Palmitate-Induced Interleukin-6 Expression in Human Coronary Artery Endothelial Cells

The endotoxin content of sterile-filtered solutions of BSA and BSA-bound fatty acids (concentrations described below) was measured with the HyCult biotechnology limulus amebocyte lysate chromogenic end point assay (Cell Sciences, Canton, MA) according to the manufacturer’s instructions. Measurements were performed in duplicate.

Human coronary artery endothelial cells (CAECs) and coronary artery smooth muscle cells (CASMCs) as well as the corresponding cell culture media were obtained from Clonetics/BioWhittaker (Verviers, Belgium). The cells, derived from healthy Caucasian donors, were cultured according to the supplier’s instructions. After reaching 80–90% confluence, the cultures were incubated for 20 h with 1.25% (wt/vol) fatty acid−free BSA (Sigma-Aldrich, Taufkirchen, Germany) for controls or the indicated concentrations of fatty acids (Sigma-Aldrich) bound to BSA, as recently described (14). In brief, fatty acids (200 mmol/l in ethanol) were diluted 1:25 into Krebs Ringer HEPES buffer containing 20% (wt/vol) BSA. This mixture was gently agitated at 37°C under nitrogen overnight. The control medium containing ethanol and BSA was prepared in a similar manner. These stock solutions were stored in aliquots under nitrogen at −20°C. At the highest fatty acid concentration used (500 μmol/l), BSA reached a concentration of 1.25% (wt/vol) in the medium. Human recombinant IL-1β and tumor necrosis factor-α (TNF-α; TEBU, Offenbach, Germany) were water soluble and therefore no carrier controls were performed. Triacsin C, fumonisin B₁, C-2 ceramide, Wy14643, troglitazone, GW501516, and etomoxir were dissolved in DMSO; DMSO was then used as the carrier control. Cycloserine and cycloheximide were water soluble.

Cells were treated as indicated. Detached cells were harvested from the supernatant by centrifugation and added to the nondetached cells harvested by trypsinization. Cells were washed with PBS and double-stained with propidium iodide (1 μg/ml) and Annexin-V-Fluos (Roche Diagnostics, Mannheim, Germany), according to the manufacturer’s instructions. Stained cells were subjected to flow cytometry (FACScalibur cytometer; Becton Dickinson, Heidelberg, Germany). The percentage of necrotic, apoptotic, and vital cells was calculated by CellQuest software (Becton Dickinson).

RNA was isolated with peqGOLD TriFast according to the manufacturer’s instructions (Peqlab, Erlangen, Germany). Total RNA treated with RNase-free DNase I was transcribed into cDNA using Avian myoblastosis virus reverse transcriptase and the first strand cDNA kit from Roche Diagnostics. Quantitative PCR was performed with SYBR Green I dye on a high-speed thermal cycler with an integrated microvolume fluorometer, according to the instructions of the manufacturer (Roche Diagnostics). The primers were obtained from Invitrogen (Karlsruhe, Germany); PCR conditions are given in Table 1. Measurements were performed in triplicate. mRNA contents are given in arbitrary units (AUs), with 1 AU corresponding to 1 ag mRNA/100 ng total RNA.

Intracellular and secreted proteins (IL-6, TNF-α, and IL-1β) as well as serum IL-6 proteins were quantified with Quantikine enzyme-linked immunosorbent assays (ELISAs) from R&D Systems (Wiesbaden, Germany). To measure cytokines in cell lysates, cell monolayers were washed with PBS and scraped off in PBS supplemented with 1 mmol/l EDTA, 1 mmol/l phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 0.5 μg/ml leupeptin, and 0.7 μg/ml pepstatin. After cells were lysed by sonication, the lysates were cleared by centrifugation. To determine TNF-α and IL-1β content, lysates were measured undiluted; to determine IL-6 content, lysates were diluted 1:10 with calibrator diluent. Measurements were performed in duplicate. To achieve standardization, the cellular protein of the lysates was determined with the Bradford method. Intracellular IL-6 protein content is given in picograms per milligram of cellular protein. Cytokines in cell culture supernatants were measured after being centrifuged in 1:100 diluted (IL-6) and undiluted (TNF-α and IL-1β) samples. Results are given in nanograms (IL-6) or picograms (TNF-α and IL-1β) per milliliter. To determine IL-6 content in serum, the samples were cleared by centrifugation and measured without dilution. Serum IL-6 content is expressed in picograms per milliliter.

