Long-chain acyl-CoA synthetase-1 mediates the palmitic acid-induced inflammatory response in human aortic endothelial cells

Abstract

Saturated fatty acid (SFA) induces proinflammatory response through a Toll-like receptor (TLR)-mediated mechanism, which is associated with cardiometabolic diseases such as obesity, insulin resistance, and endothelial dysfunction. Consistent with this notion, TLR2 or TLR4 knockout mice are protected from obesity-induced proinflammatory response and endothelial dysfunction. Although SFA causes endothelial dysfunction through TLR-mediated signaling pathways, the mechanisms underlying SFA-stimulated inflammatory response are not completely understood. To understand the proinflammatory response in vascular endothelial cells in high-lipid conditions, we compared the proinflammatory responses stimulated by palmitic acid (PA) and other canonical TLR agonists [lipopolysaccharide (LPS), Pam3-Cys-Ser-Lys4 (Pam₃CSK₄), or macrophage-activating lipopeptide-2)] in human aortic endothelial cells. The expression profiles of E-selectin and the signal transduction pathways stimulated by PA were distinct from those stimulated by canonical TLR agonists. Inhibition of long-chain acyl-CoA synthetases (ACSL) by a pharmacological inhibitor or knockdown of ACSL1 blunted the PA-stimulated, but not the LPS- or Pam₃CSK₄-stimulated proinflammatory responses. Furthermore, triacsin C restored the insulin-stimulated vasodilation, which was impaired by PA. From the results, we concluded that PA stimulates the proinflammatory response in the vascular endothelium through an ACSL1-mediated mechanism, which is distinct from LPS- or Pam₃CSK₄-stimulated responses. The results suggest that endothelial dysfunction caused by PA may require to undergo intracellular metabolism. This expands the understanding of the mechanisms by which TLRs mediate inflammatory responses in endothelial dysfunction and cardiovascular disease.

INTRODUCTION

Obesity and dyslipidemia are risk factors for cardiovascular disease and are associated with chronic and low-grade inflammatory status (33). Elevated saturated fatty acid (SFA) levels caused by metabolic disorders affect endothelial function, glucose clearance, and cardiac function, possibly due to chronic inflammatory responses (34). The vascular endothelium is the first line facing circulating fatty acids in the serum, stimulating leukocyte adhesion and inflammatory response. Circulating leukocytes adhere to the endothelial layer and establish binding before infiltrating to tissues, which is the key event for an inflammatory response in atherosclerosis and other cardiometabolic tissues. Thus, SFA-induced inflammation may contribute to both metabolic and coronary heart diseases through vascular inflammation and endothelial dysfunction (26, 28). We and others (26, 28) have shown that SFA-induced impairment of insulin signaling in the endothelium is associated with insulin resistance and endothelial dysfunction through a Toll-like receptor (TLR)-dependent mechanism.

TLRs play important roles in innate immunity and are involved in various pathological conditions, including insulin resistance, endothelial dysfunction, diabetes, obesity, cardiomyopathy, atherosclerosis, peripheral neuropathy, and nephropathy (9, 52, 56, 58). Deletion of TLR2 or TLR4 prevents high-fat diet (HFD)-induced insulin resistance and endothelial dysfunction (26, 28, 54). Moreover, SFA stimulates dimerization of TLRs, and its proinflammatory action is through a myeloid differentiation factor 88 (MyD88)-dependent mechanism in macrophages (37). This evidence suggests that SFA may directly interact with TLRs. By contrast, Lancaster et al. (35) showed that TLR4 is not a receptor for SFA but is required for SFA-stimulated inflammatory response. SFA does not form a dimer or endocytosis of TLR4 in bone marrow-derived macrophages (BMDM) (35). Furthermore, it has been proposed that SFA may stimulate TLR2 and TLR4 via endogenous ligands, including oxidized low-density lipoprotein (ox-LDL), high-mobility group protein B1 (HMG-B1), and fetuin-A through an indirect interaction with TLRs (12, 20, 48). SFA is a nutrient that is transported to the intracellular compartment and then modified and metabolized by various enzymes. Most of these intracellularly transported fatty acids are ligated to co-enzyme A (CoA) by various acyl-CoA transferases according to the length of the acyl chains. Fatty acyl-CoAs then serve as substrates in the modification of lipids and proteins, production of sphingolipids, and β-oxidation in mitochondria (15, 16, 40). These metabolic routes of SFA may affect cellular physiology and inflammatory status. A molecular simulation revealed that palmitic acid (PA), a saturated long-chain fatty acid, is not a TLR4 agonist (35). Indeed, the proinflammatory response stimulated by PA is distinct from that by lipopolysaccharide (LPS) in several ways. First, the proinflammatory response occurs across several hours by PA compared with within a few minutes by LPS or Pam3-Cys-Ser-Lys4 (Pam₃CSK₄; a synthetic TLR2 ligand). Second, a few hundred micromolar PA is required to evoke an inflammatory response compared with the nanomolar to low micromolar range of LPS and Pam₃CSK₄ in a human monocyte cell line (24). Third, PA with LPS synergistically stimulates the production of ceramide, contributing to obesity-induced insulin resistance and endothelial dysfunction (64). These notions suggest that the mechanisms underlying PA-stimulated inflammatory responses may be distinct from the canonical TLR signaling pathways. In the present study, we compared PA-induced inflammatory response to canonical TLR ligand-stimulated inflammatory response in endothelial cells, demonstrating that the intracellular metabolism of PA is involved in vascular inflammation, leukocyte adhesion, and endothelial dysfunction.

