INTRODUCTION

In human plasma, five distinct high density lipoprotein (HDL)¹ subpopulations have been identified on the basis of their apparent diameter as determined by using polyacrylamide gradient gel electrophoresis: HDL_2b (9.71-12.90 nm), HDL_2a (8.77-9.71 nm), HDL_3a (8.17-8.77 nm), HDL_3b (7.76-8.17 nm), and HDL_3c (7.21-7.76 nm) (1). The heterogeneity of HDL in terms of size distribution may be of physiopathological relevance, and Cheung and coworkers (2) reported significant alterations in HDL size in patients with symptomatic coronary artery disease as compared with healthy control subjects. Interestingly, the presence of coronary artery disease was more strongly associated with abnormalities in HDL particle size distribution than with low HDL cholesterol levels (2). Therefore, these observations raised a considerable interest in identifying the factors that can induce alterations in the relative proportions of plasma HDL_2b, HDL_2a, HDL_3a, HDL_3b, and HDL_3c subfractions in vivo.

It is now well established that plasma lipoproteins do not constitute stable entities in vivo but rather are continuously modified in the blood stream through the action of specific factors, i.e. lecithin:cholesterol acyltransferase, lipoprotein lipase, hepatic lipase, the cholesteryl ester transfer protein (CETP), and the phospholipid transfer protein (PLTP) (3). In particular, studies of the past few years demonstrated that the two distinct plasma lipid transfer proteins, CETP and PLTP, can markedly alter the size distribution of plasma lipoprotein fractions by shuttling lipid components from one lipoprotein substrate to another. CETP promotes the exchange of neutral lipids, cholesteryl esters, and triglycerides between nonequilibrated pools, leading to the net transfer of cholesteryl esters from HDL and low density lipoproteins (LDL) toward triglyceride-rich lipoproteins, with the reciprocal net transfer of triglycerides from triglyceride-rich lipoproteins toward LDL and HDL (4, 5). The subsequent hydrolysis of transferred triglycerides leads to the formation of small LDL and HDL particles (6, 7, 8). In addition, recent studies have demonstrated that CETP can promote the size redistribution or conversion of isolated HDL in the absence of other lipoprotein fractions (9, 10). Whereas CETP can also account for some phospholipid exchange activity in human plasma, the rapid net mass transfer of phospholipids from triglyceride-rich lipoproteins toward HDL during lipolysis has been shown to be mediated by the other plasma lipid transfer protein, PLTP (11). The PLTP-mediated enrichment of HDL with phospholipids is accompanied by the enlargement of the particles (11, 12, 13). Complementary studies demonstrated that PLTP can also act as an HDL conversion factor in the absence of other lipoprotein fractions by promoting the formation of large and small HDL subfractions from an initial population of HDL with intermediate size (14, 15, 16). Taken together, it appears therefore that both CETP and PLTP can promote the size redistribution of HDL particles, and then might account in vivo for the heterogeneity in the size and composition of the plasma HDL fraction.

Whereas the polyacrylamide gel electrophoresis distribution profile of HDL can be markedly transformed by lipid transfer proteins, the precise roles of CETP and PLTP in modulating the distribution of plasma HDL remain to be clearly established. That latter point was addressed in the present study by determining the effect of CETP and PLTP on the relative proportions of HDL_2b, HDL_2a, HDL_3a, HDL_3b, and HDL_3c subpopulations both in total human plasma and in reconstituted experimental mixtures containing isolated lipoproteins and purified lipid transfer proteins. More specifically, the cholesteryl ester and phospholipid transfer activities, as well as the size distribution of HDL were investigated in patients with chronic alcoholism, a pathological state that has been shown to be associated with abnormalities in plasma CETP activity (17, 18, 19, 20), HDL profile (21, 22, 23, 24, 25, 26), and as demonstrated for the first time in the present study with alterations in plasma PLTP activity. Thus, alcohol withdrawal provided us with a unique opportunity to study in vivo alterations in plasma lipid transfer activities and to analyze their effects in terms of plasma HDL distribution.

MATERIALS AND METHODS

Alcoholic patients were 33 men hospitalized for a rehabilitation program at the University Hospital of Nancy (France) (27). They did not show clinical or biological evidence of severe liver insufficiency and were dependent on alcohol according to the Diagnostic Statistical Manual III 3R dependence criteria. All of them consumed more than 100 g of pure alcohol per day until the precise day of hospitalization. The patients did not take drugs known to affect lipid metabolism.

Normal subjects were addressed to the clinical laboratory for a routine blood test.

Fasting blood samples were collected into EDTA-containing glass tubes. For alcoholic patients, blood was collected on the morning of arrival in the hospital and after 21 days of abstinence. The blood was centrifuged for 15 min at 1000 × g, and plasma aliquots were promptly stored at 196 °C until analysis.

