Molecular Mechanism of the Blockade of Plasma Cholesteryl Ester Transfer Protein by Its Physiological Inhibitor Apolipoprotein CI*

Genetically engineered mice demonstrated that apolipoprotein (apo) CI is a potent, physiological inhibitor of plasma cholesteryl ester transfer protein (CETP) activity. The goal of this study was to determine the molecular mechanism of the apoCI-mediated blockade of CETP activity. Kinetic analyses revealed that the inhibitory property of apoCI is independent of the amount of active CETP, but it is tightly dependent on the amount of high density lipoproteins (HDL) in the incubation mixtures. The electrostatic charge of HDL, i.e. the main carrier of apoCI in human plasma, is gradually modified with increasing amounts of apoCI, and the neutralization of apoCI lysine residues by acetylation produces a marked reduction in its inhibitory potential. The inhibitory property of full-length apoCI is shared by its C-terminal α-helix with significant electrostratic properties, whereas its N-terminal α-helix with no CETP inhibitory property has no effect on HDL electronegativity. Finally, binding experiments demonstrated that apoCI and to a lower extent its C-terminal α-helix are able to disrupt CETP-lipoprotein complexes in a concentration-dependent manner. It was concluded that the inhibition of CETP activity by apoCI is in direct link with its specific electrostatic properties, and the apoCI-mediated reduction in the binding properties of lipoproteins results in weaker CETP-HDL interactions and fewer cholesteryl ester transfers.

The cholesteryl ester transfer protein (CETP) 4 mediates the exchange of neutral lipids, i.e. cholesteryl esters and triglycerides between plasma lipoproteins (1,2). Through its action, CETP can influence the atherogenicity of the lipoprotein profile (2)(3)(4), and recent studies (5-7) support a potential interest in inhibiting CETP activity in vivo by means of either anti-CETP immunotherapy, antisense oligonucleotides, or specific pharmacological inhibitors. In particular, the latest pharmacologi-cal interventions with small anti-CETP molecules in human populations demonstrated that CETP inhibition markedly increases HDL cholesterol levels and also decreases low density lipoprotein (LDL) cholesterol levels (8 -10). Most interestingly, previous studies (11)(12)(13) in human populations are consistent with the association of high CETP with an increase in coronary heart disease, in particular in subjects with elevated triglycerides.
Besides interventional studies with exogenous compounds, a number of studies (14 -19) indicated that plasma CETP activity can be modulated by endogenous factors, among them the apolipoprotein components of circulating lipoproteins. Recently, apolipoprotein CI, i.e. a 6.6-kDa HDL apolipoprotein, was identified as a potent CETP inhibitor (20). In contrast to other putative apolipoprotein candidates that were identified only through in vitro experiments, the ability of apoCI to decrease specific CETP activity was documented in vivo through studies in CETPTg/apoCI-KO and CETPTg/HuapoCITg (transgenic mouse to both human CETP and human apolipoprotein CI) mice (21,22). Although it was concluded that apoCI constitutes the major physiological inhibitor of CETP in the plasma compartment, the molecular mechanism of the blockade of the cholesteryl ester transfer reaction by apoCI remained to be identified. The CETP-mediated lipid transfer reaction is a complex process, with at least two rate-limiting steps. In the first step, CETP binds to lipoproteins through electrostatic interactions with negative charges localized at the lipoprotein surface (23)(24)(25). In the second step, and after a conformational change of CETP, one neutral lipid molecule binds to an hydrophobic site in the C terminus of the protein prior to be transferred to the lipoprotein acceptor (26,27). In concordance with the two steps of the CETP-mediated transfer reaction, at least two distinct ways of CETP blockade were reported in previous studies. First, CETP inhibition may result from either insufficient or excessive binding of CETP at the lipoprotein surface in step 1 of the lipid transfer process, and both weak CETP-lipoprotein interaction (25,28) and strong, sometimes irreversible, CETP-lipoprotein association (10,25,27) results in significant inhibition of the lipid transfer reaction. Second, CETP inhibition may result from the blockade of the neutral lipid binding site in step 2 of the lipid transfer process, resulting in an abnormal production of irreversibly associated CETP-lipoprotein complexes in this case (8,27,29,30).
In the present study, the effect of apoCI on the lipid transfer process was determined in a systematic way. Concordant in vitro observations indicate that the inhibitory property of apoCI is in a direct link with its electrostatic charge properties and its ability to produce significant changes in CETP-lipoprotein interactions.

