| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Atherosclerosis and Lipoproteins |
From the Atherosclerosis and Metabolism Unit, Department of Cardiovascular Diseases (P.H., P.C.D., D.D.K., E.D.), Katholieke Universiteit Leuven, Belgium; CliniGenetics (M.D., E.B., K.B., N.A., N.B., G.M.), Nimes, France; and Department of Pathology and Immunology (M.L.B., G.G.), University of Geneva, Switzerland.
Correspondence to Paul Holvoet, PhD, Atherosclerosis and Metabolism Unit, Department of Cardiovascular Diseases, Katholieke Universiteit Leuven, Herestraat 49, PB 705, B-3000 Leuven, Belgium. E-mail paul.holvoet{at}med.kuleuven.be
| Abstract |
|---|
|
|
|---|
Methods and Results Gene expression in macrophages isolated from coronary plaque macrophages from hypercholesterolemic swine was measured on Agilent Human cDNA microarrays. Compared with a universal reference, 1653 transcripts were deregulated. The expression of 11 genes correlated positively and the expression of 5 genes correlated negatively with plaque oxLDL. Interferon regulatory factor-1 (IRF1; R2=0.69) and toll-like receptor 2 (TLR2; R2=0.18) were the strongest positive correlates of oxLDL. Superoxide dismutase 1 (SOD1) was the strongest inverse correlate of oxLDL (R2=0.57). Immunohistochemical analysis showed colocalization of IRF1, TLR2, and SOD1 protein in macrophages and confirmed the RNA expression data. OxLDL-induced foam cell formation in THP-1 macrophages was associated with increased expression of IRF1 and TLR2 and decreased expression of SOD1.
Conclusions Our data support the hypothesis that oxLDL is a proinflammatory stimulus that induces the expression of TLR2 and IRF1, 2 important gene regulators of innate immune response, and inhibits the expression of the antioxidant SOD1.
We searched for genes of which their expression in macrophages correlates with the complexity and oxidized LDL content of plaques in coronary arteries of hypercholesterolemic swine. Among 1653 deregulated genes, the interferon regulatory factor 1 (IRF1; R2=0.69) and toll-like receptor 2 (TLR2; R2=0.18) were the strongest positive gene correlates of both complexity and oxidized LDL content. Superoxide dismutase 1 (SOD1; R2=0.57) was the strongest inverse correlate of oxLDL. IRF1 and TLR2 protein correlated positively, and SOD1 protein correlated negatively with plaque oxLDL. OxLDL-induced THP-1 foam cell generation was associated with increased IRF1 and TLR2 and decreased SOD1 expression.
Key Words: atherosclerosis genes lipoproteins plaque
| Introduction |
|---|
|
|
|---|
Oxidized low-density lipoprotein (LDL) has been shown to play an important role in the pathogenesis of atherosclerosis, independent of LDL cholesterol.35 Several groups have demonstrated an association between established coronary heart disease (CHD) and oxidation of LDL.68 Finally, in the Health, Aging and Body Composition (Health ABC) cohort, a high CHD risk status based on Framingham score before CHD events was associated with high levels of circulating oxidized LDL (oxLDL)9 that predicted future myocardial infarction.10
Although it is generally accepted that macrophages oxidize LDL and that oxLDL affects macrophage function beyond foam cell generation, it remains to be investigated which genes in macrophages correlate with oxLDL in vivo. Once we have identified such genes, it has to be determined whether their expression could be affected directly by oxLDL.
To identify genes in plaque macrophages that correlate with oxLDL, we isolated macrophages from atherosclerotic plaques in the coronary arteries of hypercholesterolemic miniature swine by laser capture microdissection (LCM) and determined their gene expression profile with Agilent Human cDNA microarrays. We selected this model because hypercholesterolemia induces the progression of early to complex lesions in their coronary arteries, which are similar to human plaques.11 Furthermore, dietary cholesterol lowering reduced coronary plaque load and increased plaque stability,12 as has been shown in man. Once we had identified genes correlating with plaque complexity and oxLDL content, we studied the effect of oxLDL on the expression of these genes in THP-1 cells.
| Methods |
|---|
|
|
|---|
Miniature pigs were obtained by cross-breeding Göttinger and Yucatan miniature pigs (Charles River Laboratories, Cléons, France) as described previously.1113 Twelve pigs were on normal chow (control group). Twenty-eight age-matched pigs were on an atherogenic diet containing 4% cholesterol, 14% beef tallow, and 2% hog bile at a daily amount of 1 kg per day starting at the 4 months of age: 9 for 6 weeks, 10 for 12 weeks, and 9 for 24 weeks.
