OUP user menu

Aspirin protects human coronary artery endothelial cells against atherogenic electronegative LDL via an epigenetic mechanism: a novel cytoprotective role of aspirin in acute myocardial infarction

Po-Yuan Chang, Yi-Jie Chen, Fu-Hsiung Chang, Jonathan Lu, Wen-Huei Huang, Tzu-Ching Yang, Yuan-Teh Lee, Shwu-Fen Chang, Shao-Chun Lu, Chu-Huang Chen
DOI: http://dx.doi.org/10.1093/cvr/cvt062 137-145 First published online: 21 March 2013


Aims L5 is the most negatively charged subfraction of human low-density lipoprotein (LDL) and is the only subfraction of LDL capable of inducing apoptosis in cultured vascular endothelial cells (ECs) by inhibiting fibroblast growth factor-2 (FGF2) transcription. We examined whether plasma L5 levels are elevated in patients with ST-segment elevation myocardial infarction (STEMI) and whether aspirin provides epigenetic protection of human coronary artery ECs (HCAECs) exposed to L5.

Methods and results Plasma L5 levels were compared between patients with STEMI (n = 10) and control subjects with chest pain syndrome but a normal coronary arteriogram (n = 5). L5 was isolated from the plasma of STEMI patients and control subjects, and apoptosis, FGF2 expression, and FGF2 promoter methylation were examined in HCAECs treated with L5 and aspirin. Plasma L5 levels were significantly higher in STEMI patients than in control subjects (P < 0.001). Treatment of HCAECs with L5 resulted in reduced survival and FGF2 expression and increased CpG methylation of the FGF2 promoter. Co-treatment of HCAECs with L5 and a physiologically relevant, low concentration of aspirin (0.2 mM) attenuated the adverse effects of L5 on HCAEC survival, FGF2 expression, and FGF2 promoter methylation. In contrast, high concentrations of aspirin (≥1.0 mM) accentuated the effects of L5.

Conclusions Our results show that L5 levels are significantly increased in STEMI patients. Furthermore, L5 impairs HCAEC function through CpG methylation of the FGF2 promoter, which is suppressed in the presence of low-concentration aspirin. Our results provide evidence of a novel mechanism of aspirin in the prevention of MI.

  • Aspirin
  • DNA methylation
  • Genes
  • Lipoproteins
  • Myocardial infarction

1. Introduction

The antiplatelet effect of aspirin (acetylsalicylic acid) that results from its inhibition of cyclooxygenase enzymes has been well described and is the pharmacologic basis of aspirin use in the prevention of acute myocardial infarction (MI), including ST-segment elevation MI (STEMI) and non-STEMI.1,2 Aspirin has also been shown to have cytoprotective functions that are unrelated to its antiplatelet activity,3,4 such as improving endothelium-dependent arterial relaxation by inducing the release of nitric oxide from the vascular endothelium5 and reducing apoA levels in human hepatocytes by suppressing apoA gene transcription.6 Thus, aspirin may protect against MI through mechanisms independent of its antiplatelet activity.

Electronegative low-density lipoprotein (LDL) is a class of naturally occurring atherogenic lipoproteins.7 Human plasma can be chromatographically resolved into five subfractions, L1–L5, with increasingly negative charge.8,9 L5, the most electronegative subfraction, is the only subfraction capable of inducing marked atherogenic changes in cultured vascular endothelial cells (ECs) and is moderately elevated (up to 5–6% of LDL) in asymptomatic individuals with increased cardiac risks, including hypercholesterolaemia, active smoking, type 2 diabetes, and metabolic syndrome.8,1013 L5 differs from experimentally derived oxidized LDL (oxLDL) physically and chemically,11,14 but both forms of LDL share similar functional properties and signal through the lectin-like oxLDL receptor-1 (LOX-1).1517 In cultured vascular ECs, L5 and oxLDL induce apoptosis by inhibiting the expression of fibroblast growth factor-2 (FGF2),8,18,19 a pleiotropic protein that maintains normal endothelial physiology20 and protects against MI.21,22 We have previously shown that homocysteine, another protein that may be linked to cardiovascular disease, suppresses FGF2 transcription through CpG methylation of the FGF2 promoter.23 However, it is not known whether CpG methylation of the FGF2 promoter is inducible by atherogenic LDL.