MATERIALS AND METHODS

Materials.

Anti-phosophorylated (p-)NF-κB was obtained from Cell Signaling (Beverly, MA), and anti-p-JNK was obtained from Thermo Fisher Scientific (Carlsbad, CA). Anti-p65 NF-κB and anti-JNK antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). DsiRNA for human siRNA for ACSL1, nontargeted scrambled siRNA, and the primers for the RT-PCR were purchased from Integrated DNA Technologies (Coralville, IA). Triacsin C was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). LPS, Pam₃CSK₄, and macrophage-activating lipopeptide-2 (MALP2) were obtained from Invivogen (San Diego, CA). Bovine serum albumin (BSA; cat. no. 30-AB79) was obtained from Fitzgerald Industries International (Acton, MA).

Cell culture and transfection.

Human aortic endothelial cells (HAECs) were obtained from Lonza (Walkersville, MD) and were grown in F-12K medium containing EGM-2 single-quote supplements (Lonza). All experiments were conducted before their sixth passage. HAECs were serum starved for 2 h and then treated with BSA, BSA-conjugated PA, LPS, Pam₃CSK₄, or MALP2 at indicated concentrations and duration found in the figure legends. HAECs were transiently transfected with 100 nM siRNA duplex oligonucleotides using Lipofectamine 2000 (Life Technologies, Carlsbad, CA) according to the manufacturer’s instructions. Two days after transfection, cells were serum starved for 2 h and then treated with BSA or PA as indicated in figure legends.

Preparation of PA.

BSA (10.5%) was dissolved in 25 mM HEPES-DMEM and syringe filtered (0.22 μM, Millipore). Sodium palmitic acid (100 mM) was heated until dissolved in endotoxin-free water and rapidly added to warmed BSA solution (6:1 molar ratio). Then, this BSA-conjugated PA was added to reach the indicated concentration of PA. We used endotoxin-free reagents, and we tested all the reagents we used, including BSA, PA, media, and reagent diluents.

Animals and functional assessment for isolated mesentery arterioles.

All animal procedures were performed in accordance with the Animal Use and Care Committee at The University of Alabama at Birmingham. C57BL/6J mice were purchased from Jackson Laboratory (Bar Harbor, ME). All animals were maintained in a temperature-controlled facility with a 12-h:12-h light-dark cycle. Mice were fed normal chow diet ad libitum. Mice were anesthetized with an intraperitoneal injection of pentobarbital sodium (50 mg/kg). After the mesenteric vascular bed was removed, the mouse was euthanized by exsanguination and removal of the heart. Mesenteric arterioles were excised from the vascular bed and placed in a cooled (4°C) chamber containing dissection buffer [145 mM NaCl, 4.7 mM KCl, 2 mM CaCl, 1.2 mM MgSO₄, 1 mM NaH₂PO₄, 5 mM glucose, 3 mM 3-(N-morpholino)propanesulfonic acid (MOPS) buffer, 2 mM pyruvate, 0.02 mM EDTA, and 1% BSA, pH 7.4]. The isolated arterioles were then cannulated with glass micropipettes with 10-0 monofilament sutures and mounted in a custom-designed tissue chamber (Living System Instrumentation, Burlington, VT). The arterioles were pressurized to 45 mmHg intraluminally with the same buffer without flow and superfused with buffer without albumin. The vessel preparations were positioned on the stage of an inverted microscope. The vessel segments were gradually warmed to 37°C during a 30-min equilibration period. Next, the vessels were intraluminally perfused with triacsin C (1 μM) 30 min before BSA or PA (200 mM, 4 h). The intraluminal pressure was kept at 45 mmHg. After baseline diameter was established, arterioles were exposed to phenylephrine (1 µmol/L) until the maximal contraction was achieved (5 min). The vessels were subsequently stimulated with insulin or sodium nitroprusside (10⁻¹¹–10⁻⁵ mol/L, 3 min per concentration). Vasodilation was observed and recorded. Measurement of vessel diameter (in µm) was performed with an electronic video caliper (Living Systems Instrumentation, St. Albans, Vermont), and recorded using a software, Laboratory Chart (ADInstruments, Inc., Colorado, Springs, CO).

