Gabrielle Paulsson-Berne, Department Medicine, Karolinska Institutet, Experimental Cardiovascular Research Unit, Center for Molecular Medicine L8:03, Karolinska Hospital, 17176 Stockholm, Sweden. (fax: +46 (0)8 313147; e-mail: Gabrielle.Berne@ki.se).
Abstract. Agardh HE, Folkersen L, Ekstrand J, Marcus D, Swedenborg J, Hedin U, Gabrielsen A, Paulsson-Berne G (Department of Medicine, Experimental Cardiovascular Research, Karolinska Institutet; Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden). Expression of fatty acid–binding protein 4/aP2 is correlated with plaque instability in carotid atherosclerosis. J Intern Med 2011; 269: 200–210.
Objective. The molecular basis for atherosclerotic plaque vulnerability with high risk of plaque rupture and thromboembolism is complex. We investigated whether clinical estimates of plaque stability correlate with differentially expressed mRNA transcripts within the lesion.
Methods and results. Endarterectomy samples from patients undergoing surgery for symptomatic and asymptomatic carotid stenosis were prospectively collected and clinical parameters recorded in the Biobank of Karolinska Carotid Endarterectomies. mRNA expression profiling (n = 40) and quantitative RT-PCR (n = 105) revealed increased levels of fatty acid–binding protein 4 (FABP4/aP2) in lesions from patients with recent symptoms of plaque instability compared to asymptomatic patients (array: FC = 2, P < 0.05; RT-PCR: P < 0.05). At the mRNA level, FABP4/aP2 correlated with the cell markers CD36, CD68 and CD163 of monocyte/macrophage lineage as well as with CD4-positive T cells. FABP4/aP2 mRNA expression was also correlated with enzymes of the leukotriene pathway, 5-lipoxygenase and leukotriene A4 hydrolase. In addition, analysis of transcript profiles identified CD52 and adipophilin as the mRNAs with the highest correlation with FABP4/aP2. Expression of FABP4/aP2 by macrophages and CD52 by T cells in the lesion was confirmed by immunohistochemistry.
Conclusions. Expression of FABP4/aP2 is increased at the mRNA level in unstable carotid plaques. Immunohistochemical analyses showed localization of FABP4/aP2 to macrophage populations. These FABP4/aP2-positive macrophages constitute an important and prevalent phenotype and could provide a new link between scavenging-mediated lipid uptake and cellular metabolic stress in plaque. In addition FABP4/aP2 correlates with other important signs of inflammation and plaque instability, such as T cells and leukotriene enzymes. Taken together, these results indicate that FABP4/aP2 is a key factor connecting vascular and cellular lipid accumulation to inflammation.
Symptomatic carotid disease is characterized by embolization and symptoms of cerebral ischaemia. Lesions with increased risk of plaque rupture and atherothrombotic complications are morphologically defined by a large lipid core, a thin fibrous cap and a high degree of infiltrating inflammatory cells . The risk of a permanent stroke is highest within the first days after a transient ischaemic attack (TIA) . Randomized clinical trials have demonstrated that carotid endarterectomy (CEA) effectively reduces the risk of stroke if performed soon after the appearance of symptoms of plaque instability ; this is supported by morphological studies showing reduced inflammation in plaques from patients with more distant symptoms . Plaque ruptures may also heal as a result of proliferation and matrix deposition by smooth muscle cells together with the regression of inflammation . This natural course of lesion stabilization provides an opportunity to analyse lesions with respect to the time between symptom development and surgery.
Analyses of the lesion at the molecular level have shown that lipid- and/or lipoprotein-dependent macrophage and T-cell activation is a cause of chronic inflammation. In conjunction with this, increased vascular wall remodelling will favour fibrous cap degradation, leading to unstable or vulnerable plaques and thromboembolic complications [5, 6]. The links between lipid accumulation, immune activation and vascular remodelling are currently being intensively investigated for translation into therapeutic options.
Disturbed cellular metabolism may promote inflammatory and/or immune-specific responses within the atheroma and affect cardiovascular disease. Fatty acid–binding proteins (FABPs) are present in the cytosol. They function as lipid chaperones and modulate several lipid-signalling cascades. Experimental studies initially showed that FABP4/aP2 was expressed in adipose tissue and involved in insulin resistance . More recent studies have shown that FABP4/aP2 is important for several macrophage functions, including coordinating cholesterol trafficking, inflammatory activity and endoplasmic reticulum stress [8, 9].
