Inability of HDL from abdominally obese subjects to counteract the inhibitory effect of oxidized LDL on vasorelaxation.

Abdominal obesity is associated with a decreased plasma concentration of HDL cholesterol and with qualitative modifications of HDL, such as triglyceride enrichment. Our aim was to determine, in isolated aorta rings, whether HDL from obese subjects can counteract the inhibitory effect of oxidized low density lipoprotein (OxLDL) on endothelium-dependent vasodilation as efficiently as HDL from normolipidemic, lean subjects. Plasma triglycerides were 74% higher (P < 0.005) in obese subjects compared with controls, and apolipoprotein A-I (apoA-I) and HDL cholesterol concentrations were 12% and 17% lower (P < 0.05), respectively. HDL from control subjects significantly reduced the inhibitory effect of OxLDL on vasodilation [maximal relaxation (Emax) = 82.1 ± 8.6% vs. 54.1 ± 8.1%; P < 0.0001], but HDL from obese subjects had no effect (Emax = 47.2 ± 12.5% vs. 54.1 ± 8.1%; NS). In HDL from abdominally obese subjects compared with HDL from controls, the apoA-I content was 12% lower (P < 0.05) and the triglyceride-to-cholesteryl ester ratio was 36% higher (P = 0.08)). Emax(OxLDL + HDL) was correlated with HDL apoA-I content and triglyceride-to-cholesteryl ester ratio (r = 0.36 and r = −0.38, respectively; P < 0.05). We conclude that in abdominally obese subjects, the ability of HDL to counteract the inhibitory effect of OxLDL on vascular relaxation is impaired. This could contribute to the increased cardiovascular risk observed in these subjects.

HDL particles have multiple antiatherosclerosis properties. Among them is the beneficial effect they exert on endothelium-dependent vasorelaxation. Indeed, HDL particles can counteract the inhibitory effect of OxLDL on endothelium-dependent vasodilation in isolated rabbit aorta rings (7). Moreover, the intravenous administration of reconstituted HDL either to hypercholesterolemic men or to ATP binding cassette A1-deficient heterozygotes, both with an initially impaired endothelium-dependent vasorelaxation, restored a normal vasorelaxation in these subjects (8,9).
Endothelium dysfunction, a predictor for cardiovascular events, has been observed in overweight patients with visceral obesity and was demonstrated to be associated with body fat distribution (10)(11)(12). Because of qualitative modifications, HDL from these subjects may be less protective against impairments of endothelium-dependent vasorelaxation than HDL from lean normolipidemic subjects. Therefore, the purpose of our study was to determine whether or not HDL from abdominally obese nondiabetic patients could protect the vessel from the inhibitory effect of OxLDL on vasorelaxation.

Subjects
Nineteen nondiabetic abdominally obese patients and 16 lean normolipidemic and normoglycemic control subjects were included in this study. This study was approved by the Ethics Committee of Dijon University Hospital, and written informed consent was obtained from each subject before the study.
Obese and control subjects were not taking any antioxidant or any drug that could affect lipid metabolism. Controls were included in our study if they had a waist circumference of ,88 cm for women and ,102 cm for men, fasting glycemia , 110 mg/dl, serum triglycerides , 150 mg/dl, and HDL cholesterol of .40 mg/dl for men and .50 mg/dl for women. Obese subjects were included if they had fasting glycemia , 110 mg/dl and waist circumference of .88 cm for women and .102 cm for men.

Preparation of lipoproteins
The different lipoprotein subclasses were isolated from serum or plasma by sequential flotation ultracentrifugation according to their density. The LDLs (1.019 , d , 1.063 g/ml) used to form OxLDL were prepared in a 70.Ti rotor on an XL-90 ultracentrifuge (Beckman, Palo Alto, CA) by sequential flotation ultracentrifugation from 170 ml of fresh citrated plasma drawn from normolipidemic donors. Densities of plasma were adjusted by the addition of KBr. After a 15 h run at 50,000 rpm, the top fraction of the sample containing lipoproteins with d , 1.019 g/ml was removed. Finally, LDLs were prepared from the infranatant adjusted to a 1.063 g/ml density by a 24 h run at 45,000 rpm. HDLs (1.063 , d , 1.21 g/ml) of control and obese subjects were prepared from 20 ml of EDTA plasma in a 70.Ti rotor on an XL-90 ultracentrifuge (Beckman). After a 20 h run at 50,000 rpm, the top fraction of the sample containing lipoproteins with d , 1.063 g/ml was removed. Finally, HDLs were prepared from the infranatant adjusted to a 1.21 g/ml density by a 24 h run at 50,000 rpm. The different lipoprotein fractions were dialyzed overnight against TBS (10 mmol/l Tris and 150 mmol/l NaCl, pH 7.4). To avoid oxidation, native lipoproteins were supplemented with EDTA (final concentration, 200 mmol/l) and butylated hydroxytoluene (final concentration, 20 mmol/l) and kept at 4jC.