The data are expressed as means ± SE. Insulin-induced vasodilation was calculated as percent relaxation from the phenylephrine constriction according to the following equation:

$$\text{Relaxation \%} = \frac{\left({\text{ID}}^{\text{treat}}-{\text{ID}}^{\text{PE}}\right)\times 100}{\left({\text{ID}}^{\text{w/oCa}++}-{\text{ID}}^{\text{PE}}\right)}$$

where ID^treat is the diameter obtained when the vessel was treated with insulin, ID^PE is the diameter obtained when the vessel was constricted with phenylephrine, and ID^w/oCa++ is the maximal passive diameter observed when the vessel was fully dilated in buffer containing 2 mM EGTA and 100 μM adenosine without Ca⁺⁺.

Preparation of cell lysates and immunoblotting.

Before lysis, cells were briefly washed with ice-cold PBS. Cells were then scraped in lysis buffer containing 50 mM Tris (pH 7.2), 125 mM NaCl, 1% Triton X-100, 0.5% NP-40, 1 mM EDTA, 1 mM Na₃VO₄, 20 mM NaF, 1 mM Na pyrophosphate, and complete protease inhibitor cocktail (Roche Applied Science). Cell debris was pelleted by centrifugation of samples at 17,000 g for 10 min at 4°C. Supernatants were boiled with Laemmli sample buffer for 5 min, and proteins were resolved by 10% SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted with specific antibodies, as described in figures, using standard methods. Immunoblots were visualized and quantified with a UVP Biospectrum Imaging System and an image analysis software (Vision Works LS, UVP, Inc.).

Gene expression analysis by reverse transcription-polymerase chain reaction.

The cells were treated as described in the figure legends. Total RNA was prepared by using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. One microgram of total RNA was used for cDNA synthesis by using Omniscript RT Kit (Qiagen, Valencia, CA). Then, the cDNA was subjected to semiquantitative PCR analysis by using the Hot Star Taq Master Mix kit (Qiagen, Valencia, CA). The sequences of the primers were the following: E-selectin: forward (F)-GGTTTGGTGAGGTGGCT, reverse (R)-TGATCTTTCCCGGAACTG; IL-1β F-GGGCTCAAGGAAAAGAATC, R-TTCTGCTTGAGAGGTGCTGA; cyclooxygenase 2 (COX-2) F-CAAATCCTTGCTGTTCCCACCCAT, R-GTGCACTGTGTTTGGAGTGGGTTT; acyl-CoA synthetase-1 (ACSL1) F-TCAAGGGTGCTTCAACCCACTGTCT, R-GCTGTTGTTTCTGATGATGCCGCT. PCR product was visualized with a fluorescent dye (Envirosafe DNA/RNA Stain, Helixx Technologies, Inc., ON, Canada), and images were analyzed and quantified with a UVP Biospectrum Imaging System and image analysis software (Vision Works LS, UVP, Inc.).

Statistical analysis.

Values are presented as means ± SE. Statistical significance was assessed by two-tailed Student’s t tests, one-way ANOVA followed by Newman-Keuls post hoc tests, or two-way ANOVA followed by Bonferroni post hoc tests. Statistical analyses for vasodilation were assessed by repeated measurements of two-way ANOVA followed by Turkey’s post hoc tests. Differences were considered statistically significant if the P value was less than 0.05. Statistical analyses were performed using GraphPad Prism v. 8 (GraphPad, Inc., San Diego, CA).

RESULTS

Profiles of inflammatory response stimulated by PA are distinct from those by other canonical TLR2 and TLR4 ligands.