In our study, we hypothesized that L5 levels are elevated in the plasma of STEMI patients. In addition, we examined in human coronary artery ECs (HCAECs) whether aspirin attenuates the effects of L5 on cell survival and FGF2 expression and whether CpG methylation is involved. Our results provide evidence of a novel protective mechanism of low-dose aspirin independent of its antiplatelet activity and substantiate the importance of aspirin therapy for protecting against MI.

2. Methods

2.1 Study subjects

This study was approved by the institutional review board of the National Taiwan University Hospital and conforms with the principles outlined in the Declaration of Helsinki. All participants gave written informed consent. We analysed 10 subjects with STEMI undergoing primary coronary intervention and 5 control subjects with chest pain syndrome but a normal coronary arteriogram. The criteria for STEMI were defined according to the consensus definition of the American Heart Association/American College of Cardiology.24

2.2 Cell culture

HCAECs (Clonetics, Lonza Group Ltd., Switzerland) were maintained in EGM-MV medium supplemented with 20% foetal bovine serum and antibiotics (Clonetics). Cell cultures (passages 4–7) grown to subconfluence were washed three times with the serum-free medium and were maintained under serum-free conditions for 6 h before being treated with various agents according to the protocols determined by preliminary experiments. Eighteen hours after the designated treatment, cells were incubated with phosphate-buffered saline (PBS; lipoprotein-free control), L1, or STEMI L5 (50 μg/mL each) for 24 h in the presence or absence of aspirin (0–5 mM). At least three independent experiments (each in triplicate) were performed for each treatment group.19,23

2.3 LDL preparations

Plasma LDL obtained from STEMI patients and control subjects was resolved into subfractions L1–L5 by fast protein liquid chromatography equipped with an anion-exchange column, as described previously.8,9 OxLDL was prepared by copper oxidation of L1 from control subjects as described previously.18,19 Control LDL was L1 from control subjects. Protein concentration was estimated by the Lowry method. Agarose gel electrophoresis of LDL preparations was performed by using the Beckman Paragon system (Beckman, Palo Alto, CA, USA).

2.4 Analysis of FGF2 protein levels and DNA synthesis

HCAECs (1 × 106 cells/well) were seeded in 12-well Corning cell culture plates (Corning, Lowell, MA, USA) and treated with increasing concentrations of aspirin (0, 0.2, 1, 3, or 5 mM) in the presence of PBS, L5 from STEMI patients (STEMI L5; 50 μg/mL), or oxLDL (50 μg/mL) for 24 h. DNA synthesis was quantified by measuring 3H-thymidine (Moravek Biomedicals, Brea, CA, USA) incorporation.18,19 FGF2 protein concentrations were measured with an enzyme-linked immunosorbent assay (ELISA) by using a Quantikine kit (R&D Systems, Minneapolis, MN, USA).

2.5 Internalization of L5 by HCAECs

The internalization of L5 into HCAECs was studied by using L5 labelled with 1,1′-dioctadecyl-3,3,3′,3′ tetramethylindocarbocyamine perchlorate (DiI; Sigma-Aldrich, St Louis, MO, USA).25 HCAECs were pre-treated with aspirin (0.2 mM), anti-LDL receptor (LDLR; 0.1 μg/mL; R&D Systems), or anti-LOX-1 (0.1 μg/mL; R&D Systems) at 37°C for 16 h. Cells were washed with the fresh medium, followed by incubation with 50 μg/mL DiI-labelled L5 at room temperature for 1 h. Intracellular DiI fluorescence was calculated relative to that in cells pre-treated with PBS.

2.6 Analysis of cell viability after the treatment of HCAECs with pharmacologic inhibitors

HCAECs (5 × 104 cells/well) were dispensed into 24-well plates and incubated with L5 (50 μg/mL) or PBS for 24 h in the presence or absence of aspirin (0.2 mM), Gi protein inhibitor pertussis toxin PTX (100 ng/mL), methylation inhibitor 5-aza-deoxycytidine (5-aza-dC; 0.4 μg/mL), or Akt inhibitor 1L6-hydroxymethyl-chiro-inositol-2-(R)-2-O-methyl-3-O-octadecyl-sn-glycerocarbonate (Calbiochem, San Diego, CA, USA) (1 μg/mL). The index of EC viability was determined by the colorimetric MTT assay (Sigma-Aldrich). Absorbance was measured at a wavelength of 540 nm by using a microplate reader (Thermo Electron Corporation, Waltham, MA, USA). Cell viability was calculated relative to that in the PBS-treated control.