Blood Analysis
Peripheral venous blood from pigs was drawn from an ear vein. LDL and high-density lipoprotein (HDL) cholesterol and triglyceride levels were determined by high-performance liquid chromatography. Triglyceride levels were measured by enzymatic methods (Boehringer Mannheim). Glucose was measured with a glucometer (Menarini Diagnostics) and plasma insulin with a porcine insulin ELISA (Mercodia). Insulin resistance was calculated by a homeostasis model assessment: fasting serum insulin (mU/L)xfasting blood glucose (mmol/L)/22.5.
Plaque Load, Composition, and Complexity
Sections of the proximal left anterior descending (LAD) were stained with hematoxyline-eosine to assess lesion size. An average of 18 sections spanning a 3-mm segment of the LAD were measured and averaged. Morphometric analysis of sections was performed using the Leica Quantimet 600 image analysis system (Leica). Total lipid deposition in the lesions was determined in oil red-Ostained sections.1113 On frozen sections, macrophages were stained with anti-CD18 antibody, oxLDL with the monoclonal antibody 4E6,7
-smooth muscle (SM) actin with a mouse monoclonal IgG2a recognizing
-SM actin, and collagen with Sirius red. Interferon regulatory factor-1 (IRF1), toll-like receptor 2 (TLR2), and superoxide dismutase (SOD) were immunostained with rabbit antibodies sc-497, sc-10739, and sc-11407, respectively (Santa Cruz Biotechnology). Stained areas were measured using the Quantimet 600 image analyzer in color detection mode as described previously.1113 Coronary artery lesions were classified according their composition using the Stary classification.14
Laser Capture Microdissection
Macrophages (&1000 cells) were microdissected from between 20 and 100 sections spanning a 3-mm proximal segment of the LAD with a PixCell II LCM system using Capture HS LCM caps (Arcturus Engineering) as described in the detailed online protocol (available online at http://atvb.ahajournals.org). Macrophages were identified on the basis of histological appearance and polyploidy. The adjacent sections were used for plaque phenotyping.
Microarray Analysis of Gene Expression in Porcine Coronary Plaque Macrophages
The extraction of total RNA, the amplification of RNA, and cDNA synthesis were performed as described in the detailed online protocol. Gene expression in macrophages isolated from plaques in the first and second segment of the proximal LAD was analyzed on Agilent Human cDNA microarrays (see detailed online protocol). Quality control of the slides was performed with Datalighter software (CliniGenetics). Gene expression in macrophages was compared with that in the universal reference (UR). This contained equimolar amounts of the RNA (first and second round of amplification) extracted from 8 control organs from 6 control swine: heart, brain, lung, liver, kidney, spleen, thymus, and aorta.
Human Plaques
We obtained abdominal atherosclerotic plaques from 5 patients 74±12 years of age at 18±1.8 hours postmortem, with the approval of the ethical committee of the Geneva University Hospital, Switzerland. From 4 patients, we collected fibrolipidic and complex atheromatous plaques. From the other, we obtained only fibrolipidimic plaques.
Effect of Phorbol Myristate Acetate and OxLDL on Gene Expression in THP-1 Cells
Human LDL was isolated by density-gradient centrifugation, and copper-oxidized oxLDL was prepared as described previously.15 phorbol myristate acetate-induced differentiation and oxLDL-induced foam cell generation was performed as described previously15 (see detailed online protocol).
Real-Time Polymerase Chain Reaction Analysis
Quantitative real-time polymerase chain reaction (PCR) was performed using Sybr Green master mix according to the supplier protocols (Applied Biosystems) as described in the detailed online protocol.
Statistics
Groups were compared with KruskalWallis test (Graph Pad Prism version 4.0) followed by the Dunn Multiple Comparisons test. Spearman correlation and univariate and multivariate regression analysis was performed using the Statistical Package for the Social Sciences (SPSS for Windows; release 10.0.5). A P value of <0.05 was considered statistically significant. Genes significantly deregulated in plaque macrophages versus UR were calculated via t test (1%) after LOWESS-based normalization (see detailed online protocol). We obtained 1653 genes with a significantly different expression in macrophages. Principal component analysis (PCA) was performed on these 1653 genes (see detailed online protocol). PCA allows us to classify swine according to their significant gene expression ratios (t test 1%). Kmeans analysis was used to group the 1653 genes according to expression profiles (data centered by mean and reduced by SD). The number of clusters k was derived calculating the secondary derivative on the figure of merit (k=10; Euclidean metrics). All calculation and diagrams were performed using Datalighter software.