To examine proinflammatory responses in endothelial cells, HAECs were incubated with various TLR ligands for different time points. The expression of E-selectin (gene name: SELE) by canonical TLR agonists, such as LPS (TLR4 ligand), Pam₃CSK₄ (TLR2/TLR1 ligand), and MALP2 (TLR2/TLR6 ligand) plateaued within 2 h. In contrast, E-selectin expression stimulated by PA increased more slowly for up to 6 h (Fig. 1, A and B). Next, we examined the downstream signaling pathways of TLR2 and TLR4 (26, 29, 43, 46). Cell lysates of TLR ligand-treated HAECs were subjected to Western blots for proinflammatory signaling molecules, such as phosphorylations of JNK and NF-κB. Phosphorylation of JNK by PA peaked at 1 h, whereas phosphorylation by LPS, Pam₃CSK₄, and MALP2 peaked at 2 h, 30 min, and 1 h, respectively (Fig. 2). Phosphorylation of NF-κB by PA increased constantly for 6 h, whereas phosphorylation by LPS, Pam₃CSK₄, and MALP2 peaked at 2 h, 20 min, and 30 min, respectively. The results suggest that the mechanism by which PA stimulates proinflammatory response may be distinct from those employed by other TLR ligands.

Fig. 1.

Expression profiles of E-selectin by palmitic acid (PA) are distinct from those by other canonical Toll-like receptor 2 (TLR2) and TLR4 ligands. Human aortic endothelial cells were incubated with serum-free medium for 2 h and then incubated with PA (200 μM), LPS (1 μg/mL), Pam3-Cys-Ser-Lys4 (Pam₃; 2 μg/ml), or macrophage-activating lipopeptide-2 (MALP2; 10 ng/ml) for the indicated durations. Total RNA was isolated from cells and analyzed for expression of E-selectin by semiquantitative RT-PCR analysis. Expression level was normalized by β-actin and the maximum response by each stimulator was set to 100%. Quantification of 3 independent experiments is shown (means ± SE).

Fig. 2.

Profiles of inflammatory signaling pathways stimulated by palmitic acid (PA) are distinct from those by other canonical Toll-like receptor 2 (TLR2) and TLR4 ligands. Human aortic endothelial cells were incubated with serum-free medium for 2 h and then incubated with PA (200 μM; A), LPS (1 μg/mL; B), Pam3-Cys-Ser-Lys4 (Pam₃; 2 μg/mL; C), or macrophage-activating lipopeptide-2 (MALP2; 10 ng/mL; D) for indicated durations. Cell lysate was isolated and subjected to immunoblotting with indicated antibodies. Density of phosphorylated (p-)JNK was normalized by JNK, and density of p-NF-κB was normalized by β-actin bands. Quantification of 3 independent experiments is shown (means ± SE). Values for significant differences from basal level are indicated as following: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Inhibition of ACSLs by triacsin C blunted the expression of proinflammatory genes and signaling pathways that are stimulated by PA but not by other canonical TLR2 and TLR4 ligands.

Immediately after fatty acids are transported intracellularly, most of them are ligated to a CoA by ACSL for further intracellular utilization, including oxidation of fatty acid, synthesis of various lipid mediators, and modification of proteins and lipids. Because PA-stimulated proinflammatory responses are slower than other TLR ligand-stimulated responses, we hypothesized that PA might require intracellular processing to exhibit the proinflammatory responses. To test this hypothesis, we inhibited long-chain fatty acid synthetases (ACSL) by using triacsin C, an ACSL inhibitor, and examined whether ACSL affects TLR ligand-stimulated proinflammatory response. Pretreatment with triacsin C inhibited the PA-stimulated expression of E-selectin, IL-1β, and COX-2, whereas it did not affect LPS- or Pam₃CSK₄-stimulated expression of E-selectin, IL-1β, and COX-2 (Fig. 3, A–D). Next, we examined whether inhibition of ACSL by triacsin C affects TLR-stimulated signaling pathways. Pretreatment with triacsin C inhibited PA-stimulated phosphorylation of both NF-κB and JNK, whereas it did not affect LPS- or Pam₃CSK₄-stimulated phosphorylation of NF-κB and JNK (Fig. 3, E–G). Because expression of E-selectin and proinflammatory signaling pathways contribute to leukocyte adhesion, we examined the effect of ACSL inhibition on the leukocyte adhesion in vitro. Consistent with the aforementioned results, inhibition of ACSL by triacsin C reduced PA-induced leukocyte adhesion, whereas it did not affect LPS- or Pam₃CSK₄-induced leukocyte adhesion (Fig. 4). The results suggest that PA-induced, but not other canonical TLR ligand-induced, proinflammatory responses require intracellular processing of PA via ACSL activities.

