Lipid Droplet Biogenesis and Function in the Endothelium

Abstract

Rationale

Fatty acids (FA) are transported across the capillary endothelium to parenchymal tissues. However, it is not known how endothelial cells (EC) process a post-prandial surge of FA.

Objective

This study was designed to characterize lipid droplet (LD) formation and function in EC.

Methods and Results

LD form and degrade in EC in vivo after FA loading. In cultured EC, LD synthesis and turnover is dynamic and function to protect EC from lipotoxic stress and provide FA for metabolic needs.

Conclusions

Our results delineate endothelial LD dynamics for the first time, demonstrating their protective role in lipotoxicity, FA utilization and mobilization.

INTRODUCTION

Endothelial cells (EC) line the entire circulatory system and play a critical role in controlling the transport of oxygen and nutrients. Fatty acid (FA) transport across blood vessels lining metabolically active tissues release FA from triglyceride (TG)-rich lipoproteins via metabolism by lipoprotein lipase (LPL), which docks on the luminal surface of capillary endothelium together with GPIHBP1¹. After TG metabolism to FA, CD36 can serve as a transporter for FA into endothelial cells providing an important energy source for metabolically active tissues such as the heart and skeletal muscle². Several factors have been reported to regulate FA uptake and transport in EC including vascular endothelial growth factor B (VEGF-B)³, peroxisome proliferator activated receptor gamma (PPAR-γ)⁴, and 3-hydroxy-isobutyric acid⁵, implying dynamic regulation of FA by locally produced substances. However, there is little information investigating how EC store, metabolize and utilize FA.

Lipid droplets (LD) are intracellular compartments that serve as fat reservoirs⁶. Upon exposure to excessive amounts of FA, cells can esterify FA into TG rich LD via the Kennedy pathway to prevent lipotoxicity⁷. Since all of the enzymes involved in TG synthesis, including glycerol-3-phosphate acyltransferases (GPATs), acyl-CoA: 1-acyl-glycerol-3-phosphate acyltransferase (AGPAT), lipins, and diacylglycerol O-acyltransferases (DGATs), reside on the endoplasmic reticulum (ER), the de novo generation of LD is believed to occur in the ER membrane⁸. Continuous accumulation of TG between the ER leaflets leads to their detachment from the ER, forming cytosolic LD and dynamics of LD are tightly regulated by their associated proteins⁹. Conversely, lipids in LD can be subsequently hydrolyzed by a subset of lipases, namely adipose triglyceride lipase (ATGL), hormone sensitive lipase (HSL) and monoglyceride lipase (MGL), in a process called lipolysis. FA released during lipolysis can further be used to provide energy or substrate for cellular lipid synthesis¹⁰.

While adipocytes are the most active cells in storing and metabolizing LD, LD formation can occur in all eukaryotic cells tested, especially under pathological conditions of FA excess. For instance, obese individuals with chronically elevated levels of circulating non-esterified FA develop LD within skeletal muscle and liver, and the presence of LD are posited to promote insulin resistance¹¹. Although the presence of neutral lipids (TG and cholesterol esters) in lipid globules have been documented in EC lining mammalian atheromas^12,13,14,15, the biogenesis and metabolism of LD in EC has not been thoroughly investigated.

Thus, the purpose of this study is to characterize the ability of EC to generate and metabolize LD. Here, we show that EC readily form and degrade LD in response to changing levels of TG, in vivo. Using both cultured EC and en face imaging of EC in large vessels, we demonstrate that EC rely on DGAT1 for TG synthesis during LD formation and ATGL for lipolysis of LD. Mechanistically, LD formation in EC provides a protective mechanism from lipotoxic ER stress. Moreover, FA hydrolyzed from LD during lipolysis can be utilized as a source of energy or can be released extracellularly and esterified into LD by skeletal muscle in co-culture experiments. These findings document for the first time the dynamics of LD and implicate an active role of LD as key organelles regulating intracellular lipid homeostasis in EC.

METHODS

Detailed, expanded Methods are included in the Online Supplement

Gavage experiments

C57BL/6 mice of 8–12 weeks were fasted for 16 hrs. Blood plasma samples for TG measurement were collected by retro-orbital bleeding prior to gavage (0 time point). Next, mice were orally gavaged with olive oil (10mL/kg body weight). Thoracic aortae and blood plasma samples were collected at designated times for future examination of LD in vessels and TG measurements.

