Thrombotic occlusion of an epicardial coronary artery on the grounds of atherosclerotic plaque is considered the ultimate step in AMI (acute myocardial infarction). However, the precise pathophysiological mechanisms underlying acute coronary occlusion are not fully understood. We have analysed proteomic profiles of systemic plasma and plasma derived from the site of coronary plaque rupture of non-diabetic patients with STEMI (ST-segment elevation myocardial infarction). Label-free quantification of MS/MS (tandem MS) data revealed differential regulation of complement cascade components and a decrease in anti-thrombotic PEDF (pigment epithelium-derived factor) between CS (culprit site)-derived plasma and systemic plasma. PEDF, which is known to have a protective role in atherothrombosis, was relatively decreased at the CS, with a level of expression inverse to local MMP-9 (matrix metalloproteinase-9) activity. CS plasma displayed enhanced proteolytic activity towards PEDF. Proteomics of coronary thrombus aspirates indicate that PEDF processing is associated with coronary plaque rupture.
- acute myocardial infarction
- matrix metalloproteinase-9 (MMP-9)
- pigment epithelial-derived factor (PEDF)
- plaque rupture
Thrombotic occlusion of an epicardial coronary artery upon atherosclerotic plaque rupture is considered the ultimate and key step in AMI (acute myocardial infarction) . The propensity of plaques to rupture and cause thrombus formation depends on a complex cascade of events involving endothelium, inflammatory cells, cytokines, endothelin, complement system, apoptosis, T-lymphocytes, metalloproteinases, PLA2 (phospho-lipase A2), cholesterol and platelets . Although some of these factors have served as predictors of outcome [3–6], the molecular and cellular mechanisms underlying vessel occlusion in AMI are not fully elucidated.
Current knowledge of atherosclerotic plaque rupture and thrombus formation has mainly been derived from post-mortem material or from animal model systems . Rupture-prone lesions usually consist of a thin fibrous cap and an abundant lipid core, and are enriched with activated inflammatory cells . Inflammatory active processes within the plaque are crucial for rupture of the fibrous cap resulting in mural thrombus formation and subsequent thrombotic vessel occlusion [9–11]. As plaque rupture and coronary thrombosis occur simultaneously, the biology of coronary thrombosis is also poorly understood. However, the harvest of fresh coronary thrombus became feasible with the introduction of thrombectomy devices, which were developed to aspirate intra-coronary thrombus material for the prevention of distal embolization during PCI (percutaneous coronary intervention) in the setting of AMI. Thus insights into the biological processes of coronary plaque rupture and thrombosis are possible.
According to an update of the guidelines of American College of Cardiology and American Heart Association for the management of STEMI (ST-segment elevation myocardial infarction) , timely recanalization of the occluded artery by aspiration thrombectomy is recommended during PCI, with a beneficial impact on cardiac mortality and morbidity [13–15].
The molecular and cellular composition of coronary thrombi was previously analysed by non-proteomic techniques [3,16]. Several studies have investigated local concentrations of selected soluble molecules at the CS (culprit site) in patients with acute coronary syndrome [17–19]. We have published the first proteomic comparison of coronary CS-derived compared with systemic plasma by two-dimensional PAGE .
In the present study, we intended to analyse pathophysiological mechanisms associated with coronary plaque rupture. Therefore we have analysed coronary thrombus aspirates with one-dimensional PAGE followed by GeLC-MS/MS (tandem MS). Because of an altered synthesis and processing of many proteins in patients with diabetes , we excluded diabetic patients.
Our findings corroborate our previous description of local complement activation at the site of plaque rupture . In the present study, we detected a local down-regulation of PEDF (pigment epithelium-derived factor) at the culprit lesion site. Our data suggest a causal relationship of local complement activation, MMP (matrix metalloproteinase)-9 activity and the decrease in PEDF at the site of coronary thrombus formation.