MATERIALS AND METHODS
Plasma Samples-Fresh citrated plasmas from normolipidemic subjects were provided by the Etablissement Français du Sang (Hôpital du Bocage, Dijon, France).
Isolation of HDL Particles-Total HDL were ultracentrifugally isolated from normolipidemic human plasmas as the 1.070 Ͻ d Ͻ 1.210 g/ml fraction, with one 24-h, 45,000 rpm spin at the lowest density and one 24-h, 50,000 rpm spin at the highest density in a 70.Ti rotor in an L90-K ultracentrifuge (Beckman Instruments). HDL 2 and HDL 3 were isolated as the 1.070 Ͻ d Ͻ 1.125 g/ml and the 1.125 Ͻ d Ͻ 1.210 g/ml fractions, respectively. Isolation of HDL 2 and HDL 3 was conducted in an NVT-90 rotor in an L90-K ultracentrifuge (Beckman Instruments), with a 2.5-h, 90,000 rpm spin at density 1.070 g/ml, a 3-h, 90,000 rpm spin at density 1.125 g/ml, and a 3.5-h, 90,000 rpm spin at density 1.210 g/ml. Densities were adjusted by the addition of solid KBr. Isolated HDL, HDL 2 , and HDL 3 were dialyzed overnight against 10 mmol/liter Trisbuffered saline, pH 7.4, with 3 mmol/liter NaN 3 (TBS buffer).
Purification of Apolipoprotein CI and CETP-ApoCI was purified from delipidated HDL apolipoproteins by using the chromatofocusing method of Tournier et al. (31). This method takes advantage of the high isoelectric point of apoCI as compared with other HDL apolipoprotein components. Purified apoCI, which appeared as a homogeneous band on polyacrylamide gel, was dialyzed against TBS buffer. CETP was purified from human plasma by using a sequential chromatography procedure as described previously, and CETP preparation was deprived of both lecithin/cholesterol acyltransferase and phospholipid transfer protein activities (32,33).
Anti-apoCI Immunoaffinity Chromatography-ApoCI was removed from total human HDL by passage through an anti-apoCI immunoaffinity column as described previously (20). ApoCI-poor HDL that did not bind to the immunosorbent column were washed off with TBS buffer, and their ability to exchange cholesteryl esters was compared with total HDL.
Apolipoprotein CI Acetylation-ApoCI was acetylated as described previously (34). Briefly, 100 g of purified apoCI was incubated for 1 h at room temperature with a saturated solution of sodium acetate in the presence of increasing amounts of acetic anhydride (range, 0 -21 mol). At the end of the treatment, acetylated preparations were dialyzed overnight against TBS buffer containing 1 mmol/liter EDTA.
Matrix-assisted Laser Desorption Ionization-Time-of-Flight (MALDI-TOF) Mass Spectrometry of Native and Acetylated ApoCI-Prior to mass spectrometry analysis, native and acetylated apoCI were desalted on ZipTip microcolumns (ZipTip C18, Millipore) as described by the manufacturer. Briefly, 1 g of protein in 10 l was acidified with trifluoroacetic acid (final concentration, 0.1% trifluoroacetic acid). ZipTips were pre-wetted with 10 l of 50% acetonitrile in MilliQ grade water and equilibrated with 10 l of 0.1% trifluoroacetic acid. Bound proteins were washed twice with 10 l of 0.1% trifluoroacetic acid, and they were finally eluted with 4 l of 0.1% trifluoroacetic acid, 50% acetonitrile in MilliQ grade water. One microliter of eluted protein was mixed with 4 l of a saturated solution of ␣-cyano-4-hydroxycinnamic acid (Bruker Daltonique S.A., Wissembourg, France) in a 0.1% acetonitrile/trifluoroacetic acid (1:2, v/v) solution. Finally, 1 l of the mixture was spotted on a MTP 384 ground steel target plate (Bruker Daltonique S.A.).
