INTRODUCTION

In human plasma, the cholesteryl ester transfer protein (CETP)¹ promotes the exchange of cholesteryl esters and triglycerides between various lipoprotein fractions (1). In vivo, CETP activity results in the net mass transfer of cholesteryl esters from the antiatherogenic high density lipoproteins (HDLs) toward the proatherogenic apoB-containing lipoproteins, i.e. very low density lipoproteins and low density lipoproteins (LDLs) (1). The mechanism of action of human CETP has now been clearly established, and it involves two main steps. In a first step, positively charged amino acids of CETP can interact with negative charges of lipoprotein particles from which, in a second step, it picks up or deposits neutral lipid molecules (for a review, see Ref. 2). The CETP-lipoprotein binding has been shown to constitute one major determinant of the CETP-mediated neutral lipid transfer reaction (3), and both insufficient and excessive binding of CETP to lipoproteins resulting from alterations in the electrostatic charge at the lipoprotein surface were shown to decrease the cholesteryl ester transfer rate (4, 5). Recent studies revealed that two distinct positively charged amino acid residues, i.e. Lys₂₃₃ and Arg₂₅₉, are essential for the binding of CETP to lipoproteins, and consistent observations were made by using point mutagenesis (6) and specific monoclonal antibodies (7). However, whereas the lipoprotein binding domain of CETP has now been characterized, the lipoprotein components involved in the interaction with CETP remain to be clearly identified. During the last 15 years, both lipids and apolipoproteins were described as putative modulators of CETP activity (2). In particular, in vitro studies showed that the replacement of apoA-I by apoA-II in plasma HDL can modulate the rate at which cholesteryl esters are transferred between HDL and LDL (8), and the latter phenomenon was associated with apolipoprotein-mediated changes in the electrostatic charge of HDL (5). It is noteworthy however that the effect of apolipoproteins remains controversial, and HDL apolipoproteins have been alternatively described as neutral (9, 10, 11), inhibitory (8, 12), or activating factors (13, 14) of the cholesteryl ester transfer process. In fact, several studies reported the presence of a specific lipid transfer inhibitor protein (LTIP) in plasmas and lipoproteins from several species, including rats (15), baboons (16), and humans (10, 17-20). Although it has been recently proposed that baboon LTIP could consist in the association of one molecule of apoA-I with one 38-amino acid-long, N-terminal fragment of apoC-I, LTIP has not been characterized in human or rodent plasmas. Nevertheless, concordant observations have demonstrated that lipid transfer inhibitor activity could constitute an important point in determining plasma CETP activity (18, 19).

Recently, transgenic mice expressing human apolipoproteins in addition to enzymes or transfer proteins provided new insights into the role of HDL apolipoproteins in lipoprotein metabolism (21-29). In particular, the expression of human apoA-I in transgenic mice was shown to enhance the activity of coexpressed CETP (30-33), lecithin:cholesterol acyltransferase (28) or phospholipid transfer protein (29). However, the precise mechanism that accounts for the effect of human A-I was not fully elucidated, and in particular it has not been clearly established whether reported variations relate directly to specific properties of human apoA-I or are secondary to alterations in the size, the lipid content, or the electrostatic charge of plasma HDL. In addition, the study of the putative effect of human versus mouse apolipoproteins in vivo was complicated by the fact that different mouse lines present different HDL levels independently of the expression of human CETP, lecithin:cholesterol acyltransferase, or phospholipid transfer protein (28, 29, 31).

In the present study, HDL particles were isolated from plasma of transgenic mice that expressed either only human apoA-I (HuAITg mice), only human apoA-II (HuAIITg mice), or both human apoA-I and human apoA-II (HuAIAIITg mice) but that did not express human CETP. Subsequently, human HDL, control mouse HDL, HuAITg HDL, HuAIITg HDL, and HuAIAIITg HDL were analyzed for their composition and electronegative charge. The ability of the distinct HDL fractions to exchange cholesteryl esters with radiolabeled LDL in the presence of purified CETP was studied in parallel.