Oxidative modification of LDL
The oxidation of LDL was performed by incubating freshly prepared LDL, adjusted to a 1.2 g/l final concentration of proteins in TBS, with a copper sulfate solution (final concentration, 5 mmol/l) for 24 h at 37jC. At the end, oxidation was stopped by the addition of EDTA (final concentration, 200 mmol/l) and butylated hydroxytoluene (final concentration, 20 mmol/l), and OxLDLs were kept at 4jC (13).

Serum and HDL biochemical measurements
Glycemia was measured on a Vitros analyzer (Ortho Clinical Diagnostics, Rochester, NY). Total cholesterol and HDL cholesterol, triglycerides, proteins, apolipoprotein A-I (apoA-I), and apoB were quantitated on a Dimension analyzer with dedicated reagents (Dade Behring, Newark, DE). LDL cholesterol was calculated using the Friedewald formula. Free cholesterol and phospholipids were measured on the same analyzer with Wako reagents (Neuss, Germany). Insulin was quantitated by radioimmunoassay (CisBio International, Gif sur Yvette, France).

Insulin resistance evaluation
The insulin resistance level was estimated using the homeostasis model assessment (HOMA) method (14). HOMA was cal-culated with the following formula: HOMA 5 glucose 3 insulin/ 22.5, where glucose is the fasting glucose millimolar concentration and insulin is the fasting insulin concentration expressed in milliunits per liter.

Measurement of paraoxonase activity
Paraoxonase (PON) activity was measured as described in fresh HDL samples within 2 days of isolation (15). It was assayed in tubes containing Tris-HCl buffer (150 mmol/l NaCl and 1 mmol/l CaCl 2 , pH 7.4) and 100 mg of proteins of HDL (final volume, 800 ml). The reaction was initiated by adding paraoxon (1 mmol/l), and the absorbance was monitored at 412 nm for 1 h. The amount of 4-p-nitrophenol formed was calculated from the molar extinction coefficient (12,800 mol 21 cm 21 ). PON activity was expressed as nmol of 4-p-nitrophenol formed per minute per milligram of apoA-I under the experimental conditions.

Vasoreactivity on rabbit aorta rings
Vasoreactivity experiments were performed on rabbit (Charles River, l'Arbresle, France) aorta rings as described previously (13). Our protocol was approved by the local ethics committee for animals. Aortic rings were first contracted by 0.3 mmol/l arterenol hydrochloride, a concentration giving 75% of the maximal contraction. Then, the rings were relaxed by cumulative additions of acetylcholine in the 1 nmol/l to 0.01 mmol/l concentration range. After washout and a 30 min recovery period, aortic rings were incubated for 2 h with either OxLDL or OxLDL 1 HDL. The concentration of proteins in the organ bath (native or oxidized lipoproteins) was constantly 1 g/l for each lipoprotein subclass, as used in many previous studies conducted with rabbit aortic rings (7,13,16). At the end of the incubation period, aortic segments were again contracted with arterenol hydrochloride and progressively relaxed with acetylcholine or sodium nitroprusside. Sodium nitroprussiate, a nitric oxide (NO) donor, was used to check that the defects in vasodilation we observed were endothelium-dependent.
The maximal relaxation (E max ) induced by acetylcholine and expressed as a percentage of the contraction to arterenol hydrochloride (0.3 mmol/l) was determined from experimental data. To limit variations in the results attributable to intrinsic properties of the vessels, experiments were performed in quadruplicate on aorta rings from four different rabbits and the results were averaged.