Total lipoproteins from normal subjects were ultracentrifugally isolated as the d < 1.210 g/ml plasma fraction by one 7-h, 100,000-rpm (386,000 × g) ultracentrifugation in a TLA-100 rotor in a TL-100 ultracentrifuge (Beckman, Palo Alto, CA). Normolipidemic, fresh, and citrated plasma for individual lipoprotein preparation was provided by the Centre de Transfusion Sanguine (Hôpital du Bocage, Dijon, France). HDL₃ were isolated from total plasma as the fraction of density 1.130-1.210 g/ml with two 24-h, 45,000-rpm (149,000 × g) spins at the lower density, and one 30-h, 55,000-rpm (223,000 × g) spin at the higher density. Finally, the HDL₃ fraction was washed with one 6-h, 90,000-rpm (561,000 × g) spin at the density 1.210 g/ml in an NVT-90 rotor on a Beckman XL-90 ultracentrifuge.

HDL₃ containing tritiated cholesteryl esters ([³H]CE-HDL₃) were biosynthetically labeled as described previously (28). Briefly, the d > 1.13 g/ml fraction from 20 ml of normolipidemic human plasma was incubated for 24 h at 37 °C with 10 nmol of [1,2-³H]cholesterol (specific activity, 46 Ci/mmol; Amersham Corp.). At the end of the incubation, radiolabeled HDL₃ were recovered by sequential ultracentrifugation as described above. As judged by thin layer chromatography, more than 95% of radioactivity was recovered in the cholesteryl ester moiety.

Radiolabeled phosphatidylcholine liposomes ([¹⁴C]PC-liposomes) were prepared according to the general procedure described by Damen et al. (29, 30). Briefly, 10 µmol of egg phosphatidylcholine containing 10 nmol of ¹⁴C-phosphatidylcholine (specific activity, 100 mCi/mmol; Amersham Corp.) were dried under a stream of nitrogen and recovered in 1 ml of a 10 mmol/liter Tris, 150 mmol/liter NaCl, pH 7.4, buffer containing 0.2 g/liter NaN₃ and 1 mmol/liter EDTA. Finally, [¹⁴C]PC-liposomes were obtained by dispersing phospholipids with a sonifier (VibraCell 72434, Sonics Materials). After low speed centrifugation, the lipid dispersion was optically clear and was shown to be constituted of small, unilamellar phospholipid liposomes (29, 30).

CETP activity in total human plasma was evaluated by measuring the rate of transfer of radiolabeled cholesteryl esters from [³H]CE-HDL₃ toward the apoB-containing lipoprotein plasma fraction according to the procedure previously described (31). Briefly, mixtures containing a 25-µl aliquot of total plasma, [³H]CE-HDL₃ (2.5 nmol of cholesterol), and iodoacetate (75 nmol) in a final volume of 50 µl were incubated for 3 h at 37 °C. Incubated mixtures were subsequently subjected to ultracentrifugation at a density of 1.068 g/ml. Cholesteryl ester transfer activity was measured as the rate of radiolabeled cholesteryl esters transferred from the d > 1.068 g/ml to the d < 1.068 g/ml fractions during a 3-h incubation at 37 °C as compared with control samples kept at 4 °C. Results were expressed in percentage of ³H-CE transferred per hour per milliliter of plasma. Under the experimental conditions described above, CETP activity measurements were shown to be independent of the relative dilution of [³H]CE-HDL₃ in the plasma HDL pool (31). In addition, as demonstrated by adding increasing concentrations of purified CETP to normolipidemic human plasma, the cholesteryl ester transfer assay allows us to measure plasma CETP activity over a wide range of values (32).

CETP mass concentrations were measured by using a competitive enzyme-linked immunosorbent assay on a Biomek 1000 Biorobotic System (Beckman Instruments) (33). CETP mass concentration values were determined in quadruplicate from a calibration curve obtained with a frozen plasma standard.