Measurement of Cholesteryl Ester Transfer Activity-CETP activity was determined in microplates by a fluorescent method using donor liposomes enriched with nitrobenzoxadiazol (NBD)-labeled cholesteryl esters (phospholipid/cholesterol/NBD-cholesteryl ester molar ratio, 1:1:1) (Roar Biomedical). For measurements of cholesteryl ester transfer activity with isolated CETP, donor liposomes (phospholipids, 5 mol/ liter) and purified CETP (range, 2-16 g) were incubated in the presence of HDL (range, 0.5-4.0 mol/liter) or LDL (7 g of cholesterol), as indicated. For measurement of cholesteryl ester transfer activity in total plasma, incubation media contained donor liposomes (phospholipids, 5 mol/liter) and 10 l of plasma. Final volumes were adjusted to 250 l with TBS (unless specified), and incubations were conducted in triplicate for 3 h at 37°C in a Victor 2 1420 multilabel counter (PerkinElmer Life Sciences). The CETP-mediated transfer of NBD-cholesteryl esters from self-quenched donors to acceptor lipoproteins was monitored by the increase in fluorescence intensity (excitation, 465 nm; emission, 535 nm), and results were expressed in fluorescence arbitrary units after deduction of blank values that were obtained with control mixtures without CETP.
SDS-PAGE-HDL apolipoproteins were delipidated with ethanol/ ether (3:2), diluted in the sample buffer, and applied on 4 -12% NuPAGE® BisTris Novex SDS-polyacrylamide gels as recommended by the manufacturer (Invitrogen). Proteins were stained with Coomassie Brilliant Blue G-250, and the apparent molecular weights of individual protein bands were determined by reference to protein standards (Mark12, Invitrogen).
Agarose Gel Electrophoresis-The electrophoretic mobility (U in the following equations) of HDL was determined by electrophoresis on 0.5% agarose gels (Paragon Lipo kit, Beckman Instruments) according to the method of Sparks and Phillips (35). Briefly, gels were cast in a Sebia Tank K20 system, and electrophoresis was performed for 45 min at 100 V in barbital buffer, pH 8.6. After electrophoresis, the gels were successively fixed for 5 min in an ethanol/acetic acid/water 60:10:30 solution, dried, stained for 5 min in a 0.07% solution of Sudan Black B in ethanol/water 55:45, and destained for 10 min with a solution of ethanol/water 45:55. In parallel, gel portions containing purified bovine serum albumin, which was used as an internal standard, were stained with a 0.8 g/liter solution of Coomassie Brilliant Blue G-250 in a methanol/acetic acid/ water 10:1:10 solution and destained in a solution of methanol/acetic acid/water 2:3:40. Mean migration distances were obtained by using the GelDoc analysis software (Bio-Rad).
Calculation of Electrophoretic Mobility and Surface Potential of HDL Particles-Surface charges of HDL were determined as described previously (35). Briefly, electrophoretic mobilities (U) were calculated by dividing the electrophoretic velocity (mean migration distance (mm) per time in seconds) by the electrophoretic potential (voltage per gel distance in centimeters). To correct the pI-dependent retardation effects, the Equation 1 was applied (35), The surface potentials of HDL were calculated by using the Henry's Equation 2 (see Ref. 35), where n is the coefficient of viscosity (0.0089 poise), and D is the solvent dielectric constant.
Formation of CETP-Lipoprotein Complexes-Ultracentrifugally isolated HDL were covalently bound to CNBr-activated Sepharose 4B (Amersham Biosciences) at a ratio of 10 mg of HDL proteins per g of gel as recommended by the manufacturer. The HDL-Sepharose phase was resuspended in PBS containing 3 mmol/liter NaN 3 . Each incubation mixture contained 100 l of the HDL-Sepharose suspension in PBS, corresponding to 70 g of HDL cholesterol. In a first step, purified CETP (final concentration, 3.75 mg/liter) was bound to the HDL-Sepharose suspension during a 1-h incubation at room temperature under mild agitation. At the end of the incubation period, the HDL-Sepharose phase with bound CETP was resuspended in 200 l of PBS, which contained increasing amounts of either purified apoCI, amino acids 4 -25 N-terminal apoCI fragment, or amino acids 34 -54, C-terminal apoCI fragment (concentration range, 0 -2 mol/liter). Mixtures were incubated for 1 h at room temperature under mild agitation, and supernatants containing unbound CETP that was released from the HDL-Sepharose phase were finally recovered after a gentle, 1-min low speed centrifugation. The amount of CETP released in the supernatant was determined by a specific immunoassay with TP1 anti-CETP monoclonal antibodies. Briefly, proteins in the incubation supernatants were applied to a nitrocellulose membrane (Hybond ECL, Amersham Biosciences) by using a dot blot vacuum system (Bio-Rad). The resulting blots were blocked for 30 min at 37°C in 3% low fat dried milk in Tris-buffered saline containing 0.1% Tween, and they were washed in Tris-buffered saline/Tween. CETP was revealed by successive incubations with TP1 anti-CETP antibodies (Heart Institute, Ottawa, Canada) and horseradish peroxidase-coupled second antibodies as described previously (36). Blots were finally developed with an ECL kit (Amersham Biosciences). Band intensities were measured by using the GelDoc analysis software (Bio-Rad). The amount of CETP in each supernatant was determined by comparison with a calibration curve that was obtained with known amounts of purified CETP that were applied to the membrane together with the samples.