We recruited 54 healthy nondiabetic subjects (determined according to World Health Organization criteria) (15) from the Tübingen Family Study for type 2 diabetes, a cohort of ∼1,200 anthropometrically (sex, age, BMI, waist-to-hip ratio, and percentage of body fat [determined by bioimpedance]) and metabolically (oral glucose tolerance test, hyperinsulinemic-euglycemic clamp) characterized Caucasians. The subjects were not taking any medication known to affect glucose tolerance, insulin sensitivity, or insulin secretion. All subjects gave informed written consent before the drawing of blood samples. Subject characteristics are presented in Table 2. The study was approved by the local ethics committee.

Fasting blood parameters were determined after a 10-h overnight fast. Serum IL-6 content was measured as described above. C-reactive protein was photometrically determined with the wide range C-reactive protein assay for the high-throughput chemistry system ADVIA 1650 (Bayer Diagnostics, Tarrytown, NY). Plasma FFA concentrations were determined by an enzymatic method (NEFAC kit; Waco Chemicals, Neuss, Germany). The fatty acid composition of the plasma FFA fraction was determined as recently described in detail (16,17). In brief, total plasma lipids were extracted and separated into five fractions (phospholipids, triglycerides, diglycerides, cholesterol esters, and FFAs) by thin-layer chromatography. The FFA fraction was scraped off and esterified. The fatty acid methyl esters were analyzed by gas-liquid chromatography. Individual and grouped (saturated and unsaturated) FFAs are given as the percentage of the total FFA fraction.

In vitro data were analyzed by unpaired Student’s t test. Data from the human cohort were log transformed (ln) to approximate a normal distribution. These data were analyzed by linear regression analysis and ANOVA. Adjusted values were derived from a multivariate linear regression model. These transformations and tests were performed with the statistical software package JMP 4.0 (SAS Institute, Cary, NC). P < 0.05 was considered statistically significant. Data are given as means ± SE.

The solution of 1.25% (wt/vol) fatty acid−free BSA (control condition) contained 2.0 ng/ml endotoxin. The 500-μmol/l palmitate solution (in 1.25% BSA) revealed an endotoxin content of 2.8 ng/ml. These very low endotoxin concentrations are, according to earlier studies in human umbilical vein endothelial cells (18) and human CAECs (19), not expected to induce significant amounts of IL-6. The 500-μmol/l linoleate solution (in 1.25% BSA) exhibited an endotoxin content of 19.7 ng/ml. For this still low concentration, however, we could not exclude minor endotoxin effects on IL-6 production.

The expression of IL-6, IL-6R, and the co-receptor protein gp130 was studied in human CAECs and CASMCs after they were treated with BSA (control), 500 μmol/l linoleate, or 500 μmol/l palmitate for 20 h. Microscopically, no significant cytotoxic FFA effects could be detected under these conditions. As assessed by flow cytometry, the viable cell fractions (≥80%), the apoptotic and necrotic cell fractions (∼ 4% each), as well as the fractions containing undistinguishable late apoptotic/necrotic cells (∼ 10%) of BSA- and palmitate-treated cultures were not significantly different. As presented in Fig. 1, neither cell type expressed remarkable amounts of IL-6R mRNA (≤6 fg/μg total RNA). However, gp130 mRNA was expressed in both cell types and its level was at least twice as high in CASMCs as in CAECs. IL-6 mRNA was also readily detectable in both cell types. Although the IL-6R components, IL-6R and gp130, were not affected by any FFA in both cell types, the saturated FFA palmitate induced an 18-fold increase of IL-6 mRNA expression in CAECs (n = 3, P = 0.0003) (Fig. 1A) and a 7.5-fold increase in CASMCs (n = 3, P = 0.0089) (Fig. 1B). The absolute IL-6 mRNA levels reached under these conditions in CAECs were about fourfold higher than those seen in CASMCs. The unsaturated FFA linoleate did not significantly elevate IL-6 mRNA amounts in both cell types (Fig. 1).