2.7 Luciferase reporter gene assay

The reporter gene assay was performed by using a dual-luciferase expression system (Promega). Human FGF2 5′-flanking sequences26 were PCR amplified from human genomic DNA and inserted into the firefly luciferase reporter vector, pGL3-basic, and were completely sequenced. FGF2 constructs contained the FGF2 promoter sequence (−126 to +43 or −126 to +179) or FGF2 gene sequence with the promoter deleted (+24 to +179). HCAECs were grown to 80% confluence in plastic 12-well plates and were transfected with 0.75 μg of pGL3-basic or an equimolar amount of each pGL3-FGF2 construct along with 0.5 μg of the Renilla luciferase expression vector, phRL-TK, by using Superfect reagent according to the manufacturer's instructions (Qiagen, Valencia, CA, USA). Cells were treated 24 h later with PBS or STEMI L5 (50 μg/mL) for 24 h in the presence or absence of 0.2 mM aspirin or 0.2 mM indomethacin. Cell lysates were prepared for luciferase assays by using luciferin and a luminometer (Packard Harvester; Packard Instrument, Meriden, CT, USA). The promoter activity of the reporter construct was normalized to the promoter activity of phRL-TK and expressed as a fold-increase relative to that in cells transfected with pGL3-basic.

2.8 Bisulfite genomic DNA sequencing

HCAECs (cultured in triplicate) were grown to 80 to 90% confluence and then treated with STEMI L5 (50 μg/mL) in the presence or absence of aspirin (0.2 mM) for 24 h, followed by genomic DNA extraction using standard procedures. The EpiTect bisulfite kit (Qiagen) was used to convert all unmethylated cytosine residues in genomic DNA (2 μg) to uracil. Bisulfite-modified DNA was amplified with FGF2-specific primers (see Supplementary material online, Table S1) by using the following cycling conditions: 15 min at 95°C followed by 40 cycles of 10 s at 95°C and 45 s at 55°C.23 DNA sequencing was performed for 10 plasmid clones from each treatment group.

2.9 Statistical analysis

The significance of the difference between mean values was assessed by using a two-way Student t-test for single comparisons and the Bonferroni test for multiple comparisons. Probability values <0.05 were considered significant. Results are expressed as the mean ± standard error of the mean.

3. Results

3.1 Plasma L5 levels are significantly elevated in STEMI patients

The characteristics of STEMI patients and control subjects are shown in Table 1. Total cholesterol and LDL-cholesterol levels were similar between STEMI and control groups. However, the percentage of L5 LDL (L5/LDL%) and thus the concentration of plasma L5 ([L5]) were significantly higher in STEMI patients than in control subjects (P < 0.001) (Table 1, Figure 1A). Agarose gel electrophoresis confirmed the electronegativity of STEMI L5 (Figure 1B).

View this table:
Table 1

Patient characteristics

Age (years)58 ± 361 ± 50.55
 Female, n (%)2 (20)1 (20)
 Male, n (%)8 (80)4 (80)
Risk factor
 Hypertension, n (%)3 (30)2 (40)
 Smoking, n (%)2 (20)1 (20)
 Diabetes, n (%)2 (20)0 (0)
 None, n (%)3 (30)2 (40)
Total cholesterol (mg/dL)187 ± 15180 ± 90.705
LDL cholesterol (mg/dL)124 ± 8111 ± 100.342
L5/LDL (%)11.6 ± 1.60.4 ± 0.1<0.001
[L5] (mg/dL)14.1 ± 1.90.4 ± 0.1<0.001
L1 + L2/LDL (%)88.4 ± 5.799.6 ± 2.50.04
L3 + L4/LDL (%)NDND
  • Values are n or mean ± SD, unless otherwise indicated. LDL, low-density lipoprotein; STEMI, ST-segment elevation myocardial infarction; ND, not determined due to undetectable levels of L3 and L4.