MATERIALS AND METHODS

Fresh citrated plasma from fasting normolipidemic subjects was provided by the Centre de Transfusion Sanguine (Hôpital du Bocage, Dijon, France). Fasting plasma samples from mice were collected into EDTA-containing tubes. Four distinct mouse lines were used in the present study: C57BL/6 control mice, transgenic mice AI-2 expressing human apoA-I (HuAITg mice) (24), transgenic mice AI-2/AIIFu-9 expressing both human apoA-I and human apoA-II (HuAIAIITg mice) (25), and transgenic mice phAT-III/hapoA-II expressing human apoA-II (HuAIITg mice) (34).

HDL were ultracentrifugally isolated from mouse and human plasmas as the 1.070 < d < 1.210 g/ml fraction, with one 7-h, 120,000 rpm spin at the lowest density and two 10-h, 120,000 rpm spins at the highest density in a TL 120.2 rotor in an optima TLX ultracentrifuge (Beckman, Palo Alto, CA). Densities were adjusted by the addition of solid KBr. The isolated lipoproteins were dialyzed against a 10 mmol/liter Tris, 150 mmol/liter NaCl, 3 mmol/liter NaN₃, pH 7.4, buffer (TBS buffer).

Lipoprotein-deficient plasma was obtained after precipitation of total plasma lipoproteins with dextran sulfate, MnCl₂, and BaCl₂ according to the general procedure described by Morton et al. (18). Lipoprotein-deficient plasma fractions were dialyzed overnight against TBS buffer.

Ultracentrifugally isolated HDL were delipidated by using butanol-diisopropyl ether 40:60 (v/v) according to the general procedure of Cham and Knowles (35). The aqueous solution containing delipidated HDL apolipoproteins was dialyzed overnight against TBS buffer.

The electrophoretic mobility (U) of HDL particles was determined by electrophoresis on 0.5% agarose gels (Paragon Lipo kit, Beckman) according to the method of Sparks and Phillips (38) and modified as described previously (5). Surface charges of HDL were estimated by using the equations given by Sparks and Phillips (38). U values were calculated by dividing the electrophoretic velocity (mean migration distance (mm)/time (s)) by the electrophoretic potential (voltage (V)/gel distance (cm)) (38). To correct the pI-dependent retardation effects, the following equation was applied.

(Eq. 1)

Finally, the surface potentials of lipoproteins were calculated by using Henry's equation (38),

(Eq. 2)

where D is the solvent dielectric constant and  is the coefficient of viscosity (0.0089 poise).

HDL (protein concentration, 1 g/liter) were incubated for 30 min at 37 °C in the presence of SDS (10 g/liter). Subsequently, HDL apolipoproteins were analyzed by SDS electrophoresis in 80-250 g/liter polyacrylamide gradient gels (Phastsystem, Pharmacia Biotech Inc.) as described by the manufacturer. After Coomassie staining, the distribution profile of HDL apolipoproteins was obtained by image analysis on a BIO-RAD GS-670 imaging densitometer. The apparent molecular weights of individual protein bands were determined by comparison with protein standards (Pharmacia Low Molecular Weight calibration kit). The relative abundance of apoA-I and apoA-II was obtained by determining the ratio of corresponding areas under the densitometric curve.

CETP was purified from 2500 ml of citrated, normolipidemic human plasma by using the sequential procedure previously described (39). The active fractions were pooled, aliquoted, and stored at 80 °C. The CETP preparation was deprived of both lecithin:cholesterol acyltransferase and phospholipid transfer protein activities (39). The mass concentration of CETP in active fractions was determined by using an enzyme-linked immunosorbent assay as described previously (40).

No CETP activity was detected in isolated lipoprotein fractions or mouse plasma samples used throughout the study, and cholesteryl ester transfers were induced in all the experiments by the addition of purified human CETP.