LD Detection by BODIPY 493/503

Neutral lipids in EC lining aortae or in cultured EC were detected using the fluorescent dye BODIPY 493/503 (Invitrogen). Intact vessels from mice gavage with olive oil or aortae from mice incubated with OA (1mM) overnight were fixed and en face immunostained as described in detailed methods. Cells were grown to confluence on coverslips pre-coated with 0.1% gelatin in PBS solution. After designated treatments, cells were washed 3× with PBS and fixed with 4% paraformaldehyde in PBS solution for 15 mins. Fixed samples were washed 3× with PBS and stained with BODIPY 493/503 diluted in PBS at the final concentration of 0.1mg/mL for 15 mins to delineate LD and with DAPI (Sigma, 0.1ng/ml) to highlight nuclei. Coverslips were mounted with Fluoromount™ Aqueous Mounting Medium (sigma) and imaged by laser-scanning confocal microscopy (Leica SP5) in the sequential scan mode with HCX PL APO lambda blue 63×/1.40 oil objective lens at room temperature.

Lipid droplet purification

LD in EC were purified based on published methods ^16,17 with modifications and detailed in Supplemental Methods.

Fatty acid release assay

Human Dermal Micovasvular Endothelial Cells (HDMEC) (passage 10–14) were cultured on 0.4μm pore transwell inserts (Corning CLS3460) for 4 days until the cells formed compact monolayers. C2C12 myoblasts were cultured in a separate 12-well plate until confluent and differentiated into myotubes by 2% horse serum in DMEM for 4 days. One day prior to the experiment, HDMEC were loaded with 1mM OA added to the upper and lower chambers overnight to induced LD formation. On the day of experiment, transwell inserts with LD-rich HDMEC were place into the 12-well plate with or without differentiated C2C12 myotubes. Equal volume of EBM-2 medium containing fatty acid-free albumin (50μM) was added into both upper and lower chambers. 2% of medium was collected from both chambers after 6 and 24 hours. FFA concentration was determined using a fluometric-based assay kit (Cayman Chemicals) according to instructions. To normalized FFA release, HDMEC were lysed by protein lysis buffer (See Detailed Methods) and total protein amount was determined by DC™ Protein Assay Kit (Bio-Rad). C2C12 myotubes were washed 3× with PBS, fixed with 4% paraformaldehyde in PBS for 15mins and stained by BODIPY 493/503 diluted in PBS at the final concentration of 0.1mg/mL for 15 mins to delineate LD and with DAPI (Sigma, 0.1ng/ml) to highlight nuclei. Fixed C2C12 samples were imaged by epifluorescence microscope (Zeiss, Axio Vert 200M) with 10× objective lens at room temperature.

RESULTS

Characterization of LD dynamics in EC

To determine if EC form LD in response to elevated TG levels in blood, we gavaged a single bolus of olive oil to WT (C57BL/6) mice that were fasted overnight. Maximum TG accumulation in blood occurred 90 mins post gavage and TG levels returned to baseline after 180 mins (Figure 1A). Despite TG levels peaking at 90 mins, prominent LD formation in EC of thoracic aortae were observed at 180 mins post gavage by en face imaging of esterified neutral lipids in EC using the dye BODIPY 493/503. At 270 mins post gavage, LD were no longer detectable implying active degradation of LD in EC (Figure 1B and C). This result suggests that EC of large blood vessels can dynamically form and degrade of LD in EC after transient changes of TG levels in blood.

Figure 1. LD are dynamically regulated in EC in vivo.

WT (C57BL/6) mice were fasted for 16 hrs and gavaged with olive oil for plasma TG and endothelial LD analysis. (A) Plasma TG levels peak at 90 mins whereas (B) LD in EC layer were detected via en face immunostaining at 180 mins post-gavage, as quantified in (B) by numbers of LD. n=2–4 mice, 7 images per mouse. Scale bar, 50μm. Data in (A) and (C) are expressed as mean ± SEM.