As depicted in Fig. 2, the palmitate-induced increase in IL-6 mRNA expression was concentration dependent in both cell types. Significantly elevated IL-6 mRNA levels, as compared with control conditions, were reached with palmitate concentrations as low as 250 μmol/l (CAECs: n = 4, P = 0.0357; CASMCs: n = 4, P = 0.0014). Further, we compared the IL-6 mRNA levels reached with 500 μmol/l palmitate with those induced by commonly used concentrations of IL-1β (2 ng/ml) and TNF-α (100 ng/ml), two well-known stimuli of IL-6 expression. Under these conditions, palmitate turned out to be 3- and 10-fold more potent than IL-1β (n = 3, P = 0.0021) and TNF-α (n = 3, P = 0.0002), respectively, in CAECs (Fig. 3A). In CASMCs, palmitate was two- and threefold more effective than IL-1β (n = 4, NS) and TNF-α (n = 4, P = 0.0171), respectively (Fig. 3B). To further assess the fatty acid specificity of IL-6 gene induction, we treated CAECs and CASMCs with other saturated and unsaturated long-chain FFAs, such as stearate (C18:0), palmitoleate (C16:1 ω7), oleate (C18:1 ω9), and the poorly metabolizable palmitate analog 2-bromo-palmitate, at 500 μmol/l each. At this concentration, stearate turned out to be toxic to both cell types and, therefore, was not evaluated. In CAECs, palmitoleate (n = 4, P = 0.0163) and oleate (n = 4, P = 0.0339) significantly increased IL-6 mRNA expression five- and eightfold, respectively, but were far less potent than palmitate (Fig. 4A). 2-Bromo-palmitate showed no inductive effect at all (Fig. 4A). All FFAs except linoleate significantly raised the IL-6 mRNA content of CASMCs (n = 4, P < 0.01 each) (Fig. 4B). It is worth noting, however, that the induction factors seen in CASMCs (two- to threefold) were markedly lower than those observed in CAECs, as was already shown for palmitate (Fig. 1).

As expected from the mRNA data, palmitate (500 μmol/l) also promoted IL-6 protein expression (Fig. 5). Under control conditions, no IL-6 protein was detectable. The protein concentrations reached after palmitate treatment in CAECs were about twofold higher than those seen in CASMCs (Fig. 5A and B). Furthermore, the palmitate-induced IL-6 levels were comparable with those stimulated by IL-1β in both cell types (Fig. 5). Thus, the significant difference in IL-6 mRNA contents between palmitate- and IL-1β−treated CAECs (Fig. 3A) was not observed at the protein level. This result probably reflected differences in the kinetics of IL-6 induction, possibly caused by differences in the underlying molecular mechanisms triggered by these two stimuli. Consistent with the mRNA data, TNF-α was a less potent inducer of IL-6 protein expression in both cell types (Fig. 5). The unsaturated FFA linoleate (500 μmol/l) did not induce IL-6 protein expression in CAECs (Fig. 5A). It was surprising that linoleate provoked IL-6 protein expression to some extent in CASMCs (Fig. 5B). This induction probably reflected the minor nonsignificant increase in IL-6 mRNA induced by linoleate in these cells (Fig. 1B). Furthermore, we tried to determine secreted IL-6 protein in the conditioned media of cells treated with BSA, FFAs (palmitate and linoleate), and cytokines (TNF-α and IL-1β). However, for still unresolved technical reasons, we were not able to quantify IL-6 protein in the media containing BSA or BSA-bound FFAs. TNF-α (100 ng/ml) and IL-1β (2 ng/ml), however, provoked high levels of secreted IL-6 protein in culture supernatants of CAECs (1.07 ± 0.16 and 12.27 ± 3.37 ng · ml⁻¹ · 20 h⁻¹, respectively) and CASMCs (6.77 ± 4.41 and 9.46 ± 4.21 ng · ml⁻¹ · 20 h⁻¹, respectively). In addition, we examined whether palmitate and linoleate (500 μmol/l each) stimulate TNF-α and IL-1β protein expression or release. In both cell types, neither intracellular nor secreted TNF-α protein was detectable after BSA or FFA treatment. Moreover, CAECs did not produce measurable amounts of intracellular or secreted IL-1β protein. By contrast, CASMCs expressed relevant levels of intracellular IL-1β protein (not shown) that was also secreted into the culture medium (73.5 ± 10.8 [control] vs. 130.2 ± 14.9 [palmitate] vs. 85.8 ± 11.0 [linoleate] pg · ml⁻¹ · 20 h⁻¹). Palmitate, but not linoleate, induced a significant twofold rise in secreted IL-1β protein (n = 4, P = 0.0216).