Cholesteryl ester transfer activity was determined by measuring the transfer of radiolabeled cholesteryl esters from [³H]CE-LDL to unlabeled acceptor HDL (8). Human LDL were biosynthetically labeled according to the procedure previously described (8). Briefly, isolated HDL fractions were incubated for 3 h at 37 °C with [³H]CE-LDL (2.5 nmol of cholesterol) in the presence of partially purified CETP (4.5 µg) in a final volume of 50 µl. At the end of the incubation, the d < 1.068 and the d > 1.068 g/ml fractions were separated by ultracentrifugation and transferred into counting vials containing 2 ml of scintillation fluid. The radioactivity was assayed for 2 min in a Wallac 1410 liquid scintillation counter (Pharmacia). The recovery of total radioactivity in the d < 1.068 and in the d > 1.068 g/ml fraction was greater than 95%. Cholesteryl ester transfer was expressed as the percentage of total radioactivity transferred from [³H]CE-LDL toward the d > 1.068 g/ml fraction, after deduction of blank values from control mixtures which were incubated at 37 °C without CETP.

Binding assays were performed by incubating CETP (2.8 mg/liter) with different concentrations of HDL particles in PBS buffer. After 30 min of incubation at 37 °C, bis(sulfosuccinimidyl)suberate (BS₃) (10 mmol/liter) was added to the mixture, and the incubation was extended for 30 min at room temperature. The cross-linked protein complexes were separated by SDS electrophoresis in 40-150 g/liter gels (Phastsystem) and transferred to a nitrocellulose membrane by using a Phast semidry electrophoretic transfer system (Pharmacia) according to the general procedure described by the manufacturer. The resulting blots were blocked for 1 h in 10% low fat dried milk in PBS containing 0.1% Tween and washed with PBS Tween. CETP-containing protein bands were revealed after a 1-h incubation of nitrocellulose blots with TP1 anti-CETP antibodies, which were kindly provided by Dr. Milne and Dr. Marcel (Heart Institute, Ottawa, Canada). Finally, nitrocellulose membranes were washed with TBS Tween, incubated with horseradish peroxidase-coupled second antibodies, and then developed with a 0.5 g/liter 4-chloro-1-naphtol (Bio-Rad), 20% methanol, 0.02% H₂O₂ solution. The distribution profiles of cross-linked proteins were obtained by image analysis on a Bio-Rad GS-670 imaging densitometer, and the relative abundance of monomeric, non-cross-linked CETP was obtained by determining the ratio of the monomeric CETP peak to the total area under the densitometric curve. In control mixtures containing only purified CETP, more than 95% of CETP appeared as one single, sharp band corresponding to monomeric unbound CETP.

All chemical assays were performed on a Cobas-Fara Centrifugal Analyzer (Hoffmann-La Roche). Protein concentration was measured with bicinchoninic acid reagent (Pierce). Total cholesterol and unesterified cholesterol concentrations were measured by enzymatic methods using Boehringer Mannheim reagents. Triglyceride and phospholipid concentrations were measured by enzymatic methods using Roche and Biomérieux reagents, respectively.

One-way analysis of variance was used to determine the significance between data means.

RESULTS

As shown in Table I, some differences appeared in the lipid and protein composition of the different HDL fractions studied. As compared with control mice, expression of either only human apoA-I, only human apoA-II, or both human apoA-I and human apoA-II significantly increased the protein content of HDLs and significantly reduced their phospholipid content. Overall, the expression of human apolipoproteins in mice tended to bring the surface lipoprotein parameters toward values similar to what is observed in normal human HDL. In contrast, the neutral lipid core of HDL from transgenic mice as well as from control mice did not match the neutral lipid core of human HDL that contained higher amounts of triglycerides (Table I). Interestingly, the coexpression of human apoA-I and human apoA-II in transgenic mice produced HDL particles with apoA-II to apoA-I plus apoA-II ratios similar to those observed in human HDL (Table I).