Lipid and Protein Analyses-All assays were performed on a Victor 2 1420 multilabel counter (PerkinElmer Life Sciences). Total cholesterol was measured by the enzymatic method using the Cholesterol 100 reagent (ABX Diagnostics). Phospholipids and triglycerides were determined by enzymatic methods, as described previously (21). Protein concentration was measured by using the bicinchoninic acid reagent (Pierce).
Statistical Analyses-Mann-Whitney U test was used to determine the statistical significance between data means.

Comparative Effects of Increasing Levels of Human HDL Versus Human CETP on the ApoCI-mediated Inhibition of the Cholesteryl Ester
Transfer Reaction-When the amount of fluorescent liposome donors and HDL acceptors were kept constant in the CETP activity assay, the  cholesteryl ester transfer rate measured over a 3-h incubation period increased gradually with the amount of purified CETP added (from 2 g in Fig. 1A to 16 g in Fig. 1D). In all cases, and in accordance with previous in vitro and in vivo studies (20,21), the addition of apoCI was accompanied by a marked inhibition of the lipid transfer reaction. The inhibitory potential of a given amount of apoCI appeared to be independent of the amount of CETP added, in all cases with a constant ϳ70% reduction in cholesteryl ester transfer rate as compared with control mixtures with no apoCI added. Conversely, the inhibitory potential of purified human apoCI was markedly affected by the amount of HDL acceptors that were added to the incubation mixture, with CETP kept constant. In the latter case, the capacity of human apoCI to block the lipid transfer process, approximating 80% of inhibition in incubation mixtures with the lowest HDL levels, was completely abolished with the highest HDL concentration studied (Fig. 2). CETP inhibition by HDL is the consequence of a direct and specific property of apoCI, because apoCI-poor HDL displayed a much weaker ability to block the lipid transfer reaction in the HDL concentration range studied (Fig. 3). The apolipoprotein profiles of apoCI-poor HDL as prepared by anti-apoCI immunoaffinity chromatography and of native HDL differed only by their apoCI content (Fig. 3). Percent composition of the lipid moiety did not differ significantly between total and apoCI-poor HDL from three distinct preparations (total cholesterol %, 35.9 Ϯ 2.3 versus 31.9 Ϯ 1.6, respectively; triglyceride %, 8.2 Ϯ 0.5 versus 9.9 Ϯ 1.1, respectively; and phospholipid %, 55.9 Ϯ 2.4 versus 58.1 Ϯ 0.9, respectively). Overall, the results showed that the capacity of apoCI to block the lipid transfer reaction was independent of the amount of active CETP, but it was tightly dependent on the amount of HDL acceptors in the incubation mixtures.
Effect of Purified ApoCI on the Electrostatic Charge Properties of HDL-As illustrated in Fig. 4, apoCI produced a significant change in the surface charge properties of plasma HDL. Mean surface potential of native HDL increased gradually from Ϫ11.7 to Ϫ8.8 mV as the apoCI to HDL molar ratio increased from 0 to 6.4 (Fig. 4). Complementary experiments demonstrated that apoCI is able to modify the charge characteristics of both isolated HDL 2 and HDL 3 , however, with a greater impact on HDL 3 than on HDL 2 at the apoCI to HDL molar ratio of 1.6 and 3.2 (Fig. 5).