Figure 1

Analysis of low-density lipoprotein (LDL) subfractions in control subjects and patients with ST-segment elevation myocardial infarction (STEMI). (A) Representative fast protein liquid chromatography analysis showing the greater distribution of the L5 LDL subfraction in STEMI patients than in control subjects. (B) A representative agarose gel showing the increased electrophoretic mobility of STEMI L5. OxLDL, oxidized LDL; mAU, milli (10−3) absorbance units.

3.2 Aspirin attenuates the effects of L5 on FGF2 protein expression and DNA synthesis

In the clinical setting, peak plasma concentrations of aspirin only reach ∼0.15 mM, even after the oral administration of high-dose aspirin (650 mg).27 However, aspirin concentrations previously used in several in vitro studies have been in the mM range. Thus, in our experiments, we examined the effects of low- and high-concentration aspirin. Low-concentration aspirin (0.2 mM) alone did not affect intracellular FGF2 protein levels, but high-concentration aspirin (1–5 mM) alone reduced intracellular FGF2 protein levels in a concentration-dependent manner (Figure 2A; see Supplementary material online, Figure S1). Furthermore, 50 μg/mL STEMI L5 or oxLDL each decreased intracellular FGF2 protein levels by approximately 80%—an effect that was alleviated by low-concentration aspirin but not high-concentration aspirin (Figure 2A and B). Parallel with these findings, low-concentration aspirin (0.2 mM) alone had no effect on DNA synthesis in HCAECs, whereas high-concentration aspirin (1–5 mM) alone inhibited DNA synthesis in HCAECs in a concentration-dependent manner (Figure 2C). STEMI L5 (50 μg/mL) decreased DNA synthesis in HCAECs by about 40%—an effect that was prevented by low-concentration aspirin but not high-concentration aspirin (Figure 2C).

Figure 2

Effects of aspirin on fibroblast growth factor-2 (FGF2) protein expression and DNA synthesis in the presence or absence of atherogenic low-density lipoprotein (LDL). Intracellular FGF2 protein levels assessed by an enzyme-linked immunosorbent assay in human coronary artery endothelial cells (HCAECs) treated with increasing concentrations of aspirin in the presence of (A) 50 μg/mL oxidized LDL (oxLDL), (B) 50 μg/mL STEMI L5, or phosphate-buffered saline (PBS). (C) DNA synthesis assessed by 3H-thymidine incorporation in HCAECs treated with increasing concentrations of aspirin in the presence of PBS or STEMI L5 (50 μg/mL). All values are presented as the mean ± SEM and are representative of three independent experiments. *P < 0.05, **P < 0.01 vs. PBS; P < 0.05 vs. oxLDL or L5 alone. ASA, aspirin.

3.3 Low-concentration aspirin attenuates the internalization of L5 by HCAECs and L5-induced LOX-1 mRNA expression

To determine whether aspirin affects the internalization of L5 by HCAECs, we used DiI-labelled L5 to visualize intracellular L5. In HCAECs pre-treated with aspirin (0.2 mM), the internalization of DiI-labelled L5 (50 μg/mL) was significantly blocked (P < 0.05) (Figure 3A). As expected, pre-treatment with anti-LOX-1, which neutralizes the LOX-1 receptor that mediates the endocytosis of L5, also attenuated the uptake of L5 in HCAECs, whereas pre-treatment with anti-LDLR did not (Figure 3A). To determine whether aspirin effects LOX-1 mRNA expression, we examined LOX-1 mRNA levels in HCAECs treated with L1, L5, or L5 + 0.2 mM aspirin. Real-time PCR analysis showed that L5, but not L1, increased LOX-1 mRNA expression and that aspirin partially reversed the effects of L5 (see Supplementary material online, Figure S2).