Table I. Composition (mass percent) of HDL from human and mouse plasmas Lipid and protein concentrations were determined as described under "Materials and Methods," and A-II/A-I + A-II ratios were determined by densitometric scanning of 8-25% SDS polyacrylamide gradient gel as described under "Materials and Methods." CE, cholesteryl esters; FC, free cholesterol; TG, triglycerides; PL, phospholipids; ND, not detectable. HDL source CE FC TG PL Protein A-II/A-I + A-II % HuAITg mice 15.2  ± 0.7 4.9  ± 0.4 1.9  ± 0.3 30.3  ± 1.5 47.7  ± 2.7 ND HuAIAIITg mice 10.2  ± 0.7a,b 4.3  ± 1.4 1.5  ± 0.9 31.6  ± 0.3 52.4  ± 0.9b 27.5  ± 1.7 Hu AIITg mice 9.9  ± 1.0a,b 4.2  ± 0.3 2.2  ± 0.3 31.1  ± 1.5 52.6  ± 0.8b 40.8  ± 4.0 Control mice 13.8  ± 2.3 6.9  ± 1.5b 2.2  ± 0.4 36.8  ± 1.8a,b,d,e 40.2  ± 1.7a,b,d,e 12.0  ± 3.0 Human 15.2  ± 0.9 3.8  ± 0.7 4.8  ± 1.2a,d,e,f 30.1  ± 0.1 46.2  ± 0.6 20.5  ± 2.0 a Significantly different from HuAITg mice, p < 0.05. b Significantly different from human, p < 0.05. c Apo AII corresponds to the sum of human + mouse apoAII. d Significantly different from HuAIAIITg mice, p < 0.05. e Significantly different from HuAIITg mice, p < 0.05. f Significantly different from control mice, p < 0.05.

As shown in Table II, mean surface potentials of human HDL and control mouse HDL were 11.7 ± 0.1 mV and 12.7 ± 0.3 mV, respectively. HDL from HuAITg mice exhibited the highest electronegative charge, with a mean surface potential of 13.5 ± 0.5 mV. In contrast, expression of human apoA-II as well as coexpression of human apoA-I and human apoA-II in transgenic mice produced HDL particles with low electronegative charge (mean surface potentials: 11.9 ± 0.3 and 12.2 ± 0.3 mV, respectively). Overall, HDL fractions containing human apoA-II, i.e. HDL from human, HuAIAIITg mouse, and HuAIITg mouse plasmas presented significantly lower surface potentials than HDL from HuAITg mice (p < 0.05 in all cases).

Table II. Electronegativity and relative affinity of CETP for HDL from human and mouse plasmas Electronegativity of the particles was estimated by their surface potential, which was calculated from their electrophoretic mobilities on 0.5% agarose gel as described under "Materials and Methods." Relative affinity of CETP for the different particles was estimated by densitometric scanning of Western blot as described under "Materials and Methods." EC50 values are the HDL concentrations for which 50% of the CETP is not associated with lipoprotein particles. Data are mean ± S.D. of three determinations. HDL source Mean surface potential EC50 mV nmol/liter HuAITg mice 13.5  ± 0.5 40  ± 20 HuAIAIITg mice 12.2  ± 0.3a 130  ± 20 HuAIITg mice 11.9  ± 0.3a NDb Control mice 12.7  ± 0.3 70  ± 25 Human 11.7  ± 0.1a,c 120  ± 30 a Significantly different from HuAITg mice, p < 0.05. b ND, not determined. c Significantly different from control mice, p < 0.05.

Fig. 1 shows the cholesteryl ester transfer rates obtained with purified CETP (final concentration, 0.17 µg/ml) and with various HDL concentrations (HDL cholesterol concentration range, 20-400 nmol/ml). With all of the HDL fractions studied, cholesteryl ester transfer rates progressively increased until a maximal transfer value was obtained with an HDL cholesterol concentration of 100 nmol/ml (Fig. 1). At the optimal concentration of 100 nmol/ml, cholesteryl ester transfer rates were significantly lower with human HDL than with control mouse HDL (p < 0.05) (Fig. 1). Striking differences between distinct HDL fractions appeared, while the HDL levels kept to increase above the optimal value of 100 nmol/ml. On the one hand, raising the concentration of HDL from HuAITg or HuAIAIITg mice did not modify further the cholesteryl ester transfer rate with a plateau corresponding to the maximal transfer value being maintained up to the highest concentration studied (Fig. 1). On the other hand, increasing the concentration of human HDL and control mouse HDL above the 100 nmol/ml optimal value induced a marked, progressive, and parallel inhibition of cholesteryl ester transfer activity (Fig. 1). At a concentration of 400 nmol/ml, cholesteryl ester transfer rates measured with control and human HDL were significantly lower than values obtained with identical concentrations of HDL from HuAITg and HuAIAIITg mice (p < 0.05 in all cases). As shown in Fig. 2, the maximal transfer rate measured with HuAIITg HDL (100 nmol/ml of cholesterol) was significantly higher than the maximal transfer value measured with the same level of HuAITg HDL. In agreement with Fig. 1, cholesteryl ester transfer rates measured with 100 and 400 nmol/ml of HuAITg HDL did not differ significantly. In contrast, raising the level of HuAIITg HDL and control mouse HDL from 100 up to 400 nmol/ml was associated with significant reductions in CETP activity (Fig. 2).