Statistical procedures
Statistical analysis was performed using the Statview statistical software. Data are expressed as means 6 SD. The comparison of quantitative data was performed with the Mann-Whitney U-test. Correlation coefficients for the whole population were determined by linear regression analysis. We also performed a multiple regression analysis in which the dependent variable was the maximal vasorelaxation and the independent variables were the HDL triglyceride-to-cholesteryl ester ratio and the HDL apoA-I percentage. P , 0.05 was considered statistically significant.

Clinical and biochemical characteristics
The clinical and biochemical characteristics of obese and control subjects are shown in Table 1. No statistically significant differences between obese individuals and controls were observed for total cholesterol and LDL cholesterol. The plasma triglyceride level was 41% higher (P , 0.005) in the obese group compared with the control group, and plasma apoA-I and HDL cholesterol were 13% and 17% lower, respectively (P , 0.05). Table 2 illustrates the composition of HDL in the two groups of subjects. Compared with HDL from lean control subjects, HDL from obese patients displayed a 42% higher triglyceride content (P , 0.05) and a 12% decrease in apoA-I content (P , 0.001). The HDL triglycerideto-cholesteryl ester ratio was 36% higher in obese patients compared with controls, with borderline significance (P 5 0.08). No significant difference was observed between the two groups for PON activity.

HDL composition
OxLDL influence on endothelium-dependent vasorelaxation As described previously, preincubation of isolated arterial segments with OxLDL inhibited the endothelium-dependent relaxation induced by acetylcholine: the E max to acetylcholine decreased from 101.2 6 6.9% in segments incubated with Krebs solution to 54.1 6 8.1% in segments treated with OxLDL (P , 0.0001) (Fig. 1).

Ability of HDL to counteract the inhibitory effect of OxLDL on endothelium-dependent vasodilation
In control subjects, the E max of rabbit aortic rings was significantly greater after preincubation with OxLDL 1 HDL than after preincubation with OxLDL alone (E max 5 82.1 6 8.6% vs. 54.1 6 8.1%; P , 0.0001). On the contrary, the addition of HDL from obese patients to OxLDL did not modify the E max compared with incubation with OxLDL (E max 5 47.3 6 12.5% vs. 54.1 6 8.1%; NS) ( Fig. 1).
With HDL alone, E max was 94.8 6 4.7% in control subjects and 92.0 6 7.2% in obese patients (difference not significant). In all experiments, sodium nitroprussiatemediated endothelium-independent relaxation was not modified (data not shown).

Correlations
The E max obtained after coincubation of the vessels with OxLDL and HDL [E max (OxLDL 1 HDL)] was correlated positively with the percentage of apoA-I in HDL (r 5 0.36, P , 0.05) and negatively with the HDL triglyceride-tocholesteryl ester ratio (r 5 20.38, P , 0.05) (Fig. 2). The correlation between E max (OxLDL 1 HDL) and HDL triglycerides was borderline significant (r 5 0.30, P 5 0.09). E max (OxLDL 1 HDL) was not correlated with the percentage of HDL free cholesterol, phospholipids, total proteins, or cholesteryl esters. In multiple regression analysis including HDL apoA-I percentage and HDL triglyceride-to-cholesteryl ester ratio in the model, both  variables were associated with E max (OxLDL 1 HDL) (r 5 0.41, P , 0.05 and r 5 20.42, P , 0.01, respectively).