Gene microarray technology can be used to investigate global mRNA expression to identify mRNA populations that exhibit differential regulation in disease processes, thus providing important clues to the underlying molecular pathology. The aim of the present work was to determine how the molecular carotid plaque phenotypes, as defined by global expression analysis, correlate with clinical estimates of plaque stability. Endarterectomy samples from patients undergoing surgery for carotid stenosis were prospectively collected, and their clinical parameters were recorded in a relational database (Biobank of Karolinska carotid endarterectomies; BiKE). By linking the global gene expression of the lesions to the clinical data, we were able to identify increased FABP4/aP2 expression in the unstable carotid atheroma. The enhanced expression of FABP4/aP2 correlates with an increase in CD36, CD68, CD163 and T-cell markers. In conclusion, our findings reveal a possible important role of FABP4/aP2 in coupling lipid accumulation to inflammation, T cells and plaque instability.
Material and methods
Biobank of Karolinska Carotid Endarterectomies includes all patients diagnosed with >70% carotid artery stenosis (according to European Carotid Surgery Trial criteria) referred to the Karolinska Hospital for surgical treatment. Patients are prospectively enrolled in the study since 2001. All participants provided informed consent, and the study was approved by the Local Ethical Committee of Northern Stockholm.
Plaques were collected in a standardized fashion, and the following clinical parameters were recorded in the database: symptoms, plaque morphology as assessed by ultrasonography, blood sample analyses including inflammatory markers and lipid profiles and risk factors such as smoking, ongoing medication and comorbidities (Table 1). The clinical database was linked to global gene expression patterns from each plaque sample using in-house software. In this study cohort, patients with symptoms of cerebral embolization (TIA, minor stroke and/or amaurosis fugax) were evaluated for stroke-preventive carotid intervention according to existing European and American guidelines [3, 4, 10]. At present, there is no available clinical practice to exclude the possibility of other sources of emboli. Patients with atrial fibrillation undergoing carotid surgery were not included in the study cohort because it could not be excluded that their emboli causing symptom was of cardial origin rather than carotid.
Table 1. Anthropometric data and laboratory measurements for all patients included in the study
Total n = 112
Asymptomatic (n = 31)
Symptoms <1 month ago (n = 44)
Symptoms more than 1 month ago (n = 37)
Age, years ± SD
67 ± 9.6
71 ± 8.8
74 ± 7.4
BMI (kg m−2)
26 ± 2.6
26 ± 3.4
25 ± 3.2
Diabetes, % (n)
Symptoms, % (n)
Laboratory values ± SD
S-Cholesterol mmol L−1
4.8 ± 1.1
4.5 ± 0.9
4.6 ± 1.2
Triglycerides mmol L−1
1.9 ± 1.4
2.1 ± 1.5
1.5 ± 0.9
HDL mmol L−1
1.1 ± 0.3
1.1 ± 0.3
1.3 ± 0.5
LDL mmol L−1
2.9 ± 0.9
2.6 ± 0.8
2.8 ± 1.0
CRP mg L−1
2.7 ± 2.5
7.5 ± 14.9
6.1 ± 9.8
Fibrinogen g L−1
3.5 ± 0.7
3.8 ± 0.9
3.9 ± 1.1
Medications % (n)
Patients underwent standard endarterectomy under sedation and local anaesthesia. Lesions were immediately rinsed in ice-cold sterile saline and cut transversely over the most prominent region of the plaque. Three-quarters of the specimen was instantly frozen on dry ice for later isolation of RNA. One-quarter, including half of the culprit lesion extending into the internal carotid artery, was mounted for later cryosection for morphological analysis.
The frozen plaques were disrupted in RLT lysis buffer (Quiagen, Sollentuna, Sweden) and phenol using a FastPrep-FP120 machine (Qbiogene, Carlsbad, CA, USA) according to the manufacturer’s instructions. Downstream purification was performed using RNeasy total RNA isolation kit (Quiagen), including a DNase treatment step. The quantity and quality of RNA were determined .