Figure 3

Aspirin-regulated L5 signal transduction pathways. (A) Effects of aspirin on the internalization of L5 in human coronary artery endothelial cells (HCAECs). Representative photomicrographs of HCAECs that were incubated with 50 μg/mL DiI-labelled L5 after pre-treatment with phosphate-buffered saline (PBS), 0.2 mM aspirin, 0.1 μg/mL anti-LOX-1 (lectin-like oxidized low-density lipoprotein receptor-1) antibody, or 0.1 μg/mL anti-LDLR (low-density lipoprotein receptor) antibody (original magnification ×20). Values are presented as the mean ± SEM and are representative of three independent experiments. *P < 0.05, **P < 0.01 vs. PBS. ASA, aspirin. (B) Effects of signal transduction inhibitors on aspirin-regulated L5 cytotoxicity. Cell viability assessed by the MTT assay in human coronary artery endothelial cells (HCAECs) incubated with 50 μg/mL STEMI L5, either alone or with signalling pathway modifiers, as indicated (treatments 1–8). Values are presented as the mean ± SEM and are representative of three independent experiments. *P < 0.05 vs. L5 alone (treatment 1); **P < 0.05 between the indicated pair. ASA, aspirin; PTX, pertussis toxin; 5-aza-dC, 5-aza-deoxycytidine; Akt inh, Akt inhibitor; PBS, phosphate-buffered saline.

3.4 Akt mediates aspirin's regulation of L5-induced cytotoxicity

To characterize the pathway through which aspirin regulates L5-induced cytotoxicity, we examined HCAEC viability in the presence or absence of STEMI L5, aspirin (0.2 mM), or pharmacologic inhibitors by using the MTT assay (Figure 3B; see Supplementary material online, Figure S3). Relative to PBS, STEMI L5 alone reduced cell viability by about 40% (treatment 1), which was significantly reversed in cells incubated with PTX (treatment 3) or 5-aza-dC (treatment 4) (Figure 3B). Aspirin (0.2 mM) also reversed the cytotoxic effect of L5 (treatment 2); however, co-treatment with the Akt inhibitor 1L6-hydroxymethyl-chiro-inositol-2-(R)-2-O-methyl-3-O-octadecyl-sn-glycerocarbonate blocked the effect of aspirin (treatment 8 vs. treatment 2). These results indicated that DNA methylation contributed to the cytotoxicity of L5, which could be attenuated by 0.2 mM aspirin. In addition, aspirin's regulation of L5-induced cytotoxicity was mediated by Akt (Figure 3B).

3.5 L5 and aspirin regulate the FGF2 promoter

To determine whether L5 and aspirin regulate FGF2 protein expression at the transcriptional level, we evaluated FGF2 mRNA levels in HCAECs treated with L5 in the presence or absence of aspirin. Low-concentration aspirin (0.2 mM) but not high-concentration aspirin (5 mM) attenuated STEMI L5-induced down-regulation of FGF2 mRNA levels (see Supplementary material online, Figure S4). Furthermore, we examined the regulation of FGF2 promoter activity by using a luciferase reporter gene assay (Figure 4A). In the absence of STEMI L5, constructs containing FGF2 sequences –126/+43 and –126/+179 induced 120-fold more luciferase activity than did the pGL3 vector alone, confirming the presence of a basal promoter between –126 and +43 of FGF2. Addition of STEMI L5 (50 μg/mL) reduced the luciferase activity of constructs –126/+43 and –126/+179 by 30 to 40%, respectively (P < 0.05). This reduction was reversed in the presence of aspirin (0.2 mM) but not indomethacin (a cyclooxygenase inhibitor with antiplatelet activity) (Figure 4B). Thus, in HCAECs, physiologic levels of aspirin suppressed the effects of L5 on FGF2 transcription through a cyclooxygenase-independent mechanism. Furthermore, the attenuation of L5-induced FGF2 promoter suppression by aspirin was partially prevented by Akt inhibitor (see Supplementary material online, Figure S5), suggesting the involvement of Akt in the regulation of FGF2 by aspirin and L5.

Figure 4

Transcriptional regulation of the fibroblast growth factor (FGF2) gene by aspirin and L5. (A) Diagram showing the makeup of pGL3-FGF2 reporter constructs. (B) Human coronary artery endothelial cells (HCAECs) were co-transfected with the indicated reporter construct and phRL-TK (internal control) after incubation with STEMI L5 (50 μg/mL) in the presence or absence of 0.2 mM aspirin or 0.2 mM indomethacin. Luciferase activity is expressed relative to that with pGL3-basic alone. Values are presented as the mean ± SEM and are representative of three to five independent experiments. *P < 0.05 vs. PBS. PBS, phosphate-buffered saline.