Regarding the molecular mechanism underlying palmitate-induced IL-6 mRNA expression, we examined the involvement of mitochondrial β-oxidation, ceramide formation, and activation of peroxisome proliferator−activated receptors (PPARs), a class of FFA-activated nuclear receptors and transcription factors. We did so by treating the cells with the carnitine palmitoyl transferase 1 inhibitor etomoxir, inhibitors of the ceramide synthesis pathway (triacsin C, cycloserine, and fumonisin B₁), C-2 ceramide, and finally isoform-specific PPAR agonists (Wy14643, troglitazone, and GW501516). None of these treatments, however, revealed an important role for any of these pathways (data not shown). Because we demonstrated that palmitate induces IL-1β production in CASMCs, we also tested whether IL-6 expression depends on IL-1β production. To examine this, we incubated the cells with the protein synthesis inhibitor cycloheximide (20 μg/ml) before and during palmitate (500 μmol/l) treatment. Cycloheximide significantly reduced the amount of IL-1β protein secreted into the medium (n = 4, P = 0.0273) (Fig. 6A) but did not reduce IL-6 mRNA expression (Fig. 6B), thereby excluding the possibility that IL-1β production mediates IL-6 gene expression. By contrast, cycloheximide significantly increased the IL-6 mRNA content of palmitate- (n = 4, P = 0.0014) as well as BSA- (n = 4, P < 0.0001) treated CASMCs (Fig. 6B). Similar effects of cycloheximide were also found in CAECs (data not shown).

To investigate the relation of individual FFAs and IL-6 in vivo in humans, fatty acid patterns of the fasting plasma FFA fraction, fasting plasma FFA concentrations, and fasting serum IL-6 concentrations were determined in 54 healthy participants of the Tübingen Family Study for type 2 diabetes. In these subjects, no indications for acute inflammation were seen (leukocytes: 5,737 ± 262 μl⁻¹; C-reactive protein: 0.20 ± 0.03 mg/dl). Of the total FFA concentration, 44% were saturated and 56% were unsaturated (39% monounsaturated, 17% polyunsaturated) (Table 2). Among the saturated FFAs, palmitate and stearate prevailed (64 and 28% of all saturated FFAs, respectively). Oleate and linoleate were the most frequent unsaturated FFAs (56 and 23% of all unsaturated FFAs, respectively). All other fatty acids were <5% of the total FFA fraction (Table 2). Although the circulating FFA concentration was not associated with serum IL-6 (Fig. 7A), there was a significant, positive correlation between serum IL-6 and the palmitate content of the FFA fraction (Fig. 7B). This correlation persisted when IL-6 levels were adjusted for the percentage of body fat, an important determinant of circulating IL-6 concentrations (Fig. 7C). IL-6 did not correlate with stearate (unadjusted: r = 0.03, P = 0.8; adjusted: r = 0.10, P = 0.5), oleate (unadjusted: r = −0.09, P = 0.5; adjusted: r = −0.14, P = 0.3), or linoleate (unadjusted: r = 0.03, P = 0.8; adjusted: r = 0.01, P = 1.0). To assess whether the association between serum IL-6 levels and the palmitate content of the FFA fraction was independent of the total FFA concentration, we adjusted the IL-6 levels for circulating FFA levels. As depicted in Fig. 7D, this association remained significant. Again, IL-6 was not correlated with stearate (r = 0.03, P = 0.8), oleate (r = −0.08, P = 0.6), or linoleate (r = 0.05, P = 0.7) after adjusting for total FFAs. Furthermore, none of the minor fatty acids (<5% of total FFAs) (Table 2) were associated with serum IL-6 concentrations. Finally, no correlation was seen between IL-6 and the whole saturated or unsaturated FFA subfraction, respectively.

Recent clinical data support a role for FFAs in vascular inflammation, endothelial dysfunction, and atherosclerosis (20–22). The molecular mechanisms underlying these associations are, however, far from clear. Whereas mono- and polyunsaturated FFAs, such as oleate (C18:1 ω9), linoleate (C18:2 ω6), arachidonate (C20:4 ω6), eicosapentaenoate (C20:5 ω3), and docosahexaenoate (C22:6 ω3), have been convincingly demonstrated to modulate endothelial functions (23–28), very little is known about the role of saturated FFAs in vascular complications.