Gene array hybridization and analysis
Affymetrix HG-U133A Genechip® arrays were performed on 10 μg total RNA, according to a standard protocol (http://www.affymetrix.com). The computer data files (*.cel) were generated with Affymetrix software. Following CEL-file generation, RMA (Robust Microarray Analysis) algorithm (R, http://www.bioconductor.org) data analysis platform was chosen for calculation of the hybridization signals. Gene expression was compared between different clinical phenotypes; that is, symptomatic versus asymptomatic patients. Background noise was removed, and data were normalized in accordance with the log scale RMA analysis method for all 40 arrays . Processed gene expression data were analysed using in-house software (KIGeneConnect). t-test was used for statistical analysis, and a P-value <0.05 was considered significant.
Correlation between FABP4/aP2 mRNA levels and transcripts of cell markers
To evaluate expression relative to different cell types, we correlated the mRNA expression values of FABP4/aP2 (Affymetrix 203980_at) with all transcripts on the HG-U133A array, using GeneSpring software. The markers CD36 (209555_s_at), CD68 (203507_at; monocyte/macrophages), CD163 (216233_at; macrophages), CD3 (205456_at, T cells) and smooth muscle α-actin (200974_at; smooth muscle cells) were used to identify common cell types in the plaque. In addition we analysed the correlation between FABP4/aP2 expression and the leukotriene enzymes 5-lipoxygenase (LOX-5; 204445_s_at, 204446_s_at 214366_s_at) and leukotriene A4 hydrolase (LTA4H; 208771_s_at). We also determined the correlation between FABP4/aP2 and these common leucocyte markers using Affymetrix list of annotation numbers for the leukocyte transcripts on the HGU133A arrays (Appendix 1: Signature genes for different blood celltypes, http://media.affymetrix.com/support/technical/technotes/bloodappendix_technote.pdf, Affymetrixcom).
cDNA synthesis for TaqMan real-time RT-PCR
Quantitative Taqman RT-PCR was performed with the ABI Prism 7700 sequence detector and software (Applied Biosystems, Foster City, CA, USA). Samples were run in duplicate for 40 cycles to determine the threshold cycle (Ct). The primer/probe pairs for FABP4/aP2 analysis used ID: HS00609791_m1. Threshold cycle (Ct) values were normalized to cyclophilin A (ID: HS99999904_m1) to obtain ΔCt values. Relative gene expression values are presented as ΔΔCt values (calculated by subtracting mean ΔCt value of the reference patient group) converted to linear expression values by the formula y = 2−ΔΔCt as recommended by ABI.
Serial cryosections (10 μm) of human CEA specimens (n = 10) were incubated with primary antibodies: Rabbit anti-human FABP4/aP2 (1 : 600) (Abcam, Cambridge, UK), CD3 (1 : 300) (DAKO, Glostrup, Denmark), mouse anti-human CD52 (1 : 800), CD36 (1 : 200) (all, Abcam), von Willebrand factor (1 : 1000), CD68 (1 : 100) (DAKO) or CD163 (1 : 100) (all from GeneTex, San Antonio, TX, USA) at 4 °C overnight. Samples were detected with biotinylated secondary antibodies and developed with diaminobenzidine (Vector Laboratories, Burlingame, CA, USA).
Sections were incubated overnight with primary antibodies and blocked with 10% serum. The following secondary antibodies were used: goat anti-rabbit IgG AlexaFluor488, goat anti-rabbit TexasRed, goat anti-mouse TexasRed, goat anti-mouse AlexaFluor488, anti-rabbit biotin and TexasRed streptavidin (all Molecular Probes, Eugene, OR, USA). Autofluorescence was blocked with 0.03% Sudan Black (BDH Laboratory Supplies, Poole, Dorset, UK), and DAPI (Sigma Saint-Louis, MO, USA) was used for nuclear staining. Sections were analysed using a Leica TCS-SP5 confocal microscope. No difference in staining pattern was observed between consecutive single- or double-stained sections. Isotype controls were used for all antibodies.