DISCUSSION
In this study, we have demonstrated that, contrary to HDL from lean control subjects, HDL from abdominally obese patients is unable to counteract the inhibitory effect of OxLDL on endothelium-dependent vasodilation. Moreover, we have shown that both the triglyceride enrichment of HDL particles and the reduction in HDL apoA-I observed in abdominally obese patients are associated with the decrease in the vasorelaxant effect of HDL.
The deleterious effect of OxLDL on vascular reactivity is related mainly to a decreased bioavailability of NO, a powerful vasorelaxant molecule. Indeed, lysophosphatidylcholine, a product of the oxidation of phosphatidylcholine, activates the production of superoxide anion that reacts with NO to produce peroxynitrites, decreasing NO bioavailability (17,18). Moreover, lysophosphatidylcholine can be incorporated into cytoplasmic membranes, inducing the uncoupling of receptors for vasodilatory ligands and G proteins, which decreases NO production (19). Furthermore, OxLDL can deplete caveolae of free cholesterol, leading to the displacement of endothelial nitric oxide synthase (eNOS) to an intracellular compart-ment where it cannot be activated (20). Moreover, OxLDL particles decrease the stability of eNOS mRNA (21).
HDL particles counteract the inhibitory effect of OxLDL on vascular relaxation both by their antioxidant properties and by stimulating the production of NO by endothelial cells. HDLs are removed from LDL lipid peroxides and oxidized derivatives of cholesterol via cholesteryl ester transfer protein (22). PON, platelet-activated factor-acetyl hydrolase (PAF-AH), and LCAT, three HDL-associated enzymes, hydrolyze oxidized cholesterol or hydroperoxides (23). ApoA-I contributes to the removal of oxidized lipids from LDL and promotes the formation of phosphatidylcholine core aldehydes, which are hydrolyzed by PON or PAF-AH (24,25). Furthermore, HDL allows eNOS to stay in caveolae by preventing the depletion of free cholesterol induced by OxLDL (26). As far as NO production is concerned, it can be stimulated by HDL. Indeed, HDL increases the expression of the eNOS gene and upregulates eNOS activity by inducing its phosphorylation through different signaling pathways involving Akt, mitogen-activated protein kinases, and calcium-calmodulin-dependent kinases (27)(28)(29)(30).
The inability of HDL from abdominally obese subjects to counteract the inhibitory effect of OxLDL on endotheliumdependent vasodilation indicates that one or several of the HDL protective effects detailed above is deficient.
Previous works have reported impaired antioxidative activity of HDL in patients with decreased HDL cholesterol concentration and in patients with the metabolic syndrome (31,32). In those studies, the activities of PON, PAF-AH, and LCAT were not decreased, but HDLs were triglyceride-enriched. Our abdominally obese subjects are similar to the patients from those studies in the sense that they presented a low HDL cholesterol concentration and a triglyceride enrichment of HDL. It was recently demonstrated that the replacement of cholesteryl esters by triglycerides in HDL abrogates the HDL-mediated protection against LDL oxidation (32). Enrichment in triglycerides reduces the exposure of apoA-I to the aqueous phase as a result of its increased penetration into the lipid core of HDL and thus could decrease the capacity of apoA-I to transfer oxidized lipids from LDL (33). Thus, a decreased antioxidative capacity of HDL in our patients could contribute to the decreased capacity of HDL to counteract the inhibitory effect of OxLDL on vasorelaxation. In our work, the decreased antioxidative capacity of HDL in obese subjects may be reinforced by the fact that we measured a decreased percentage of apoA-I in HDL. Indeed, as detailed above, apoA-I has its own antioxidative capacity.
The modification of apoA-I conformation induced by HDL triglyceride enrichment could also have consequences on NO production by endothelial cells. Indeed, binding of HDL particles to cell surface receptors, such as scavenger receptor class B type I, implied in the signaling pathways activating eNOS could be impaired (28).
We recently demonstrated that HDL from type 2 diabetic patients was also unable to counteract the inhibitory effect of OxLDL on vascular relaxation and that the HDL triglyceride enrichment appeared to play an important and triglyceride-to-cholesteryl ester ratio (B) and maximal endothelium-dependent relaxation in rabbit aortic rings after preincubation with HDL and OxLDL. role in the decreased vasorelaxant effect of HDL (34). The results of our present study indicate that the loss of the vasorelaxant effect of HDL in type 2 diabetic patients appears at a very early stage, before the development of hyperglycemia, when plasma triglyceride concentration is very moderately increased, in a less important manner than in type 2 diabetes mellitus.
In conclusion, our study demonstrated the inability of HDL from abdominally obese subjects to counteract the inhibitory effect of OxLDL on vascular relaxation. The HDL triglyceride enrichment and the decreased apoA-I content could be major determinants of this abnormality. Thus, in abdominally obese subjects, HDLs appear less atheroprotective for at least two reasons: because their plasma concentration is decreased and because their compositional changes impair their intrinsic antiatherogenic properties, especially their vasorelaxant effect, as we demonstrate here.