As shown in Fig. 3, delipidated HDL apolipoproteins from control mice induced a concentration-dependent inhibition of the CETP-mediated transfer of cholesteryl esters from [³H]CE-LDL to HDL₃. The progressive inhibition of cholesteryl ester transfer rates no longer appeared when HDL apolipoproteins were preheated for 1 h at 56 °C, and cholesteryl ester transfer rates measured in the presence of 100 µg of control mouse HDL apolipoproteins were significantly higher with preheated HDL apolipoproteins than with homologous preparations maintained at 4 °C prior to lipid transfer determination (p < 0.001). In contrast to data obtained with delipidated HDL apolipoproteins from control mice, no significant inhibition of the CETP-mediated cholesteryl ester transfer reaction was observed when adding either heated or nonheated HDL apolipoproteins from HuAITg mice (Fig. 3).

Whereas the same isolated lipoprotein and purified CETP fractions were used in all incubation mixtures, striking differences in CETP-mediated cholesteryl ester transfer rates could be noted in the presence of nonheated lipoprotein-deficient plasmas from either control mice or HuAITg mice (Fig. 4). Interestingly, preheating of lipoprotein-deficient plasma fractions for 1 h at 56 °C produced opposite effects on cholesteryl ester transfer rates. Indeed, as shown in Fig. 4, preheating markedly and significantly increased CETP activity in lipoprotein-deficient plasma from control mice, whereas it significantly reduced CETP activity in lipoprotein-deficient plasma from HuAITg mice. Following preheating treatment, cholesteryl ester transfer rates were remarkably similar in both control and HuAITg lipoprotein-deficient plasmas, and they were clearly higher than those measured in nonheated lipoprotein-deficient plasma from control animals (Fig. 4).

The affinity of CETP for individual HDL fractions was evaluated by using semiquantitative densitometric analysis (41), and cross-linking was used in the present study to rule out any dissociation of CETP-lipoprotein complexes during electrophoresis. As shown in Fig. 5, in the absence of HDL, CETP appeared as one single band with an approximately 74-kDa molecular mass (CETP sample), and cross-linking of purified CETP did not affect the electrophoretic migration of the monomeric protein (CETP+BS₃ sample). In contrast, when incubated in the presence of human HDL (CETP+HDL+BS₃ sample), CETP was detected in two distinct bands with mean molecular masses of approximately 74 and 200 kDa, corresponding to free CETP and to CETP associated with HDL apolipoproteins, respectively. As shown in Fig. 6, with human HDL, control mouse HDL, HuAITg HDL, and HuAIAIITg HDL, the relative proportion of free, unbound CETP was gradually reduced as the HDL amount added to the experimental mixtures increased. The apparent EC₅₀ values were determined for each HDL fraction as the HDL concentration at which 50% of CETP remained unassociated (Fig. 6). As shown in Table II, EC₅₀ values calculated for individual HDL fractions were in the following order: HuAIAIITg HDL > human HDL > control mouse HDL > HuAITg HDL. In fact, EC₅₀ values measured with HDL from human and HuAIAIITg mouse plasmas were higher than EC₅₀ values obtained with HDL from control and HuAITg mice (120 ± 30 and 130 ± 20 nmol/liter versus 70 ± 25 and 40 ± 20 nmol/liter, respectively), suggesting that CETP has a lower affinity for human HDL and HuAIAIITg HDL than for control mouse HDL and HuAITg HDL.

