Hypercholesterolemia Promotes a CD36-dependent and Endothelial Nitric-oxide Synthase-mediated Vascular Dysfunction*

Numerous studies have implicated either the presence or absence of CD36 in the development of hypertension. In addition, hypercholesterolemia is associated with the loss of nitric oxide-induced vasodilation and the subsequent increase in blood pressure. In the current study, we tested the hypothesis that diet-induced hypercholesterolemia promotes the disruption of agonist-stimulated nitric oxide generation and vasodilation in a CD36-dependent manner. To test this, C57BL/6, apoE null, CD36 null, and apoE/CD36 null mice were maintained on chow or high fat diets. In contrast to apoE null mice fed a chow diet, apoE null mice fed a high fat diet did not respond to acetylcholine with a decrease in blood pressure. Caveolae isolated from in vivo vessels did not contain endothelial nitric-oxide synthase and were depleted of cholesterol. Age-matched apoE/CD36 null mice fed a chow or high fat diet responded to acetylcholine with a decrease in blood pressure. The mechanism underlying the vascular dysfunction was reversible because vessels isolated from apoE null high fat-fed mice regained responsiveness to acetylcholine when incubated with plasma obtained from chow-fed mice. Further analysis demonstrated that the plasma low density lipoprotein fraction was responsible for depleting caveolae of cholesterol, removing endothelial nitric-oxide synthase from caveolae, and preventing nitric oxide production. In addition, the pharmacological removal of caveola cholesterol with cyclodextrin mimicked the effects caused by the low density lipoprotein fraction. We conclude that the ablation of CD36 prevented the negative impact of hypercholesterolemia on agonist-stimulated nitric oxide-mediated vasodilation in apoE null mice. These studies provide a direct link between CD36 and the early events that underlie hypercholesterolemia-mediated hypertension and mechanistic linkages between CD36 function, nitric-oxide synthase activation, caveolae integrity, and blood pressure regulation.

CD36 is a class B scavenger receptor that is expressed in cardiac myocytes, adipocytes, macrophages, platelets, and microvascular endothelial cells (1,2). Recent studies with mice and rats that lack CD36 demonstrate a role for this protein in atherosclerosis (3), angiogenesis (4), and diabetes (5)(6)(7). Genetic linkage studies implicate the absence of CD36 in the development of hypertension (8), although direct experimental evidence is lacking. One of the confounding factors in the linkage studies is that genes involved in blood pressure regulation such as endothelial nitric-oxide synthase, leptin, and neuropeptide Y are linked to CD36 within chromosome 4, making it difficult to attribute changes in blood pressure to CD36 (8,9). Studies by Pravenec et al. (8) demonstrate that the transfer of a segment of chromosome 4 to spontaneous hypertensive rats (lack CD36) normalized blood pressure; however, when just CD36 was transferred, blood pressure did not normalize (10). In contrast, Greenwalt et al. (11) demonstrate that CD36 was increased in animal hypertensive models, suggesting that CD36 may negatively impact blood pressure. Because of the numerous genetic alterations in spontaneous hypertensive rats, it has been difficult to assign a specific role to CD36 in blood pressure regulation. To directly examine the role of CD36 in blood pressure regulation, we used a targeted CD36 null mouse model (3,12).
CD36 is enriched in plasma membrane structures called caveolae (13,14). In endothelial cells, caveolae are enriched in cholesterol, sphingomyelin, endothelial nitric-oxide synthase (eNOS), 1 and caveolin (15)(16)(17). Caveolin is a 22-kDa protein that directly binds to eNOS in caveolae and maintains the enzyme in an inactive state (18,19). Agonist-induced increases in intracellular calcium cause calcium-calmodulin complexes to displace caveolin from eNOS, thereby activating the enzyme. Once the intracellular calcium returns to basal levels, caveolin displaces calmodulin and inactivates the enzyme. The localization of eNOS to caveolae plays a role both in the inactivation and activation of the enzyme. We have used an endothelial cell line to demonstrate that the depletion of caveola cholesterol will result in the re-localization of eNOS from caveolae to an internal membrane compartment, where it cannot be stimulated with extracellular agonists (15,16).
Endothelial nitric-oxide synthase is a critical regulator of vasomotor tone. Nitric oxide binds to soluble guanylate cyclase in vascular smooth muscle cells, which results in a net decrease in intracellular calcium, relaxation of smooth muscle cells, dilation of the vessels, and a decrease in blood pressure. Interestingly, a dysfunction of eNOS is associated with both hypertension and atherosclerosis, and clinically these two diseases have a strong associative correlation (20,21). The mechanism for this interaction is not known; however, recent studies with a heterologous in vitro system demonstrated that copper-oxidized LDL could deplete caveola cholesterol in a CD36-dependent manner, which resulted in the re-localization of eNOS and an inhibition of agonist-stimulated nitric oxide generation (15,16). Furthermore, it was demonstrated that HDL could restore the cholesterol concentration of caveolae, the localization of eNOS to caveolae, and consequently the ability of agonists to stimulate eNOS.