PLTP activity in total human plasma was evaluated by measuring the transfer of radiolabeled phosphatidylcholine from [¹⁴C]PC-liposomes to the plasma HDL fraction. This latter experimental system was proven to be specific for PLTP activity measurement, since CETP does not transfer phosphatidylcholine from liposomes toward HDL (34). Briefly, 30-µl aliquots of total plasma were incubated for 15 min up to 120 min at 37 °C with [¹⁴C]PC-liposomes (125 nmol of phosphatidylcholine) and iodoacetic acid (120 nmol) in a final volume of 80 µl. In blank controls, mixtures were maintained at 4 °C. At the end of the incubation period, 320 µl of TBS were added to each incubation mixture and phospholipid liposomes were precipitated by the addition of 300 µl of a solution containing 500 mmol/liter NaCl, 215 mmol/liter MnCl₂, 445 units/ml heparin. ``Total count'' controls received 300 µl of TBS buffer instead of the MnCl<sub>2</sub>/heparin precipitant reagent. After removal of precipitates by low speed centrifugation, resulting supernatants were mixed with 2 ml of scintillation fluid (OptiScint Hisafe 3, Pharmacia, Uppsala, Sweden), and the radioactivity was assayed for 2 min in a Wallac 1410 liquid scintillation counter (Pharmacia Biotech Inc.). Phospholipid transfer activity was calculated as the rate of radiolabeled phospholipids (total count control) transferred from [<sup>14</sup>C]PC-liposomes to the plasma HDL fraction after deduction of ``4 °C blank'' control values. Consistently less than 5% of total radioactivity was recovered in the supernatant of 4 °C blank control samples. Results were expressed in percentage of [¹⁴C]PC transferred per hour per milliliter of plasma.

CETP and PLTP were purified from 300 ml of citrated human plasma (Centre de Transfusion Sanguine, Dijon, France) according to the sequential procedure previously described (34). Briefly, the d > 1.21 g/ml plasma fraction was subjected successively to hydrophobic interaction chromatography on a phenyl-Sepharose CL-4B column (Pharmacia), to affinity chromatographies on Heparin-Ultrogel A4R (Sepracor, Villeneuve-la-Garenne, France) and Cibacron Blue-Trisacryl (Sepracor) columns, and to anion exchange chromatography on a Mono-Q HR 5/5 column (Pharmacia). The experimental procedure allowed two lipid transfer protein preparations to be obtained, both of which were deprived of lecithin:cholesterol acyltransferase activity. As compared with starting material, the CETP-containing fraction represented an approximately 5000-fold increase in specific transfer activity of cholesteryl esters from [³H]CE-HDL₃ toward LDL and an approximately 1500-fold increase in specific transfer activity of phospholipids from [¹⁴C]PC-HDL₃ toward LDL (34). In contrast, the CETP-containing fraction did not catalyze the transfer of phospholipids from [¹⁴C]PC-liposomes toward HDL₃ (34). The PLTP-containing fraction represented an approximately 2500-fold increase in the specific transfer activity of phospholipids from [¹⁴C]PC-HDL₃ toward LDL and an approximately 1000-fold increase in the specific transfer activity of phospholipids from [¹⁴C]PC-liposomes toward HDL₃. In contrast, the PLTP-containing fraction did not catalyze the transfer of cholesteryl esters from [³H]CE-HDL₃ toward LDL. In summary, the sequential chromatographic procedure allowed us to obtain in parallel one protein fraction containing active CETP but devoid of both PLTP and lecithin:cholesterol acyltransferase and one protein fraction containing active PLTP but devoid of both CETP and lecithin:cholesterol acyltransferase.

Apparent hydrodynamic diameters of HDL subpopulations were determined after separation by electrophoresis on nondenaturing polyacrylamide gradient gels according to the general procedure previously described (1). Briefly, the d < 1.21 g/ml fraction isolated either from total plasma or from incubated reconstituted mixtures containing isolated lipoproteins and purified lipid transfer proteins were subjected to electrophoresis in 15-250 g/liter polyacrylamide gradient gels. The electrophoretic migration was conducted at 70 V during the first hour and then at 150 V for 20 h in a 90 mmol/liter Tris, 80 mmol/liter boric acid, pH 8.3, buffer containing 3 mmol/liter Na-EDTA and 3 mmol/liter NaN₃. At the end of the electrophoresis, the gels were stained with Coomassie Brilliant Blue G (1).

The distribution profiles of HDL were obtained by analysis of polyacrylamide gradient gels on a BIO-RAD GS-670 imaging densitometer. The mean apparent diameters of HDL subfractions were determined by comparison with a calibration curve constructed with albumin, lactate deshydrogenase, and ferritin (Pharmacia high molecular weight calibration kit), which were subjected to electrophoresis together with the samples.

The relative proportions of plasma HDL subfractions (HDL_2b, 9.71-12.90 nm; HDL_2a, 8.77-9.71 nm; HDL_3a, 8.17-8.77 nm; HDL_3b, 7.76-8.17 nm; HDL_3c, 7.21-7.76 nm) (1) were obtained by determining the relative areas under the scan curve and by relating them to the total area corresponding to the entire plasma HDL fraction (2, 31, 35).

Assays were performed on a Cobas-Fara Centrifugal Analyzer (Roche). Protein concentrations were measured with the bicinchoninic reagent (Pierce) according to Smith et al. (36). Total cholesterol, unesterified cholesterol, and triglyceride concentrations were measured by enzymatic methods using Boehringer Mannheim reagents. The plasma HDL-cholesterol concentrations were measured after selective precipitation of apoB-containing lipoproteins with Boehringer phosphotungstic acid/MgCl₂ reagent as recommended by the manufacturer. The plasma VLDL + LDL cholesterol concentrations were calculated as the difference between total cholesterol and HDL-cholesterol levels.