Fig. 3.

Inhibition of acyl-CoA synthetases (ACSL) by triacsin C blunted the expression of proinflammatory genes and signaling pathways that are stimulated by palmitic acid (PA) but not by other canonical Toll-like receptor 2 (TLR2) and TLR4 ligands. Human aortic endothelial cells (HAECs) were incubated with serum-free medium for 2 h and then incubated with PA (200 μM) for 6 h, LPS (1 μg/mL) for 4 h, or Pam3-Cys-Ser-Lys4 (Pam₃; 2 μg/mL) for 2 h. HAECs were pretreated with triacsin C (10 μM) 30 min before treatment with PA, LPS, and Pam₃. A: representative gel is shown. Total RNA was isolated from cells and analyzed for expression of E-selectin (B), IL-1β (C), and cyclooxygenase-2 (COX-2; D) by semiquantitative RT-PCR analysis. Amount of gene expression was normalized by β-actin expression. Cell lysate was isolated from cells and subjected to immunoblotting with indicated antibodies. E: a representative image is shown. Density of phosphorylated (p-)NF-κB was normalized by NF-κB (F) and density of the p-JNK by JNK (G). Quantification of 3 independent experiments is shown (means ± SE). ns, Not significant; NT, not treated. Student’s t test was performed between untreated and triacsin C-treated samples. ***P < 0.001 and ****P < 0.0001.

Fig. 4.

Inhibition of acyl-CoA synthetases (ACSL) by triacsin C blunted the leukocyte adhesion stimulated by palmitic acid (PA) but not by other canonical Toll-like receptor 2 (TLR2) and TLR4 ligands. Human aortic endothelial cells were plated in a 6-well plate and incubated with serum-free medium for 2 h and then treated with PA (200 μM) for 6 h, LPS (1 μg/mL) for 4 h, or Pam3-Cys-Ser-Lys4 (Pam₃; 2 μg/mL) for 2 h. Triacsin C (10 μM) was pretreated 30 min before treatment with PA, LPS, and Pam₃. Cells were then washed with PBS, and calcein-AM (fluorescent dye)-loaded human monocytes (U937; 5 × 10⁵/well) were incubated for 30 min. Unbound cells were washed with PBS and fixed with ethanol for 10 min. Five fields per each experimental condition were imaged (A), and fluorescent cells were counted and averaged per each experimental condition (B). Scale bar, 25 μm. ns, Not significant; NT, not treated. Student’s t test was performed between untreated and triacsin C-treated samples. ***P < 0.001.

ACSL1 is involved in PA-stimulated proinflammatory responses.

To identify the specific ACSL isotype for the PA-stimulated inflammatory response, ACSL1 was knocked down using siRNA (Fig. 5, A and B). The knockdown of ACSL1 blunted the PA-stimulated expression of E-selectin and IL-1β (Fig. 5, C and D), but not that of COX-2 (Fig. 5E). The knockdown of ACSL1 blunted PA-stimulated phosphorylation of NF-κB and JNK (Fig. 5, F–I). In contrast, the knockdown of ACSL1 did not affect LPS- or Pam₃CSK₄-stimulated expression of E-selectin (Fig. 6, A and B). To examine the role of ACSL1 in leukocyte adhesion, HAECs were transfected with scrambled siRNA or siRNA for ACSL1. The knockdown of ACSL1 reduced PA- but not LPS- or Pam₃CSK₄-stimulated leukocyte adhesion (Fig. 6, C and D). The data suggest that ACSL1 in HAECs plays a major role in PA-stimulated proinflammatory responses.

Fig. 5.

Knockdown of acyl-CoA synthetase-1 (ACSL1) blunted palmitic acid (PA) -induced proinflammatory genes and signaling pathways. Human aortic endothelial cells were transfected with scrambled or siRNA for ACSL1 and then incubated for 48 h. Cells were treated with BSA or PA (PA, 200 μM) for 6 h. Total RNA was isolated from cells and analyzed for expression of E-selectin, IL-1β, and cyclooxygenase-2 (COX-2) by semiquantitative RT-PCR analysis. Representative image is shown (A), and ACSL1 mRNA was significantly reduced (B). Expression of E-selectin (C), IL-1β (D), and COX-2 (E) was normalized by β-actin. Cell lysate was isolated and subjected to immunoblotting with indicated antibodies. Representative image is shown (F), and the level of ACSL1 was reduced by siRNA (G). Density of phosphorylated (p-)JNK was normalized by JNK (H) and density of p-NF-κB by NF-κB (I). Quantification of 3 independent experiments is shown (means ± SE). ns, Not significant. One-way ANOVA was performed with Bonferroni’s multiple comparison post hoc tests. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 6.