Loss of CETP Inhibitory Property of Acetylated ApoCI-To bring more insight into the role of basic, positively charged residues of apoCI in mediating its lipid transfer inhibitory property, purified human apoCI was incubated with increasing concentrations of acetic anhydride, i.e. a neutralizing agent of positively charged groups of lysine residues (34). As checked by MALDI-TOF analysis, incubation of human apoCI in the presence of 0.5 mol of acetic anhydride produced 42-Da increments in FIGURE 4. Mean surface potential of HDL in the presence of purified apoCI. Human HDL were isolated by ultracentrifugation, and apoCI was purified by chromatofocusing (see "Materials and Methods"). Mixtures containing HDL and various amounts of apoCI (range of apoCI to HDL molar ratio: 0 -6.4) were incubated for 1 h in PBS at room temperature. Subsequently, 5 l of the mixtures were submitted to electrophoresis in 0.5% agarose gels. After staining of the HDL bands, migration profiles were analyzed on an imaging densitometer, and the surface potential of HDL was calculated as compared with bovine serum albumin that was run on each gel as a standard (see "Materials and Methods"). Experiments were done in triplicate and gave similar results with three different HDL preparations. Sudan Black-stained gel in the upper panel shows a typical migration pattern. Vertical bars in the lower panel are the mean Ϯ S.D. of triplicate determinations with three different HDL preparations. *, significantly different from HDL incubated without apoCI, p Ͻ 0.05; Mann-Whitney test. FIGURE 5. Comparison of the ability of apoCI to induce changes in electronegative charge of HDL 2 and HDL 3 . Human HDL 2 and HDL 3 were isolated by ultracentrifugation, and apoCI was purified by chromatofocusing (see "Materials and Methods"). Mixtures containing HDL and various amounts of apoCI (range of apoCI to HDL molar ratio, 0 -6.4) were incubated for 1 h in PBS at room temperature. Subsequently, 5 l of the mixtures were submitted to electrophoresis in 0.5% agarose gels. After staining of the HDL bands, migration profiles were analyzed on an imaging densitometer, and the surface potential of HDL was calculated as compared with bovine serum albumin that was run on each gel as a standard (see "Materials and Methods"). Plotted data correspond to changes as compared with homologous control values with no apoCI added. They are mean Ϯ S.D. of triplicate determinations. a, significantly different from homologous HDL incubated without apoCI, p Ͻ 0.05; Mann-Whitney test. b, significantly different from HDL2 incubated with the same amount of apoCI, p Ͻ 0.05; Mann-Whitney test.  6C). As illustrated in Fig. 7, acetylation of apoCI in the presence of acetic anhydride led to the emergence of new isoforms as observed by isoelectrofocusing, with a gradual shift of apparent pI from 8.3 with native apoCI down to 3.2 with apoCI treated with the highest amount of acetic anhydride (Fig. 7). In the meantime, CETP activity rose from 10 (with native apoCI) to 77% (with acetylated apoCI) of control transfer values measured in the absence of apoCI, indicating that the inhibitory potential of apoCI is markedly reduced as a result of the neutralization of its lysine residues (Fig. 7).
Despite a significant reduction of apoCI chromogenicity after acetylation (Fig. 8A), acetylated apoCI was found to readily associate with HDL, and as for native apoCI, association with HDL did not induce profound changes in the apolipoprotein composition of the particles as shown by polyacrylamide gradient gel electrophoresis (Fig. 8B). Most strikingly, the binding of acetylated apoCI to HDL, unlike the binding of native apoCI (Fig. 4), did not produce a marked reduction in HDL electronegativity, even with an opposite tendency to shift toward lower surface potential values in the presence of increasing levels of acetylated apoCI (Fig. 9).
Differential Effects of Full-length or Fragmented ApoCI on CETP Activity and HDL Electronegativity-Two apolipoprotein CI fragments, i.e. fragment 34 -54 and fragment 4 -25 corresponding to the two distinct ␣-helices of human apoCI, were synthesized (Fig. 10A). Their ability to inhibit CETP activity was monitored either in reconstituted mixtures containing isolated lipoproteins and purified CETP (Fig. 10B) or in total human plasma (Fig. 10C). Again, and with the two experimental systems, full-length human apoCI was able to block the lipid transfer reaction in a concentration-dependent manner. The inhibitory property of apoCI was shared, however, with a lower efficiency by the C-terminal fragment of apoCI (amino acid residues 34 -54). Conversely, no change in CETP activity was observed with the N-terminal fragment of apoCI (amino acid residues 4 -25).