Although the in vitro data are suggestive of a mechanism that may link atherosclerosis and hypertension at the molecular level, data demonstrating that CD36 and caveolae are involved in blood pressure regulation does not exist. To address this issue we used apoE null and apoE/CD36 null mice to determine whether CD36 is involved in blood pressure regulation. The data demonstrate that CD36 deficiency protected mice from hypercholesterolemia-induced vascular dysfunction. CD36 deficiency also protected caveolae from cholesterol depletion, prevented nitric-oxide synthase re-localization, and maintained agonist-induced nitric oxide generation. In addition, experiments with isolated vessels demonstrated that the plasma LDL fraction is responsible for the reversible inhibition of agonist activation of eNOS. These studies provide the first molecular link between CD36 function, caveolae integrity, nitric-oxide synthase activity, and blood pressure regulation. In addition, these data imply that endothelial CD36 may be a useful clinical target in the management of atherosclerosisinduced vascular dysfunction.
Animals and Instrumentation-The following animals were used in the experiments: C57BL/6, apoE null, CD36 null, and apoE/CD36 null. At 6 weeks of age the animals were put on either a normal chow (0% cholesterol, 5.7% fat, Harlan Tekland #2018), or a high fat diet (0.2% cholesterol, 21% fat, Harlan Tekland #88137) and maintained on the diet for 6 weeks. Both male and female mice were used in the in vivo blood pressure studies and generated similar data; however, for the ease of presentation only the female mouse data are presented. Preliminary studies established the appropriate concentration for each drug used (data not shown).
Each mouse was acutely instrumented in three ways, catheterization of the femoral artery for direct measurement of blood pressure, catheterization of the carotid artery for drug administration, and placement of an ultrasonic flowprobe around the descending aorta for blood flow measurements. Catheters were constructed using polyethylene tubing with an outer diameter of 0.61 mm that was warmed and stretched to taper to an outer diameter of ϳ0.4 mm. The larger end was fitted onto a 30-gauge 0.5-inch needle, and a bevel was cut into the tapered end. The finished catheters ranged in length from ϳ20 mm for femoral catheters to 30 mm for carotid catheters. A 0.5-V series Transonic flow probe was used for abdominal aortic blood flow measurements.
Mice were anesthetized with 100 mg/kg ketamine and 10 mg/kg xylazine intraperitoneal. After the onset of surgical anesthesia, (gauged by lack of direct response to toe pinch, muscular relaxation, and loss of corneal reflexes), the mouse was placed in dorsal recumbency, and the surgical field was visualized using a dissecting microscope. A covered Deltaphase isothermal pad was used for temperature stabilization during the procedure. Anesthesia was maintained throughout the procedure by giving one-third of the ketamine dose (33 mg/kg intraperitoneal) to effect. For the femoral artery catheter placement, a 10-mm incision was made over the left femoral artery and vein. The fat pad overlying the vessels was reflected and held in place by a retractor. The femoral artery was then gently dissected away from the vein and femoral nerve using Dumoxel 5 micro-dissecting forceps. Once a 4.5-mm section of the vessel was isolated, 3 ligatures were placed around the artery. The first ligature (5-0 silk) was placed most distally, close to where the femoral artery branches to the fat pad. This ligature was tied completely to occlude the blood vessel. A second 5-0 silk ligature was placed proximally on the artery, near the abdominal wall; a half-surgical knot was tied and left loose in this ligature. A third ligature of 6 -0 Ethilon was placed between the two, and again, a half-surgical knot loosely tied. To facilitate cannulation, a pair of hemostatic forceps was clamped to the most distal ligature to apply tension to the artery and to the most proximal ligature to occlude blood flow into the middle portion of the artery. A small opening was made into the lumen of the vessel just proximal to the distal ligature using a 30-gauge needle. The proximal edge of the opening was then elevated using the tip of the microdissecting forceps, and the catheter was inserted and advanced until the tip rested against the most proximal ligature. The middle ligature was then tightened around both the vessel and catheter. The proximal ligature was then released and tied around both the vessel and the catheter. The catheter was further secured in place by bringing the most distal ligature around the catheter itself and tying it and by placing a fourth ligature of 5-0 silk through the thigh muscle and around the catheter. An injection cap was placed on the catheter, and it was flushed with heparin saline solution (2 units/ml) to ensure patency.
The carotid artery was approached by a 4.5-mm incision on the cervical midline and then dissecting through the muscles on the left side of the trachea. The carotid artery, which lies just lateral and dorsal to the trachea, was carefully isolated by gentle dissection, and the catheter was placed using the three-ligature technique described above.
For placement of the flowprobe, a ventral midline abdominal incision was made and extended through the linea alba into the abdominal cavity. The intestines were deflected to the right of the mouse to allow visualization of the abdominal aorta and the left kidney. A 3.4-mm segment of the aorta was isolated by dissection just caudal to the kidneys. The 0.5-V probe was placed around the vessel and held in place by a micromanipulator. The acoustic window was then filled with HR lubricating jelly to achieve the best possible signal.
The femoral artery catheter was then attached to a Transpac IV transducer (Abbott Laboratories) that had been primed with heparin saline. The transducer and probe were then connected to the Transonic Systems T206 small animal blood flow monitor. Windaq software was used to continuously record aortic blood flow (ml/mm) and blood pressure (mm of Hg). During recording, mice were given three doses of acetylcholine (0.5, 1.0, and 7.0 g/kg; only the 1.0 g/kg data are shown) and sodium nitroprusside (1.0, 15.0, 30.0 g/kg; only the 15.0 g/kg data are shown). Mice were continuously monitored during the procedure, and blood pressure was allowed to return to base line and then stabilize for ϳ30 s between doses. Catheters were flushed with heparin saline between drugs as well to prevent clot formation. Upon completion of data collection, mice were euthanized by CO 2 administration.