Knockdown of acyl-CoA synthetase-1 (ACSL1) blunted the expression of E-selectin and leukocyte adhesion stimulated by palmitic acid (PA) but not by other canonical Toll-like receptor 2 (TLR2) and TLR4 ligands. Human aortic endothelial cells (HAECs) were transfected with scrambled or siRNA for ACSL1 and then incubated for 48 h. HAECs were serum starved for 2 h and then incubated with PA (200 μM) for 6 h, LPS (1 μg/mL) for 4 h, or Pam3-Cys-Ser-Lys4 (Pam₃; 2 μg/mL) for 2 h. Total RNA was isolated from cells and analyzed for the expression of E-selectin by semiquantitative RT-PCR analysis. Representative image is shown (A), and expression of E-selectin (B) was normalized by β-actin. Quantification of 3 independent experiments is shown (means ± SE). HAECs were plated in a 6-well plate and incubated with serum-free medium for 2 h and then treated with PA (200 μM) for 6 h, LPS (1 μg/mL) for 4 h, or Pam₃CSK₄ (2 μg/mL) for 2 h. Next, cells were washed with PBS, and calcein-AM (fluorescent dye)-loaded human monocytes (U937; 5 × 10⁵/well) were incubated for 30 min. Unbound cells were washed with PBS and fixed with ethanol for 10 min. Five fields per each experimental condition were imaged (C), and fluorescent cells were counted and averaged per each experimental condition (D). NT, not treated. Scale bar, 12.5 μm. Quantification of 3 independent experiments is shown (means ± SE). ns, Not significant. Student’s t test was performed between scrambled siRNA (sc) and si-RNA for ACSL1-treated samples. ***P < 0.001.

Inhibition of ACSL by triacsin C restored the insulin-induced vasodilation that was impaired by PA.

We (26, 29) previously reported that PA inhibits insulin-induced vasodilation by TLR2- and TLR4-mediated mechanisms. To examine whether inhibition of ACSL affects insulin-induced vasodilation, the vessels were preinfused with triacsin C for 30 min before PA infusion. PA suppressed the vasodilator action of insulin, and inhibition of ACSL by triacsin C almost restored the vasodilation in response to insulin (Fig. 7A). However, neither PA nor triacsin C affected the sodium nitroprusside-stimulated (endothelium-independent vasodilation) vasomotor activity (Fig. 7B). These data demonstrate that PA suppressed the vasodilator action of insulin and that inhibition of ACSL activity by triacsin C restored the vasodilatory action of insulin. The results suggest that PA-induced inflammatory response contributing to endothelial dysfunction requires the intracellular processing of PA by ACSL.

Fig. 7.

Triacsin C restored insulin-induced vasodilation that was impaired by PA. Mesenteric arteries from 8 wk old male C57BL/6J mice were isolated. Arteries were prepared as described in materials and methods. Arteries were intraluminally perfused with triacsin C (1 μM) 30 min before BSA or PA (200 μM) for 4 h. Vessels were preconstricted with phenylephrine (1 μM) and then exposed to increasing concentrations of insulin (A) or sodium nitroprusside (SNP; B). Vasodilations in response to insulin or SNP without or with PA in the presence or absence of triacsin C were observed and recorded using a data acquisition system. Insulin-stimulated vasodilation was impaired in PA-perfused vessels but not in BSA-perfused vessels. Pretreatment with triacsin C restored the insulin-stimulated vasodilation that was suppressed by PA. However, there was no difference among vessels pretreated with PA or BSA in the presence or absence of triacsin C when the vessels were stimulated with SNP. The percentage of vasorelaxation was calculated as described in materials and methods and presented as means ± SE. Data were analyzed by repeated-measures two-way ANOVA combined with Tukey’s post hoc test; n = 7–9. Values showed significant difference as follows: *P < 0.05, **P < 0.01, and ***P < 0.001, BSA vs. PA; #P < 0.05 and ##P < 0.01 Triacsin C+PA vs. PA.