Data are expressed as means ± S.D. Student's t test and the Wilcoxon signed rank test were used to determine the significance of the difference between data means as indicated. Coefficients of correlation r were calculated by using linear regression analysis. Multiple regression analysis was used to determine the contribution of HDL cholesterol and triglycerides to the correlation with CETP activity.

RESULTS

Ultracentrifugally isolated HDL₃ were incubated for 24 h at 37 °C in the presence or absence of either CETP or PLTP, which were partially purified from total human plasma as described under ``Materials and Methods.'' When HDL₃ were incubated in TBS buffer in the absence of lipid transfer protein supplementation (control samples), one population of HDL with a mean apparent diameter of 8.4 nm corresponding to HDL_3a predominated, with the presence of minor discrete subpopulations both of larger size with mean apparent diameter of 9.1 nm corresponding to HDL_2a, and of smaller size with mean apparent diameters of 7.8 and 7.4 nm, corresponding to HDL_3b and HDL_3c, respectively. Incubation of isolated HDL₃ particles in the presence of increasing concentrations of either CETP or PLTP induced marked changes in the size distribution of HDL subpopulations as compared with homologous control samples. In both cases, the major initial HDL_3a subpopulation progressively disappeared (Fig. 1). Despite some similarities in the CETP-mediated and the PLTP-mediated HDL conversion, some distinction between the two processes could be made. Indeed, whereas the formation of small HDL_3c particles (mean apparent diameter, 7.4 nm) tended to be privileged in the presence of CETP, PLTP favored the appearance of both small and large particles (Fig. 1). In addition, the formation of a large HDL subpopulation, which corresponded to HDL_2b (mean apparent diameter, 10.1 nm), and which did not appear with CETP, could be noted with the highest PLTP concentration studied (Fig. 1).

In order to confirm further the ability of CETP and PLTP to redistribute HDL particles, the HDL₃ fractions ultracentrifugally isolated from five distinct normolipidemic plasmas were incubated for 24 h at 37 °C in the presence of either purified CETP or purified PLTP, which were added at the highest level used in Fig. 1 (15.00 and 3.75 µg in a final volume of 50 µl, respectively). After incubation, the changes in the relative proportion of various HDL subpopulations were calculated as the difference from control, homologous HDL, which were incubated in TBS buffer. Statistical analysis by using the Wilcoxon signed rank test revealed that purified CETP induced a significant decrease in the relative proportion (mean ± S.D.) of HDL_3a (8.4 ± 6.3%; p < 0.05) and HDL_3b (3.7 ± 1.4%; p < 0.05) and a significant increase in the relative abundance of HDL_3c (+8.9 ± 1.9%; p < 0.05). In contrast, no significant amounts of HDL_2b and HDL_2a were generated during the incubation of HDL₃ with CETP (changes: +2.7 ± 3.2% (not significant), and +0.6 ± 5.6% (not significant), respectively). Purified PLTP induced significant decreases in the relative proportions of HDL_3a (27.2 ± 5.3%; p < 0.05) and HDL_3b (16.1 ± 3.2%; p < 0.05), and significant increases in the relative proportions of HDL_2b (+24.5 ± 1.2%; p < 0.05), HDL_2a (+3.0 ± 0.7%; p < 0.05), and HDL_3c (+15.9 ± 3.9%; p < 0.05).

In order to bring more insight into the role of PLTP and CETP in promoting the size redistribution of HDL in total human plasma, total lipoprotein fractions isolated from nine normal subjects (plasma total cholesterol, 217 ± 50 mg/dl; plasma HDL-cholesterol, 50 ± 16 mg/dl; plasma triglycerides, 130 ± 60 mg/dl) were incubated for 24 h at 37 °C in the absence or in the presence of either purified PLTP or purified CETP. As shown in Table I, PLTP and CETP induced significant alterations in the relative proportions of HDL subpopulations. As compared with TBS control, PLTP induced both a significant increase in the relative abundance of HDL_2b and a significant decrease in the relative abundance of HDL_3a. In the meantime, the relative proportions of HDL_2a, HDL_3b, and HDL_3c were not significantly modified. In contrast to observations made with PLTP, CETP induced a significant decrease in the relative abundance of HDL_2a and a significant increase in the relative abundance of HDL_3b and HDL_3c, while the proportions of HDL_2b and HDL_3a were not significantly modified (Table I).

