Local Non-Esterified Fatty Acids Correlate With Inflammation in Atheroma Plaques of Patients With Type 2 Diabetes

A total of 40 consecutive patients undergoing carotid endarterectomy (>70% carotid stenosis) at the Vascular Surgery Units of Hospital Clínico San Carlos and Fundación Jimenez Díaz were studied. The study was approved by the local ethics committees in accordance with international guidelines, and informed consent was obtained before enrollment.

Basic patient characteristics are shown in Table 1. There were no differences between patients with type 2 diabetes and nondiabetic subjects in age, sex, or prevalence of hypertension, hypercholesterolemia, or smoking. Atherosclerotic plaques obtained during surgery were immediately processed for further studies. The first four diabetic and first four nondiabetic subject samples were snap-frozen immediately and named the “test group” and the next 32 collected were embedded in paraffin and named the “validation group.” Both groups had similar clinical characteristics. Clinical data from the TOF-SIMS group are shown in Table 2. Blood samples for biochemical analysis were drawn before the surgical procedure. NEFAs were determined using the NEFA C enzymatic assay kit (WAKO, Neuss, Germany).

Samples were kept at −80°C until 10-μm slices were cut using a cryostat (CM1900; Leica) at a constant temperature of −25°C. Tissue sections were deposited onto a stainless steel plate (15-7PH; Goodfellow) and stored again at −80°C. After drying under a pressure of a few hPa for 15 min, they were directly analyzed in a TOF-SIMS V mass spectrometer (IonTof, Germany) fitted with a bismuth cluster ion source located at the Parque Cientifico de Barcelona.

The primary ions impinge the surface of the tissue section with a kinetic energy of 25 keV. The primary ion dose was between 4.7 × 10¹¹ and 10¹² ions/cm2. The secondary ions were extracted with 2 keV energy and postaccelerated to 10 keV just before hitting the detector surface (single-channel plate followed by a scintillator and a photomultiplier). A low-energy electron flood gun was activated to neutralize the surface during the analysis. The effective ion flight path is ∼2 m using a reflectron, and the mass resolution is ∼6,300 (full width at half-maximum) at a mass-to-charge ratio (m/z) of 35 and 10,000 at 795.7 m/z.

The field of view is 8,000 × 8,000 μm² (512 × 512 pixels). The name of the compounds or the m/z value of the peak centroid, the maximal number of counts in a pixel, and the total number of counts are written below each image. The color scales correspond to the interval [0, maximal number of counts in a pixel].

Mass calibration and ion peak identifications were performed as has previously been described (13). Briefly, initial calibration of monatomic hydrogen was performed, followed by sequential calibration of known molecules up to 882 Da. In this study, we focus only on previously identified quasimolecular ions (14). Total arterial wall NEFA was calculated as the mean of the different NEFA values normalized by the surface of the sample.

Frozen endartery tissue was sectioned at 10 μm and mounted on a slide covered with a polyethylene naphthalate membrane (PALMZeiss Microlaser Technologies). Tissue slices were stained with hematoxylin-eosin. RNase-free conditions were maintained as completely as possible.

TOF-SIMS–directed areas of interest were collected by laser microdissection and pressure catapulting, which was performed using a PALM MicroLaser system (PALM-Zeiss) containing a PALM MicroBeam (driven by PALM Micro-Beam software) and a PALM RoboStage. A typical setting used for laser cutting was a beam size of 30 μm and laser strength of 30 mV under a 5× ocular lens.

Dissected artery sections were catapulted directly into 25 μl RNAqueous microlysis solution, and total RNA was isolated according to the manufacturer's recommendations (RNAqueous micro RNA isolation kit, Ambion) and stored at −80°C until use. Samples from four patients were studied in each group (diabetic vs. nondiabetic) with at least three areas of interest per patient, with a total surface of 120,000 μm².

Human aortic VSMCs (CRL-1999; ATCC) were cultured in Ham's F-12 supplemented with 10% FBS, 2 mmol/l glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, ITS (2.5 μg/ml insulin from bovine pancreas, 2.5 μg/l human transferrine, and 2.5 ng/ml sodium selenite), and 30 μg/ml endothelial growth supplement at 37°C in 5% CO₂ as previously described (15). Cells between passages three and seven were used for all experiments. Certain experiments were replicated in primary cultures of rat VSMC.

Total RNA from VSMC cultures was isolated using TRIzol Reagent (Invitrogen). One microgram of RNA was reverse transcribed with a High Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA). RNA from tissue was isolated as described above. All real-time PCR reactions were performed on an ABI Prism 7500 sequence detection PCR system (Applied Biosystems) according to the manufacturer's protocol using the DeltaDelta Ct method (16). Expression levels are given as ratios to 18S. Predeveloped primer and probe assays were obtained for human 18S and MCP-1 and LPL from Applied Biosystems.

Carotid atherosclerotic plaques were stored in paraformaldehyde for 24 h and later in ethanol until paraffin embedding. Immunohistochemistry was performed as previously described (17). Primary antibodies were mouse anti-LPL antibody (Abcam), rabbit anti–MCP-1 antibody (Immugenex), anti-smooth muscle cell α-actin (1A4; Dako), or anti-human macrophages (HAM-56; Dako). Negative controls using the corresponding IgG were included to check for nonspecific staining.

Computer-assisted morphometric analysis with the Olympus semiautomatic image-analysis system Micro Image software (version 1.0 for Windows) was performed by a pathologist who was blinded to the patient's group as previously described (18). Results are expressed as the percentage of positive staining per square millimeter.

The distribution and DNA-binding activity of NF-κB in situ was detected using a digoxigenin-labeled double-stranded DNA probe with a specific NF-κB consensus sequence. Competition assays with 100-fold excess of unlabeled probe were used as negative controls. For colocalization studies, immunohistochemistry for macrophages was carried out on slides directly from the final wash of the Southwestern histochemistry protocol without allowing them to dry as previously described (19).

Electrophoretic mobility shift assay (EMSA) for NF-κB DNA binding activity was performed with nuclear protein extracts from VSMC as previously described (20). The specificity of the assay was tested with a 100-fold excess of unlabeled NF-κB consensus oligonucleotide added to the ³²P-labeled probe-binding reaction.

Equal amounts (30 μg) of cell lysate protein were loaded onto 12.5% polyacrylamide gels, eletrophoresed, and transfered to nitocellulose membranes. Then, membranes were blocked with 7% milk powder in TBS-Tween for 1 h and incubated with anti-LPL (goat polyclonal sc-32382 used at 1:200; Santa Cruz Biotechnology), anti-IκBα (rabbit polyclonal sc-371 used at 1:1,000; Santa Cruz Biotechnology) antibodies overnight at 4°C. Membranes were washed with TBS-Tween and incubated with anti-goat, anti-mouse, or anti-rabbit antibodies (1:2000) for 1 h at room temperature. The signal was detected using a chemiluminiscence kit (GE Healthcare). ELISA for MCP-1 (BD Biosciences) detection in culture supernatant was performed according to the manufacturer's protocol.

Statistical analysis was performed with SPSS for Windows software package (version 17; SPSS, Chicago, IL). Results are expressed as means ± SD. In vitro experiments were performed at least three times. Statistical testing was performed with a two-tailed α level of 0.05. Differences between diabetic and nondiabetic endarterectomy samples were assessed by the Mann-Whitney nonparametric test. Spearman correlation coefficients were calculated for continuous characteristics.

Atheroma plaque samples subjected to TOF-SIMS imaging analysis using bismuth clusters as the primary ion source render multiple secondary ions corresponding with some of the most abundant metabolites present on their surface (Fig. 1). Ionization using liquid ion guns is biased toward more hydrophobic metabolites rendering secondary ions. Therefore, we focused on surface analysis of lipids and lipid derivatives.

As shown in Fig. 1,A, a density plot of the inorganic phosphate (79 m/z) molecules can be created similarly to doing so for any conventional histological study. The presence of phosphate ion correlates with the presence of biological tissue and is comparable with the optical image in Fig. 2. Thus, the morphometric measurement of inorganic phosphate can be used to estimate the size of the specimens. All metabolite measurements are expressed as the ratio between the metabolite area and phosphate area in order to normalize size variation. The phosphate area average value was 1.835 vs. 2.54 e⁶ ± 873,000 μm² in diabetic versus nondiabetic subjects, respectively (P = not significant).

Quantization of the relative abundance of molecules can be calculated because their ionization efficiency is constant for the whole rastering. Significant differences between the molecules in diabetic and nondiabetic atheroma plaques were found in only 3 of the 16 molecules. These three molecules were linoleic acid, palmitic acid, and oleic acid.

The linoleate quasimolecular ion (m/z) was consistently increased in all diabetic samples analyzed (C18:2) (0.135 ± 0.034 vs. 0.083 ± 0.013 in diabetic and nondiabetic subjects, respectively; P < 0.05). Also, palmitic acid (C16:0) (0.211 ± 0.036 vs. 0.144 ± 0.033; P < 0.05) and oleic acid (C18:1) (0.184 ± 0.078 vs. 0.106 ± 0.017; P < 0.05) were significantly increased in plaques from the diabetic samples. As a control, synthetic triglycerides were irradiated at the ion dose used for analysis. Quantification of the in-source fragmentation rendered a negligible amount of NEFA (<10% of base peak). This effectively ruled out artifacts in NEFA measurement. The distribution pattern was similar between the three fatty acids but did not fully overlap with triglyceride distribution.

Other lipids widely regarded as major players on atherosclerotic pathogenesis, such as cholesterol, did not significantly differ in quantity between diabetic and nondiabetic samples (Table 3). Furthermore, there were no differences in triglyceride accumulation (Table 3). Dyslipidemia in diabetes is mainly characterized by increased plasma triglyceride and NEFA levels (21,22). To test whether NEFA plasma levels correlate with arterial wall NEFA, we measured enzymatically both protein-bound and protein-free plasma fatty acids. Total fasting plasma NEFAs were higher in diabetic than in nondiabetic subjects (24.37 ± 20.16 vs. 14 ± 5.73 mg/dl, respectively; P < 0.05) (Table 1). We did not find any correlation between plasma NEFA and any of the lipid molecules measured by TOF-SIMS at the arterial wall (supplemental Table 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db09-0848/DC1). The sum of NEFA present on the tissue correlated with glycemia but not with NEFA plasma levels (supplemental Table 1).

Several lipases promote the release of NEFA from complex lipids, mainly triglycerides. LPL is the major contributor to fatty acid hydrolysis in muscle and adipose tissue (23). High LPL expression by foam cells (24) plays a major role in local NEFA release in atherosclerotic vessels.

We used laser-capture microdissection in slices consecutive to those used for TOF-SIMS to quantify mRNA expression in NEFA-rich areas identified by TOF-SIMS imaging (Fig. 2,A). A consistent increase (twofold; P < 0.05) in LPL mRNA expression was found in all diabetic NEFA-rich areas studied (Fig. 2 B) in comparison with nondiabetic NEFA-rich areas and control diabetic NEFA-poor areas.

The expression of MCP-1 provides an insight into plaque-associated inflammation. MCP-1 mRNA and protein were quantitated by laser-capture microdissection qPCR (Fig. 2,D) and by immunohistochemistry (Fig. 3,A, C, and E), respectively. MCP-1 mRNA expression was higher in diabetic NEFA-rich areas (Fig. 2 D). Diabetic subjects had a significantly higher percentage of MCP-1–immunostained neointimal area when compared with nondiabetic subjects (40 ± 5 vs. 23 ± 4%, respectively; P < 0.05).

Immunohistochemical analysis of LPL expression was carried out in paraffin-embedded sections; the area immunostained for LPL was markedly higher in diabetic than in nondiabetic patients (17.1 ± 7.8 vs. 11.8 ± 4.8% of neointimal area; P < 0.05) (Fig. 3 B, D, and F). Negative controls did not stain for LPL (not shown).

In relation to nondiabetic atherosclerotic samples, the diabetic group showed a significant increase of the percentage of neointima staining positive for macrophages (22 ± 3 vs. 15 ± 3%; P < 0.05) (Fig. 4,A and C) and T-cells (4.4 ± 2 vs. 1.8 ± 0.5%; P < 0.05) (Fig. 4 B and D), which is consistent with the increased chemokine expression. Those results are similar to those found in the literature (25).

NF-κB activation was localized to NEFA-rich areas by Southwestern in frozen samples consecutive to those used for TOF-SIMS, allowing an overlay between them (Fig. 5,A and B). In addition, a higher number of nuclei staining positive for NF-κB activation with Southwestern histochemistry was noted in paraffin-embedded diabetic atherosclerotic plaques than in nondiabetic subjects (5,140 ± 512 vs. 3,738 ± 316 stained nuclei per mm²; P < 0.05) (Fig. 5 C–E). Nuclear staining was absent in negative controls (not shown).

Additionally, we tested in cultured VSMC the possible functional relation between the observed colocalization of higher NEFA concentrations and NF-κB activation and LPL and MCP-1 upregulation. For these studies we chose linoleic acid, the NEFA with a highest increment in diabetic plaques (60% increase over that in nondiabetic subjects).