DISCUSSION

We and others (26, 28, 29) previously reported that SFA causes proinflammatory response through the TLR2- and TLR4-mediated mechanisms in vascular endothelial cells. Inflammatory responses in the vascular endothelium affect HFD-induced insulin resistance and impairment of insulin-induced vasodilation (26, 28, 29). Although SFA and canonical TLR agonists stimulate both JNK and NF-κB, the mechanisms by which SFA and the TLR agonists stimulate inflammatory responses may be distinct in HAECs. In the present study, we compared the proinflammatory responses activated by PA and canonical TLR agonists. We observed that PA-stimulated inflammatory responses are mediated, at least in part, by ACSL1 activity. In contrast, other TLR agonists do not require the ACSL1 activity to exhibit proinflammatory response in HAECs. This suggests that intracellular fatty acyl-CoA contributes to the expression of cell adhesion molecules and leukocyte binding in endothelial cells. One of the potential mechanisms by which fatty acyl-CoA stimulates proinflammatory responses is the production of reactive oxygen species and proinflammatory lipid mediators including prostaglandins and ceramides (5, 13, 64). The second potential mechanism is the acylation of proteins and other lipids, which may affect inflammatory responses (14, 21, 31, 41, 63). The third potential mechanism is the acylation of various proteins and nuclear factors that regulate gene expression, protein function, and subcellular localization (15, 18, 22). These potential mechanisms require fatty acyl-CoA, which is produced by the metabolic processing of fatty acids.

We demonstrate that pharmacological inhibition of ACSL or knockdown of the ACSL1 blunted PA-induced proinflammatory response. Moreover, other studies demonstrate that ACSL1 in myeloid cells plays a critical role in atherosclerosis (27). In contrast, Anderson et al. (1) demonstrated that triacsin C potentiates the inflammatory response by stearic acid (C₁₈ saturated fatty acid) in mouse macrophages, while Lancaster et al. (35) showed that triacsin C suppresses PA-stimulated JNK phosphorylation. These discrepancies between studies could be due to differences in experimental conditions or cell types. Nonetheless, our results suggest that intracellular fatty acyl-CoA is a source for metabolic reprogramming, which may contribute to PA-induced inflammatory response (35).

We observed that pretreatment with triacsin C reduced the expression of proinflammatory markers in endothelial cells such as E-selectin, IL-1β, and COX-2 (Fig. 3, A–D). By contrast, knockdown of ACSL1 decreased E-selectin and IL-1β, but not COX-2 (Fig. 5, A–E). This suggests that PA-induced expression of E-selectin and IL-1β is regulated by ACSL1 whereas expression of COX-2 is regulated by other isotypes of ACSL. Because SFA-induced Cox-2 expression is mediated through a TLR4/MyD88/IL-1 receptor associated kinase-1/TNF receptor-associated factor 6-dependent mechanism in macrophages (36), one or more of the ACSL isotypes seems to be downstream molecule(s) of the TLR signaling pathways. Cox-2 regulates the production of prostacyclin (PGI₂) and other prostanoids that are metabolites of arachidonic acids (53). ACSL3 and ACSL4 have a substrate preference for unsaturated fatty acids over SFA (15). Thus, it is conceivable that the expression of COX-2 can be regulated by other isotypes of ACSL in HAECs. Further investigation about the role of each ACSL isotype in the regulation of the proinflammatory genes and the detailed underlying mechanisms is warranted.

Cluster of differentiation (CD)36 and lipoprotein lipase (LPL) facilitate fatty acid transport in endothelial cells and increase inflammatory responses (55, 57). CD36 plays an important role in atherosclerosis and foam cell formation in conjunction with TLR activation (7), and SFA or oxidized lipids stimulate TLR via CD36 or LPL activity (25, 32, 49, 50). However, it is unknown whether the fatty acid uptake is directly associated with SFA-stimulated inflammatory responses or not. Intracellular fatty acyl-CoAs can be converted into other proinflammatory lipid mediators including diacylglycerol, prostaglandin E₂, arachidonic acids, leukotrienes, and ceramides (6, 23, 53). One potential mechanism for the SFA-induced inflammatory response is that SFA may alter membrane fluidity, affecting the activity of TLRs or assembly of TLR with other components of TLR coreceptors, including CD14, MD2, TLR1, and TLR6 (36, 51, 65). Because polyunsaturated fatty acid (PUFA) blunts the SFA-induced inflammatory response, membrane fluidity and TLR activation may contribute, at least in part, to SFA-induced inflammatory responses. The next possibility is that fatty acid oxidation increases oxidative stress in endothelial cells, which in turn stimulates the proinflammatory response (13, 60). Furthermore, ACSL1 activity is crucial for mitochondrial fatty acid oxidation in some tissues such as brown adipose tissue, skeletal muscle, liver, and heart (16, 17, 39, 66). Thus, the metabolic processing of fatty acid by ACSL1 may be necessary for PA-induced inflammation in conjunction with TLR priming by other endogenous agonists (35). Further studies clarifying the distinct mechanisms by which these TLR activators regulate the proinflammatory responses are warranted.