MATERIALS AND METHODS
Non-diabetic patients in the setting of primary PCI were included on the basis of a de novo lesion of a native coronary vessel with TIMI 0 flow, and an intraluminal filling defect suggestive of thrombus within 50 mm of the respective ostium. Patients gave written informed consent under an approval of the Ethics Committee of the Medical University of Vienna, Vienna, Austria.
Sample collection was performed as described in . In brief, STEMI patients were heparinized with unfractionated heparin at an activated coagulation time ≥300 s, after taking 250 mg of aspirin and 600 mg of clopidogrel. A total of 20 ml of CS blood was aspirated with a Medtronic Export Aspiration Catheter (Medtronic Vascular) or the Pronto® V3 thrombus extraction catheter (Vascular Solutions). Because of a 0.9% NaCl flush of the aspiration catheter prior to thrombectomy, we normalized specimens by haematocrit resulting in equal total protein concentrations . Particulate coronary thrombus material was separated with a cell strainer (pore filter size 40 μm, BD Falcon, Becton Dickinson). The particulate thrombus was immediately frozen at −80°C for later analysis. In parallel, systemic blood was drawn from the femoral sheath. Blood samples collected into EDTA K30 and heparin lithium tubes were processed for routine blood analysis, real-time RT–PCR (reverse transcription–PCR) and for generation of platelet-poor plasma. Plasma aliquots were stored at −80°C until measurement.
In the first part of the present study, plasma samples from 12 representative patients [SP1 (study population 1)] were analysed to determine the local proteomic pattern at the acute coronary thrombus site. Semi-quantitative proteomic techniques were employed to compare plasma derived from aspirates of the culprit lesion site with plasma simultaneously harvested from systemic arterial blood. The complexity of the plasma proteome  requires extensive and laborious sample processing; hence, the experiments were carefully performed in a small but representative study population.
In the second part of the study, proteomically identified protein alterations were confirmed in a larger independent study population [SP2 (study population 2)] by high-throughput quantitative analyses. PEDF plasma levels were measured by ELISA analyses, whereas MMP-2 and MMP-9 activities were determined by zymographic assays. SP2 consisted of 42 patients, who were enrolled according to the same inclusion criteria.
Qualitative analysis of proteins expressed at the site of coronary plaque rupture was achieved by GeLC-MS/MS analysis from a total of 12 patient plasma sample pairs. To identify quantitative differences in local protein enrichment at the CS, CS-derived plasma was compared with systemic plasma of the same patients. Plasma sample pairs were depleted by the use of the 6-Hu multiple affinity removal system (Agilent) according to the manufacturer's instructions prior to GeLC-MS/MS.
A total of 30 μg of depleted plasma were separated on a 6 cm 6% polyacrylamide stacking gel followed by a short (1 cm) 12% polyacrylamide separating gel as described previously . After electrophoresis, gels were silver stained and entire gel lanes were cut into ten bands as described previously . Bands were destained with 15 mM K3Fe(CN)6/50 mM Na2S2O3 and reduced and alkylated prior to gel digestion with 0.1 mg/ml trypsin (Roche Diagnostics) overnight at 37°C . The digested peptides were extracted in 5% formic acid/50% acetonitrile and concentrated in a vacuum centrifuge prior to nanoLC-MS/MS analysis .