Most interestingly, changes in the electrostatic charge properties observed with full-length or fragmented apoCI paralleled those observed in lipid transfer assays. Indeed, the progressive inhibition of CETP activity by full-length apoCI was associated with a progressive shift of native HDL toward particles with weaker electronegativity (i.e. with higher electrostatic potential values) (Fig. 11). A significant shift of HDL toward particles of lower electronegativity could also be observed by adding increasing concentrations of the C-terminal fragment of apoCI. In contrast, the N-terminal fragment of apoCI with no lipid transfer inhibitory property produced no effect on the electrostatic charge of HDL (Fig. 11).
Dissociation of CETP-Lipoprotein Complexes by ApoCI-The effect of apoCI and apoCI fragments on HDL-CETP association was assessed as their ability to dissociate preformed CETP-HDL complexes over a 1-h incubation at room temperature. As shown in Fig. 12, human apoCI could dissociate CETP-lipoprotein complexes in a concentration-dependent manner with a progressive, apoCI-mediated rise in unbound CETP in incubation mixtures containing constant amounts of HDL and purified CETP. Again, the C-terminal 34 -54 apoCI fragment was found to retain, at least partially, the property of full-length apoCI, and unlike the N-terminal 4 -25 fragment, the C-terminal fragment of apoCI could also dissociate CETP-HDL complexes (Fig. 12). Most interestingly, and as compared with native apoCI, the approximate 2-3-fold lower ability of the C-terminal apoCI fragment to dissociate CETP-HDL complexes (Fig. 12) paralleled its approximate 2-3-fold weaker ability to inhibit CETP activity and to decrease HDL electronegativity (Figs. 10 and 11).

DISCUSSION
Over the last decade, CETP arose as a new target in the prevention and treatment of atherosclerosis, and recent human studies with specific pharmacological inhibitors supported the potent effect of CETP inhibition on the plasma lipoprotein profile, including both a significant rise in HDL cholesterol and a significant decrease in LDL cholesterol (8 -10). Besides interventional studies with exogenous compounds, plasma is known to contain a specific lipid transfer inhibitory protein, i.e. apolipoprotein CI (20 -22). ApoCI in HDL can suppress the CETPmediated lipid transfer process in a concentration-dependent manner, and in vivo studies in apoCI-knocked out mice expressing human CETP supported the ability of plasma apoCI to act as a potent and physiological inhibitor of CETP activity. These observations indicate that variation in apoCI concentration may constitute a major yet unrecognized determinant of neutral lipid exchange in human plasma. In addition, they suggest that the elucidation of the molecular mechanism of CETP inhibition by apoCI may help to figure new means of CETP blockade. The last point was addressed in the present study through the exhaustive analysis of the implication of apoCI in the complex process of CETP-mediated lipid transfer. It appears that the inhibitory effect of human apoCI is a direct consequence of its unique electrostatic properties that lead to alteration of the HDL-CETP interactions.
In the light of earlier kinetic studies (37,38), cholesteryl ester transfer reaction in plasma is dependent on both the amount and the specific activity of CETP, as well as on the concentration and intrinsic properties of lipoprotein substrates (in particular HDL). Kinetic analysis in the present study, as conducted through gradual changes in the level of FIGURE 7. Effect of acetylation on the electrostatic charge and CETP inhibitory properties of apoCI. Five g of apoCI treated or not with acetic anhydride (0, 125, 500, 2,000, or 21,000 nmol) were isoelectrofocused as described under "Materials and Methods" (right panel). Purified human CETP (0.44 g) was incubated with apoCI (1 mol/liter) treated with 0, 125, 500, 2,000, and 21,000 nmol of acetic anhydride (left panel). CETP activity was expressed as compared with CETP activity measured without apoCI. Horizontal bars are the mean Ϯ S.D. of three distinct experiments. FIGURE 8. Effect of native and acetylated apoCI supplementations on the apolipoprotein composition of HDL. A, apoCI was purified to homogeneity as described under "Materials and Methods," and the same amounts (1.7 g) were analyzed on SDSpolyacrylamide gradient gel prior (lane 1) or after (lane 2) acetylation in the presence of 21 mol of acetic anhydride. B, HDL were incubated for 1 h at room temperature in the presence of native apoCI or acetylated apoCI, and HDL-bound apolipoproteins were analyzed on SDS-polyacrylamide gradient gel after delipidation. Lane 1, no addition; lane 2, HDL incubated with native apoCI (apoCI to HDL ratio, 3.2); lane 3, HDL incubated with native apoCI (apoCI to HDL ratio, 6.4); lane 4, HDL incubated with acetylated apoCI (acetylated apoCI to HDL ratio, 3.2); lane 5, HDL