Table I. Effect of the incubation of total plasma lipoproteins in the presence of either purified CETP or purified PLTP on the relative abundance of HDL subpopulations Total lipoproteins were ultracentrifugally isolated as the d < 1.210 g/ml fraction from nine human plasmas. Subsequently, isolated lipoproteins (cholesterol, 112.5 µg) were incubated for 24 h at 37 °C in the absence (TBS control) or in the presence of either purified CETP (3.75 µg) or purified PLTP (3.75 µg) in a final volume of 75 µl. At the end of the incubation period, the size distribution of HDL was analyzed by using native polyacrylamide gradient gel electrophoresis, and the relative abundance of individual HDL subpopulations was expressed as percentage of the total HDL fraction (see ``Materials and Methods''). Data are mean ± S.D. of nine distinct subjects. HDL subpopulation TBS PLTP CETP HDL2b 7.87  ± 1.83 8.66  ± 2.34a 7.76  ± 2.54 HDL2a 37.56  ± 6.43 39.00  ± 5.94 33.03  ± 2.53a HDL3a 37.87  ± 2.62 32.76  ± 3.42b 34.64  ± 3.88 HDL3b 15.58  ± 7.75 17.62  ± 5.85 21.36  ± 6.97a HDL3c 1.13  ± 0.56 1.96  ± 2.05 3.21  ± 4.84b a  Significance of the difference from TBS control: p < 0.01 (Wilcoxon test). b  Significance of the difference from TBS control: p < 0.05 (Wilcoxon test).

Table II shows the plasma lipid parameters in alcoholic patients before and after 3 weeks of abstinence. Whereas total cholesterol and triglyceride concentrations were not significantly altered after alcohol withdrawal, marked changes appeared in the distribution of cholesterol among distinct plasma lipoprotein fractions. Indeed, HDL cholesterol levels were significantly decreased by about 45%, whereas the cholesterol levels in the apoB-containing lipoproteins (i.e. VLDL and LDL) were significantly increased by about 15% (Table II). As a consequence, the VLDL + LDL to HDL cholesterol ratio was approximately doubled after 3 weeks of abstinence (Table II).

Table II. Plasma lipid parameters before and after alcohol withdrawal Plasma lipid parameters were determined in total plasma from 33 alcoholic men before and after 3 weeks of abstinence (see ``Materials and Methods''). Data are means ± S.D. Before After p Total cholesterol (g/liter) 2.19  ± 0.62 2.09  ± 0.48 0.1578 HDL cholesterol (g/liter) 0.76  ± 0.42 0.42  ± 0.13 0.0001 VLDL + LDL cholesterol (g/liter) 1.45  ± 0.64 1.66  ± 0.50 0.0118 Ratio of VLDL + LDL to HDL cholesterol 2.14  ± 0.38 4.32  ± 1.78 0.0001 Triglycerides (g/liter) 1.68  ± 1.99 1.71  ± 0.74 0.4557

CETP mass concentration, CETP activity, and PLTP activity were measured in total plasma from alcoholic patients before and after 3 weeks of abstinence (see ``Materials and Methods''). As shown in Table III, an approximately 29% increase in CETP mass concentration was observed after alcohol withdrawal. Consistently, the rate of radiolabeled cholesteryl esters transferred from a tracer dose of [³H]CE-HDL₃ to the plasma VLDL + LDL fraction (CETP activity) was increased by approximately 29% after alcohol withdrawal (Table III). CETP mass concentration and CETP activity correlated significantly in alcoholic patients before (r = 0.53; p = 0.0015) and after (r = 0.62; p = 0.0001) alcohol withdrawal. Since specific CETP activity, calculated as the ratio of plasma CETP activity to plasma CETP mass concentration did not vary significantly before and after abstinence (Table III), it appears that alterations in CETP mass concentrations mainly accounted for alterations in plasma cholesteryl ester transfer activity.

Table III. Plasma CETP and PLTP activities before and after alcohol withdrawal Plasma lipid transfer activities were determined in fasting plasma from 33 alcoholic men before and after 3 weeks of abstinence (see ``Materials and Methods''). CETP mass concentration was determined by enzyme-linked immunosorbent assay. Plasma CETP activity was evaluated by measuring the rate of radiolabeled cholesteryl esters transferred from [3H]CE-HDL3 to the plasma VLDL + LDL fraction, and specific CETP activity was calculated as the ratio of the rate of cholesteryl ester transferred to the plasma CETP mass concentration. PLTP activity was evaluated by measuring the rate of radiolabeled phosphatidylcholine transferred from [14C]PC-liposomes to the plasma HDL fraction. Data are means ± S.D. Before After p CETP mass (µg/ml) 1.98  ± 0.55 2.56  ± 0.58 0.0001 CETP activity (%/h/ml) 173.5  ± 70.5 223.2  ± 69.3 0.0007 Specific CETP activity (%/h/µg) 90.0  ± 30.1 87.1  ± 22.2 0.2743 PLTP activity (%/h/ml) 473.9  ± 203.7 312.7  ± 148.4 0.0001