Next, LD formation was characterized in cultured mouse lung endothelial cells (MLEC, abbreviated as EC) by supplementation of the monounsaturated (18:1) fatty acid, oleic acid (OA) into the medium. OA-induced LD formation in EC in a time- (Figure 2A and B) and dose-dependent (Figure 2C and D) manner, reaching maximum cellular TG content (285.86±41.05 nmol TG/mg protein) after 24hrs (1mM OA loading). OA supplementation increased the mRNA levels of enzymes involved in TG synthesis, including GPAT4, AGPAT2, Lipin2, DGAT1 and DGAT2, whereas the expression of sterol ester synthetic enzymes, ACAT1 and ACAT2, did not change (Figure 2E). In addition to the synthetic pathway, EC express lipases for TG degradation, including ATGL, HSL and MGL (Figure 2F, lane 1). In response to OA supplementation, the protein levels of ATGL and HSL were increased (Figure 2F), implicating active lipolysis during LD formation. These data indicate that EC store excessive FA into TG and baseline lipolysis is also activated to maintain TG homeostasis.

Figure 2. EC utilize TG synthesis enzymes for LD formation and canonical lipases for lipolysis.

(A) OA supplementation (1mM) increases LD formation in MLEC over time, as quantified in (B). (C) OA-induced LD formation in MLEC is dose-dependent and maximal at 1mM OA incubation overnight as quantified in (D). (E) mRNA expressions of TG synthetic enzymes (GPAT4, AGPAT2, Lipin2, DGAT1 and DGAT2 expressions) are induced in MLEC responding to OA treatment over time, whereas sterol ester synthetic enzymes (ACAT1 and ACAT2) are not. (F) Western blotting analyses show that ATGL and HSL levels increase in response to OA treatment whereas MGL expression remains unchanged in MLEC. (n=3) (G) MLEC express PLN family proteins (PLN2 and PLN3) shown by agarose gel of qPCR products. (n=3) (H) LD fractionation from EAhy 926 cells untreated (“−” lanes) or treated with OA “+” lanes) show that PLN2, PLN3, ATGL, CGI-58, and Cav-1 are enriched in LD fraction whereas other proteins examined, FLOT-1, Grp94 and Hsp90, are not. (n=3) Scale bars, 25μm. Data in (B), (D), and (E) are expressed as mean ± SEM (n=3, 5 images per experiment in (B) and (D)) and analyzed by two-way ANOVA. *p<0.05, **p<0.01, ***p<0.005.

Since LD associated proteins regulate LD dynamics, we examined LD associated proteins in EA.hy 926 cells pre-incubated with OA overnight. These cells were used since large numbers of cells are required for the purification of LD from non-adipocytes. RT-qPCR analyses showed that of the 5 known perilipins (PLN) only PLN2 and PLN3 are expressed in EAhy 926 cells (Figure 2G). Subcellular fractionation of LD from other cellular constituents using sucrose gradient fractionation demonstrated LD fractions are indeed enriched with PLN2, PLN3, and lipolytic enzymes, including ATGL and its co-activator, CGI-58 (Figure 2H). As a control, the lipid raft protein Flot-1, ER localized Grp94 and abundant cytoplasmic protein Hsp90 were not detected in purified LD. We also observed an enrichment of caveolin-1 (Cav-1) in LD fractions similar to previous work in adipocytes^{18, 19}. These biochemical data show that purified endothelial LD are decorated with several known LD-associated proteins that can regulate LD dynamics.

In the final step of TG synthesis, DGATs add the last fatty acyl group to diacylglycerides (DAG), forming TG. Mammalian cells express two DGAT isoforms, DGAT1 and DGAT2²⁰. To assess the function of individual DGAT in EC, we used specific chemical inhibitors that block DGAT-1 (A922500²¹) or DGAT-2 (PF-06424439) during LD formation. Co-treatment of OA with A922500 blocked the majority of TG synthesis in EC with only 16.9±8.1% of LD remaining, while PF-06424439 inhibited TG synthesis to a lesser extent. Combining both inhibitors almost completely abolished LD formation in EC (Figure 3A and quantified in B), indicating that DGATs are essential for TG synthesis in EC with a greater dependency on DGAT1 than DGAT2.

Figure 3. DGAT1 and ATGL are key enzymes regulating LD dynamics in EC.