| Results |
|---|
|
|
|---|
|
Porcine Coronary Plaque Load and Composition
Figure 1 shows the increase of plaque load (Figure 1A) and the change of the macrophage (Figure 1B), smooth muscle cell (SMC; Figure 1C), and oxLDL (Figure 1D) area with the duration of atherogenic diet feeding. Macrophage staining overlapped with lipid and oxLDL staining. SMC staining overlapped with collagen staining. The lower SMC content, especially in the 24-week swine, in association with the higher macrophage content, resulted in a lower SMC-to-macrophage ratio: 1.9±1.9 compared with 20±14 (P<0.01).
|
Figure 2 shows representative lesions of type I, II, and III (Figure 2A). Nine hypercholesterolemic swine showed type I lesions (5 6-week, 2 12-week, and 2 24-week swine), 11 swine showed type II lesions (2, 5, and 4), and 8 had type III lesions (2, 3, and 3). The macrophage (Figure 2B) and the oxLDL (Figure 2C) area increased and correlated with the plaque complexity. Rs values were 0.83 and 0.80, respectively (P<0.0001 for each). Macrophage area correlated positively with oxLDL area (Rs=0.70; P<0.001). The SMC (Figure 2D) and collagen (Figure 2E) area was lower in type III lesions.
|
Microarray Analysis
First, we validated the heterologous hybridization between human probes present on the microarray and pig RNA samples by estimating the normalized UR human microarray coverage. Of the 10 741 individual genes present on the chip, 85% were consistently hybridized with a spot intensity >2x. Under similar hybridization conditions, 95% of the 261 positive controls present on the microarray were positively hybridized. Thus, heterologous hybridization accounted for 95% of the microarray potential. Then we successfully isolated good-quality RNA from macrophages from coronary plaques of 20 hypercholesterolemic swine (7 with type I lesions, 8 with type II, and 5 with type III lesions). We compared gene expression in macrophages with that in the UR. Of the 10 741 probe sets imprinted on the chips, 1653 transcripts were statistically deregulated (supplemental Table I, available online at http://atvb.ahajournals.org). PCA was used to rank the different swine according to their significant plaque gene expression ratio differences for these 1653 transcripts (t test 1%). The swine were ranked according to their specific transcriptomic signatures (ie, 1653 descriptors per pig reduced in 3 dimensions by PCA). Strong correlations were observed between Stary groups (I, II, and III), the plaque content (oxLDL % and macrophages %), and the genomic complexity evidenced by PCA ranking (supplemental Figure I). However, two swine were out of range in this analysis. One pig exhibited small plaques (Stary type I) with minimum lipid and macrophage deposition and a high gene expression ratio. A second pig exhibited large plaques (Stary type III) with a low gene expression ratio. They were therefore considered as outliers and were not included in the pig stratification. Subsequent expression profiling was performed with 18 swine ranked in 3 coherent and robust groups according to their genomic and phenotypic properties. Based on this stratification, the expression of the 1653 significant deregulated genes was analyzed to identify increasing and decreasing profiles. Increasing expression profiles contained genes that correlated positively with plaque complexity; decreasing profiles contained genes that correlated negatively with plaque complexity. Increasing and decreasing clusters were clearly sorted out with a Kmeans analysis for K=10 clusters. On these 10 clusters, we found 3 increasing and 3 decreasing gene expression profiles (supplemental Table I).
Correlation of Gene Expression With OxLDL Content of Porcine Coronary Plaques
Supplemental Figure IB and IC shows the Eisen diagrams for genes that correlated with oxLDL, together with the cluster to which they belong. Table 2 shows the correlation coefficients with oxLDL area. In multivariate linear regression analysis, using a model that contained all genes that correlated with oxLDL in univariate analysis, IRF1 (R2=0.69; P<0.0001) and TLR2 (R2=0.18; P=0.006) were the strongest positive correlates of oxLDL. The strongest negative correlate was SOD1 (R2=57; P<0.001).