Endothelial function is associated with insulin resistance and metabolism (2, 30). In this study, we demonstrate that triacsin C restores the vasodilatory effect of insulin that is impaired by PA (Fig. 7). We (26, 29) previously demonstrated that the mesentery arteries isolated from diet-induced obese mice display impaired insulin-stimulated vasodilation. Obesity-induced endothelial dysfunction was prevented by the deletion of either TLR2 or TLR4 (26, 29). Inflammatory responses inhibit endothelial nitric oxide (NO) synthase (eNOS) activity, which leads to the reduction of NO and vasodilation (26, 29). Reduced NO and impaired vasodilatory activity are associated with insulin resistance and contribute to impaired glucose uptake (3, 10, 34). This is consistent with the notion that the deletion of TLR2 or TLR4 protects mice from HFD-induced insulin resistance and glucose intolerance (26, 29). Similarly, knockdown of TLR2 or TLR4 prevented the PA-induced impairment of eNOS activation in HAECs (26, 29). In skeletal muscle, insulin-stimulated activations of eNOS and NO production are crucial for capillary recruitment (59). These vascular actions of insulin contribute to glucose uptake and are dependent on insulin sensitivity.

TLR2 and TLR4 are required for PA-induced proinflammatory responses not only in vascular endothelial cells but also in macrophages and pancreatic β-cells (24, 38, 62). Whether PA directly binds to TLR2 and TLR4 (35, 44, 47) has been a subject of controversy. Lancaster et al. (35) reported that priming of TLR4, but not direct binding of PA to TLR4, is required to stimulate PA-induced inflammatory response by metabolic reprogramming. This may be one of the reasons that the PA-stimulated inflammatory response is distinct from canonical TLR2 and TLR4 agonist-stimulated inflammatory response.

Because plasma contains various long-chain fatty acids such as saturated and unsaturated fatty acids, it has been suggested that the choice of a fatty acid influences the results of experiments (19, 61). Moreover, oleic acid or mixtures of fatty acids reverse the adverse effects of PA (11, 45). In contrast, Clerk et al. (10) and Liu et al. (42) demonstrated that infusion of intralipid-heparin to healthy young human subjects increases plasma free fatty acid levels and lowers insulin-stimulated blood flow and glucose disposal. Thus, although our experimental condition using PA may not directly implicate the physiological and pathophysiological conditions, the results suggest that intracellular processing of PA by the ACSL activity seems to play an important role in PA-induced endothelial dysfunction.

Endothelial dysfunction caused by metabolic abnormality and inflammatory responses contributes to cardiovascular complications, which increases mortality and morbidity. Herein, we demonstrate that the mechanism by which PA stimulates the inflammatory response is distinct from those employed by canonical TLR agonists (Fig. 8). This expands the understanding of the mechanisms by which TLRs mediate inflammatory responses in endothelial dysfunction and cardiovascular disease.

Fig. 8.

Schematic diagram of palmitic acid (PA) -stimulated proinflammatory response. PA stimulates proinflammatory response through an acyl-CoA synthetase-1 (ACSL1) -dependent mechanism, which is distinct from the canonical Toll-like receptor (TLR) ligand-stimulated inflammation. This metabolic inflammation inhibits insulin-stimulated vasorelaxation, which contributes to cardiovascular complications in lipid-rich environments.

GRANTS

This work was supported by the University of Alabama at Birmingham Diabetes Research Center-sponsored pilot and feasibility program supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (P30 DK-079626), National Heart Lung and Blood Institute (R01 HL-128695 to J. Kim), National Institute on Aging (R03 AG-058078 to J, Kim), and NIDDK (R00 DK-95975-03 and R01 DK-120684 to S. Bhatnagar).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.K. conceived and designed research; G.R. performed experiments; G.R., S.B., and J.K. analyzed data; G.R., S.B., D.J.H., and J.K. interpreted results of experiments; G.R. and J.K. prepared figures; G.R. and J.K. drafted manuscript; G.R., S.B., D.J.H., and J.K. edited and revised manuscript; G.R., S.B., D.J.H., and J.K. approved final version of manuscript.