Plasma PLTP activity was measured as the rate of transfer of radiolabeled phosphatidylcholine from [¹⁴C]PC-liposomes toward the plasma HDL fraction, an experimental procedure that has been shown to evaluate specifically the phospholipid transfer activity of PLTP independently on the phospholipid exchange activity catalyzed by CETP (34). As shown in Fig. 2, the incubation of total plasma with [¹⁴C]PC-liposomes induced a time-dependent transfer of radiolabeled PC toward the plasma HDL fraction until a plateau was reached for incubation times exceeding 75 min. In the present study, PLTP activity in various plasma samples was evaluated by incubating plasma/[¹⁴C]PC-liposome mixtures for 30 min at 37 °C. Unlike CETP activity, plasma PLTP activity was significantly reduced by approximately 34% after alcohol withdrawal (Table III). Plasma CETP activity and PLTP activity correlated negatively in alcoholic patients before abstinence (r = 0.43; p = 0.0115), but not after abstinence (r = 0.09; not significant).

Significant relationships were observed between plasma lipid parameters and either CETP activity or PLTP activity. Indeed, as shown in Table IV, CETP activity in plasma from alcoholic patients correlated positively and significantly with VLDL + LDL cholesterol and triglycerides. Consistent but weaker correlations were observed after alcohol withdrawal (Table IV). As shown in Table V, PLTP activity correlated positively and significantly with HDL cholesterol before as well as after alcohol withdrawal. Before alcohol withdrawal, multiple regression analysis revealed that when HDL cholesterol and triglyceride levels are combined in a two-variable model, only triglyceride (p = 0.0001), and not HDL cholesterol (p = 0.2485), levels related significantly to variations in CETP activity. After alcohol withdrawal, when HDL cholesterol and triglyceride levels are combined in a two-variable model, relationships with variations in CETP activity reached the significance level neither with triglycerides (p = 0.0755) nor with HDL cholesterol (p = 0.1710) levels.

Table IV. Correlations between CETP activity and plasma lipid parameters before and after alcohol withdrawal Plasma CETP activity and lipid concentrations were determined in fasting plasma from 33 alcoholic men before and after 3 weeks of abstinence. CETP activity in total plasma was evaluated by measuring the rate of transfer of radiolabeled cholesteryl esters from a tracer dose of [3H]CE-HDL3 to the plasma apoB-containing lipoproteins (see ``Materials and Methods''). Values are coefficients of correlation (r). Before After Total cholesterol 0.387 0.254 HDL cholesterol  0.428  0.347 VLDL + LDL cholesterol 0.676a 0.357 Triglycerides 0.823a 0.528b a  p = 0.0001. b  p = 0.0016.

Table V. Correlations between PLTP activity and plasma lipid parameters before and after alcohol withdrawal Plasma PLTP activity and lipid concentrations were determined in fasting plasma from 33 alcoholic men before and after 3 weeks of abstinence. PLTP activity in total plasma was evaluated by measuring the transfer of radiolabeled phosphatidylcholine from a tracer dose of [14C]PC-liposomes to the plasma HDL fraction (see ``Materials and Methods''). Values are coefficients of correlation (r). Before After Total cholesterol 0.370 0.080 HDL cholesterol 0.743a 0.646b VLDL + LDL cholesterol  0.126  0.097 Triglycerides  0.305  0.251 a  p = 0.0001. b  p = 0.0002.

The distribution of HDL subpopulations was determined by native polyacrylamide gradient gel electrophoresis in plasma from alcoholic patients before and after 3 weeks of abstinence (see ``Materials and Methods''). As shown in Fig. 3, significant differences were observed in the relative abundance of distinct HDL subpopulation before and after alcohol withdrawal. The relative proportions of the largest HDL particles (i.e. HDL_2b and HDL_2a) were significantly reduced after abstinence. In contrast, the relative proportions of the smallest particles (i.e., HDL_3b and HDL_3c) were significantly increased after abstinence. The relative abundance of HDL_3a remained unchanged (Fig. 3).

In order to determine whether lipid transfer activities could account at least in part for alterations in the size distribution of plasma HDL, we searched for correlations between changes in the relative abundance of individual HDL subpopulations and changes in either plasma CETP activity or plasma PLTP activity. As shown in Fig. 4, changes in plasma CETP activity correlated positively with changes in the proportion of HDL_3b but negatively with changes in the proportion of HDL_2a. In contrast, changes in PLTP activity due to alcohol withdrawal correlated positively with changes in HDL_2b but negatively with changes in HDL_3a (Fig. 4).