incubated with acetylated apoCI (acetylated apoCI to HDL ratio, 6.4). FIGURE 9. Mean surface potential of HDL in the presence of acetylated apoCI. Mixtures containing HDL and various amounts of acetylated apoCI (range of acetylated apoCI to HDL molar ratio, 0 -6.4) were incubated for 1 h in PBS at room temperature. Subsequently, 5 l of samples were submitted to electrophoresis in 0.5% agarose gels as described in legend to Fig. 4. NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45 either purified CETP or HDL, indicated that the inhibitory potential of apoCI is dependent on the amount of HDL but not on the amount of CETP added. These observations support the hypothesis that apoCI may inhibit CETP through its ability to modify the HDL substrate, rather than through a specific blockade of the CETP molecule. This observation led us to consider in greater detail the specific molecular properties of apoCI and, in particular, the physicochemical changes it induces when combined with HDL, i.e. its main carrier in plasma (39). It must be emphasized at this stage that apoCI, the smallest molecule among the apolipoprotein family, is a highly basic apolipoprotein molecule with a cluster of lysine residues in its C-terminal region (31). The peculiar electrostatic properties of apoCI were shown in vitro to be able to produce a significant change in HDL electronegativity, which is recognized today as a leading parameter determining both the strength of CETP-HDL interactions and the velocity of CETP-mediated lipid transfers (23)(24)(25)40). In accordance with earlier transfer studies with native or chemically modified HDL subpopulations (25,40), alteration in the mean surface potential of apoCI-enriched HDL corresponded to a shift from optimal values in native HDL particles (mean surface potential of Ϫ11.7 mV) to inappropriate values in apoCI-enriched particles. The resulting tremendous changes in both CETP interaction and neutral lipid exchange velocity are in agreement with previous data that indicated that lipid transfer exchange with HDL subpopulations becomes minimal when surface potential is above Ϫ11.0 mV (25). In direct link with the implication of positively charged amino acid residues of apoCI in mediating its inhibitory effect, the inhibitory potential of apoCI was lost by acetylation of lysine residues.

Inhibition of CETP by ApoCI
Although the CETP inhibitory potential of apoCI is biologically sig- FIGURE 11. Effect of apoCI fragments on the electronegative charge of HDL. The charge of HDL that were incubated in the presence of increasing amounts of purified apoCI (see "Materials and Methods") or synthetic apoCI fragments was determined as described in Fig. 4. Results are expressed as the change in electrostatic charge as compared with HDL incubated in the absence of apoCI. Each point is the mean Ϯ S.D. of triplicate determinations with three different HDL preparations. Asterisks show significant differences compared with the initial value with no addition, p Ͻ 0.05; Mann-Whitney test. nificant, with more pronounced CETP-mediated changes in plasma lipoproteins in CETPTg/apoCI-KO mice versus CETPTg mice (21), apoCI overexpression in apoCI transgenic mice increases the circulating levels of CETP (22). This is explained by an indirect, stimulatory effect on the CETP gene in response to marked hyperlipidemia that is due to the inhibition by the elevated levels of apoCI of both lipoprotein lipase activity and binding of VLDL to their receptors (41)(42)(43)(44)(45). It results that overexpression of native apoCI does not represent a suitable method for decreasing total cholesteryl ester transfer activity in vivo despite a beneficial impact on the specific activity of CETP. In contrast to initial studies with full-length native apoCI, the structure-function analysis conducted in the present study revealed that a C-terminal fragment of apoCI with no hyperlipidemic potential (43) is sufficient to exert a significant inhibitory effect on CETP. Again, the ability of the C-terminal fragment of apoCI to block the lipid transfer reaction paralleled concomitant changes in the electrostatic charge of HDL. It is noteworthy that the N-terminal ␣-helix of apoCI with no CETP inhibitory properties had no effect on HDL electronegativity, indicating that differences in the effectiveness of C-terminal and N-terminal apoCI fragments reflect at least in part differences in their intrinsic physicochemical properties. Most interestingly, and as observed with native apoCI, only the C-terminal fragment and not the N-terminal fragment was able to undergo self-association in aqueous buffer and in the concentration range used in the present studies (results not shown). The biological relevance of a specific C-terminal apoCI fragment both in the inhibition of plasma CETP activity and in the improvement of the plasma lipoprotein profile will deserve further attention in in vivo studies.