Blood was drawn from healthy volunteers, and their peripheral blood mononuclear cells (PBMCs) were analysed by flow cytometry using a CyAn-ADP cytometer (Beckman Coulter, High Wycombe, UK). The following mouse anti-human antibodies were used: CD3-Cascade Yellow (DAKO), CD4-APC-Cy7 (BD, Franklin Lakes, NJ USA), CD8-PE-Dyomics590 (ExBio, Vestec, Czech Republic), CCR6-PE-Cy7 (eBioscience, San Diego, CA, USA), CCR4-FITC (R&D Systems, Minneapolis, MN, USA), CXCR3-PE (R&D Systems) and CD52-APC (BioSite, San Diego, CA, USA).
Time interval between symptoms of plaque instability and surgery correlate with expression of FABP4/aP2
We prospectively collected endarterectomy samples from patients undergoing surgery for carotid stenosis, and clinical parameters were recorded in the database BiKE (Table 1). BiKE also includes complete mRNA expression profiles of carotid plaques and tissue for histology. The time interval between last recorded symptomatic thromboembolic episode (defined as ocular/cerebral TIA or minor stroke) and surgery was selected as a clinical parameter indicating plaque instability, assuming that lesions from patients with recent symptoms are more unstable and inflammatory compared with lesions from patients with past symptoms. Although there was a difference between the three groups (patients with recent, past or no symptoms) regarding age (P = 0.01) and proprotion of female patients (P < 0.0001) (Table 1), the expression of FABP4/aP2 was not associated with age or gender.
RNA from plaques from 40 patients was analysed by global gene expression analysis using the Affymetrix HG-U133A platform. Patients were grouped according to the time of most recent symptom qualifying for CEA: (i) 1 month or less (n = 14); (ii) more than 1 month (n = 17); and (iii) no symptoms (n = 9). Gene expression profiles in plaques removed within 1 month of a qualifying symptom were compared with those removed later, as well as with plaques from asymptomatic patients. Expression data demonstrated a 2-fold increase in FABP4/aP2 (P < 0.05) mRNA levels in plaques from patients with recent symptoms (Table 2) compared to plaques from patients in the other two groups. We then confirmed this in a larger group of patients (n = 105) using real-time quantitative RT-PCR. Within this group, 42 patients had symptoms within 1 month, 34 had symptoms more than 1 month ago and 29 were asymptomatic. In this cohort, plaque FABP4/aP2 mRNA expression was significantly higher in the group of patients with recent symptoms compared to the asymptomatic group (P < 0.05) (Fig. 1).
Table 2. Upregulated genes in patients with symptoms within 1 month (n = 14) compared to patients with symptoms more than 1 month ago (n = 17) and asymptomatic patients (n = 9)
≤1 month versus >1 month
≤1 month versus asymptomatic
Array data HGU-133A (*P < 0.05).
Fatty acid–binding protein 4 (FABP4/aP2)
MHC, class II, DQ alpha
IQ motif containing GTPase-activating protein 1
WD-repeat and SOCS box-containing 1
Correlation between FABP4/aP2 mRNA and selected cell markers
To further characterize the molecular phenotype of plaques with increased FABP4/aP2 expression, markers of cell types typically found in plaques were selected from the Affymetrix GeneChip arrays (n = 50). Strong positive correlations were seen with the macrophage markers CD68 and CD163. A negative correlation was observed between FABP4/aP2 expression levels and the smooth muscle cell marker α-actin, which is an indicator of vessel wall healing. With regard to T-cell markers, CD4 showed a significant correlation (Fig. 2), whereas CD3 and CD8 did not show significant correlations with FABP4/aP2 (CD4, r = 0.48, P < 0.0005; CD3, r = 0.23, n.s.; CD8, r = 0.06, n.s.). We have previously observed upregulation of the enzymes LOX-5 and LTA4H in unstable plaques . Both transcripts coding for these enzymes were correlated with FABP4/aP2: LOX-5, 204445_s_at (r = 0.40; P =0.0037), 204446_s_at (r = 0.54; P =5.2e-05), 214366_s_at (r = 0.49; P =0.00026) and LTA4H, 208771_s_at (r = 0.50; P =0.00019).
Expanded correlation of FABP4/aP2 mRNA levels with transcript profiles
To relate the level FABP4/aP2 mRNA to other predominant transcripts in the atheroma, the correlation between FABP4/aP2 mRNA and all leucocyte cell markers was examined on the Affymetrix GeneChip array. The genes exhibiting the highest degree of positive correlation to FABP4/aP2 were the scavenger receptor CD36 and the lymphocytic marker CD52 (Fig. 2). This confirmed the finding that FABP4/aP2 increases with cellular markers of instability and high levels of inflammation. In addition, correlation analysis between all transcripts on the array and FABP4/aP2 showed that adipophilin had the greatest degree of correlation (r = 0.74; P > 0.05) closely followed by the previously identified markers CD36 and CD52.
Localization of FABP4/aP2 in plaque
The localization of the FABP4/aP2 protein within the lesion was analysed by immunohistochemistry of endarterectomized carotid plaques (n = 10). Markers for specific cell types were stained in consecutive sections with FABP4/aP2 or CD52: von Willebrand factor (endothelial cells), α-actin (smooth muscle cells), CD3 (T cells), CD36 (scavenger receptor), and CD68 and CD163 (macrophages). FABP4/aP2 exhibited staining in regions of dense inflammatory cell infiltrates. CD52-positive cells also colocalized mainly to these regions (Fig. 3). In addition, double immunofluorescence staining demonstrated that FABP4/aP2 was primarily localized in macrophages expressing CD36, CD68 or CD163 (Fig. 4c–e), as well as in a limited number of endothelial cells (Fig. 5a). Moreover, additional double staining was performed for CD52 with selected cell markers (CD3, CD68 and CD163) to understand the relationship. CD52 exclusively colocalized with CD3-positive T cells with no overlap with the other markers. CD52 and FABP4/aP2 were not observed in the same cells but were localized in neighbouring cell populations in the same regions within the plaque (Fig. 5b–c).
CD52 expression on T cells
To clarify whether CD52 expression was linked to specific T-cell populations, we performed flow cytometry with circulating lymphocytes from healthy individuals. Our analysis showed that CD52 was ubiquitously expressed on CD4- and CD8-positive cells. The expression did not differ between T-cell subset because Th1 (CXCR3+), Th2 (CCR4+) and Th17 (CCR6+) were all equally positive for CD52 (Fig. 6). In the atheroma, in contrast to peripheral lymphocytes, we only detected a correlation between CD52 and CD4 mRNA expression (r = 0.77; P < 0.0001), whereas CD3 and CD8 did not show any significant correlation indicating a more pronounced expression on CD4-positive T cells.
By using our biobank of human carotid endarterectomies (BiKE) in combination with gene array and clinical data, we have been able to identify differentially expressed gene transcripts in vulnerable plaques. Recent versus earlier or no thromboembolic symptoms were selected as a clinical variable corresponding to varying degrees of plaque stability and were furthermore correlated with global gene expression profiles. Data from expression studies are relative, depending on the comparison between transcript profiles from a target tissue and a control, and may be skewed depending on the tissue used for comparison. To avoid the inherent problems of establishing acceptable control tissue, our comparisons regarding the transcript profiles were made between plaques from patients with different clinical phenotypes. Based on this analysis, we demonstrated increased FABP4/aP2 expression in lesions from patients with recent symptoms of plaque instability. It is important to point out that lesions from patients who are asymptomatic or from those with earlier symptoms are likely to represent lesions at diverse stages. Lesions from asymptomatic individuals may either be stable, in a stage prior to instability-induced rupture and thromboembolic symptoms, or in a healing phase. From our array analysis comparing the expression in lesions from patients with differing times of symptoms of plaque instability, it was surprising to find that previously suggested markers for plaque instability, such as CD68, did not differ significantly between the patient groups. This may indicate that FABP4/aP2 is a more sensitive marker for plaque instability.
Fatty acid–binding proteins, which include at least nine proteins, mediate intracellular transport of lipids and regulate lipid-mediated cellular responses. FAPB4/aP2 was initially associated with adipocytes , but is also expressed by monocytes and macrophages . Several studies in humans have linked FABP4/aP2 to coronary artery disease (CAD) and its risk factors . For example, reduced FABP4/aP2 expression, as a result of a polymorphism in its promoter region, leads to a reduction in CAD events . In line with this, experimental studies have provided strong evidence for the importance of FABP4/aP2 in the pathogenesis of cardiovascular disease. Of particular interest is its capacity to mediate the effects of inflammatory eicosanoids . We recently reported increased expression and activity of LOX-5 and LTA4H, two key enzymes involved in leukotriene biosynthesis, in symptomatic lesions from the BiKE database . In the present study, we have shown that FABP4/aP2 correlates well with these two enzymes. FABP4/aP2 may therefore add an additional component to inflammatory lipid signalling by binding and retaining both prostaglandins and leukotrienes, as it has previously been shown that FABPs have the ability to stabilize LTA4H in vitro .
In addition, the results of a recent elegant study by Erbay et al.  emphasize the importance of FABP4/aP2 during the development of atherosclerotic lesions, highlighting a central link between expression of FABP4/aP2 and development of macrophage stress in the lesions. However, their study was performed in mouse models, so although a detailed molecular analysis was provided, this may not be analogous to the clinical setting. Our study therefore adds important data regarding the link between patients with symptomatic carotid stenosis and FABP4/aP2 expression, contributing to the complexity of inflammation and plaque instability.
We have shown that the FABP4/aP2 protein mainly colocalized to areas with inflammatory cell infiltrates. To investigate which precise cell types express FABP4/aP2, we performed double staining with specific cell markers. We used three different monocyte/macrophage markers, CD36, CD68 and CD163, to costain with FABP4/aP2. For each of these markers, we found coexpression, which is supported by the correlation at the RNA level. FABP4/aP2 is expressed by some but not all CD68- and CD163-positive cells, which may indicate a difference in macrophage phenotypes. Recently, the existence of several, complex macrophage populations has been demonstrated . Moreover, we observed a correlation between FABP4/aP2 gene expression and CD52 and CD4 gene expression on T cells, suggesting a novel link between cellular lipid accumulation and T-cell accumulation in the development of plaque instability. However, immunohistochemical analysis showed that FABP4/aP2 was not expressed on CD52-positive cells. The role of CD52 in terms of vascular disease has not been widely investigated . In our in vitro studies, we detected CD52 on all subsets of T cells. This may indicate that CD52 is a stably expressed marker on most T cells; however, at the mRNA level in plaque, CD52 is only significantly correlated with CD4 and not CD8. In the context of the present study in which FABP4/aP2 was correlated with CD52, one can only speculate that this augmented expression in the unstable plaque could in part lead to more T cells being activated, as reduced FABP4/aP2 has been shown to reduce T-cell proliferation and interferon-γ production . In addition, it has been demonstrated that lipid antigens may activate T cells in atherosclerosis. We have previously demonstrated lipid antigen activation of natural killer (NK)T cells in experimental atherosclerosis . Thus, an increase of FABP4/aP2 implies an increased load and handling of lipids and thus, a risk for activating NKT cells.
We also detected a correlation at the transcript level between FABP4/aP2 and adipophilin, which has been shown to participate in foam cell formation and increase at the levels of transcript and protein in symptomatic plaques . Macrophage infiltration and lipid core size are major risk factors for developing symptomatic atherosclerotic lesions, and the importance of understanding the different macrophage phenotypes for future therapies has been emphasized [20–22]. These arguments are underscored by the present investigation, because they indicate a further need for understanding the stimuli that cause a shift in monocytes to develop into FABP4/aP2-positive macrophages with lipid accumulation.
In conclusion, the results of this investigation demonstrate increased expression of FABP4/aP2 in symptomatic and unstable carotid lesions. At the mRNA level, we found a correlation between FABP4/aP2 and macrophage markers as well as LOX-5 and LTA4H. In addition, we observed cellular coexpression of FABP4/aP2 with CD36, CD68 and CD163. Taken together, these results provide strong indications that FABP4/aP2 is a factor connecting vascular and cellular lipid accumulation to inflammation. This suggests that increased FABP4/aP2 expression in the atherosclerotic plaque is a risk factor for unstable carotid vascular disease with atherothrombotic complications.
Conflict of interest statement
No conflict of interest was declared.
This study was supported by grants from AFA Insurance, the Swedish Heart-Lung Foundation, the Swedish Foundation for Strategic Research, the Swedish Research Council and the European Union FP6-Eicosanox project.