Cholesterol Mass Quantification-Cholesteryl heptadecanoate was added to each vessel preparation to serve as an internal standard, and the samples were then extracted with isopropanol-hexane (22). The extracted lipid was derivatized by suspending the dried lipid in N,Obis(trimethylsilyl) trifluoroacetamide/trimethylchlorosilane/acetonitrile (89:1:10). The material was heated at 80°C for 5 min, dried, suspended in iso-octane, and used for gas chromatography (Protocol T496125B, Supelco, Sigma Aldrich). Authentic cholesterol was dissolved in iso-octane and used as a standard for the retention time of cholesterol. The samples were injected (splitless) onto an Agilent 6890 GC G2579A system (Agilent, Palo Alto, CA) equipped with a SGE HT5 aluminum clad fused silica capillary column (12 m ϫ 0.32 mm ϫ 0.1 m, Supelco, Bellefonte, PA). The GC temperature program was as follows. The initial temperature was 220°C for 3 min then increased to 310°C (20°C/min) and then to 400°C (10°C/min) and held for 3.5 min. A model 5973 mass-selective detector (Agilent Technologies) was used in both scan and selected ion-monitoring modes to identify the samples.
Radiolabeled Cholesterol Determination-Thin-layer chromatography and liquid scintillation counting was used to measure the amount of [ 3 H]sterol in each sample (23). Each sample was adjusted to a volume of 1 ml with distilled water. Methanol (1.2 ml) containing 2% (v/v) acetic acid was added to the sample before vortexing 2 times, 30 s each. Chloroform (1.2 ml) was then added, and the sample was vortexed 2 times, 30 s each. The organic and aqueous phases were separated in a Beckman clinical centrifuge at 3750 rpm, 15 min, 25°C. The organic phase was dried under nitrogen and then suspended in 50 l of the solvent system (80:20:1; petroleum ether/ethyl ether/acetic acid). Pure cholesterol was dissolved in the solvent system and used as a standard (5 g/spot). Lipids were visualized by charring with sulfuric acid-ethanol and heating at 180°C for 10 min. Unlabeled cholesterol was added to each fraction to facilitate visualization. The appropriate spots were scraped, and the amount of radiation was quantified by liquid scintillation counting.
Caveolae Isolation-Caveolae were isolated from mouse endothelium as described by Schnitzer and et al. (24 -26) with the modifications described below. A mouse was anesthetized with ketamine/xylazine (see above), then a vertical incision was made into the abdomen. A catheter was placed in the isolated descending aorta just distal to the renal artery and vein. To allow the perfusion to flow into the abdominal cavity, a small incision is made with scissors into the caudal vena cava. The perfusate went into the aorta, through the femoral arteries and capillaries of each leg, and then returned to the abdomen through the femoral veins and the caudal vena cava.
An infusion pump was used to perfuse 10 ml of each of the following solutions (2 ml/min): 1) 0.2 g/ml nitroprusside in Ringers (Sigma K-4002) at room temperature, 2) 0. The vessels were then transferred to a large glass tissue grinder, suspended in 1.7 ml of sucrose/HEPES/KCl/protease inhibitors, and homogenized with 20 strokes on ice, taking care to not grind bead complexes. 2.3 ml of 90% Nycodenz was mixed with the homogenized vessels and overlaid onto a linear 55-70% Nycodenz gradient. The preparation was then overlaid with 0.5 ml of sucrose/KCl/HEPES/protease inhibitors and centrifuged at 38,580 ϫ g in an SW 41 swinging bucket rotor for 30 min at 4°C. The endothelial plasma membrane-bead complex was collected from the bottom of the tube by aspirating the supernatant and suspending the pellet in 2.0 ml of Tricine buffer (0.25 M sucrose, 1 mM EDTA, 20 mM Tricine, pH 7.8). The caveolae/rafts were then isolated from the plasma membranes as we previously described (23,27). Caveolae were then separated from lipid rafts by immunoisolation with caveolin IgG as described (26,28,29).
SDS-PAGE and Immunoblotting-Samples were concentrated by trichloroacetic acid precipitation and washed in acetone. Pellets were suspended in sample buffer that contained 1.2% (v/v) ␤-mercaptoethanol and heated at 95°C for 3 min before being loaded onto gels. Proteins were separated in a 12.5% SDS-polyacrylamide gel using the method of Laemmli (30). The separated proteins were then transferred to PVDF. The PVDF was blocked in TBS that contained 5% dry milk for 1 h at room temperature. Primary antibodies were diluted in TBS that contained 1% dry milk and incubated with the PVDF for 1 h at room temperature. The PVDF was washed 4 times, 10 min each in TBS plus 1% dry milk. The secondary antibodies (all conjugated to horseradish peroxidase) were diluted 1/20,000 in TBS plus 1% dry milk and incubated with the PVDF for 1 h at room temperature. The PVDF was then washed, and the bands were visualized by chemiluminescence.
