaque rupture and thrombosis and that they may play a role in either thrombosis and fibrinolysis, the extent of reperfusion, and perhaps long-term clinical outcomes. An acute increase in plasminogen levels post-AMI was previously documented, although the clinical implications remain undefined (26). Interestingly, a similar pattern of changes in Lp(a), OxPL/apoB, and autoantibodies to oxidized low density lipoprotein (OxLDL) were previously noted in the same subjects, as well as in percutaneous coronary intervention patients in another cohort (6,12). Since plasminogen (like apoB) is usually in excess to apo(a), except in cases of very high Lp(a) levels, it suggests that plasminogen may represent a larger carrier of OxPL than Lp(a). We are currently evaluating the changes in OxPL/plasminogen and plasminogen in large ACS studies and assessing their predictive value.
Interestingly, animal and human studies have suggested, but not proven definitively, that elements of the fibrinolytic system may have proatherogenic properties (27). In fact, experimental studies have suggested that plasminogen mediates proatherogenic effects in mouse models (28,29). Although these data need confirmation, one may postulate that some of these effects may be mediated through the OxPL and CD36 macrophage scavenger receptor pathway (30,31). The presence of OxPL on plasminogen may also activate smooth muscle cells, enhance endothelial cells growth, or induce release of proinflammatory cytokines that theoretically may be beneficial in injured tissues, but detrimental in atherosclerotic lesions. The identification of specific lysine receptors on a number of cell surfaces and bacteria have implicated plasminogen in additional functions such as facilitating tissue remodeling, enhancing wound healing, mediating angiogenesis and embryogenesis, and inhibiting infection, tumor growth, and metastasis (1). The presence of OxPL on plasminogen may mediate clearance of such pathogens by recruiting additional arcs of the innate immune system, such as natural antibodies, scavenger receptors, C-reactive protein, and potentially, Lp(a), which has also been postulated to be involved in wound healing (1,32). Plasminogen-deficient patients (33) and murine models of plasminogen deficiency support many of these functions of plasminogen manifested by diffuse fibrin deposition leading to multiple organ failure (34,35).
Substantial experimental, clinical, and genetic evidence have established Lp(a) as an independent risk factor for premature CAD, death, myocardial infarction, stroke, and peripheral arterial disease, although the mechanisms underlying its proatherogenic potential are not fully established (3–5). Lp(a) has also been shown to inhibit the fibrinolytic properties of plasminogen in vitro through several mechanisms, including inhibiting tPA-mediated activation of glu-plasminogen (18,24), inhibition of plasminogen and tPA lysine-dependent binding to fibrin surfaces (36), and inhibiting the action of plasmin in converting glu-plasminogen to the activated form lys-plasminogen (23). In a large series of clinical and experimental studies (9), we have established that a key component of the atherogenicity of Lp(a) and its value in predicting new cardiovascular events (3,13) may be its unique property, among lipoproteins, to preferentially bind and transport proinflammatory OxPL. Measuring OxPL on apoB particles (OxPL/apoB), which primarily reflects OxPL on Lp(a), appears to uniquely measure the biological activity and proatherogenic potential of Lp(a), particularly of the most atherogenic small isoforms of apo(a) associated with high Lp(a) levels (14). As opposed to plasminogen, Lp(a) contains, not only covalently bound OxPL, but also OxPL present in its lipid phase (10).
It appears that only Lp(a) and plasminogen contain significant amounts of OxPL in plasma proteins, as no other immunoreactive bands can be seen in immunoblots of all species tested. There was no evidence that plasminogen was attached to OxPL-containing Lp(a) or apoB particles circulating in plasma. Kringles of the plasminogen-prothrombin gene family share conformational epitopes with each other and with apo(a), but plasminogen and Lp(a) are the only kringle-containing structures containing OxPL, although all coagulation and growth factors were not tested in this study. This suggests that it is not the kringle structures per se that mediate OxPL binding, but other as yet unidentified motif(s) common to both apo(a) and plasminogen. Edelstein et al. (37) recently reported that commercial and cell culture sources of human plasminogen contained covalently bound OxPL as detected by several techniques, including immunoreactivity of the antibody T15 (37), which our group initially demonstrated to be identical in the variable region to the IgM E06 (38), inorganic phosphate analysis, and presence of lyso-PC from plasminogen following treatment with lipoprotein-associated phospholipase A2. Their study did not evaluate non-human sources of plasminogen or evaluate its role in lipoprotein disorders and CVDs, nor did it differentiate Lp(a)/apo(a) versus plasminogen pools in plasma. Studies are underway to define the specific binding sites and amino acids on plasminogen and Lp(a) that bind OxPL. This would also allow a determination of the affinity and relative proportion of plasminogen molecules that carry OxPL.