DISCUSSION

Compared with transgenic mice expressing only human CETP (HuCETPTg mice), the coexpression of human apoA-I and human CETP (HuAICETPTg mice) was shown to induce a greater decrease in both plasma HDL cholesterol levels and HDL particle size (30). The latter data were explained in terms of an enhanced interaction of human CETP with lipoproteins containing human apoA-I in vivo, and a stronger association of CETP with plasma HDL was reported in HuAICETPTg mice versus HuCETPTg mice (30). These observations suggest that alterations in the plasma HDL pool, and in particular in its content in human apoA-I, might affect the plasma cholesteryl ester transfer reaction independently of the plasma CETP mass concentration. However, it is not clear at this stage whether the CETP-modulating potency of plasma HDL, and in particular of HuAICETPTg HDL, is due to a specific interaction of human CETP with human apoA-I, to specific changes in the properties of human apoA-I-containing HDL, or simply to the higher plasma HDL concentrations resulting from the overexpression of human apoA-I in transgenic animals. In addition to apoA-I, the expression of human apoA-II in transgenic mice was shown to result in increased triglyceride content of HDL particles containing both human apoA-I and human apoA-II when HDL were isolated in the presence of the lipase inhibitor E600 (31). These in vivo observations were explained in terms of the ability of human apoA-II to inhibit hepatic lipase activity in HuAIICETPTg animals but not in terms of the ability of apoA-II to modulate CETP activity (31). In the present study, five sources of HDL, i.e. human HDL, control mouse HDL, HuAITg mouse HDL, HuAIITg mouse HDL, and HuAIAIITg mouse HDL were used, and their abilities to interact with purified human CETP were studied in parallel. The present in vitro approach allowed us to study in greater detail the CETP-HDL interactions independently of uncontrolled variations in the size of the HDL pool, in CETP mass concentrations, and in other enzyme activities that are likely to occur in vivo from one mouse line to another.

When the ability of increasing amounts of various HDL preparations to act as substrates for CETP was studied, maximal cholesteryl ester transfer rates were obtained in all cases with similar HDL:CETP ratios. Contrary to all expectations, human HDL constituted less efficient lipoprotein substrates in the CETP reaction than control mouse HDL. These observations do not support the possibility of a preferential ability of human versus mouse apolipoproteins to mediate interaction with human CETP, and they indicate that the differential properties of human versus mouse apolipoprotein A-I (42) do not directly affect CETP activity. Whereas previous in vitro studies indicated that the replacement of apoA-I by apoA-II in human HDL leads to a substantial inhibition of the CETP-mediated lipid transfer reaction (5, 8), maximal cholesteryl ester transfer values measured with HuAITg HDL and HuAIAIITg HDL were similar, and HuAIITg HDL with the highest A-II to A-I+A-II ratio constituted some better substrates for CETP, suggesting that the relative proportion of apoA-I and apoA-II in plasma HDL is probably not a major determinant of CETP activity in vivo (5, 9, 31).