Nitric-oxide Synthase Assay-NOS activation was determined in intact cells or vessels as described (31,32). Briefly, the cells were plated into 12-well plates at 5,000 cells/well and grown to 60% confluency, or one vessel was placed per well. The cells were treated as described under "Results," and the measurement of the enzymatic activity of eNOS initiated by adding 0.75 Ci/ml L-[ 3 H]arginine/well. The NOS reaction was terminated by adding 500 l of ice-cold 1 N trichloroacetic acid to each well. The cells were freeze-fractured twice in liquid nitrogen for 2 min with thawing at 37°C for 5 min and then scraped with a rubber spatula. The contents of each well were transferred to ice-cold glass test tubes. Ether extraction was performed 3 times with watersaturated ether to remove the trichloroacetic acid. The samples were neutralized with 1.5 ml of 25 mM HEPES, pH 8, applied to Dowex AG50WX-8 (Tris form) columns, and eluted with 1 ml of 40 mM HEPES buffer, pH 5.5, containing 2 mM EDTA and 2 mM EGTA. L-[ 3 H]Citrulline was collected in scintillation vials and quantified by liquid scintillation counting. In individual experiments performed in 12-well plates, 4 wells were used for each treatment group. Findings were confirmed in at least 3 independent experiments. NOS activation in the intact cells was completely inhibited by 2 mM L-NNA.
Statistics-Least squares analysis of variance was used to evaluate the data with respect to sample, treatment, time, and their interaction using the analysis of variance procedure of Stastica. When appropriate, fractions were compared within a given time using the Tukey's honestly significant difference test. The means were considered different at p Ͻ 0.01.

RESULTS
Acetylcholine-induced Vasodilation-To test the hypothesis that CD36-deficient mice are protected from hypercholesterolemia-induced vascular dysfunction, we used age and sexmatched mice fed either a chow or high fat diet for 6 weeks. The animals used in these studies were in a C57BL/6 background strain and consisted of C57BL/6, apoE null, CD36 null, and apoE/CD36 null. At the end of the study the animals were acutely prepared to measure basal and agonist-modified blood pressure, and plasma was collected for the determination of total cholesterol levels. C57BL/6, apoE null, CD36 null, and CD36/apoE null mice fed a chow diet and C57BL/6 and CD36 null mice fed a high fat diet had similar plasma cholesterol levels (322 Ϯ 39 mg/dl). In contrast, apoE null and CD36/apoE null mice fed a high fat diet had significantly increased cholesterol levels, and importantly, similar total plasma cholesterol levels (842 Ϯ 48 and 862 Ϯ 57 mg/dl, respectively).
The basal, resting blood pressure for all of the animals regardless of being fed a chow or high fat diet was 84 Ϯ 5 mm of Hg. Infusion of 15 g/kg of sodium nitroprusside induced a transient decrease in blood pressure of 26 Ϯ 8 mm Hg in all of the study groups, indicating that the vessels were capable of dilating (Fig. 1, A and B). As expected, the nitric-oxide synthase (NOS) inhibitor, L-NNA, did not inhibit the response to sodium nitroprusside because sodium nitroprusside chemically breaks down to nitric oxide independent of NOS enzymatic activity. All of the chow-fed animals and C57BL/6 and CD36 null mice fed a high fat diet had a transient decrease in blood pressure of 24 Ϯ 9 mm of Hg after infusion with 1 g/kg of acetylcholine (Fig. 1, A and B). In striking contrast, apoE null mice fed a high fat diet did not response to acetylcholine infusion with a decrease in blood pressure (Fig. 1B). Importantly, apoE/CD36 null mice fed a high fat diet responded to acetylcholine with a transient decrease in blood pressure similar to control animals (Fig. 1B). Infusion of L-NNA completely inhibited the acetylcholine-induced decrease in blood pressure. Similar data were obtained when 1 g/kg bradykinin was used to induce vasodilation (data not shown).
eNOS Re-localization-We next began to dissect the molecular mechanism responsible for 1) the lack of a response to acetylcholine in apoE null mice fed a high fat diet and 2) the protective effect afforded by CD36 deficiency. The data presented above indicated that acetylcholine infusion induces vasodilation in a NOS-dependent manner. Thus, one possible explanation could be the loss of eNOS in high fat-fed apoE null mice but not in high fat-fed apoE/CD36 null mice. To test this possibility femoral artery vessels were isolated from similar groups of animals as those used for Fig. 1 and processed to generate protein lysates. The lysates were then resolved by SDS-PAGE and immunoblotted with eNOS IgG. Because eNOS is not expressed in other vascular cells, the immunoblot signal can be associated with endothelial cells. Both apoE null mice and apoE/CD36 null mice fed a chow or high fat diet contained similar amounts of eNOS in the whole vessel ( Fig. 2A, Whole  Vessel).
Because caveolae can influence eNOS activity (19) and because the total immunodetectable eNOS in the vessel was not altered, we next determined if the subcellular localization of eNOS changed. To determine whether eNOS was localized to endothelial caveolae in apoE null and apoE/CD36 null mice fed a high fat diet, we used a published method to isolate caveolae in vivo (24,25). In brief, silica beads were perfused through the femoral artery and then cross-linked before extracting the vessel. Endothelial plasma membranes were then isolated, and caveolae were obtained by differential density gradient centrifugation followed by immunoisolation with caveolin IgG (26,29,31). The proteins associated with caveolae were resolved by SDS-PAGE and immunoblotted with eNOS ( Fig. 2A) and caveolin (Fig. 2B) IgG. Chow-fed apoE null and apoE/CD36 null mice along with apoE/CD36 null mice fed a high fat diet contained eNOS in caveolae. In contrast, apoE null mice fed a high fat diet did not have any immunodetectable eNOS associated with caveolae despite the fact that the vessels contain similar amounts of eNOS.