3.6 L5 and aspirin regulate the FGF2 promoter via methylation of a CpG island

Genomic DNA sequence analysis was used to identify an 1877-bp CpG island in the human FGF2 gene that starts at –532 in the 5′-flanking region of FGF2 and extends through exon 1 into the first intron. This portion of the human FGF2 gene contains the L5-responsive promoter that we show in Figure 4B can be regulated by aspirin (Figure 5A). Because CpG methylation is a key factor in FGF2 gene expression,23 we investigated whether L5 and aspirin regulate CpG methylation of the FGF2 promoter. The methylation status of 20 CpG dinucleotides in the FGF2 promoter was characterized by using bisulfite genomic DNA sequencing in HCAECs (Figure 5B). None of the 20 cytosine residues was methylated in the PBS-treated control cells. In contrast, the treatment of cells with STEMI L5 (50 μg/mL) resulted in the methylation of all 20 cytosine residues. When 0.2 mM aspirin was added to L5-treated HCAECs, the methylation of these cytosine residues was markedly reduced: only 8 of the 20 remained methylated. In addition, we examined the mRNA levels of DNA methyltransferase (DNMT)1, DNMT3A, and DNMT3B by real-time PCR and observed a modest L5-induced increase in the levels of DNMT1 and DNMT3B mRNA—an effect that could be attenuated by aspirin (see Supplementary material online, Figure S6). These data indicated that L5 represses FGF2 transcription in ECs by promoting the methylation of CpG dinucleotides in the FGF2 promoter and that this effect is significantly reduced by the addition of aspirin.

Figure 5

Analysis of CpG methylation of the fibroblast growth factor-2 (FGF2) promoter. (A) The human FGF2 promoter contains 20 CpG sites, as indicated. The primers used for bisulfite sequencing are shown. (B) The methylation status is illustrated for each of the 20 CpG dinucleotides in untreated [phosphate-buffered saline (PBS) control] human coronary artery endothelial cells (HCAECs) and in HCAECs treated with STEMI L5 (50 μg/mL) in the presence or absence of 0.2 mM aspirin. The data are representative of three independent experiments; for each experiment, HCAECs were cultured in triplicate. Identical CpG methylation patterns were observed in each experiment. ASA, aspirin.

4. Discussion

We have shown for the first time that L5 is significantly increased in the plasma of STEMI patients when compared with that of control subjects. In our STEMI patients, the mean plasma L5 level was increased to ∼12% of total LDL. The mean plasma [L5] in these patients was approximately 150 μg/mL, which far exceeds the toxic threshold of L5 (25–50 μg/mL) determined in vitro,8,12,13,16 helping to explain why severe coronary endothelial dysfunction occurs in STEMI.28 We have found that patients with non-ST elevation myocardial infarction (NSTEMI) have lower and more variable L5 levels than patients with STEMI, ranging between 0.5 and 5% of total LDL (unpublished data). In addition, we found that L5 isolated from STEMI patients can impair coronary endothelial function by inducing methylation of CpG sites in the FGF2 promoter. Importantly, we showed that low-concentration aspirin can suppress the L5-induced methylation of FGF2. Equally as important, high-concentration aspirin not only lost this protective capability but accentuated the harmful effects of L5. Our findings are summarized in Figure 6.

Figure 6

A schematic illustration of fibroblast growth factor-2 (FGF2) gene regulation by L5 and aspirin in human coronary artery endothelial cells (HCAECs). The transcription start site of the FGF2 gene is indicated as +1. Methylation of the cytosine residues of CpG dinucleotides is indicated by –CH3. Arrows indicate stimulation or direction of transportation. Lines with bars indicate inhibition. The Akt pathway has been described elsewhere.10 LOX-1, lectin-like oxidized low-density lipoprotein receptor-1.

An indicator of EC survival,29 FGF2, is a potent angiogenic factor involved in all aspects of angiogenesis (e.g. EC proliferation and migration and vascular differentiation). Previously, we showed that both oxLDL and electronegative L5 LDL down-regulated endothelial FGF2 by inhibiting Akt phosphorylation.10 In this study, the ability of an Akt inhibitor to attenuate the effects of aspirin on L5-reduced FGF2 expression further supports the important role of Akt in the L5 signalling pathway. Furthermore, involvement of the Akt and Gi signalling pathways in the regulation of the FGF2 promoter by L5 and aspirin also implies a complex interplay between cellular survival and different signalling pathways in the response to atherogenic L5 and aspirin.