It must be emphasized that maximal cholesteryl ester transfer rates measured with HDL from various sources were obtained by using low HDL:CETP ratios, which did not correspond to conditions normally occurring in vivo. Interestingly, with higher HDL:CETP ratios, striking differences in the ability of the various HDL fractions to act as lipoprotein substrates in the CETP-mediated lipid transfer reaction were observed. Indeed, the transfer rates measured with human HDL, control mouse HDL, and HuAIITg HDL significantly decreased with elevated HDL concentrations, whereas no inhibition was observed with either HuAITg HDL or HuAIAIITg HDL. While an inhibitory effect of high HDL levels has been previously reported by others using human and rodent plasma HDL (8, 15, 43, 44), the molecular mechanism of CETP inhibition remains a matter of controversy, due at least in part to the complexity of lipid transfer kinetics (45). At least two distinct explanations can be proposed to account for the inhibitory effect observed in the present study: (i) an excess of HDL might reduce the interaction of CETP with the other lipoprotein particles through a substrate inhibition mechanism (43, 45), and (ii) HDL might contain a specific inhibitor of the CETP-mediated transfer reaction (10, 15, 16). The former hypothesis, derived from the general model of lipid transfer (45), holds that high concentrations of HDL can sequester CETP, resulting in a slower transfer rate. In other words, the lack of lipid transfer inhibition with human apoA-I-containing, transgenic mouse HDL along the concentration range studied might relate to apoA-I-induced alterations of transgenic HDL that would prevent the expected sequestration of CETP. Interestingly, Tollefson et al. (15) reported a concentration-dependent inhibition of CETP activity by some human plasma HDL fractions containing LTIP, whereas no inhibitory effect was observed with LTIP-free human plasma HDL fractions. The latter observations came in support of the second hypothesis, and they suggest that in the present study lipid transfer inhibitory activity might be present in HDL from human, HuAIITg mouse, and nontransgenic mouse plasmas, but it would be absent in HDL from transgenic mice expressing human apoA-I. To investigate further the latter hypothesis, delipidated HDL apolipoproteins were obtained from control and HuAITg mouse plasmas, and their ability to inhibit CETP activity was investigated with or without preheating treatment. Indeed, previous studies clearly demonstrated that LTIP is an heat-labile factor (10, 18), and this property has been recently applied to the evaluation of lipid transfer inhibitory activity in human plasma (18). The use of a similar experimental approach in the present study suggests that a heat-labile lipid transfer inhibitory activity is associated with the HDL protein moiety of control mice, whereas it is virtually absent from the protein moiety of HuAITg HDL. Complementary investigations revealed further that lipid transfer inhibitory activity is also detectable in the lipoprotein-deficient fraction from control mouse plasma. In contrast, lipid transfer inhibitory activity was absent from the lipoprotein-deficient fraction from HuAITg mouse plasma, and the heating treatment led to a significant decrease, rather than a significant increase, in cholesteryl ester transfer rate, reflecting possibly some uncontrolled degradation of activating plasma components. Although the reason for the absence of specific lipid transfer inhibitory activity in plasma from transgenic mice expressing human apoA-I is unclear, it might relate to the removal of the putative LTIP from HDL during their assembly as it has been proposed to occur for mouse apoA-I in HuAITg mice (24). In contrast, plasma lipid transfer inhibitory activity was still detectable in HDL from mice expressing only human apoA-II, which, unlike human apoAI, does not produce alterations in the levels of endogenous mouse apolipoproteins A-I and A-II (25).

It is now well established that the binding of CETP at the lipoprotein surface is driven in part by electrostatic interactions that constitute a rate-limiting step in the CETP-mediated lipid transfer process (3-5). In fact, during the last decade, alterations in lipoprotein electrostatic charge, as produced by using various experimental protocols (nonesterified fatty acid (46) or retinoic acid (47) enrichments, chemical modifications (4), in vitro changes in apolipoprotein composition (5, 8), and digestions with phospholipase A2 (48) or lipoprotein lipase (49)) were shown to result in dramatic changes in CETP activity. As shown in the present study, the electrostatic charge of control mouse HDL was intermediate between the electrostatic charge of HuAITg HDL and HuAIAIITg HDL (two HDL fractions showing no lipid transfer inhibitory activity), and the electrostatic charge, as well as the lipid composition of HuAIITg HDL and HuAIAIITg HDL (two HDL fractions with and without lipid transfer inhibitory activity, respectively) were remarkably similar, suggesting that the electronegativity and lipid content of HDL are unlikely to account for differences in inhibitory effects. Although the general procedure used to quantitate CETP-HDL binding in the present study was a nonequilibrated method, complementary affinity binding experiments confirmed, in agreement with previous studies (4, 5, 41), a good correspondence between the electronegativity of human, control mouse, HuAITg mouse, and HuAIAIITg mouse HDL fractions and their ability to bind CETP. However, differential lipid transfer inhibitory potencies of distinct HDL cannot be explained by the observed differences in CETP-HDL binding, and, as observed above with lipoprotein electrostatic charge, the apparent EC₅₀ mean value determined with control mouse HDL was intermediate between EC₅₀ mean values measured with HuAITg HDL and HuAIAIITg HDL.

In conclusion, data of the present study suggested new and concordant arguments in favor of the presence of a specific lipid transfer inhibitory activity in human, control mouse, and HuAIITg mouse HDL, an inhibition that was not detected in HDL from HuAITg and HuAIAIITg transgenic mice. Although affinity binding experiments demonstrated that the interaction of CETP is driven by the electronegativity of HDL, alterations in CETP-lipoprotein binding are unlikely to account for the observed differences in lipid transfer inhibitory activities in plasma samples from various sources.