DISCUSSION

To our knowledge, the present study is the first to compare in parallel the ability of purified CETP and PLTP fractions to induce the size redistribution of human plasma HDL. As suggested by previous studies (9, 10, 14, 15, 16), we observed some similarities in the CETP-mediated and the PLTP-mediated HDL conversion processes when experimental mixtures contained only isolated HDL and purified lipid transfer proteins. However, while the relative abundance of HDL_3a and HDL_3b was reduced in the presence of either CETP or PLTP, some differences were observed between the two conversion processes. The formation of small HDL_3c from the initial HDL_3a subpopulation was privileged in the presence of CETP, while PLTP favored the appearance of both small (HDL_3c) and large (HDL_2b and HDL_2a) conversion products. In support of the role of PLTP in the HDL enlargement, earlier studies demonstrated that the transfer of phospholipids from phosphatidylcholine vesicles or triglyceride-rich lipoproteins toward plasma HDL is accompanied by the appearance of large HDL_2b and HDL_2a particles, with significant decrease in HDL₃ subclasses (11, 12, 13). Overall, previously published data as well as the observations made in the present study agree in indicating that CETP might preferentially facilitate the formation of small HDL subpopulations whereas PLTP might be responsible for the preponderance of subpopulations of larger size. In support of that latter view, Pulcini and co-workers (37) recently demonstrated by incubating plasma and HDL₃ containing a radioiodinated diacyl lipid-associating peptide that CETP accounts for the formation of HDL_3c-like particles whereas PLTP favors the formation of HDL_2b-like particles.

In order to address further the precise roles of PLTP and CETP in modulating the size distribution of various HDL subpopulations in total human plasma, the total plasma lipoprotein fraction isolated from healthy subjects was incubated in the present study with either purified CETP or purified PLTP, and resulting changes in the size distribution of HDL were analyzed. Interestingly, CETP and PLTP appeared to affect specifically some, but not all, of the individual HDL subpopulations. Incubation of total plasma lipoproteins with PLTP significantly increased the proportion of HDL_2b and significantly reduced the proportion of HDL_3a, while the other HDL subpopulations (i.e. HDL_2a, HDL_3b, and HDL_3c) remained virtually untouched. In turn, incubation of total plasma lipoproteins with CETP altered significantly the proportions of HDL_2a, HDL_3b, and HDL_3c but not the proportions of HDL_2b and HDL_3a. In the presence of CETP, the relative proportion of HDL_2a was significantly reduced, while both HDL_3b and HDL_3c became significantly more abundant.

Whereas the roles of lipid transfer proteins in HDL remodeling have been studied only independently in previous studies, the combined roles of CETP and PLTP in determining in vivo the relative proportions of the plasma HDL_2b, HDL_2a, HDL_3a, HDL_3b, and HDL_3c subpopulations remained to be clearly established. That important issue was addressed in the present study by following simultaneously CETP activity, PLTP activity, and HDL size distribution in total plasma from alcoholic patients before and after alcohol withdrawal. We chose to study the latter parameters in alcoholic patients entering a cessation program for the following reasons: (i) alcohol consumption is known to be associated with marked, significant changes in the level and distribution of plasma HDL (21, 22, 23, 24, 25, 26); (ii) alterations in plasma lipid transfers are suspected to contribute to the alcohol-induced rise in plasma HDL-cholesterol (17, 18, 19, 20); (iii) plasma lipid parameters (18, 22, 27), distribution of HDL subpopulations (20, 38), and plasma lipid transfer activity (17, 19, 20, and this study) are significantly modified after alcohol withdrawal.

In good agreement with previous data (18, 20, 39), the characteristic decrease in plasma HDL cholesterol levels observed after alcohol cessation was accompanied by a shift of HDL particles toward the small range. More precisely, we observed significant reductions in the relative proportions of the two largest HDL subpopulations, i.e., HDL_2b and HDL_2a, with concomitant significant rises in the relative proportions of the two smallest HDL subpopulations, i.e., HDL_3b and HDL_3c. These observations are in good agreement with data from other groups indicating that alcohol withdrawal is accompanied by significant and specific decrease in the cholesterol content of HDL_2b and HDL_2a subfractions (20). In turn, several groups (26, 40) reported that alcohol intake in adult males and females can be associated with specific increases in large HDL subfractions as determined by using native polyacrylamide gradient gel electrophoresis. In agreement with previous observations (19), plasma cholesteryl ester transfer activity was significantly increased after alcohol withdrawal due to a significant increase in plasma CETP mass. Concomitantly the distribution of plasma cholesterol among lipoprotein fractions was significantly modified after alcohol withdrawal with a significant decrease in HDL cholesterol levels and a significant increase in VLDL + LDL cholesterol levels. In agreement with previous observations in healthy normolipidemic subjects (31) or in alcoholic patients (17) CETP activity measured in total plasma from alcoholic patients, before as well as after alcohol withdrawal, correlated negatively with plasma HDL cholesterol levels but positively with cholesterol, triglyceride, and VLDL + LDL cholesterol levels. One of the main findings of the present study is that not only CETP activity, but also PLTP activity, is significantly affected by withdrawal therapy in alcoholic patients. However, unlike cholesteryl ester transfer activity, plasma phospholipid transfer activity was significantly reduced after alcohol cessation. Interestingly, correlations of the cholesterol content of individual lipoprotein fractions with phospholipid transfer activity were in contrast with those observed with cholesteryl ester transfer activity.