Our findings demonstrate that plasminogen contains covalently bound OxPL and that plasminogen represents a second major pool of OxPL in human plasma in addition to Lp(a). OxPL on plasminogen seem to facilitate fibrinolysis and that patients with ACS develop elevated levels of both plasminogen and OxPL/plasminogen. The presence of OxPL on plasminogen may have pathophysiological implications in atherothrombosis and clinical events and await future basic and clinical investigations.
For an expanded Methods section and supplemental figure, please see the online version of this paper.
Oxidized Phospholipids Are Present on Plasminogen, Affect Fibrinolysis, and Increase Following Acute Myocardial Infarction
This study was funded by a grant from the Fondation Leducq (to Drs. Chapman, Witztum, and Tsimikas), the Swiss National Science Foundation (to Dr. Leibundgut), National Institutes of HealthHL0888093, HL086559 (to Drs. Witztum and Tsimikas), and ES013125 (to Dr. Yin). Dr. Chapman has received research funding from Merck, Kowa, and Pfizer; is on the speaker's bureau of Merck and Kowa; and is on the advisory board of AstraZeneca, Merck, Kowa, and Danone. Dr. Tsimikas is a consultant for ISIS and Quest; and has received grants from Merck and Pfizer. Drs. Tsimikas and Witztum are inventors of patents owned by the University of California for the clinical use of oxidation-specific antibodies. All other authors have stated that they have no relationships relevant to the contents of this paper to disclose. Stephen Nicholls, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- acute coronary syndrome(s)
- acute myocardial infarction
- analysis of variance
- apolipoprotein (a)
- apolipoprotein B-100
- coronary artery disease
- cardiovascular disease
- enzyme-linked immunosorbent assay
- familial hypercholesterolemia
- liquid chromatography–mass spectrometry
- low-density lipoprotein
- lipoprotein (a)
- oxidized phospholipids containing phosphocholine
- oxidized phospholipids
- phospholipase A2
- relative light units
- tissue-type plasminogen activator
- Received October 26, 2011.
- Revision received December 16, 2011.
- Accepted December 22, 2011.
- American College of Cardiology Foundation
- Tsimikas S.,
- Mallat Z.,
- Talmud P.J.,
- et al.
- Tsimikas S.,
- Bergmark C.,
- Beyer R.W.,
- et al.
- Edelstein C.,
- Pfaffinger D.,
- Hinman J.,
- et al.
- Bergmark C.,
- Dewan A.,
- Orsoni A.,
- et al.
- Tsimikas S.,
- Kiechl S.,
- Willeit J.,
- et al.
- Tsimikas S.,
- Lau H.K.,
- Han K.R.,
- et al.
- Kiechl S.,
- Willeit J.,
- Mayr M.,
- et al.
- Merki E.,
- Graham M.J.,
- Mullick A.E.,
- et al.
- Faghihnia N.,
- Tsimikas S.,
- Miller E.R.,
- Witztum J.L.,
- Krauss R.M.
- Schneider M.,
- Witztum J.L.,
- Young S.G.,
- et al.
- Gillotte K.L.,
- Horkko S.,
- Witztum J.L.,
- Steinberg D.
- Hancock M.A.,
- Boffa M.B.,
- Marcovina S.M.,
- Nesheim M.E.,
- Koschinsky M.L.
- Jenkins G.R.,
- Seiffert D.,
- Parmer R.J.,
- Miles L.A.
- Meltzer M.E.,
- Doggen C.J.M.,
- de Groot P.G.,
- Rosendaal F.R.,
- Lisman T.
- Folsom A.R.,
- Aleksic N.,
- Park E.,
- Salomaa V.,
- Juneja H.,
- Wu K.K.
- Kremen M.,
- Krishnan R.,
- Emery I.,
- et al.
- Xiao Q.,
- Danton M.J.S.,
- Witte D.P.,
- et al.
- Boullier A.,
- Friedman P.,
- Harkewicz R.,
- et al.
- Greenberg M.E.,
- Li X.M.,
- Gugiu B.G.,
- et al.
- Miller Y.I.,
- Choi S.H.,
- Wiesner P.,
- et al.
- Tefs K.,
- Gueorguieva M.,
- Klammt J.,
- et al.
- Ploplis V.A.,
- Carmeliet P.,
- Vazirzadeh S.,
- et al.
- Rouy D.,
- Grailhe P.,
- Nigon F.,
- Chapman J.,
- Angles-Cano E.