Although the data suggested that the inability of acetylcholine to induce vasodilation in apoE null mice fed a high fat diet was due to the mis-localization of eNOS, it was also possible that some other component of the signaling pathway was missing. To address this possibility we also examined the total vessel levels and caveola levels of CD36 (Fig. 2C), acetylcholine receptors (Fig. 2D), and hetero-trimeric G protein (Gq) (Fig.  2E). Fig. 2 demonstrates that CD36 was present in apoE null mice but not apoE/CD36 null mice, whereas acetylcholine receptors and Gq were enriched in caveolae isolated from all of the animals.
Numerous groups have reported that the in vitro and phar-FIG. 1. Effects of nitroprusside and acetylcholine on blood pressure. C57BL/6, apoE, CD36, and CD36/apoE mice were fed either a chow (A) or high fat diet (B) for 6 weeks (n ϭ 11-15/group). At the end of the study the animals were acutely prepared to measure blood pressure and then infused with 15 g/kg sodium nitroprusside and 1 g/kg acetylcholine. Where indicated, the animals were perfused with 2 mM L-NNA 5 min before the infusion of sodium nitroprusside and acetylcholine. The basal, resting blood pressure for all of the animals regardless of being fed a chow or high fat diet was 84 Ϯ 5 mm of Hg. Preliminary studies with different concentrations of sodium nitroprusside and acetylcholine demonstrated that the amounts used here were sufficient for a maximal response (data not shown). Results reflect the mean Ϯ S.E., n ϭ 11-15; *, p Ͻ 0.01 with respect to C57BL/6 in each group.

FIG. 2. Immunoblots of proteins associated with whole vessels and endothelial caveolae.
Entire femoral arteries were isolated from similar groups of animals as described in Fig. 1 and processed to generate protein lysates (Whole Vessel). Alternatively, an established procedure was used to isolate caveolae from the femoral arteries of similar sets of animals (24,25). The caveolae were further purified by immunoisolation (26,29,31). 20 g of the whole vessel protein lysate and 20 g of caveolae were resolved by SDS-PAGE and immunoblotted with IgGs for eNOS, caveolin, CD36, acetylcholine receptor, and Gq. Because equal amounts of proteins were loaded in each lane, the greater intensity associated with the caveolae lanes indicate that the protein was enriched in caveolae compared with whole vessels. In addition, the similar intensities in the whole vessel lanes indicate that the vessels contained similar amounts of the immunoblotted protein regardless of localization to caveolae. The presented data are from two animals (one for whole vessels and one for caveolae) and is representative of vessels and caveolae isolated from eight animals in each group. E, apoE mice; E/36, apoE/CD36 mice.

CD36-dependent Vascular Dysfunction
macological depletion of caveola cholesterol results in the disruption of caveola-mediated signal transduction (34 -38). However, it has never been demonstrated that depletion of caveola cholesterol is involved in eNOS-mediated signaling in a physiological relevant model of disease. Therefore, to determine whether a decrease in caveola cholesterol may be responsible for the movement of eNOS out of caveolae, we used the silica bead method (24, 25) followed by immunoisolation (26,29,31) to isolated in vivo caveolae from each group of animals and quantified the amount of total cholesterol by gas chromatography (39). Caveolae isolated from the vessels of chow and high fat-fed animals had similar amounts of cholesterol with the exception of caveolae isolated from apoE null mice fed a high fat diet. Caveolae isolated from apoE null mice fed a high fat diet were depleted of cholesterol (Fig. 3).
Reversibility of Hypercholesterolemia-induced eNOS Dysfunction-The data presented thus far demonstrate that the lack of CD36 protects apoE null mice fed a high fat diet from an eNOS-dependent vascular dysfunction. One of the most striking differences between apoE null mice fed a chow and high fat diet is that high fat-fed mice are hypercholesterolemic. If hypercholesterolemia is responsible for CD36-dependent eNOS dysfunction, then a return of plasma cholesterol levels to normal should restore vascular function; however, apoE null mice fed a high fat diet will remain hypercholesterolemic even when placed on a chow diet (40). Therefore, we conducted the experiment in situ by isolating femoral arteries from apoE null and apoE/CD36 null mice fed chow or high fat diets. Plasma was also isolated from apoE null (Fig. 4A) and apoE/CD36 null (Fig.  4B) mice fed chow or high fat (hypercholesterolemic) diets. Arteries isolated from apoE null and apoE/CD36 null mice fed chow or high fat diets were incubated with chow plasma or high fat plasma for 16 h then processed to determine the amount of eNOS present in the vessel (V) and the amount of eNOS associated with caveolae (C). Consistent with the in vivo studies (Fig. 2), the total amount of eNOS present in the vessels did not change regardless of the treatment. The caveolae in apoE null chow vessels incubated with apoE null chow (Fig. 4A) or apoE/ CD36 null chow (Fig. 4B) plasma were enriched in eNOS. In contrast, when apoE null chow vessels were incubated with apoE null (Fig. 4A) or apoE/CD36 null (Fig. 4B) high fat plasma, eNOS no longer associated with caveolae. Caveolae isolated from apoE null high fat vessels maintained in apoE null or apoE/CD36 null high fat plasma did not contain eNOS. However, incubation of apoE null high fat vessels with apoE null or apoE/CD36 null chow plasma allowed eNOS to associate with caveolae. Importantly, apoE/CD36 null vessels were unaffected by any of the treatments.