NanoLC-MS/MS data were acquired in three replicates from all plasma sample patient pairs. In order to minimize bias due to instrumental drift, the injection scheme of ten gel bands per sample was alternated between ‘CS-band 1', ‘systemic band 1', followed by ‘CS-band 2', ‘systemic band 2' etc., until band 10. Peptides were separated using the Agilent Chip technology combined with an Agilent 1100 nanoLC system as described previously . Then, peptides were trapped on a 40 nl Zorbax 300SB-C18 column (Agilent) at 250 nl/min with mobile phase-A: 2% acetonitrile/0.2% formic acid. Peptides were separated on 75 mm×150 mm Zorbax 300SB-C18 column (Agilent) at 400 nl/min using a 60 min gradient from 3% to 50% mobile phase-B: 98% acetonitrile/0.2% formic acid. Peptide identification was accomplished by MS/MS fragmentation analysis with an iontrap mass spectrometer (XCT-ultra, Agilent) equipped with an orthogonal nanospray ion source. MS/MS data were interpreted by the Spectrum Mill MS Proteomics Workbench software (Agilent; version A.03.03.081), including peak list generation and a search engine, allowing for two missed cleavages and searched against the Swiss-Prot Database for human proteins (version 15.1) allowing for precursor mass deviation of 1.5 Da, a product mass tolerance of 0.7 Da and a minimum matched peak intensity [%SPI (percentage scored peak intensity)] of 70%. Carbamidomethylation of cysteine residues was set as the fixed modification. Oxidized methionine was set as the variable modification.
Estimated false discovery rates were calculated from reversed database searches as previously described . Reversed scores are listed in the publicly available PRIDE-XML files. Consistently, less than 1% of peptides scoring higher than 13 were found in the searches using the reversed database in comparison with the reference database. Therefore we automatically assigned a protein valid if it was identified with at least one specific peptide scoring above 13.0, which was not found existing in any other human protein by BLAST searches. Protein identifications not fulfilling that requirement were selected manually as described previously  whereby selection of protein isoforms was essentially performed as described by Zhang et al. . Only peptides scoring above 9.0 were entered into the database. Identified proteins were quantified calculating the emPAI (exponentially modified protein abundance index) . The emPAI value of an individual protein is applied to determine the protein content within the plasma in molar fraction percentages. Mean emPAI values, obtained from 12 patient sample pairs, were used to calculate ratios between CS and systemic plasma samples for comparative analysis.
Determination of PEDF levels in plasma
CS and systemic plasma concentrations of PEDF were determined by commercially available ELISA kits (Chemikine PEDF Sandwich ELISA Kit, Chemicon International). Plasma samples were thawed on ice and after adding urea to a final concentration of 8 M, they were incubated on ice for 1 h. Subsequently, the assay was performed according to the manufacturer's instructions. All tests were run in duplicate. Results were expressed as Δ levels (defined as the relative difference between CS and systemic plasma).
Determination of mRNA expression for PEDF in blood
Total RNA isolation from CS and systemic blood samples was performed by using the QIAamp RNA Blood Mini kit (Qiagen). PEDF primers and probes (Assay-on-Demand Hs01106937_m1) and a eukaryotic 18S RNA endogenous control were from Applied Biosystems. The real-time ABI Prism 7000 sequence detection system (Applied Biosystems) was used.
Determination of MMP activity in plasma
We estimated the MMP-2 and MMP-9 activity in heparinized CS plasma by gelatinase zymography as described previously . Premade 10% polyacrylamide gels containing 0.1% gelatin (Invitrogen) were loaded with 1 ng of recombinant MMP-9, 2 ng of recombinant MMP-2 (Sigma–Aldrich) and 1 μl of CS plasma and were run with Tris/glycine running buffer (Invitrogen). After electrophoresis, gels were incubated in renaturing buffer (Invitrogen) at room temperature (20°C) for 30 min and then incubated in developing buffer (Invitrogen) at 37°C overnight. Subsequently, gels were stained with PageBlue™ Protein Staining Solution (Fermentas Life Sciences) and scanned on the Image Scanner II (GE Healthcare).
We compiled a standard curve of recombinant MMP-9 and MMP-2, and expressed the proteolytic activity using ImageJ 1.42q software (Wayne Rasband, National Institutes of Health) as described previously . In brief, plasma level activities were calculated using the following equation: where Iobs and Istd are the intensities of lytic gel areas of plasma samples and standard MMPs respectively and Wstd is the amount (ng) of standard MMPs loaded on to the gel. Zymographic data are expressed as plasma MMP activity corresponding to μg/ml of the recombinant MMP reference standard.