(A) DGAT1 inhibition by A922500 (5μM) has a greater effect on OA-induced LD formation in MLEC than DGAT2 inhibition by PF-06424439 (10μM). Co-incubation with both inhibitors completely abolishes LD formation, as quantified in (B). (C) Basal lipolysis is significantly delayed over time by Atglistatin (10 μM) in MLEC pre-treated with OA (1mM) overnight, as quantified in (D) by TG content. (E) Thoracic aortae incubated with OA (1mM) overnight show LD formation in EC layer via en face immunostaining, where DGAT-1 inhibitor, A922500 (5μM) abolishes LD formation and ATGL inhibition by Atglistatin (10μM) enhances LD accumulation, as quantified in (F) by BODIPY intensity. (n=3) Scale bars in (A) and (C), 10μm. Scale bar in (E), 50μm. Data in (B), (D) and (F) are expressed as mean ± SEM (n=3, 5 images per experiment) and analyzed by unpaired student-t test. *p<0.05

ATGL has been identified as the key lipase initiating TG hydrolysis in adipose tissue²². As ATGL levels increased in response to OA supplementation and ATGL was enriched in LD, we tested if ATGL plays a similar role in EC as in adipocytes. After loading of OA, inhibition of ATGL activity by Atglistatin²³ delayed LD degradation in comparison to non-treated cells with 56.9±7.8% of LD remaining after 48 hours (Figure 3C and D), and also significantly inhibited FA release from LD-rich EC (Online Figure IA). In addition, co-treatment with Atglistatin during OA supplementation increased TG accumulation (Online Figure IB). Inhibition of HSL (with CAY10499) and MGL (with URB602) critical lipases during subsequent steps of TG hydrolysis reduced FA release from EC but had no effects on TG accumulation (Online Figure IA and B). These data suggest that ATGL acts as the rate-limiting step for TG hydrolysis.

To validate these findings in intact blood vessels, isolated aortae were supplemented OA for 12 hrs in the presence of inhibitors of either DGAT1 or ATGL, and EC neutral lipids were imaged en face. Indeed, DGAT1 inhibition by A922500 abolished a majority of OA-induced LD in EC whereas blocking lipolysis with Atglistatin increased TG accumulation (Figure 3E and F). Collectively, these results in cultured EC and ex vivo blood vessels indicate that EC form LD largely via DGAT1 and degrade LD via ATGL.

Examination of LD functions in EC

Next, we examined several potential functions of LD in EC including: 1). LD formation as a mechanism to protect against lipotoxicity; 2). LD metabolism of TG as a source for FA for mitochondrial function; and 3). LD metabolism as a mechanism to transiently store and liberate FA for subsequent metabolism by skeletal muscle. To test if LD formation protected cells from lipotoxicity, EC were treated with saturated FA, including palmitic acid (PA, 16:0) and stearic acid (SA, 18:0) in the absence or presence of OA. In contrast to OA loading, equimolar concentrations of PA and SA weakly induced LD formation (Figure 4A). Instead, the saturated FA promoted ER stress assessed by BiP and CHOP induction (Figure 4B, lanes 5 and 9), indicating these saturated FA were utilized differently than OA in EC. When saturated FA (PA or SA) were mixed with equimolar amounts of OA and supplemented to EC, LD formation was promoted (Figure 4A, bottom panels) and ER stress levels were diminished (Figure 4B, lanes 7 and 11). Inhibition of DGAT1 and 2 promoted ER stress induced by OA and abrogated the protective effect of OA on PA/SA induced ER stress (Figure 4B, lanes 4, 8, and 12). Interestingly, inhibition of DGAT1, but not DGAT2, was sufficient to promote ER stress induced by OA alone (Online Figure IIA), but not by combinations of OA and PA/SA (Online Figure IIB, lanes 4, 8, and 12), implying the differential substrate utilization between DGAT1 and DGAT2. Collectively, these data suggest that the capability of EC to form LD is critical for their protection from lipotoxic ER stress.

Figure 4. LD formation protects EC from lipotoxcity and lipolysis-derived FA function as an energy resource in EC or for release to parenchymal cells.

(A) MLEC were treated with OA (C18:1), PA (C16:0), SA (C18:0) all at 1mM, or mixtures of OA and PA/SA (0.5mM of each) overnight. LD formation was detected in EC treated with OA or OA with PA/SA. (n=3) (B) OA incubation induces ER stress in MLEC when both DGAT1 (5μM, A922500) and DGAT2 (10μM, PF-06424439) are inhibited, as accessed by BiP and CHOP immunoblotting. PA and SA lead to increases in ER stress where the presence of OA abrogates this induction. This reduction is reversed when both DGAT1 and DGAT2 are inhibited. (n=2) (C) OA incubation has no effect on EC general mitochondrial function measured by OCR with injections of compounds indicated (−OA and +OA). Inhibition of FAO (ETOX; 40μM) in LD-rich EC shows decreased baseline OCR and mitochondria capacity (FCCP; 1μM) (+OA +ETOX) determined by FCCP (1μM) treatment, whereas ETOX has no effect on normal EC (−OA +ETOX). (D) Glycolytic parameters were determined by measuring ECAR. LD-rich EC exhibited impaired glycolytic activity via addition of glucose (10mM) (+OA). Inhibition of fatty acid oxidation (ETOX; 40μM) in LD-rich EC shows recovery of glycolytic flux (+OA +ETOX). (E) FA released from LD-rich HDMEC in a transwell insert accumulates at both sides of the transwell over time (CTRL). Inhibition of lipolysis (Atglistatin; 10μM or CAY10499; 10μM) reduces this accumulation whereas activation of lipolysis (Forskolin; 10μM and IBMX; 0.1mM) increases FA release. (F) In the co-culture system with LD-rich EC and C2C12 myotubes in the bottom well, FA release was only detected in the apical chamber, whereas (G) C2C12 myotubes accumulated LD. (n=3, 5 images of each experiment). Scale Bar in (A), 10μm. Scale Bar in (G), 50μm. Data in (C), (D), (E) and (F) are expressed as mean ± SEM (n=3) and analyzed by two-way ANOVA. *p<0.05 to −OA group. #p<0.05 to +OA group. #p<0.05 to control. N.S., non significant.

The release of FA from TG via lipolysis can provide FA for energy production in oxidative tissues such as skeletal muscle²⁴ and the heart²⁵. Since LD in EC can be hydrolyzed via ATGL, we tested if FA released from TG lipolysis can be utilized as a substrate for cellular energy. To address this, oxygen consumption rate (OCR) was measured as an index of mitochondrial respiration. LD-enriched EC (+OA) exhibited normal ATP production, maximal respiration, and non-mitochondrial respiration compared to cells treated with vehicle (−OA); Figure 4C). However, suppression of fatty acid oxidation (FAO) by inhibition of carnitine palmitoyltransferase-1 (CPT-1), with etomoxir (ETOX), baseline OCR and respiratory capacity was decreased only in LD-enriched EC (+OA + ETOX), suggesting that EC can utilize the LD derived FA for energy production in the resting state (Figure 4C). As normal EC rely primarily on glycolysis for energy, we tested if increased FAO in LD-enriched EC altered the glycolytic flux critical for ATP generation as measured by extracellular acidification rate (ECAR) reflecting glucose metabolism to lactate. LD-enriched EC (+OA) have reduced ECAR when exogenous glucose is provided (Figure 4D, +OA after glucose addition). This difference in ECAR was normalized in EC after oligomycin to induce maximum glycolytic capacity and after 2-deoxy glucose (2DG) incubation to assess non-glycolytic acidification rate (Figure 4D, after Oligo and 2-DG addition). To substantiate that the reduction in glycolysis is due to increase of FAO in LD-rich EC, we measured ECAR when FAO is inhibited by ETOX. Indeed, the impaired glycolytic flux is partially rescued by blocking FA oxidation (Figure 4D, +OA +ETOX after glucose addition), demonstrating that increased FAO decreases glycolytic flux in LD-enriched EC. Additionally, direct measurements of lactate is reduced in LD-enriched EC and could be partially rescued by ETOX treatment (Online Figure III). Together, our data indicate that the metabolized LD could provide FA as an energy resource and modulate EC metabolic flux and suppress glycolysis.

FA released from LD lipolysis could also provide extracellular FA for adipose or parenchymal tissues. Therefore, the release of free FA from LD (generated by pretreatment with OA) was assessed in EC cultured on a transwell insert. The EC monolayers were post-confluent as indexed by the lack of paracellular leakage of FITC-conjugated dextran over 24hrs of experimentation (Online Figure IV). Under these conditions, FA release was monitored on both apical and basolateral sides of EC and similar amounts of FA accumulated up to 24 hours on either side of the transwell chamber (Figure 4E, CTRL). The release of FA was attenuated by inhibition of ATGL and HSL, and enhanced by forskolin (FSK) -mediated activation of lipolysis (Figure 4E). To further determine if FA released from EC can be taken up by parenchymal cells, we performed the transwell assay with C2C12 myotubes cultured in the bottom chamber. In this co-culture system, free FA were detectable only in the apical chamber (Figure 4F) and the C2C12 myotubes were laden with LD imaged with BODIPY(Figure 4G). The accumulated FA were released from LD-rich EC as no LD were detected in C2C12 myotubes co-cultured with normal EC that were not pre-treated with OA (Online Figure V).

DISCUSSION

Our knowledge of lipid metabolism in EC is derived primarily from the role of LPL on the surface of capillary EC. LPL hydrolyzes very low-density lipoproteins (VLDL) and chylomicrons, generating FFA that are taken up via a CD36 dependent mechanism²⁶ for energy utilization in skeletal and cardiac muscle or re-esterified into TG in liver ²⁷. In contrast to this well studied process, how EC respond to changes in intracellular FA levels and the biogenesis of LD have not been described. Previous studies in vivo have reported LD in EC in mammalian atheromas^12,13,14,15 and in arteries from a patient with cardiomyopathy homozygous for a loss of function point mutation in ATGL²⁸, highlighting the significance of lipid storage in EC. Hence, the central focus of this paper is to characterize LD biogenesis and degradation in EC since little is known.

Here, we show that LD can be readily formed in intact EC lining blood vessels, both in vivo and ex vivo. The transient presence of LD in EC after a gavage of olive oil reflects the dynamic nature of LD synthesis and metabolism, a function clearly demonstrated in vessels treated with a DGAT1 inhibitor to block synthesis or an ATGL inhibitor to block metabolism. With respect to LD synthesis, DGAT1 is primarily responsible for TG synthesis in EC based on data using specific inhibitors of DGAT1 and DGAT2. Inhibition of DGAT1 during OA treatment, accounted for over 80% of TG synthesis. On the other hand, inhibition of DGAT2 activity had modest, but significant effect on TG synthesis (~20%). This is different from LD formation in adipocytes differentiated from mouse embryonic fibroblasts lacking either DGAT1 or 2 where the deletion of either DGAT did not influence TG accumulation, due to compensation by the remaining enzyme²⁹. Alternatively, it is possible that this discrepancy is due to different TG synthesis pathways involved in DGAT1 and 2. Previous studies have indicated that DGAT2 is responsible for incorporating endogenously synthesized monounsaturated FA into TG whereas DGAT1 may be involved in esterifying exogenous FA taken up by cells²⁰. Given our assays provide exogenous monounsaturated FA as substrate to EC, it is likely that DGAT1 dependent TG synthesis is favored. Nevertheless, the in vivo functional significance of DGAT 1 and 2 in EC is not known and interesting to investigate using conditional KO approaches.

In addition to the characterization of LD formation in EC, we developed a simple assay to assess the time dependent lipolysis of TG in LD, a process mediated by the canonical lipases ATGL, HSL and MGL. Blockage of ATGL markedly reduces TG degradation and FA release from LD loaded EC and the subsequent metabolism by HSL and MGL is consistent with ATGL mediated hydrolysis of TG to DAG as the rate-limiting step in LD degradation. Isolated primary EC lacking ATGL exhibit spontaneous LD accumulation under normal culture conditions, implying endogenous activity of ATGL in TG turnover (unpublished observations). Indeed, ex vivo loading of EC lining intact aortae with OA to generate LD is augmented by inhibition of ATGL confirming the importance of this lipase in vivo.

Finally, we show that LD formation in EC has at least three clear functions: to prevent lipotoxicity, provide FA to reduce glycolytic flux and for the release of FA from EC to adjacent cell types as shown in EC/myotubes co-culture experiments. Interestingly, PA and SA do not promote overt LD formation in EC but OA esterification into TG rich LD neutralizes the lipotoxic actions of these saturated FA. This may occur via enhance metabolism of saturated FA or by esterification of PA and SA into OA enriched LD as previously described in CHO cells use mass spectrometry measurements of TG acyl chains⁷. In addition, we show that in EC enriched with LD, that loading does not influence ATP production and OCR unless FA oxidation is blunted by ETOX during conditions of basal and maximal respiration (in the presence of FCCP). Moreover, the rate of glycolysis is attenuated by OA loading, in a manner reversible by ETOX, suggesting that LD lipolysis can attenuate glycolysis and impact glycolytic reserve in EC. Recent data has shown that the loss of CPT1A in EC, the target of ETOX, does not affect energetics in EC, but reduces angiogenesis due to the loss of carbon intermediates required for DNA synthesis³⁰. However, these studies were not performed in EC laden with LD. Finally, in transwell assays, we show that LD derived FA can be released non-directionally from EC and release is attenuated by inhibition of ATGL and HSL and augmented by the adenylate cyclase activator, forskolin providing evidence that LD metabolism can generate FA for utilization by other cell types. Co-culture of LD-laden EC with C2C12 myotubes results in undetectable FA on the basolateral side of EC and the accumulation BODIPY positive LD in myotubes only in EC preload with OA. These data imply that FA not only can passively diffuse across EC to provide FA for skeletal muscle oxidative metabolism, but the LD derived FA can regulate the flux of FA critical for metabolically active tissues.

In conclusion, this is the first study characterizing LD formation in EC. Indeed, EC in vivo, ex vivo or in culture readily generate and metabolize LD demonstrating the dynamic nature of unsaturated FA storage and utilization (see model in Online Figure VI). Moreover, we define new functions of LD derived FA in EC and imply that abnormal LD metabolism may contribute to vascular disease. Recent data supporting this concept shows that silencing of ATGL in human EC, increases TNF-α induced ICAM-1 expression and NF-κB activation³¹ and the loss of ATGL promotes micro and macrovascular EC dysfunction and inflammation³². Additional experiments testing the link between vascular LD and inflammatory vascular diseases are clearly warranted.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Endothelial cells (EC) are constantly exposed to lipids derived from dietary fat and lipoproteins.

• Capillary EC metabolize chylomicron and VLDL derived triglycerides (TG) to free fatty acids for energy utilization by the heart and skeletal muscle.

• Lipid droplets (LD) are intracellular hubs for lipid metabolism but never studied in EC.

What New Information Does This Article Contribute?

• EC can actively incorporate fat into TG rich lipid droplets suggesting that fatty acids have an intermediate fate and are not only passively transported across the endothelium.

• Lipid droplet metabolism protects endothelial cells from lipotoxicity and provides fatty acids for mitochondrial function and transport to adjacent cells.

Here we document that lipid droplets are a new lipid storage organelle in vascular endothelial cells regulating fatty acid metabolism and flux. Abnormalities in lipid droplet metabolism may contribute to endothelial dysfunction during conditions of hyperlipidemia.

Nonstandard Abbreviations and Acronyms

2-DG

2-deoxyglucose

AGPAT

1-acylglycerol-3-phosphate O-acyltransferase

ATGL

adipose triglyceride lipase

CPT

carnitine palmitoyltransferase-1

DAG

diacylgylcerides

DGAT

diacylglycerol O-acyltransferase

EC

endothelial cells

ECAR

extracellular acidification rate

ETOX

etomoxir

FA

fatty acids

FAO

fatty acid oxidation

FSK

forsoklin

GPAT

glycerol-3-phosphate acyltransferase

HDMEC

human dermal microvascular endothelial cells

HSL

hormone sensitive lipase

LD

lipid droplets

LPL

lipoprotein lipase

MGL

monoglycerol lipase

MLEC

mouse lung endothelial cells

OA

oleic acid

OCR

oxygen consumption rate

PA

palmitic acid

PLN

perilipin