We next used the in situ vessel system to determine whether the localization of eNOS to caveolae was required for agonist activation of eNOS. Vessels isolated from apoE/CD36 null mice fed a chow or high fat diet and incubated with apoE null (Fig.  5A) or apoE/CD36 null (Fig. 5B) chow or high fat plasma maintained the ability to maximally stimulate eNOS in response to acetylcholine (Fig. 5). The maximal eNOS response was determined by incubating the appropriate vessels in 2 g/ml of ionomycin, a calcium ionophore. ApoE null chow vessels maintained in apoE null or apoE/CD36 null chow plasma generated nitric oxide in response to acetylcholine; however, similar vessels maintained in high fat plasma did not generate nitric oxide. ApoE null high fat vessels maintained in high fat plasma were unable to produce nitric oxide in response to acetylcholine. In contrast, when apoE null high fat vessels were placed in apoE null or apoE/CD36 null chow plasma, the ability to respond to acetylcholine was restored. Importantly, all of the vessels responded similarly to ionomycin, which indicated that the vessels contain functional eNOS even if agonists did not activate the enzyme.
We next determined if the amount of cholesterol associated with caveolae could be altered by incubating vessels in chow or high fat plasma (Fig. 6). The amount of cholesterol associated with caveolae isolated from apoE/CD36 null vessels was not affected by any of the treatments. In contrast, caveolae in vessels isolated from chow-fed apoE null mice were depleted of cholesterol after incubation in apoE null (Fig. 6A) or apoE/   FIG. 3. Quantification of caveola cholesterol. An established procedure was used to isolate caveolae from similar sets of animals as described in Fig. 1 (24,25). Caveolae were further purified by immunoisolation (26,29,31). The internal standard, cholesteryl heptadecanoate was added to the isolated caveolae, and the material was extracted with isopropanol-hexane and derivatized with N,O-bis(trimethylsilyl) trifluoroacetamide. The samples were then resolved and quantified by gas chromatography. Results reflect the mean Ϯ S.E., n ϭ 6; *, p Ͻ 0.01 with respect to C57BL/6 in each group. CD36 null (Fig. 6B) high fat plasma. Caveolae isolated from apoE null high fat vessels incubated with high fat plasma were depleted of cholesterol. In contrast, incubation of apoE null high fat vessels with apoE null or apoE/CD36 null chow plasma returned the amount of cholesterol associated with caveolae to control levels.
LDL Mediates the Vascular Dysfunction-The data thus far suggested that a plasma component was responsible for the depletion of caveola cholesterol, the re-localization of eNOS, and the lack of nitric oxide production. To further determine which plasma component was responsible for these effects, plasma from chow and high fat-fed apoE null (Fig. 7A) and apoE/CD36 null (Fig. 7B) mice were separated into HDL, LDL, VLDL, and LPDP fractions. LPDP (100%) or LPDP plus a single lipoprotein component (10 g/ml) were incubated with isolated apoE null chow vessels for 16 h. At the end of the incubation, acetylcholine-stimulated eNOS activity was measured on the intact vessels. Fig. 7 demonstrates that LPDS (chow or high fat)-treated vessels produced as much nitric oxide in response to acetylcholine as vessels treated with ionomycin. The HDL and VLDL fractions from either chow or high fat plasma did not affect agonist activation of eNOS. The LDL fraction from chow-fed mice did not inhibit eNOS activation; however, the LDL fraction from high fat-fed mice completely inhibited acetylcholine-induced eNOS activity.
The data presented thus far suggest that the removal of cholesterol from caveolae promotes the inhibition of eNOS stimulation. To determine whether LDL isolated from high fat-fed mice is capable of depleting caveolae of cholesterol, vessels isolated from chow-fed apoE null mice were incubated for 16 h in chow plasma containing [ 3 H]acetate to label the cellular sterol pools. The cells were washed and incubated for 16 h in LPDP only or LPDP plus HDL, LDL, or VLDL (10 g/ml) from chow-fed or high fat-fed apoE null mice. After the incubation, the medium containing the lipoproteins were collected, the lipids were extracted, and cholesterol was separated by thin-layer chromatography and quantified by liquid scintillation spectroscopy. High fat LDL contained a large amount of radiolabeled cholesterol, whereas the chow LDL did not contain significant amounts of radiolabeled cholesterol (Fig. 8). The HDL, VLDL, and LPDP fractions from either chow or high fat-fed mice did not contain significant amounts of radiolabeled cholesterol.