PEDF proteolytic degradation solution assay and Western blot analysis
Recombinant GST (glutathione transferase)-tagged PEDF (Novus Biologicals) was incubated with heparinized CS, systemic plasma and particulate thrombus lysate from a total of four patients at 37°C for 10 and 180 min respectively. After incubation, GST-tagged PEDF was extracted from the plasma by GST Spin Traps (GE Healthcare) in accordance with the manufacturer's protocol. The degradation of PEDF was determined by Western blot analysis. Proteins were separated on a 12% (w/v) polyacrylamide gel and transferred on to 0.2 μm nitrocellulose membranes (Schleicher & Schuell) and immunoreactions were performed using a rabbit polyclonal anti-GST antibody (1:1000 dilution; Abcam). After scanning the blots (Image Scanner II; GE Healthcare), quantification of bands was performed using ImageJ 1.42q software and indicated as aU (arbitrary units).
Data analysis and statistics
All measurements were normalized for haematocrit resulting in equal total protein concentrations. The Wilcoxon signed-rank test was applied for pairwise comparison of protein levels and MMP activities. Variables are expressed as medians with inter-quartile ranges (25th–75th percentiles). Partial correlation analyses were employed for evaluating a correlation of PEDF levels with MMP-9 and MMP-2 activities in CS plasma in the bivariate Spearman correlation, while adjusting for time delay from symptom onset to first balloon inflation as potential confounders. P values ≤0.05 were considered significant. Statistical analyses were performed with version 13.0 SPSS for Windows.
The patient characteristics of SP1 and SP2 are summarized in Table 1. There was no significant difference between patient characteristics of SP1 and SP2.
Profiling of soluble proteins enriched at the site of plaque rupture
To identify soluble factors associated with coronary plaque rupture during AMI, we analysed plasma derived from the coronary CS compared with systemic plasma by GeLC-MS/MS. After filtering proteins for shared peptides , we retrieved 195 protein identities from plasma immunodepleted of high abundant proteins, 164 of which showed at least two distinct peptide identifications in at least two experiments (Supplementary Table S1 at http://www.clinsci.org/cs/123/cs1230111add.htm). LC-MS/MS results were uploaded to the PRIDE protein identification database (http://www.ebi.ac.uk/pride/).
Calculation of emPAI values revealed nine out of 164 valid proteins to be present at different concentrations between the culprit lesion site and the systemic plasma in a statistically significant fashion (Table 2 and Figure 1). Complement factor H-related protein 2 was found enriched at the coronary CS (P=0.026). Proteins found to be significantly decreased at the coronary CS included complement factor B (P=0.033), carboxypeptidase N subunit 2 (P=0.027), plasminogen (0.021), C5 (complement C5) (P=0.033), PEDF (P=0.013), serum paraoxonase/arylesterase 1 (P=0.05), complement component C7 (P=0.018) and vitamin K-dependent protein S (P=0.024).
PEDF protein levels are decreased at the site of plaque rupture
Analyses of atherosclerotic lesions suggest that PEDF may play a crucial role in stabilizing atherosclerotic plaque by controlling the angiogenic balance . Therefore we decided to further analyse the relative proteomic decrease in PEDF. ELISA of coronary thrombus aspirates confirmed decreased levels of plasma PEDF at the CS [1.87 (1.06–2.81) μg/ml, n=42] compared with corresponding systemic plasma [2.12 (1.42–3.63) μg/ml, P=0.017] (Figure 2). In contrast, real-time RT–PCR analyses of 12 patients revealed that PEDF mRNA levels are equal in circulating blood cells and cells at the site of coronary thrombus (log relative quantity −0.09).
Proteolytic degradation of PEDF by CS plasma and thrombus
On the basis of the available evidence, we hypothesized that the decrease in PEDF plasma concentration at the site of plaque rupture might be caused by a local cleavage of PEDF. PEDF is known to be degraded by the plasma metalloproteinases MMP-2 and MMP-9 . MMP-9 and C5a, a potent inducer of MMP-9, were both enriched at the coronary CS [20,31]. In order to test whether PEDF was enzymatically degraded at the coronary CS, we incubated a GST-tagged form of purified PEDF in vitro with particulate thrombus lysate, CS plasma, systemic plasma or PBS alone for 10 and 180 min. Quantification of the cleavage product (GST) by immunoblotting served as a measurement of PEDF degradation. As shown in Figure 3, particulate thrombus lysate was most efficient in degrading PEDF in vitro after 180 min (9.5±2.4 aU), compared with CS plasma (5.3±0.47 aU), systemic plasma (3.0±1.50 aU) and PBS alone (1.1±0.20 aU). The differences in GST accumulation between 10 and 180 min showed that PEDF degradation occurs in a time-dependent manner.
Correlation between PEDF level and MMP-9 activity in CS plasma
Total PEDF protein was inversely correlated with MMP-9 activity in CS plasma (r=−0.365, P=0.029 in partial correlation correcting for time delay) as shown in Figure 4, whereas MMP-2 activity showed no correlation with PEDF levels (P=0.895).
Despite recent data from advanced intravascular imaging , the precise mechanisms underlying acute coronary occlusion are unknown. Although animal model systems are useful to understand biological mechanisms, they are clearly limited in their accuracy to replicate human disease. One approach to understand plaque rupture is to investigate the acute proteome of coronary thrombosis.
The decrease in complement zymogen precursors C5, C7 and complement factor B, as well as the accumulation of complement factor H-related protein 2 corroborate our previous findings of local complement activation at the site of coronary thrombus formation . The low plasminogen and vitamin K-dependent protein S levels are in accordance with the procoagulatory state at the site of thrombus formation. The local decrease in carboxypeptidase N subunit 2 and serum paraoxonase/arylesterase 1 has to be confirmed by high-throughput analyses in a larger study population.
PEDF was found to be decreased at the coronary CS by proteomic and ELISA analyses. Even though the magnitude of the proteomic differences of PEDF was moderate, differential PEDF expression was detected in all plasma samples.
MMP-9 is known to proteolytically degrade PEDF . As expected from studies in stable coronary disease [32–33] and AMI [34–36], we identified MMP-9 in particulate coronary thrombus (results not shown). The crucial role of MMPs in destabilizing coronary plaques, by degrading extracellular matrix components of the fibrous cap, has been well investigated [37–39]. However, the precise mechanisms triggering coronary plaque rupture have not yet been elucidated.
In our analyses, local MMP-9 activity inversely correlated with local PEDF levels. We suggest that MMP-9 contributes to increased proteolytic activity of CS-derived plasma towards PEDF in vitro and the decreased PEDF levels at the CS in vivo (Figure 4).
A recent study found a co-localization of the anaphylatoxin C5a with MMP-1 and MMP-9 in human coronary plaques. In addition, C5a-induced MMP-9 and MMP-1 expression in macrophages from atherosclerotic plaques . In the present study, C5a did not affect the expression of MMP-2 mRNA, which might explain the correlation of MMP-9 but not MMP-2 activity with PEDF levels. In addition to macrophages from atherosclerotic plaque [31,38], neutrophils have been identified as the main source of MMP-9 in patients with coronary artery disease . Neutrophils at the CS [3,20] are able to express MMP-9 after incubation with C5a . In the present study, thrombus-site blood neutrophil counts showed a significant correlation with local MMP-9 activity (r=0.383, P=0.015; Pearson correlation). These data suggest a causal relationship between local complement activation, neutrophil recruitment, MMP-9 and the local decrease in PEDF at the site of coronary plaque rupture.