|
RT-PCR and Immunohistochemical Analysis
Because microarray analysis identified IRF1, TLR2, and SOD1 as the strongest correlates of oxLDL, we measured their expression in extracts from type I, II, and III lesions (n=6 each) by real-time PCR analysis. Figure 3A shows that IRF1 and TLR2 RNA expression increased with plaque complexity, whereas SOD1 RNA expression decreased. Immunohistochemical analysis showed colocalization of IRF1, TLR2, and SOD1 in association with macrophages (Figure 3B). IRF1 and TLR2 protein increased with plaque complexity (Figure 3C). They correlated positively with plaque oxLDL: RS value 0.63 and 0.56 (P<0.01 for both), respectively. SOD decreased with plaque complexity (Figure 3C) and correlated negatively with plaque oxLDL (0.70; P<0.001).
|
Quantitative RT-PCR analysis of extracts from the intima of human plaques (n=9) showed a 2.0±0.44-fold higher expression of IRF1 compared with extracts of the media (n=4). That of TLR2 was 2.1±0.42-fold higher, that of TLR4 1.9±0.42-fold higher, whereas that of SOD1 was 2.2±0.47-fold lower. There was no difference between fibrolipidimic and atheromatous plaques.
Gene Expression in THP-1 Cells
Figure 4 shows that oxLDL-induced foam cell generation in THP-1 macrophages increased the expression of IRF1 with 114%, that of TLR2 with 128%, but decreased the expression of SOD1 with 35%. Although TLR4 expression did not correlate with oxLDL in coronary plaques, oxLDL-induced foam cell generation in THP-1 macrophages was associated with a 5-fold increase in TLR4 expression.
|
| Discussion |
|---|
|
|
|---|
Genes Correlating Positively With Plaque OxLDL
The IRF1 and TLR2 transcripts were the strongest positive correlates of oxLDL. Immunohistochemical analysis confirmed that IRF1 and TLR2 correlated positively with oxLDL. IRF1 is expressed at low levels in unstimulated cells but is induced by many cytokines such as interferon-
, interferon-ß, and interferon-
, tumor necrosis factor-
, and interleukin-1 and interleukin-6. It is critical to induce the onset of Th1 cell differentiation16 and evokes appropriate innate and adaptive immune responses. TLR2 is crucial for the modulation of the innate immune response in atherosclerosis.17 The strong correlation between oxLDL and TLR2 is particularly important because in human plaques, it is associated with histological markers of plaque vulnerability.18 In addition, TLR2 stimulation induced intimal hyperplasia and atherosclerosis lesion development in mice.19 In agreement with previous data,20 IRF1 and TLR2 were overexpressed in complex human plaques, further supporting the relevance of our studies. Previously, it has been shown that TLR4 can also be involved in the specific cellular immune response to oxLDL.21 Although TLR4 expression did not correlate with plaque oxLDL, it increased with plaque complexity. This is important in view of the recent finding that lack of TLR4 reduced atherosclerosis in mice deficient in apolipoprotein E resulting from a decrease of proinflammatory cytokines and monocyte chemoattractant protein 1 and thereby a reduction of the number of macrophages in the plaque.22 That TLR4 expression did not correlate with plaque oxLDL was unexpected in view of the recent data of Witztum et al showing that minimally oxLDL induced early mRNA and protein expression of cytokines that depended on TLR4.23
Other genes that correlated with plaque oxLDL and play a role in inflammation are LITAF,24 CASP1,25 and ASM3A.26 The latter produces ceramide that contributes to SMC proliferation and migration, endothelial cell differentiation, or apoptotic cell death.2729 It also aggregates LDL; their uptake involves LDL-receptor related protein that correlated with plaque oxLDL.30 Concerning apoptosis, oxLDL correlated with annexinA1, 1 of the "eat me" signals on apoptotic cells.31 We also found a strong positive correlation between plaque oxLDL and SERPINE2 or plasminogen activator inhibitor type 1 that impairs clot lysis and thereby increases cardiovascular risk.32 In addition, oxLDL correlated positively with chemokine receptor CXCR4 and SDC2 that are involved in monocyte chemotaxis.33,34 Finally, oxLDL correlated with SLC2A3 (or glucose transporterlike protein III) that is required for the increased glucose uptake that is known to be associated with the respiratory burst in phagocytic cells.35
Genes Correlating Negatively With Plaque OxLDL
Of the deregulated transcripts, SOD1 was the one that directly related to cellular antioxidant capacity of the arterial wall. It was the strongest independent negative correlate of oxLDL. Immunohistochemical analysis confirmed the strong negative correlation between SOD1 and oxLDL. The negative correlation between SOD1 expression and oxLDL is important in view of previous findings that oxLDL predicts future myocardial infarction10 and that higher expression of vascular extracellular SOD was associated with smaller infarct size.36 This inverse correlation is also important in view of its proposed role in the relationship between oxidative stress and inflammation.37