Cyclodextrin Mimics the Effects of High Fat Plasma/LDL-The data suggest that the LDL in high fat-fed mice deplete caveolae of cholesterol in a CD36-dependent manner and that this depletion of cholesterol promotes the re-localization and subsequent inhibition of eNOS stimulation. To further validate this conclusion we used an independent method to deplete caveolae of cholesterol in intact vessels. We treated vessels isolated from chow-fed apoE null mice with 5 mM cyclodextrin for 2 h and then measured the amount of cholesterol associated with caveolae (Fig. 9A) and agonist-stimulated eNOS activity (Fig. 9B). Cyclodextrin depletes cells of cholesterol and has been widely used as a reagent to disrupt the function of caveo- lae (34 -38). Cyclodextrin treatment depleted caveolae of cholesterol to the same extent as high fat plasma (Fig. 9A). In addition, the cyclodextrin-mediated depletion of caveolae cholesterol correlated with an inability of acetylcholine to stimulate eNOS activity (Fig. 9B). DISCUSSION Genetic linkage studies and in vitro studies suggest that CD36 may be involved in the development of hypertension; however, the direct involvement of CD36 has not been demonstrated (6,8 -10,41). In the present studies, we used CD36 null mice and apoE/CD36 null mice to determine whether 1) the absence of CD36 affected blood pressure or blood pressure regulation and 2) if the absence of CD36 in a hypercholesterolemic environment affected blood pressure or blood pressure regulation. The data demonstrate that the absence of CD36 in mice fed a chow or high fat diet does not affect resting blood pressure or the ability to respond to physiological vasodilatory agonists such as acetylcholine and bradykinin. CD36 null mice have similar levels of serum cholesterol as C57BL/6 mice. In contrast, apoE null mice and apoE/CD36 null mice had slightly evaluated serum cholesterol levels when fed a chow diet, and the animals became hypercholesterolemic when fed a high fat diet. In this pro-cardiovascular disease environment apoE null mice lost the ability to vasodilate in response to acetylcholine, whereas apoE/CD36 null mice still responded to acetylcholine. These data demonstrate that the absence of CD36 protects against an eNOS-dependent vascular dysfunction, or stated FIG. 7. LDL isolated from high fat-fed mice inhibits agoniststimulation of eNOS. ApoE mice were fed a chow diet for 6 weeks, then intact femoral arteries were isolated. In addition, apoE and apoE/ CD36 mice were fed a chow or high fat diet for 6 weeks and used to isolate plasma. The plasma was fractionated into HDL, LDL, VLDL, and LPDP. The vessels were incubated with LPDP or LPDP plus a lipoprotein component (10 g/ml) for 16 h (A, apoE plasma; B, apoE/ CD36 plasma). At the end of the incubation, the vessels were incubated with L-[ 3 H]arginine and 1 M acetylcholine or 2 g/ml ionomycin for 15 min and then processed to quantify the amount of L-[ 3 H]citrulline generated. Results reflect the mean Ϯ S.E., n ϭ 6 (separate vessels and plasma isolations). *, p Ͻ 0.01 with respect to ionomycin values in each group.
FIG. 8. LDL isolated from high fat-fed apoE mice depletes caveolae of cholesterol. ApoE mice were fed a chow diet for 6 weeks, then intact femoral arteries were isolated. In addition, apoE mice were fed a chow or high fat diet for 6 weeks and used to isolate plasma. The plasma was then fractionated into HDL, LDL, VLDL, and LPDP. The vessels were incubated with 10 Ci of L-[ 3 H]acetate and chow plasma for 16 h to label the cells. The vessels were washed and then incubated for 16 h in LPDP only or LPDP plus HDL, LDL, or VLDL (10 g/ml) isolated from chow-fed or high fat-fed apoE mice. After the incubation, the medium containing the lipoproteins was collected, the lipids were extracted, cholesterol was separated by thin-layer chromatography, and the cholesterol was quantified by liquid scintillation. The plasma was isolated from apoE mice; however, identical data were obtained with plasma isolated from apoE/CD36 mice (data not shown). Results reflect the mean Ϯ S.E., n ϭ 6 (separate vessels and plasma isolations), *, p Ͻ 0.01 with respect to LPDP.
FIG. 9. Cyclodextrin mimics the action of LDL isolated from high fat-fed mice. Femoral artery vessels from chow-fed apoE mice were incubated with plasma isolated from chow or high fat-fed apoE mice for 16 h. In addition, similar vessels were incubated with chow plasma containing 5 mM cyclodextrin for 2 h. A, at the end of the incubation, caveolae were isolated, and the associated cholesterol was quantified by gas chromatography. B, at the end of the incubation, the vessels were incubated with L-[ 3 H]arginine and 1 M acetylcholine for 15 min and then processed to quantify the amount of L-[ 3 H]citrulline. Results reflect the mean Ϯ S.E., n ϭ 4 (separate vessels and plasma isolations). *, p Ͻ 0.01 with respect to chow. differently, in a hypercholesterolemic environment, CD36 promotes an eNOS-mediated vascular dysfunction.
The in vivo data clearly demonstrate that CD36 mediates an eNOS-dependent vascular dysfunction in a hypercholesterolemic environment. Earlier work by our laboratory demonstrated that copper-oxidized LDL can efflux cholesterol from caveolae and promote the translocation of eNOS out of caveolae (15,16). Extracellular agonists could not stimulate eNOS once the enzyme was no longer associated with caveolae (15,16). A major limitation with these earlier studies is that copper-oxidized LDL is not a physiological ligand, and the observed cholesterol efflux may have been the result of an in vitro artifact. By using an in vivo method to isolate endothelial plasma membranes, we now demonstrate that a high fat diet in apoE null mice promotes the translocation of eNOS away from caveolae, whereas apoE/CD36 null mice fed the same high fat diet retained eNOS in caveolae. Importantly, the diets did not affect the overall level of eNOS in the vessel but only the subcellular location of the enzyme. In addition, other possible explanations for the lack of a response to acetylcholine were ruled out, such as the level of acetylcholine receptors and hetero-trimeric G proteins. Overall, the data suggest that the re-localization of eNOS was responsible for the lack of agonist-induced eNOS activity.