PEDF is a 50 kDa secreted glycoprotein that belongs to the non-inhibitory serpin family . It was first identified in retinal pigment epithelium  and is responsible for maintaining the avascularity of the corneal tissue [44–46]. PEDF counterbalances the pro-angiogenic effect of VEGF (vascular endothelial growth factor) by inducing endothelial cell apoptosis [46,47]. PEDF is a pleiotropic protein with anti-angiogenic, neuroprotective, neurotrophic, anti-tumorigenic, antioxidant, anti-inflammatory and anti-thrombotic properties [48–51]. As most of these effects are crucial during AMI, the role of PEDF in CVD (cardiovascular disease) has been investigated frequently in recent years (reviewed in ). Neovascularization in atherosclerotic lesions is associated with plaque progression and destabilization . PEDF has been identified within human atherosclerotic plaque and may control the angiogenic balance during the progression of atherosclerotic lesions . In addition to its anti-angiogenic effects, PEDF is known to protect endothelial cells against inflammatory activation and injury by suppressing TNFα (tumour necrosis factor α)-induced IL (interleukin)-6 expression and ROS (reactive oxygen species) generation . IL-6 is the main inducer of the hepatic synthesis of CRP (C-reactive protein), which, in turn, may activate the complement system inside the coronary vessel .
The source of PEDF within coronary plaque is unclear but PEDF may diffuse from plasma on to extracellular matrix in atherosclerotic lesions . In patients with acute coronary syndrome, plasma and intraplatelet PEDF levels were decreased .
A simple take-home message from our in vivo study is that, at the site of atherosclerotic plaque rupture, MMP-9 is proteolytically active [38,55–57] and is able to degrade exogenous PEDF. As PEDF mRNA levels are equal in circulating blood cells and cells at the plaque rupture site, low native PEDF levels at the coronary CS are due to accelerated cleavage by MMP-9, thus giving way to local inflammation and oxidative stress, leading to plaque rupture.
Current proteomic techniques allow for only a relatively low-throughput. Therefore our findings are based on a limited number of representative patient samples. Quantitative high-throughput analyses were performed in a second study population. The complexity and the highly dynamic range of tissue and plasma proteomes limit access to candidate proteins. In addition, the stringent filters applied to MS/MS identification (to rule out false-positives) resulted in a limited number of proteins. The relationship of blood volumes aspirated from the CS (10–20 ml) and the actual volume of several hundred microlitres within the coronary artery up to the site of plaque rupture  may account for a relative under-representation of local factors.
We suggest that PEDF processing by MMP-9 at the CS is associated with coronary plaque rupture.
This work was supported by the Austrian National Bank [grant number ONB12162 (to I.M.L.)] and the Oesterreichische Kardiologische Gesellschaft [grant number 2009 (to K.D.)].
Klaus Distelmaier conceived and designed the study, performed analysis and interpretation, data collection, writing the paper, statistical analysis and obtained funding. Christopher Adlbrecht performed data collection, analysis and interpretation, and critical revision of the paper. Johannes Jakowitsch and Christopher Gerner provided methodological support, performed analysis and interpretation, and critical revision of the paper. Oswald Wagner performed analysis and interpretation, and critical revision of the paper. Irene Lang took overall responsibility including final approval of the paper, obtaining funding, conception and design, analysis and interpretation. Markus Kubicek took overall responsibility including final approval of the paper, conception and design, analysis and interpretation.
We thank Veronika Seidl for technical assistance, Martin Perlinger for mass spectrometer maintenance, and Helge Wimmer, Johannes Griss, Wolfgang Trittner and Editha Bayer for computational support.
Abbreviations: AMI, acute myocardial infarction; aU, arbitrary units; CS, culprit site; emPAI, exponentially modified protein abundance index; GST, glutathione transferase; IL, interleukin; MMP, matrix metalloproteinase; MS/MS, tandem MS; PCI, percutaneous coronary intervention; PEDF, pigment epithelium-derived factor; RT–PCR, reverse transcription–PCR; SP, study population; STEMI, ST-segment elevation myocardial infarction
- © The Authors Journal compilation © 2012 Biochemical Society