Linoleic acid increased the NF-κB DNA-binding activity in nuclear extracts, as assessed by EMSA (2.2-fold at 90 μmol/l linoleic acid; P < 0.05) (Fig. 6,A and B). Confocal microscopy confirmed linoleic acid–induced NF-κB activation and nuclear translocation of p65 NF-κB (Fig. 6 C).

Linoleic acid increased LPL and MCP-1 mRNA and protein, and this effect was prevented by pretreatment with 2.5 μg/ml parthenolide, an NF-κB inhibitor (28,29) (Fig. 7 A–D) (28,29). This suggests that NF-κB mediates linoleic acid actions on VSMC.

Our group recently reported the use of TOF-SIMS to study the cartography of lipids in atherosclerotic plaques (13). We now report the presence of NEFA-rich areas within human atherosclerotic plaques. These NEFA-rich areas are especially prominent in diabetic subjects, and they colocalize with areas of inflammation characterized by NF-κB activation and increased LPL and MCP-1 expression.

The presence of NEFA-rich areas on the inner side of the endarterectomy specimens and its relation to inflammation is a novel finding, and very little is known about the mechanism involved. The extent of those NEFA-rich areas did not correlate with plasma NEFA levels, suggesting a local origin in the plaque. The distribution pattern of the three main fatty acids was similar but did not fully overlap with triglyceride distribution or VLDL deposition (S. Mas, unpublished observation). This is consistent with the hypothesis that, in addition to triglycerides, other complex lipids could act as NEFA precursors. However, we cannot exclude the possibility of a triglyceride origin of NEFA. Several lipases such as LPL (30), endothelial lipase (31) and phospholipase A2 (32,33) activate and release NEFA on atherosclerotic lesions and contribute to vascular injury. LPL is considered the rate-limiting enzyme for hydrolysis of lipoprotein triglycerides (34), and it is likely to be secreted by macrophages within atherosclerotic lesions (24,35). The increased expression of LPL in the NEFA-rich areas suggests that NEFA might be generated locally by the action of this enzyme as lipid-laden macrophages become a major source of LPL (35). NEFA presence at the intima could be considered an indirect marker of macrophage infiltration. Nonetheless, image overlay shows a similar but not identical pattern for macrophages and NEFA, suggesting a contribution of other cell types, namely VSMC.

Diabetic patients had higher plaque levels of several NEFA probed (palmitic, linoleic, and oleic acids) despite cholesterol and triglyceride levels similar to those in nondiabetic subjects. In addition, diabetic subjects had evidence for higher plaque inflammation (25). Furthermore, NF-κB activation and increased expression of MCP-1 mRNA were noted in NEFA-rich areas. This raises the question of the relationship between NEFA and inflammation that may be bidirectional. LPL may be expressed in the context of inflammation (23). In addition, linoleic acid activated NF-κB and increased the expression of LPL and MCP-1 in cultured VSMC. Recently, linoleate induction of cytokine expression by monocytes has been reported (36). The data are also consistent with the in vivo observation that an acute elevation of plasma NEFA activates the NF-κB pathway (37) and increases the hepatic expression of proinflammatory cytokines (38). NEFA may also contribute to vascular injury by inducing the synthesis of proteoglycan core proteins that promote LDL accumulation at the subendothelial layers (11). Very little is known about fatty acid signal transduction. In different experimental systems, they have been postulated to bind to toll-like receptors, particularly TLR4 (39,40), peroxisome proliferator–activated receptor γ (41,42), or fatty acid–binding receptors (43) and to activate protein kinase C θ (44,45). These pathways converge on NF-κB activation and have been linked to inflammation and insulin resistance (30,46).

In conclusion, diabetic atheroma plaques have higher NEFA content than those from nondiabetic subjects. In vivo and in vitro experiments suggest that NEFA could be a key factor in the genesis of inflammation at the plaque.

This study was supported by grants from Comunidad de Madrid (S2006/GEN-0247), the European Network (HEALTH F2-2008-200647), Ministerio de Sanidad, Instituto de Salud Carlos III, Redes RECAVA (RD06/0014/0035 and RD06/0014/0008), REDINSCOR (RD06/003/0011), Ministerio de Ciencia y Tecnología (SAF2007/63648), Eurosalud (EUS2005-0365), Fundación Mutua Madrileña, and Fundación Ramón Areces. A.O. was supported by Programa de Intensificación de la Actividad Investigadora (ISCIII/Agencia “Laín Entralgo” Comunidad de Madrid).

No potential conflicts of interest relevant to this article were reported.