Our results showed that aspirin preserves EC function by suppressing FGF2 promoter methylation by L5. As we have shown in the current study, one mechanism by which aspirin may counteract the effects of L5 is by preventing the uptake of L5 into ECs. Furthermore, we showed that the CpG-rich promoter of FGF2 is heavily methylated in the presence of STEMI L5 and that this effect is suppressed by the addition of low-concentration aspirin, suggesting that aspirin may regulate FGF2 through an epigenetic mechanism of action. Promoter DNA methylation by L5 has not been previously reported and may be not limited to the promoter of FGF2. We have similarly studied promoter methylation of the apoptosis-related LOX-1 and Bcl-2 genes and found that L5-induced cytosine methylation occurred at cytosine residues in the LOX-1 promoter but not in the Bcl-2 promoter (unpublished data). The 5′-flanking region of the FGF2 promoter is responsible for many activities of this gene, including its basal transcription.26 The FGF2 promoter is located within a CpG island, which is defined as a region of DNA larger than 500 bp that has a moving average %(G+C) >55 and an observed/expected CpG dinucleotide ratio >0.65.30 The human FGF2 gene contains multiple GC boxes (GGGCGG or CCGCCC) and a basal promoter between −126 and +24, which was vulnerable to modulation by STEMI L5 and aspirin in opposite directions in our transient transfection system. Recently, chronic aspirin use (<300 mg/day) has been associated with reduced CpG methylation of the promoter of E-cadherin (an adhesion molecule involved in tumour invasion and metastasis), providing support that aspirin is capable of epigenetic regulation.31

The protective effect of low-dose aspirin (75–325 mg/day) on the cardiovascular system has been largely attributed to aspirin's anti-thrombotic activity as an irreversible inhibitor of cyclooxygenase-1 (COX-1) in platelets.2,32 However, it has been shown that aspirin has pleiotropic effects on vascular ECs that are induced through a cyclooxygenase-independent mechanism,3,4 which may explain the complex anti-atherogenic and anti-inflammatory effects of aspirin. Thus, it is plausible that the beneficial effects of aspirin that have been seen in primary prevention trials of acute coronary events were not caused by aspirin's conventional inhibitory effects on platelet aggregation, leucocyte adhesion, or smooth muscle cell proliferation, but by a novel vasoprotective action, such as improving EC survival. However, the interplay between low-concentration aspirin and coronary atherosclerosis is complex. In some clinical trials, the ‘protective’ plasma levels of 0.1–0.2 mM aspirin, which we found to be beneficial in our in vitro experiments, failed to successfully prevent restenosis after vascular angioplasty.33,34 The discrepancy between the pro-survival effect of 0.2 mM aspirin on HCAECs observed in this study and the lack of an anti-thrombotic effect of similar therapeutic concentrations of aspirin observed in clinical trials underlies the need for complex in vitro models that mimic the clinical situation more closely.

In conclusion, we have characterized a novel mechanism of aspirin in protecting the coronary endothelium against the effects of L5. Our findings suggest that aspirin's regulation of FGF2 may be as important as its anti-platelet activity in preventing MI. Moreover, our evidence that a therapeutic concentration but not high concentrations of aspirin protect ECs from L5-induced cell death provides important groundwork for developing a targeted therapeutic approach for the prevention of plaque instability.


This work was supported by grants NSC 91-2320-B-002-185, 93-2314-B-002-125, 94-2320-B-002-121, 95-2320-B-002-116, 98-2628-B-002-088, 97-2320-B-002-057-MY3 to (P.-Y.C., Y.-T.L.,J.L.) and NSC 100-2314-B-039-040-MY3 (C.-H.C.) from the National Science Council, Taipei, Taiwan; grants NTUH 92A14, 93A02, 95S342 from the National Taiwan University Hospital, Taipei, Taiwan to P.-Y.C.; Taiwan Department of Health Clinical Trial and Research Center of Excellence, DOH102-TD-B-111-004 to C.-H.C.; and research grant 1-04-RA-13 from the American Diabetes Association to C.-H.C. The authors have no relationships with industry to declare.


The authors thank Nicole Stancel, PhD, ELS, of the Texas Heart Institute at St Luke's Episcopal Hospital, for editorial assistance in the preparation of this manuscript.

Conflict of interest: none declared.


View Abstract