Here we provide evidence that the saturated FFA palmitate (C16:0), one of the most prominent FFAs in human blood (∼28% of total FFAs) (this study; 16), induces substantial amounts of the inflammatory cytokine IL-6 in CAECs and minor amounts in CASMCs. By contrast, linoleate (C18:2 ω6), a polyunsaturated FFA with controversial effects on endothelial activation (25,26), did not promote IL-6 expression in CAECs and had only a minor, possibly endotoxin-mediated, IL-6−inducing effect in CASMCs. As to the FFA specificity of this cellular response, palmitate was by far the most potent IL-6−inducing FFA in CAECs. In CASMCs, palmitate as well as the unsaturated FFAs oleate and palmitoleate induced similar levels of IL-6 mRNA expression. The absolute amounts of IL-6 induced by any of these FFAs in CASMCs was, however, markedly lower than those produced by palmitate in CAECs. Therefore, we consider the CAECs to be the main source of palmitate-provoked IL-6 production in the vasculature. Overall, our results allow us to add palmitate, as a novel potent inducer of IL-6 expression, to the list of known stimuli of IL-6 expression, such as IL-1α, IL-1β, TNF-α, and lipopolysaccharides (4,29). Whether palmitate is able to induce IL-6 in other known IL-6−producing cells, such as monocytes, Kupffer cells, and adipocytes, remains to be elucidated.

Concerning the molecular mechanism of palmitate-induced IL-6 production, we currently can exclude only the involvement of mitochondrial β-oxidation, ceramide formation, PPAR activation, and IL-1β production. One surprising finding was that inhibition of protein biosynthesis markedly enhanced IL-6 mRNA expression not only in palmitate-treated cells, but also in control cells. This observation opens the possibility that IL-6 induction represents de-repression of the IL-6 gene by enhanced degradation of an actively synthesized repressor protein. Clearly, further studies are needed to clarify this issue.

Furthermore, we found that CAECs as well as CASMCs do not express appreciable amounts of IL-6R. This finding argues against a local auto- or paracrine role of this inflammatory cytokine on vasoreactivity and atherosclerosis. Therefore, in our opinion, two other probably synergistic functions of vasculature-derived IL-6 (rev. in 8) seem to be more conceivable: 1) endocrine activation of the hepatic acute-phase response provoking the release of strong risk factors for cardiovascular disease, such as C-reactive protein (30,31) and fibrinogen (32,33); and 2) local paracrine activation of monocytes and platelets favoring fibrinogen deposition and thrombus formation in the vessel wall. In the case of an endocrine function, it could be postulated that palmitate-induced vascular IL-6 secretion influences plasma IL-6 concentrations, but this remains to be shown. However, with regard to the immense surface of the endothelium (∼350 m²) and the ubiquity of palmitate in the circulation, we assume that vascular IL-6 production in response to palmitate, as quantified herein on a cellular basis, is substantial and should indeed contribute to serum IL-6 levels.

In our analysis of the plasma FFA profile, we found, in accordance with earlier findings (16), that palmitate (C16:0), stearate (C18:0), oleate (C18:1 ω9), and linoleate (C18:2 ω6) are by far the most prominent FFA species in human plasma and, therefore, could be of systemic importance. Among these and other minor FFAs, only palmitate correlated with IL-6 levels in healthy human subjects. Moreover, this association was independent of the total FFA concentration and body fat, a well-known source of palmitate and IL-6. This observation 1) supports our in vitro finding that palmitate is an important adipose tissue−derived determinant of IL-6 production and 2) strengthens our idea that, in addition to adipose tissue, other tissues (with the vasculature being a prime candidate) also significantly contribute to plasma IL-6 concentrations. Therefore, it is tempting to speculate that in obesity-linked disorders, hypertrophic adipose tissue contributes essentially to systemic IL-6 concentrations in two ways: directly via IL-6 secretion by adipocytes and indirectly via release of palmitate, which, in turn, elevates vascular IL-6 production. Such a prominent role for adipose tissue in chronic vascular inflammation would be in keeping with many clinical findings (rev. in 2,34,35).

In summary, we have shown that the plasma FFA palmitate, in physiological concentrations, is able to induce considerable amounts of IL-6 mRNA and protein in endothelial cells of human coronary arteries. Furthermore, palmitate, but no other plasma FFAs, correlated in vivo with IL-6 concentrations independently of body fat mass. Therefore, we postulate that 1) palmitate is an important determinant of IL-6 production in vitro as well as in vivo and 2) elevated plasma concentrations of palmitate (e.g., as found in the obese state) contribute to vascular IL-6 secretion and chronic inflammation.

These studies were supported by a grant from the Deutsche Forschungsgemeinschaft (KFO 114/1-1).

We thank Claudia Péterfi, Alke Guirguis, Melanie Weisser, Joanne Mushack, and Elisabeth Metzinger for excellent technical assistance.