Overall, the observations of the present study converge on a role of apoCI in reducing concomitantly the electronegativity of HDL and the binding of CETP to the lipoproteins. The latter point was confirmed by the redistribution of CETP toward a free, nonlipoprotein-associated pool in the presence of increasing concentrations of positively charged apoCI. Most interestingly, the C-terminal fragment of apoCI that was able to reduce the electronegativity of HDL could also disrupt the binding of CETP to lipoproteins, however, with a weaker efficiency than native apoCI. The electrostatic charge of lipoproteins is a leading parameter in determining the CETP-lipoprotein interaction in the first step of the CETP-mediated lipid transfer process. In the studies of Sammett and Tall (24) with lipolyzed VLDL and HDL, conditions that were shown to increase the transfer of HDL cholesteryl esters were proven to favor the binding of CETP to lipoproteins. In addition, Nishida et al. (23) reported stronger affinity of CETP for succinylated or acetylated lipoproteins.
The molecular mechanism of CETP inhibition by apoCI and its C-terminal fragment seems to differ from the molecular mechanisms of the two pharmacological inhibitors of CETP that have been studied so far in human populations. On the one hand, the JTT-705 compound and its analogs were shown to inactivate a free sulfhydryl group in the hydrophobic site on CETP, thus resulting in an irreversible inhibition of the binding and transfer of neutral lipids (7,8,29). On the other hand, torcetrapib was shown to induce a shift of CETP from a free form to an HDL-bound, nonproductive complex (9, 10). Most interestingly, earlier studies demonstrated that similar excessive binding of CETP to lipoproteins accompanies the inhibition of the CETP-mediated lipid transfer reaction by specific anti-CETP monoclonal antibodies (27). In other words, inhibition of neutral lipid binding to CETP and enhanced association of CETP with lipoproteins might constitute two distinct ways to CETP inhibition in vivo. In addition, the present study indicates that insufficient binding of CETP to apoCI-containing HDL may also constitute another means of CETP blockade. The latter mechanism is compatible with the proposed hypothesis of the existence of an optimal interaction of CETP with native plasma HDL, with both excessive and insufficient CETP-HDL interactions resulting in lower cholesteryl ester transfer rates (25,40).
In conclusion, the inhibition of CETP activity by apoCI is dependent at least in part to its peculiar electrostatic properties that are able to shift HDL toward one profile of lower electronegativity. Based on earlier studies, relatively subtle alterations in the composition of the lipoprotein surface can influence, sometimes dramatically, the binding of CETP to lipoproteins and thereby the lipid transfer rate. Thus, it is highly probable that controlled, apoCI-mediated alteration of the binding of CETP to lipoproteins may result in fewer cholesteryl ester transfers in human plasma. FIGURE 12. ApoCI-mediated release of CETP from the HDL surface. Purified HDL were covalently bound to CNBr-Sepharose beads as described under "Materials and Methods." Incubations were conducted in the presence of 100 l of HDL-Sepharose suspension in PBS, corresponding to 70 g of HDL cholesterol. The suspensions were first incubated for 1 h at room temperature in the presence of purified CETP (final concentration, 3.75 mg/liter). After a short centrifugation, the infranatants containing the HDL-Sepharose phase and the bound CETP were resuspended in 200 l of PBS in the presence of increasing amounts of purified apoCI-(34 -54) or apoCI-(4 -25) (concentration range, 0 -2 mol/liter) and were incubated for 1 h at room temperature. After a short centrifugation, the proteins in the supernatant were applied to a nitrocellulose membrane with a dot blot system, and CETP was subsequently detected by immunoassay as described under "Materials and Methods." The amount of CETP in each supernatant was determined by comparison with a calibration curve that was obtained with serial dilutions of purified CETP that were applied to the membrane together with the samples. No cholesterol release could be detected in supernatants after exposure to apoCI, indicating that at least 98% of HDL remained bound to the beads under the experimental conditions used. Each point represents the mean Ϯ S.D. of four determinations. Asterisks show significant differences compared with HDL-Sepharose preincubated with CETP and incubated without addition, p Ͻ 0.05; Mann-Whitney test.