Numerous factors can influence the subcellular localization of eNOS including, acylation, phosphorylation, interaction with caveolin, and caveola cholesterol levels (20,42,43). Endothelial nitric-oxide synthase has been reported to be associated with both caveolae and the Golgi (44). The relationship between caveola-localized eNOS and Golgi-localized eNOS is unclear. However, recent data suggest that caveolae are assembled or begin to assemble in the Golgi (45)(46)(47), and it is possible that Golgi-localized eNOS in fact associates with nascent caveolae. The phosphorylation of eNOS and the de-acylation of eNOS in response to agonists have been demonstrated to cause eNOS to translocate to the cytosol (19, 48 -50). The data to date demonstrate that eNOS can be active while associated with caveolae, the cytosol, and the Golgi; however, the regulatory mechanisms responsible for controlling eNOS localization and the activation state of eNOS are not completely understood (19, 48 -50). Previous in vitro studies demonstrated that copper-oxidized LDL binding to CD36 did not alter the acylation or phosphorylation state of eNOS nor did it alter the total level of caveolin associated with caveolae (15). Consistent with the earlier in vitro studies, we did not detect phosphorylation of eNOS by using commercially available phospho-eNOS-specific antibodies (data not shown). However, the in vitro studies demonstrated that, in endothelial cells, CD36 mediates the efflux of caveolae cholesterol to copper-oxidized LDL. The depletion of caveola cholesterol caused eNOS to translocate to an unidentified intracellular compartment where it could not be stimulated by agonists (15). In the current study, we demonstrated for the first time that apoE null mice fed a high fat diet lack an enrichment of cholesterol in the isolated caveolae fraction. However, caveolae isolated from apoE/CD36 null mice fed a high fat diet were highly enriched in cholesterol. These data suggest that the absence of CD36 protects mice from a vascular dysfunction by protecting caveolae from the lost of cholesterol. The ability of CD36 to efflux cholesterol from endothelial cells is unique because CD36 in macrophage is generally thought of as a mechanism for the net uptake of sterol. However, endothelial cells are net exporters of cholesterol and do not accumulate large amounts of sterol, whereas macrophages have the capability to internalize and store sterols (51). The molecular mechanism of how CD36 mediates the efflux of caveola cholesterol is unclear; however, other investigators demonstrate that the protein caveolin in required for the net efflux of cholesterol from caveolae to extracellular acceptors (52,53). It is also important to note that CD36 only mediated an efflux of caveola cholesterol in mice that were fed a high fat diet.
The data implicated a plasma component in the hypercholesterolemic mice as the causative factor for the vascular dysfunction. If this was the case then we reasoned that plasma isolated from high fat-fed mice should be able to inhibit agonist-stimulated eNOS activity in a vessel isolated from a chowfed apoE mouse. Furthermore, we wanted to determine whether it was possible to restore a dysfunctional vessel to a "normal" vessel; therefore, we incubated high fat vessels with plasma isolated from chow-fed mice. High fat plasma isolated from apoE null or apoE/CD36 null mice when added to vessels isolated from chow-fed mice promoted the loss of eNOS from caveolae, inhibition of agonist-stimulated eNOS, and depletion of caveola cholesterol. Importantly, the incubation of vessels isolated from high fat-fed mice (dysfunctional) with chow plasma resulted in a return of eNOS to caveolae, a return of NOS activity, and a return of caveola cholesterol. Fractionation of the plasma demonstrated that only the LDL fraction from high fat-fed animals was capable of inhibiting eNOS stimulation and depleting cells of cholesterol. These data illustrate several important concepts. First, the fact that LDL from both apoE null and apoE/CD36 null mice caused the dysfunction indicated that the dysfunction was not due to some difference between the plasma in the two types of mice. Second, the dysfunction appeared to be caused by the depletion of caveola cholesterol, which resulted in the re-localization of eNOS. Third, because the lipoproteins were added in the presence of 100% lipoprotein-deficient plasma and control LDL from chowfed animals did not have an effect, it seems unlikely that the data were caused by modification of the LDL after isolation. The data cannot conclusively rule out the possibility that a trace component of the LDL fraction and not LDL itself was responsible for eNOS inactivation. However, the data do demonstrate that the inability to stimulate eNOS in a hypercholesterolemic environment is CD36-dependent.
The current study is the first demonstration that CD36 plays a direct role in the development of an eNOS-mediated vascular dysfunction. The data also suggest that the mechanism of CD36-mediated dysregulation is the disruption of caveolae and the re-localization of eNOS away from caveolae. Another important finding is that eNOS dysregulation is reversible; that is, when hypercholesterolemic plasma is replaced with normal plasma, the dysfunctional vessels can regain function. These findings suggest that lowering plasma cholesterol levels in patients may not only effect atherosclerotic lesions but also have direct effects on endothelium vasoreactivity. In addition, the demonstration of CD36 involvement in this process provides a potentially useful therapeutic target for maintaining vascular reactivity in hypercholesterolemic patients.