Fenofibrate induces HDL-associated PAF-AH but attenuates enzyme activity associated with apoB-containing lipoproteins.

Human plasma platelet-activating factor acetylhydrolase (PAF-AH) is an enzyme associated mainly with the apolipoprotein B (apoB)-containing lipoproteins and primarily with LDL. A small proportion of enzymatic activity is also associated with HDL. Plasma paraoxonase 1 (PON1) is an esterase exclusively associated with HDL. The effect of fenofibrate on PAF-AH and PON1 activities in patients with dyslipidemias of Types IIA, IIB, and IV were studied. Fenofibrate reduced plasma PAF-AH activity in all patient groups. In Type IIA patients, this reduction was mainly due to a fall in enzyme activity associated with the dense LDL subspecies, whereas in Type IIB and Type IV patients, it was due to the decrease in PAF-AH activity associated with both the VLDL+IDL and dense LDL subspecies. Drug therapy in Type IIB and Type IV patients significantly increased the HDL-associated PAF-AH activity due to the increase in enzyme activity associated with the HDL-3c subfraction. Fenofibrate did not affect serum PON1 activities toward paraoxon and phenylacetate in either patient group. The fenofibrate-induced elevation of HDL-associated PAF-AH activity in dyslipidemic patients of Type IIB and Type IV, as well as the reduction in enzyme activity associated with atherogenic apoB-containing lipoproteins in all patient groups, may represent a new and important antiatherogenic effect of this potent lipid-modulating agent.

Platelet-activating factor (PAF) is a potent proinflammatory lipid mediator that is implicated in atherogenesis (1). In plasma, PAF is hydrolyzed and inactivated by PAF-acetylhydrolase (PAF-AH) (EC 3.1.1.47), an enzyme associated mainly with the apolipoprotein B (apoB)-containing lipoproteins and primarily with LDL (2). A small proportion of circulating enzyme activity is also associated with HDL (2,3). PAF-AH exhibits a ␣ / ␤ hydrolase conformation (4) and has broad substrate specificity toward lipid esters containing short acyl chains (5). Thus, PAF-AH can hydrolyze short-chain diacylglycerols, triacylglycerols, and acetylated alkanols, but also displays phospholipase A 1 and A 2 activities, as well as transacetylase activity (5,6). Among them, the Ca 2 ϩ -independent phospholipase A 2 activity of PAF-AH has been principally studied, and thus this enzyme has been denoted as lipoprotein-associated phospholipase A 2 (7). Indeed, PAF-AH has marked preference for phospholipids with short-chain moieties at the sn -2 position and, with the exception of PAF, can hydrolyze proinflammatory and proatherogenic oxidized phospholipids produced by peroxidation of phosphatidylcholines containing an sn-2 polyunsaturated fatty acyl residue (8).
The role of PAF-AH in atherosclerotic disease is controversial. Data from the West of Scotland Coronary Prevention Study (Scotland, 1989(Scotland, -1995 trial suggest that plasma levels of PAF-AH mass, which mainly reflects the LDL-associated enzyme, represent an independent risk factor for coronary artery disease (9). In contrast, recent findings in the Women's Health Study (US, 1992 to present) suggest that plasma PAF-AH is not a strong predictor of cardiovascular risk in apparently healthy middle-aged women over a mean follow-up of 3 years (10). Nonetheless, loss of plasma PAF-AH activity due to a G 994 → T mutation in the PAF-AH gene may constitute a genetic determinant of atherosclerotic disease in the Japanese population (11). Despite conflicting observations concerning the potential relevance of total plasma-and LDL-associated PAF-AH to atherosclerotic disease, several lines of evidence suggest that HDL-associated PAF-AH activity, although present at low levels, may contribute to the antiatherogenic effects of this lipoprotein (12). Thus, adenoviral transfer of human plasma PAF-AH gene in apoE Ϫ / Ϫ mice significantly reduced macrophage adhesion and homing (13), and inhibited injury-induced neointima formation and spontaneous atherosclerosis (14). A contributory role in the HDL-associated PAF-AH activity may be played by paraoxonase 1 (PON1), an enzyme that is present in plasma exclusively associated with this lipoprotein (15). Indeed, PON1 exhibits PAF-AH-like catalytic activity in addition to its paraoxon and phenyl acetate hydrolytic activities (16).
We recently demonstrated that patients with primary hypercholesterolemia exhibit an alteration in the relative distribution of PAF-AH between LDL and HDL particles, resulting in a decrease in the ratio of HDL-PAF-AH to plasma-PAF-AH [or to LDL-cholesterol (LDL-C)] levels, which is proportional to the severity of the hypercholesterolemia (17). Furthermore, atorvastatin therapy partially restored such an altered PAF-AH distribution by reducing both plasma LDL-C levels and LDL-associated PAF-AH activity, although this statin did not affect plasma levels of HDL-C or HDL-associated PAF-AH activity (18).
Considered together, the above findings support the convention that HDL-associated PAF-AH may play an antiatherogenic role. It remains indeterminate, however, as to whether drugs that modify plasma levels of HDL could influence PAF-AH activity associated with this lipoprotein. Fibrates are a family of hypolipidemic drugs that may reduce plasma LDL-C levels but equally induce elevation in HDL-C levels (19). We therefore undertook the present study to investigate the effect of a potent fibrate, fenofibrate, on HDL-associated PAF-AH relative to its effects on enzyme activity associated with apoB-containing lipoproteins in atherogenic dyslipidemias of Types IIA, IIB, and IV. This question is of special interest as this fibrate induces a shift in LDL particle profile from small, dense LDL to large, buoyant particles; indeed, PAF-AH is primarily associated with dense LDL particles in plasma (20).

Patients
Unrelated hyperlipidemic patients (n ϭ 71) attending the Outpatient Lipid Clinic of the University Hospital of Ioannina participated in the study. Secondary causes of dyslipidemia (hypothyroidism, diabetes mellitus, liver or renal diseases, alcoholism, etc.) were excluded by personal history, physical examination, and appropriate laboratory tests. None of the study participants was obese (BMI Ͼ 30 kg/m 2 ), hypertensive (blood pressure Ͼ 140/90 mmHg on repeated measurements), or was taking medications known to interfere with lipid metabolism. No patient had any clinical or ECG evidence of cardiovascular disease. After the initial screening, patients gave informed consent and were advised to follow the National Cholesterol Education Program Step 1 diet for 3 months. At the end of this period, a complete laboratory baseline analysis was performed. According to their lipid levels, patients were divided into the following groups: 1 ) primary hypercholesterolemia (Type IIA dyslipidemia), consisting of 18 patients (mean age 54.2 Ϯ 10.8 years, BMI 25.8 Ϯ 3.9 kg/m 2 , seven active smokers, eight males), exhibiting plasma LDL-C levels Ͼ 160 mg/dl. 2 ) Combined hyperlipidemia (Type IIB dyslipidemia), consisting of 23 patients (mean age 51.5 Ϯ 11.4 years, BMI 26.6 Ϯ 4.5 kg/m 2 , 10 active smokers, 14 males), exhibiting plasma LDL-C levels Ͼ 160 mg/dl and triglyceride levels Ͼ 200 mg/dl. 3 ) Primary hypertriglyceridemia (Type IV dyslipidemia), consisting of 30 patients (mean age 51.1 Ϯ 11.7 years, BMI 28.1 Ϯ 4.4 kg/m 2 , 12 active smokers, 13 males), exhibiting plasma triglyceride levels Ͼ 200 mg/dl, and LDL-C levels Ͻ 160 mg/dl. Patients with familial hypercholesterolemia were excluded by appropriate genetic analysis (17). No difference in the above biological and clinical characteristics was observed among patient groups. Micronized fenofibrate (200 mg at bedtime) was initiated in all patients; after 3 months of active treatment, a second blood analysis was performed. Compliance with treatment and diet was assessed as described previously (18).
Ninety-eight age-and sex-matched apparently healthy normolipidemic subjects were selected from individuals receiving a medical check-up at our hospital and served as controls. The study was approved by the Ethics Committee of the University Hospital of Ioannina.

Effect of fenofibrate on PAF-AH production and secretion by human monocyte/macrophages
Peripheral blood monocytes from patients (before the initiation of therapy with fenofibrate) as well as from healthy volunteers were isolated and cultured as previously described (21). After 6 days of culture, the cells were treated either with fenofibrate or with fenofibric acid (dissolved in DMSO) at final concentrations ranging from 10 M to 300 M. Treatments were performed for 24 h and 48 h in RPMI medium containing 10% human serum in which endogenous PAF-AH was completely and irreversibly inactivated by preincubation with 1 mM Pefabloc for 30 min. After treatment, PAF-AH activity was determined in supernatants and cell lysates prepared as previously described (18). tration) as a substrate, whereas PON1 activities in serum and lipoprotein subfractions were determined using paraoxon and phenyl acetate as substrates (18).

Analytical methods
Serum total cholesterol, triglycerides, HDL-C, apoB, apoA-I, and apoE were determined as previously described (18). Serum LDL-C was calculated using the Friedewald formula (provided that triglyceride levels were lower than 350 mg/dl. In 15 patients with high triglyceride values, LDL-C was not determined). The total cholesterol, triglyceride, and phospholipid content in each HDL subfraction were measured enzymatically using the Bio-Merieux kit (20), whereas the protein content of the lipoprotein subfractions was measured by the BCA method (Pierce). The lipoprotein mass of each subfraction was calculated as the sum of the mass of the individual lipid and protein components (20).

Statistical analysis
Data were expressed as mean Ϯ SD. Statistical analysis was performed using paired Student's t -test for comparisons between baseline and posttreatment values, while one-way ANOVA followed by the LSD test was used for comparisons between individual groups. Correlations between PAF-AH activity and lipid parameters were estimated using linear regression analysis.

Effect of fenofibrate therapy on plasma lipid profile in phenotypes IIA, IIB, and IV
Fenofibrate significantly decreased serum total cholesterol and triglyceride levels as well as serum apoE levels in all patient groups ( Table 1 ). A significant reduction in serum apoB levels in all groups was also observed, whereas LDL-C levels were reduced in Type IIA and IIB patients, but not in Type IV patients; this finding suggests that the decrease in serum total cholesterol levels in Type IV patients was mainly due to reduction in VLDL-C levels. Most importantly, fenofibrate induced significant elevation in both serum HDL-C and apoA-I levels in Type IIB and IV dyslipidemic patients, but not in the Type IIA patient group (Table 1).

Plasma PAF-AH activity
Total plasma PAF-AH activity at baseline was higher in all patient groups as compared with controls. Furthermore, baseline values of enzyme activity in dyslipidemic Type IIB patients were significantly higher compared ei-ther to those in Type IIA or to Type IV patients ( Table 2 ). HDL-associated PAF-AH activity (HDL-PAF-AH) in dyslipidemic Type IIB and Type IV patients was significantly lower compared either to normolipidemic controls or to Type IIA patients. Furthermore, Type IV patients exhibited significantly lower HDL-PAF-AH activity as compared with Type IIB patients (Table 2). In all patient groups, HDL-PAF-AH activity was negatively correlated to plasma triglyceride levels ( r ϭ Ϫ 0.29, P Ͻ 0.02). It is important to note that the ratio of HDL-PAF-AH to LDL-C levels before treatment was significantly lower in all patient groups compared with controls, whereas no difference in this ratio was observed among the patient groups (Table 2).
Fenofibrate treatment led to a reduction in total plasma PAF-AH activity in all patient groups. Enzyme activity in Type IIA and Type IV patients was decreased by 28% and 22%, respectively, to reach control values. A decrease (27%) in enzyme activity was also observed in Type IIB patients, although it remained significantly elevated as compared with controls, even after fenofibrate therapy ( Table 2). The reduction in plasma PAF-AH activity in Type IIA and Type IIB patients, but not in Type IV patients, was positively correlated with a reduction in LDL-C levels ( r ϭ 0.45, P Ͻ 0.005 for Type IIA and r ϭ 0.53, P Ͻ 0.01 for Type IIB). In contrast, the reduction in enzyme activity in Type IV patients was positively correlated with reduction in plasma apoE levels ( r ϭ 0.45, P Ͻ 0.05). Importantly, fenofibrate treatment significantly increased HDL-PAF-AH in Type IIB and Type IV patients, although posttreatment values remained lower than levels in controls. By contrast, no change was observed in HDL-PAF-AH in Type IIA patients after fenofibrate administration (Table 2). Furthermore, the ratio of HDL-PAF-AH to LDL-C levels significantly increased in all patient groups after fenofibrate treatment (50% in Type IIA and Type IIB, and 43% in Type IV), although it still remained lower as compared with the control group (Table 2).

Serum PON1 activities
No difference was observed in baseline values of serum PON1 activity toward paraoxon in any of the groups studied. Equally, no difference was found in PON1 activity toward phenyl acetate between Type IIA or Type IV dyslipidemic patients and controls, whereas Type IIB patients exhibited lower enzyme activity compared with controls (although it did not reach statistical significance). Fenofibrate therapy did not affect enzyme activity in any patient group (Table 2).

PAF-AH activity in plasma lipoprotein subspecies
To further investigate the effect of fenofibrate on the PAF-AH activity associated with apoB-and apoA-I-containing plasma lipoprotein subspecies, we fractionated plasma lipoproteins before and after fenofibrate therapy. To study the effect of fenofibrate on enzyme activity associated with apoB-lipoprotein subspecies, total plasma was subjected to ultracentrifugation. As previously reported, a proportion of LDL-bound PAF-AH dissociates and is redistributed to HDL during ultracentrifugation of total plasma (3). Thus, to avoid any contamination of HDL-PAF-AH activity with the LDL-associated enzyme during ultracentrifugation, we studied the effect of fenofibrate on apoA-I-associated PAF-AH activity using plasma depleted of apoB-lipoproteins (see Methods).
Among the apoB-containing lipoproteins, PAF-AH activity was preferentially associated with the dense LDL-5 subfraction in all patient groups, as well as in normolipidemic controls ( Table 3). Dyslipidemic Type IIB and Type IV patients exhibited significantly higher enzyme activity associated with the VLDLϩIDL subfraction compared with controls, a phenomenon not observed in Type IIA patients. Furthermore, the Type IIB patient group exhibited significantly higher enzyme activity associated with this subfraction compared with Type IIA, but significantly lower enzyme activity compared with dyslipidemic Type IV patients (Table 3). In Type IIA and Type IIB patient groups, PAF-AH activity associated with the dense LDL-5 subfraction was higher compared either with controls or with Type IV patients, the Type IIB group displaying higher enzyme activity compared with Type IIA patients (Table 3). Furthermore, the Type IIA and Type IIB patient groups exhibited significantly higher enzyme activity associated with the LDL-4 subfraction compared either to controls or to Type IV patients, a phenomenon not observed for the other LDL subfractions.
Among the apoA-I-containing lipoproteins, PAF-AH activity in all patient groups as well as in controls was preferentially associated with the HDL-3c subfraction, which corresponds to the VHDL-1 subfraction in our previous study (20). A representative profile of the distribution of PAF-AH activity among the HDL subfractions in Type IIB patients is shown in Fig. 1A. Dyslipidemic patients of Type IIB and Type IV exhibited significantly lower enzyme activity associated with this subfraction compared either with controls (unpublished observations) or with Type IIA patients (Fig. 1B), a phenomenon not observed in the other HDL subfractions (data not shown). Furthermore, Type IV patients had significantly lower enzyme activity in HDL-3c  Values represent the mean Ϯ SD. ANOVA followed by LSD test was used for comparisons between individual groups. A P value Ͻ 0.05 was considered significant. a P Ͻ 0.05 compared with controls. b P Ͻ 0.05 compared with Type IIA at the same time (baseline or posttreatment). c P Ͻ 0.05 compared with Type IIB (baseline or posttreatment).
by guest, on July 19, 2018 www.jlr.org Downloaded from compared with Type IIB, whereas no difference was observed between Type IIA patients and controls (unpublished observations). Interestingly, fenofibrate therapy in Type IIB and Type IV patients significantly increased PAF-AH activity associated with the HDL-3c subfraction, a phenomenon that was not observed in Type IIA patients (Fig.  1B). Drug administration did not affect enzyme activity associated with the other HDL subfractions in either patient group (unpublished observations).
To further investigate the relationship between the increase in plasma HDL-C levels and HDL-associated PAF-AH induced by fenofibrate in Type IIB and Type IV patient groups, we determined the lipoprotein mass in each HDL subfraction. As shown in Table 4, baseline values of HDL-2a  and HDL-2b mass in Type IIB and Type IV patient groups  were significantly lower compared with those of Type IIA,  the Type IV having lower values compared with Type IIB. No difference was observed in the mass of any HDL subfraction between Type IIA patients and controls (data not shown). Fenofibrate therapy significantly increased the mass of HDL-2a and HDL-3a subfractions in Type IIB and Type IV patient groups, whereas it did not affect the mass of the other HDL subfractions in these patient groups as well as the mass of all HDL subfractions in Type IIA patient group.
In contrast to the increase in HDL-PAF-AH activity, fenofibrate treatment significantly reduced PAF-AH activity associated with VLDLϩIDL subfraction in Type IIB and Type IV patient groups, a phenomenon not observed in Type IIA patients ( Fig. 2A). It must be noted that the above reduction was not observed when enzyme activity was expressed per milligram of VLDLϩIDL subfraction (data not shown). Furthermore, a reduction in enzyme activity associated with LDL-4 and LDL-5 subfractions was observed after fenofibrate treatment in Type IIA and Type IIB patients. (Fig. 2B). Fenofibrate treatment in Type IV patients reduced enzyme activity associated only with the LDL-5 subfraction (Fig. 2B). Finally, in all patient groups, fenofibrate did not affect PAF-AH activity associated with intermediate and large LDL particles (data not shown).

PON1 activities in lipoprotein subspecies
PON1 activities toward paraoxon and phenyl acetate were determined in lipoprotein subspecies separated by ultracentrifugation of serum depleted of apoB-lipoproteins (see Methods). In all patient groups as well as in controls, both enzyme activities were preferentially associated with the dense HDL subfraction, HDL-3c. No differences were observed for the PON1 activities associated with each HDL subfraction among all studied groups. Fenofibrate therapy did not affect PON1 activity associated with any HDL subfraction in all patient groups. Figure 3 illustrates the PON1 activities toward paraoxon (Fig. 3A) and toward phenyl acetate (Fig. 3B) associated with HDL subfractions before and after fenofibrate therapy in dyslipidemic Type IIB patients.

Effect of fenofibrate on PAF-AH secretion from macrophages
To explore the possibility that the effects of fenofibrate on LDL-and HDL-associated PAF-AH might be due to drug effect on enzyme secretion, we studied the effect of fenofibrate on PAF-AH production and secretion by peripheral blood monocyte-derived macrophages. We used macrophages since these cells represent a major source of plasma PAF-AH (22). Cells isolated from six subjects of each group were used in these studies. Incubations were performed for 24 h and 48 h, with various concentrations of fenofibrate or fenofibric acid ranging from 10 M to 300 M. As expected, there was a steady increase in both the secreted and the cell-associated PAF-AH activity in untreated cells (incubated with DMSO) of normolipidemic controls, attaining 178 Ϯ 46 and 51 Ϯ 18 nmol/mg DNA/h, respectively, at 24 h as well as 323 Ϯ 97 and 84 Ϯ 36 nmol/ mg DNA/h, respectively, at 48 h of culture. Similar results were obtained for untreated macrophages from each of the three groups of dyslipidemic patients studied (data not shown). Cell treatment with fenofibrate or fenofibric acid at concentrations studied up to 300 M did not significantly alter the secreted or the cell-associated enzyme activity measured in untreated cells (data not shown). It must be noted that in enzyme assays performed in the presence of 50 M or 300 M of either fenofibrate or fenofibric acid using macrophage supernatant as the source of the enzyme (containing 5.1 nmol/ml/min of PAF-AH activity), no effect of both substances on PAF-AH activity was observed.

DISCUSSION
In the present study, we show for the first time that fenofibrate therapy in patients with primary hypercholesterolemia (IIA), combined hyperlipidemia (IIB), and primary hypertriglyceridemia (IV) significantly reduces PAF-AH activity associated with atherogenic apoB-containing lipoproteins. Most importantly, fenofibrate therapy in dyslipidemic patients of Type IIB and Type IV significantly increased the HDL-associated PAF-AH activity due to the increase in enzyme activity associated with HDL-3c subfraction. In contrast, fenofibrate did not affect HDL-associated PON1 activities toward paraoxon and phenyl acetate in either patient group.
The decrease in plasma LDL-C levels induced by fenofibrate could represent the major mechanism accounting for the drug-induced reduction in total plasma PAF-AH activity in hyperlipidemic patients of Type IIA and Type IIB. This is supported by the strong positive correlation observed between the decrease in enzyme activity and in plasma LDL-C levels. Fenofibrate preferentially reduced PAF-AH activity associated with the dense LDL particles, i.e., those particles carrying the majority of LDL-associated enzyme activity (20). We may suggest that such a reduction could primarily be due to the well-known drug effect on the transformation of small, dense LDL particles to the large buoyant LDL, i.e., particles that exhibit a higher rate of clearance from the circulation compared with small, dense ones (23,24). Furthermore, the druginduced reduction in enzymatic activity associated with the VLDLϩIDL subspecies may play a contributory role in the reduction of plasma PAF-AH activity that was observed in Type IIB patients.
Our results show for the first time that patients with pri-