Significant relationships between alterations in the distribution profile of plasma HDL and changes in lipid transfer activities induced by alcohol withdrawal were observed. As previously observed by Välimäki and co-workers (20), alcohol cessation was accompanied by simultaneous increase in plasma CETP activity and decrease in the cholesterol concentration of the largest HDL_2a and HDL_2b subpopulations. However, in that latter study (20) no correlations between changes in CETP activity and changes in HDL₂ cholesterol levels were demonstrated in 11 alcoholic women following a detoxication program. In contrast, we observed significant correlations between changes in lipid transfer activities and changes in the relative abundance of various HDL subpopulations in 33 alcoholic men. Overall, significant relationships were found in support of a significant contribution of CETP and PLTP in changing the size distribution of HDL. Indeed, in the present study, changes in CETP activity correlated negatively with changes in the relative abundance of HDL_2a and positively with changes in the relative abundance of HDL_3b, while changes in PLTP activity correlated positively with changes in the relative abundance of HDL_2b and negatively with changes in the relative abundance of HDL_3a. Thus, in combination with in vitro observations made with isolated lipoproteins and purified lipid transfer proteins, observations in alcoholic patients suggest that plasma PLTP can promote the formation of HDL_2b particles at the expense of HDL_3a, while plasma CETP can promote the formation of HDL_3b particles at the expense of HDL_2a.

Based on recent observations, differences in the conversion activity of CETP and PLTP might rely on differences in the molecular mechanism of HDL conversion. The CETP-mediated size reduction of HDL would involve at start the net mass transfer of core neutral lipids from HDL donors toward lipoprotein acceptors (i.e. apoB-containing lipoproteins in plasma or HDL acceptors in reconstituted mixtures containing only isolated HDL and purified CETP) (10). The depletion in HDL core in vivo might be explained by two distinct mechanisms: either (i) the net transfer of cholesteryl esters from HDL donors toward lipoprotein acceptors without equimolecular net mass transfer of triglycerides in the opposite direction (10), or (ii) the hepatic lipase-catalyzed hydrolysis of HDL triglycerides initially transferred by CETP (7). In both cases, since the amounts of surface components would remain virtually unchanged, the surface-to-core ratio of HDL would progressively increase, and apoA-I excess would finally dissociate from the HDL surface, leading to the generation of a new HDL product of smaller size (41, 42, 43). As observed in the present study in alcoholic patients undergoing a withdrawal therapy, that latter mechanism might promote in vivo mainly the shift of the plasma HDL_2a subfraction toward the small HDL_3b fraction. It is noteworthy that the interdependence between HDL_2a and HDL_3b subfractions is compatible with the apolipoprotein composition of the two types of particles that contain both apolipoprotein A-I and apolipoprotein A-II as reported either in total human plasma (44) or in reconstituted mixtures containing isolated HDL and purified CETP (41). As recently proposed by Lusa et al. (45), the PLTP-mediated HDL conversion would instead involve initially a fusion of HDL particles and not a molecular transfer of core lipid species. Whereas the PLTP-mediated phospholipid transfer activity does not account directly for the HDL enlargement, it might favor the initiating step of particle fusion (45). In a last step, the resulting fusional complex would split into both new HDL particles of large size and apoA-I molecules containing only trace amounts of lipids (14, 16, 45). As suggested by the data of the present study, PLTP might then account in vivo for the transformation of HDL_3a particles into HDL_2b particles. Since HDL_2b were shown to contain only apoA-I, and not apoA-II, it is suggested that in total plasma PLTP might preferentially interact with HDL_3a-AI rather than with HDL_3a-AIAII particles. In support of that latter view, apoA-I-containing reconstituted HDL, but not apoA-II-reconstituted HDL, were converted into large particles upon incubation in the presence of purified PLTP (45).

In conclusion, the data of the present study demonstrated that CETP and PLTP can exert opposite effects on the HDL distribution profile in human plasma. Observations in alcoholic patients before and after alcohol withdrawal suggested further that, in addition to increased synthesis of apoA-I (46, 47, 48) and changes in lipase activity (20, 23, 49, 50), the combination of reduced CETP activity and increased PLTP activity may account for the shift of the plasma HDL fraction toward the large HDL_2a and HDL_2b subpopulations in alcoholic patients.