OxLDL correlated negatively with GPC3 that inhibits cell proliferation and regulates cell survival,38 CAMKK2 that regulates the tumor necrosis factor-
mediated inhibition of apoptosis,39 SLC9A6 that is found in recycling endosomes,40 and SERPINH1 that is induced by heat shock and involved in the maturation of collagen.41
IRF1, TLR2, SOD1, and TLR4 Gene Expression in THP-1 Foam Cells
The strong correlation of oxLDL with IRF1 and TLR2 in porcine coronary plaques prompted us to investigate the effect of oxLDL on IRF1 and TLR2 expression in THP-1 monocytic cells, macrophages, and foam cells. Although TLR4 did not correlate with oxLDL in coronary plaques, we studied its expression in THP-1 cells because its expression correlated with plaque complexity and belonged to the same increasing cluster as TLR2. OxLDL-induced foam cell generation was associated with increased expression of IRF1, TLR2, and TLR4. The induction of TLR4 in foam cells was in agreement with previous data of Xu et al showing increased TLR4 mRNA expression in human foam cells.42 THP-1 foam cell formation was associated with decreased expression of SOD1. A coordinated upregulation of the glutathione and thioredoxin systems in human macrophages43 may participate in the cellular defense against oxLDL. Our cell experiments show that oxLDL itself can reduce this defense by inhibiting the expression of SOD1, in agreement with the inverse relationship between oxLDL and SOD1 in coronary plaque macrophages and the inverse relationship between SOD1 expression and plaque progression. The latter finding is in agreement with the earlier finding that overexpression of SOD1 in mice lacking apolipoprotein E indeed resulted in the retardation of atherosclerosis.44
Limitation of the Study
The outcome of gene expression studies relying on microarray analysis depends on the presence of transcripts on the chips. In our studies, we could not study the expression of important oxidant enzymes like myeloperoxidase and arachidonate 12/15-lipoxygenase. The heterologous hybridization accounted at the most for 90% of the microarray potential. Thus, we certainly missed transcripts that correlate with plaque complexity because of lack of cross-hybridization. However, there were no porcine microarrays available when we started these studies. We measured gene expression in abdominal atherosclerotic plaques that were collected postmortem. We cannot exclude that aorta and coronary artery plaques may express different genes and proteins. However, our data are in agreement with previously published data, as discussed above.
Microarray analysis has been performed on plaques from swine on a rather high-cholesterol diet containing bile acids. Meanwhile, RT-PCR analysis showed a similar association between oxLDL and IRF1, TLR2, and SOD1 in coronary plaques from swine that were fed a 2% cholesterol diet in the absence of bile acid extracts. Although our data suggest that oxLDL is important for the regulation of IRF1, TLR2, and SOD1 in coronary atherosclerotic plaques, we do not exclude that other factors can mediate their expression. We are aware that one cannot establish functional relationships on gene cluster analysis alone. However, we believe that our cell experiments strengthen the credibility of the associations identified by gene expression profiling.
Conclusion
Using a hypercholesterolemic minipig model and a PCA analysis of combined phenotypic properties and expression data, we demonstrate that IRF1, TLR2, and SOD1 (inverse) are highly correlated with oxLDL. Together with our cell experiments, our data show that oxLDL-induced foam cell generation stimulates the expression of IRF1 and TLR2, 2 important regulators of innate immune response, and reduces the expression of the antioxidant and antiatherosclerotic SOD1.
| Acknowledgments |
|---|
Sources of Funding
The Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (Program G0232.05), the Interuniversity Attraction Poles Program P5/02, the Agence Nationale pour la Valorisation de la Recherche (ANVAR), Clinigenetics (2002-2004), and the Swiss National Science Foundation 32-068034.02 supported this study.
Disclosures
None.
| Footnotes |
|---|
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
B. Ky, A. Burke, S. Tsimikas, M. L. Wolfe, M. G. Tadesse, P. O. Szapary, J. L. Witztum, G. A. FitzGerald, and D. J. Rader The Influence of Pravastatin and Atorvastatin on Markers of Oxidative Stress in Hypercholesterolemic Humans J. Am. Coll. Cardiol., April 29, 2008; 51(17): 1653 - 1662. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
ATVB Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2006 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |