Depletion of Pre-β-high Density Lipoprotein by Human Chymase Impairs ATP-binding Cassette Transporter A1- but Not Scavenger Receptor Class B Type I-mediated Lipid Efflux to High Density Lipoprotein

The ATP-binding cassette transporter A1 (ABCA1) mediates the efflux of cellular unesterified cholesterol and phospholipid to lipid-poor apolipoprotein A-I. Chymase, a protease secreted by mast cells, selectively cleaves pre-β-migrating particles from high density lipoprotein (HDL)3 and reduces the efflux of cholesterol from macrophages. To evaluate whether this effect is the result of reduction of ABCA1-dependent or -independent pathways of cholesterol efflux, in this study we examined the efflux of cholesterol to preparations of chymase-treated HDL3 in two types of cell: 1) in J774 murine macrophages endogenously expressing low levels of scavenger receptor class B, type I (SR-BI), and high levels of ABCA1 upon treatment with cAMP; and 2) in Fu5AH rat hepatoma cells endogenously expressing high levels of the SR-BI and low levels of ABCA1. Treatment of HDL3 with the human chymase resulted in rapid depletion of pre-β-HDL and a concomitant decrease in the efflux of cholesterol and phospholipid (2-fold and 3-fold, respectively) from the ABCA1-expressing J774 cells. In contrast, efflux of free cholesterol from Fu5AH to chymase-treated and to untreated HDL3 was similar. Incubation of HDL3 with phospholipid transfer protein led to an increase in pre-β-HDL contents as well as in ABCA1-mediated cholesterol efflux. A decreased cholesterol efflux to untreated HDL3 but not to chymase-treated HDL3 was observed in ABCA1-expressing J774 with probucol, an inhibitor of cholesterol efflux to lipid-poor apoA-I. Similar results were obtained using brefeldin and gliburide, two inhibitors of ABCA1-mediated efflux. These results indicate that chymase treatment of HDL3 specifically impairs the ABCA1-dependent pathway without influencing either aqueous or SR-BI-facilitated diffusion and that this effect is caused by depletion of lipid-poor pre-β-migrating particles in HDL3. Our results are compatible with the view that HDL3 promotes ABCA1-mediated lipid efflux entirely through its lipid-poor fraction with pre-β mobility.

Atherosclerosis is a disease characterized by accumulation of cholesterol in the arterial intima, the innermost layer of the arterial wall (1). Initially, cholesterol accumulates intracellularly, mostly in the intimal macrophages that scavenge oxidized low density lipoproteins. As the macrophages become cholesterol-loaded, they are transformed into foam cells, a key event in the early atherogenic process. Because macrophages are unable to down-regulate scavenger receptors or the inflow of cholesterol by this pathway, removal (efflux) of excess cellular cholesterol from these cells is of particular importance to prevent foam cell formation (2).
Efflux of cellular cholesterol is a complex and heterogeneous process promoted by many factors and requiring the presence of extracellular cholesterol acceptors (3)(4)(5). Desorption of free cholesterol by aqueous diffusion to phospholipid-containing lipoproteins is a nonspecific and relatively inefficient pathway that operates in all cells (6). A more efficient pathway of cholesterol efflux is mediated by the exchangeable apolipoproteins A, C, and E, the major apolipoproteins associated with HDL 1 (7). Lipid-free or lipid-poor apolipoproteins promote the efflux of free cholesterol by interacting with the ATP-binding cassette transporter AI (ABCA1) in the cell membrane (8). In contrast, mature HDL and other phospholipid-containing spherical particles are believed to acquire free cholesterol by simple aqueous diffusion or by facilitated diffusion mediated by the scavenger receptor class B, type I (SR-BI). Indeed, the rate of cholesterol efflux from SR-BI-expressing cells to phospholipid-containing acceptors is a linear function of the quantity and type of phospholipids present in the extracellular acceptors (9). However, it has also been reported that the presence of intact apolipoproteins may influence the ability of phospholipid-containing particles to promote cholesterol efflux (10,11).
Experimental studies indicate that the ABCA1 is involved in the efflux of free cholesterol and phospholipid from different cell types, mediating the lipidation of apoA-I, a process thought to be responsible for the formation of pre-␤-migrating HDL particles. These particles, in turn, interact with lecithin-cholesterol acyltransferase in plasma, which catalyzes the esteri-fication of free cholesterol and the subsequent maturation of the pre-␤ particles to ␣-migrating HDL with a core of cholesteryl esters (12,13). Cholesterol efflux to purified, lipid-poor apoA-I has been well established as an ABCA1-mediated pathway for the removal of cellular sterols (14 -16). However, although in some studies it has been observed that the ultracentrifugally isolated HDL 3 fraction is able to induce ABCA1mediated efflux of cholesterol (14,(17)(18)(19), the relative contributions of ␣and pre-␤-migrating particles in HDL 3 to cellular cholesterol efflux via ABCA1 or other pathways of efflux have not been well characterized.
We have demonstrated previously that several proteases found in the arterial intima can degrade HDL 3 in vitro and reduce its efficiency as an acceptor of cholesterol from macrophages. The mast cell proteases chymase (20) and tryptase (21) and some metalloproteinases (22) as well as plasmin and kallikrein (23) specifically deplete the small subpopulation of pre-␤-migrating HDL particles and so impair the efflux of cholesterol from human macrophage foam cells promoted by HDL 3 . Because cholesterol efflux to HDL 3 can be mediated by ABCA1dependent and -independent mechanisms, we were interested to find out which of these two types of efflux is affected by proteolytic enzymes capable of depleting pre-␤-migrating HDL particles. We selected human chymase as an example of a neutral protease present in the arterial intima because chymase, when bound to heparin, i.e. in its physiological form, is partially resistant to its natural inhibitors (24) and, thus, capable of proteolyzing HDL in the plasma (20) and in the aortic intimal fluid (24,25).
To be able to define which type of efflux is impaired by treatment of HDL 3 with chymase, we chose J774 murine macrophages incubated in the presence of cAMP to examine the ABCA1-dependent pathway of efflux (14,26) and Fu5AH rat hepatoma cells to examine the SR-BI pathway (27). In addition, to study in further detail the chymase-inhibition of lipid efflux, we used probucol for inhibition of efflux to lipid-free or lipid-poor apolipoproteins (26 -30). The results using the two cell lines strongly suggested that chymase-induced proteolysis of HDL 3 specifically impairs the component of efflux which is dependent on the function of ABCA1. The ABCA1 dependence of the efflux was confirmed by using brefeldin and gliburide, two inhibitors of ABCA1 (31,32). Our results also demonstrate that HDL 3 effectively promotes ABCA1-mediated efflux of lipids caused by the presence of pre-␤-migrating lipid-poor particles.
Recombinant Human Chymase-Recombinant human chymase (specific activity 80 BTEE units/g) expressed in the baculovirus insect cell system was kindly provided by Dr. T. Kamimura of Teijin Ltd., Hino, Tokyo, Japan. The preparation was dissolved in a buffer containing 150 mM NaCl, 1 mM EDTA, 5 mM Tris-HCl, pH 7.4, before proteolytic treatment of HDL 3 . Chymase activity was fully inhibited by 100 g/ml soybean trypsin inhibitor (Sigma), as demonstrated spectrophotometrically with BTEE as substrate, as described (33).
Isolation of Plasma HDL 3 and Purification of ApoA-I-Human HDL 3 (d ϭ 1.125-1.210 g/ml) was isolated from fresh plasma of normal healthy donors by sequential ultracentrifugation, using KBr for density adjustments. We have observed that this ultracentrifugally isolated HDL 3 fraction contains variable amounts of both ␣and pre-␤-migrating HDL (20 -23). ApoA-I was purified from human blood plasma, as described previously (34).

Proteolysis of HDL 3 by Recombinant Human
Chymase-1 mg/ml HDL 3 was incubated at 37°C in 5 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, pH 7.4 (TNE buffer), in the absence or presence of 0.5 g/ml recombinant human chymase, equal to 40 BTEE units/ml, for up to 24 h. After incubation, the tubes were cooled down rapidly by placing them on ice. To inhibit chymase fully, soybean trypsin inhibitor (final concentration 100 g/ml) was added to the vials. The incubation mixtures were then subjected to structural and functional studies, as described below. Degradation of HDL 3 apolipoproteins was evaluated by SDS-PAGE on 16% polyacrylamide gels (35), and identification of the apoA-I fragments was performed by apoA-I immunoblotting.
Treatment of HDL 3 with PLTP-800 g of HDL 3 and 5 g of PLTP (corresponding to phospholipid transfer activity of 2000 nmol/h) were incubated in TNE buffer (final volume 400 l) for 24 h at 37°C. The PLTP-treated HDL 3 was then incubated for 2 h at 37°C in the absence or presence of recombinant human chymase, and, after inhibition of the enzymatic activity by addition of soybean trypsin inhibitor, the cholesterol efflux-promoting activity of the HDL 3 samples was analyzed as described below. The content of pre-␤-HDL in the various HDL 3 preparations was determined as a percentage of the sum of both the ␣and pre-␤-HDL subpopulations in agarose gel electrophoresis or two-dimensional polyacrylamide gradient gel electrophoresis by densitometric analysis of the gels.
Evaluation of HDL Subclasses by Agarose Gel Electrophoresis or Two-dimensional Electrophoresis and ApoA-I Immunoblotting-Agarose gel electrophoresis of the ultracentrifugally isolated HDL 3 was performed, using the Beckman Paragon system according to the instructions of the manufacturer. Proteins were stained with Coomassie Blue for detection of the ␣and pre-␤-bands. The distribution of HDL 3 subclasses was analyzed by two-dimensional electrophoresis, where agarose gel electrophoresis was followed by nondenaturing polyacrylamide gradient gel electrophoresis and subsequent immunoblotting for apoA-I (36). In the first dimension, 5 l of plasma was separated on a 0.5% agarose gel; agarose gel strips containing the preseparated lipoproteins were then transferred to a 3-20% polyacrylamide gradient gel. Separation was performed in the second dimension at 30 mA for 4 h. Fractionated HDL 3 was then electroblotted onto a nitrocellulose membrane on which apoA-I-containing lipoproteins were detected with the help of a sheep anti-apoA-I antibody.
Cell Culture-J774 mouse macrophages were cultured in RPMI 1640 with 10% FCS. Fu5AH rat hepatoma cells were grown in Dulbecco's modified Eagle's medium with 5% FCS. The cells were incubated in 5% CO 2 at 37°C, seeded in 12-well plates, and utilized for experiments when they reached 80 -90% confluence.
cAMP Stimulation-J774 monolayers were washed with PBS and incubated for 24 h in RPMI containing 4 Ci/ml [1,2-3 H]cholesterol, as described previously (28). The labeling medium contained 1% FCS and 2 g/ml Sandoz 58-035 to ensure that the labeled cellular cholesterol was present in unesterified form. For the experiments in which the cells were enriched with cholesterol, 50 g of protein/ml of unlabeled acetylated low density lipoprotein was included in the incubation medium. After a 24-h labeling period, the cells were washed and then incubated with 0.2% BSA in RPMI with or without 0.3 mM CPT-cAMP for 12 h. After this, some wells were washed with PBS, dried, and extracted with 2-propanol to provide base-line (time 0) values for the total cellular [1,2-3 H]cholesterol content. In some experiments, 10 M brefeldin was added to the culture medium together with CPT-cAMP stimulation.
Measurement of Cholesterol Efflux-Stimulated and unstimulated monolayers of cells containing [1,2-3 H]cholesterol were washed with PBS and incubated for the 4-h efflux time in the presence of cholesterol acceptors at different concentrations or in their absence. In some experiments, 100 M gliburide was added to the medium together with the cholesterol acceptors (32). 100 g/ml soybean trypsin inhibitor was added to the culture medium to prevent activation of chymase during the efflux period. Cholesterol efflux was quantified by removing the media from the cell monolayers and centrifuging them to remove any detached cells. The radioactivity present in the incubation medium was determined by liquid scintillation counting and the percentage of radiolabeled cholesterol released (% efflux) was calculated as (cpm in medium at 4 h/cpm in cells at time 0) ϫ 100.
Measurement of Phospholipid Efflux-Phospholipid efflux was induced as described previously (37). Briefly, J774 cells were incubated for 48 h in RPMI 1640 containing 1% FCS and 4 Ci/ml [methyl-3 H]choline chloride. After the stimulation period in the presence of CPT-cAMP and the efflux time to the different acceptors (6 h), the medium was collected and centrifuged. Control cell monolayers were treated identically, except that CPT-cAMP was omitted from the incubation medium. After centrifugation, the supernatants were separated and the lipids extracted. The aqueous phase was aspirated, and, to remove any remaining free [ 3 H]choline, the chloroform phase was washed three times with 10:9 (v/v) methanol/water. The chloroform phase was then dried under a stream of nitrogen (N 2 ), and redissolved in 1 ml of toluene. A 700-l aliquot of each sample was transferred to a liquid scintillation vial and quantified by scintillation counting. To analyze the cellular lipids, the monolayers of cells were washed three times with PBS, and the lipids were extracted by the addition of 1 ml of 2-propanol. The 2-propanol extracts were dried under a stream of N 2 , the free [ 3 H]choline was extracted, and the chloroform phase was treated, as described before.
Statistical Analysis-Results are reported as the means Ϯ S.D. Statistical significance was determined by two-tailed Student's t test.

RESULTS
Ability of the Chymase-treated HDL 3 to Promote Efflux of Cellular Cholesterol from J774 and Fu5AH Cells-We treated the HDL 3 fractions isolated from human plasma with human chymase for various periods of time. The concentration of chymase used (0.5 g/ml) was chosen to match the assumed concentration of chymase in the intimal fluid (24). After fully inhibiting the chymase activity, the untreated and chymasetreated HDL 3 preparations were added to three cellular systems to test their ability to promote cellular cholesterol efflux. Aliquots of HDL 3 were added to the cell cultures in the low range of acceptor protein concentration, in which the high affinity efflux of cholesterol dominates (25), and the release of cholesterol into the medium was measured after incubation for 4 h. The cell models used in these experiments were J774 murine macrophages and Fu5AH rat hepatoma cells. Under basal conditions, J774 express low levels of ABCA1 and SR-BI and release membrane cholesterol to extracellular acceptors by passive diffusion, whereas stimulation with cAMP up-regulates ABCA1-mediated lipid efflux to apolipoproteins. In contrast, under basal conditions, Fu5AH cells express high levels of SR-BI in the plasma membrane, and thus the efflux of lipids depends on facilitated diffusion by this receptor. As shown in Fig. 1A, when unstimulated J774 cells (Ϫ cAMP) were used as cholesterol donor cells, pretreatment of HDL 3 with chymase for up to 24 h did not have any influence on the rate of cholesterol efflux. However, upon cAMP stimulation of the cells (ϩ cAMP), an increased efflux (2-fold) to the untreated preparation of HDL 3 (at 0 h) was observed, and a short preincubation of HDL 3 with chymase (2 h) was sufficient to prevent the cAMP-dependent increase in the efflux of cholesterol. In a control experiment, we used lipid-free apoA-I as cholesterol acceptor (Fig. 1B). As expected, after cAMP stimulation, the apoA-Idependent efflux was increased (by 3-fold), confirming that the ABCA1-mediated pathway was stimulated. When 1 mg/ml apoA-I was incubated with 0.5 g/ml chymase, smallsized polypeptides were formed, and the cholesterol efflux from the cAMP-stimulated cells was fully inhibited (not shown).
Analysis of HDL 3 by agarose gel electrophoresis showed that the untreated HDL 3 preparation contained both ␣and pre-␤migrating species (estimated pre-␤ content 6%), and treatment with chymase totally depleted the pre-␤-HDL subpopulation within 2 h of incubation (see Fig. 1A, inset). Throughout the whole incubation period, no change in the mobility of the ␣-migrating fraction of HDL 3 was observed. The ability of chymase to deplete pre-␤-HDL was also confirmed by twodimensional polyacrylamide gradient gel electrophoresis analysis (see Fig. 5). Analysis of the untreated HDL 3 by SDS-PAGE showed two major bands corresponding to apoA-I and apoA-II (not shown). As reported previously (25), treatment of HDL 3 with chymase results in degradation of apoA-I with high efficiency and in degradation of apoA-II with low efficiency. ApoA-I immunoblotting of the gel revealed the appearance of an additional band with apparent molecular mass of 26 kDa which reflected degradation of apoA-I (32 Ϯ 1%, data not shown). In this experiment, prolongation of chymase treatment beyond 2 h did not lead to further significant degradation of apoA-I.
In contrast to the results obtained with cAMP-stimulated J774 cells, chymase degradation of HDL 3 for up to 24 h did not have any influence on the efflux of cholesterol from the Fu5AH cells (Fig. 2). Fu5AH cells express high levels of SR-BI; hence, these results indicate a lack of effect of chymase proteolysis on the cholesterol acceptor function of HDL 3 in conditions where the SR-BI-facilitated cholesterol efflux mechanism dominates.
Because an effect of chymase treatment on HDL 3 was observed only when using the J774 cells, we studied in further detail the efflux of cellular cholesterol from these cells. First, we determined the rate of efflux from control or cAMP-stimulated cell cultures in the presence of various concentrations of HDL 3 (up to 50 g/ml) (Fig. 3). The efflux of cholesterol from nonstimulated cells to untreated and chymase-treated HDL 3 was similar and of low affinity (apparent K m about 50 g/ml). In sharp contrast, the efflux of cholesterol from stimulated cells to untreated HDL 3 was far more efficient (K m of 18 g/ml; V max 20%/4 h) (Prism, GraphPad Inc., San Diego). Treatment of HDL 3 with chymase for 2 h dramatically decreased the efficiency of this efflux process down to the level observed in the unstimulated cells. Because ABCA1 expression is also up-regulated in cholesterol-loaded macrophages, we then studied the effect of chymase treatment of HDL 3 on the efflux of cholesterol from cholesterol-loaded unstimulated J774. The results showed that chymase treatment for 2 h reduced cholesterol efflux from these cells by 40% (Student's t test: p ϭ 0.052) (data not shown).  3 -To measure the phospholipid efflux, the phospholipids in J774 cells were radiolabeled. Stimulation of the radiolabeled J774 cells with cAMP caused a 10-fold and a 3-fold increase in phospholipid efflux to apoA-I and to untreated HDL 3 (at 0 h), respectively. Chymase treatment of HDL 3 for 2 h prevented the cAMP-induced increase in the efflux of phospholipids, i.e. it remained at a level similar to that observed when nonstimulated cells were used (Fig. 4).

ABCA1-mediated Phospholipid Efflux from J774 Macrophages to Chymase-treated HDL
Influence of the Content of Pre-␤-HDL on the HDL 3 -promoted Efflux from cAMP-treated J774 Macrophages-Because the observed inhibitory effect of chymase on ABCA1-dependent efflux was associated with depletion of the pre-␤-HDL particles present in the HDL 3 preparation (see Fig. 1), we studied whether HDL 3 preparations containing different quantities of pre-␤-HDL would promote different degrees of ABCA1-dependent efflux. Analysis of preparations obtained from different donors contained pre-␤-HDL ranging from 4 to 14%, and none of the HDL 3 preparation used in this study (n ϭ 3) was totally devoid of pre-␤ particles. To increase the content of pre-␤ particles, HDL 3 was preincubated for 24 h at 37°C in buffer alone or with PLTP, demonstrated previously to promote the size conversion of HDL particles and to generate pre-␤-HDL (38). As reported previously (39), PLTP treatment of HDL 3 for 24 h increased the content of pre-␤-HDL in the HDL 3 preparation (in this experiment from 9 to 18%; Fig. 5, A and B, left). In contrast, chymase treatment for 2 h, which fully depleted pre-␤-HDL from the native HDL 3 , also drastically reduced the content of pre-␤-HDL in the PLTP-treated HDL 3 (on average by about 70%; range 65-75%).
We then incubated the untreated, PLTP-treated, and chymase-treated HDL 3 with cAMP-stimulated J774 macrophages and measured the rates of cholesterol efflux (see description of the various preparations in the legend to Fig. 6). These data, using a system in which ABCA1-dependent efflux dominates, confirmed that the ability of HDL 3 to stimulate cholesterol efflux from cholesterol-loaded macrophages is increased after PLTP treatment (39). More importantly, using HDL 3 preparations with graded pre-␤ content it could be demonstrated that the rate of cholesterol efflux from the cells varied in parallel with the content of pre-␤ particles in HDL 3 (Fig. 6A). However, no pre-␤-dependent variation of the efflux was observed when the various HDL 3 preparations were incubated with unstimulated J774 macrophages (Fig. 6B). These results indicated that the magnitude of the ABCA1-dependent efflux could be modified by enhancing the content of pre-␤-HDL (by PLTP treatment) or by reducing it (by chymase treatment). However, neither the increase in the content of pre-␤-HDL nor the depletion of pre-␤-HDL particles by chymase pretreatment had any effect on the efflux from Fu5AH cells (not shown).

Effect of Chymase Treatment of HDL 3 on Their Ability to Promote Cellular Cholesterol Efflux from cAMP-stimulated J774 Cells under Conditions When Efflux to Lipid-poor Apolipoproteins or Efflux Facilitated by ABCA1 Is Inhibited-To
demonstrate further that the lipid-poor, pre-␤-migrating component present in the HDL 3 preparation was responsible for the ABCA1-dependent efflux, we studied the effect of chymase treatment of HDL 3 on efflux from cAMP-stimulated J774 cells in the absence or presence of 1) probucol, an inhibitor of the lipid efflux promoted by lipid-poor particles (28 -30), and 2) brefeldin and gliburide, two inhibitors of ABCA1 (31, 32). The results shown in Fig. 7 indicate that probucol is able to reduce Eagle's medium with 5% FCS in the presence of 2 g/ml acyl-CoA: cholesterol acyltransferase inhibitor. Cell were then incubated for 12 h with 0.2% BSA. After this treatment, monolayers were incubated for 4 h with either 12.5 g/ml untreated HDL 3 or 12.5 g/ml chymase-treated HDL 3 as cholesterol acceptors (treatment with chymase is described in Fig. 1). Data are the means Ϯ S.D. from a representative experiment with triplicate wells (n ϭ 3). Cell were then incubated for 12 h with 0.2% BSA in the presence or absence of 0.3 mM CPT-cAMP followed by incubation with the indicated acceptors for 6 h. HDL 3 was treated with chymase for 2 h under the conditions described in Fig. 1. Fractional efflux at 6 h was determined for 25 g/ml lipid-free apoA-I, 12.5 g/ml untreated HDL 3 , and 12.5 g/ml chymase-treated HDL 3 . Open bars denote untreated control monolayers; solid bars denote monolayers pretreated with CPT-cAMP. Data are the means Ϯ S.D. from a representative experiment with triplicate wells (n ϭ 3). the efflux promoted by untreated HDL 3 . In contrast, cholesterol efflux induced by the chymase-treated HDL 3 was insensitive to probucol, a finding consistent with the hypothesis that a protease-sensitive lipid-poor component was present in HDL 3 and was responsible for the ABCA1-mediated efflux. When the ABCA1 pathway of efflux was blocked by either of the two inhibitors, brefeldin or gliburide, the efflux to untreated HDL 3 was also reduced (compare with control). Importantly, when ABCA1 was blocked, no further reduction of efflux was observed upon treatment of HDL 3 with chymase, confirming that the ABCA1-mediated efflux of cholesterol fully accounted for the chymase effect. DISCUSSION In the present study, we attempted to identify the cellular cholesterol efflux pathway associated with the high affinity component of the efflux to HDL 3 , which is lost upon treatment of the acceptors with chymase (25). For this purpose, we studied the cholesterol efflux promoted by the chymase-treated HDL 3 from 1) J774 cells that express ABCA1 upon stimulation with cAMP, and 2) Fu5AH cells that physiologically express the SR-BI. Limited proteolysis of HDL 3 by chymase produced full depletion of the pre-␤-migrating particles and completely abolished the ABCA1-mediated component of cholesterol efflux to the acceptor particles. Moreover, chymase treatment of HDL 3 inhibited phospholipid efflux, a process specifically mediated by ABCA1 (15,16,40).
Our results are compatible with the view that pre-␤ particles do not contribute to the ability of HDL 3 to promote cholesterol efflux by either simple or facilitated diffusion. This conclusion is also consistent with the observation that pre-␤ HDL does not participate in the cholesterol efflux to HDL 3 from cell systems poorly expressing ABCA1 (41). The fact that chymase treatment did not reduce the two diffusional efflux processes stimulated by phospholipid-rich acceptors indicates that chymase cleaved a pool of lipid-poor particles. This hypothesis was confirmed by our present observation that only efflux to untreated HDL 3 , but not to chymase-treated HDL 3 (from which lipid-poor  Fig. 1 and "Experimental Procedures"). The cells were then incubated for 12 h with 0.2% BSA in the presence (A) or absence (B) of 0.3 mM CPT-cAMP followed by incubation for 4 h with various HDL 3 preparations (all at 12.5 g/ml) containing different amounts of pre-␤-HDL. HDL 3 was preincubated in the absence (ϪPLTP) or presence (ϩPLTP) of PLTP for 24 h followed by incubation in the absence (Ϫchymase) or presence (ϩchymase) of chymase for 2 h, as described under "Experimental Procedures." Pre-␤ contents: 0% ϭ ϪPLTP/ϩchymase; 4% ϭ native HDL 3 (from the donor with the lowest pre-␤ content in this study); 6% ϭ ϩPLTP/ ϩchymase; 9% ϭ ϪPLTP/Ϫchymase; 18% ϭ ϩPLTP/Ϫchymase.

FIG. 5. Effect of chymase treatment on the particle composition of untreated HDL 3 (A) and PLTP-pretreated HDL 3 (B)
. Native HDL 3 and PLTP-pretreated HDL 3 were incubated with chymase for 2 h under the conditions described in Fig. 1. At the end of the incubation, the distribution of HDL subclasses was analyzed by twodimensional electrophoresis, where agarose gel electrophoresis was followed by nondenaturing polyacrylamide gradient gel electrophoresis. Fractionated HDL 3 was then electroblotted onto a nitrocellulose membrane on which apoA-I-containing lipoproteins were detected with the use of a sheep anti-apoA-I antibody. particles had been depleted), was inhibited by probucol, an antagonist of cholesterol efflux to lipid-poor particles (28 -30). The report that trypsin-labile apolipoproteins in preparations of HDL 3 are responsible for an apolipoprotein-dependent component of cholesterol efflux (19) is also consistent with the present findings. We and others (42,43) observed recently that the C-terminal domain of apoA-I is necessary to promote ABCA1-mediated cholesterol efflux. Of note, chymase treatment of HDL 3 produces a 26-kDa N-terminal fragment of apoA-I, with cleavage of the C terminus (44). Altogether, these findings indicate that the previously demonstrated effect of chymase treatment on high affinity efflux of cholesterol to HDL 3 is caused by the digestion of the pre-␤-migrating lipidpoor particles contained in HDL 3 which specifically interact with the ABCA1 system.
In sharp contrast to the observations with cAMP-stimulated J774 cells, no effect of chymase treatment was observed when cholesterol efflux was measured from unstimulated J774 cells or from Fu5AH cells expressing SR-BI. Because proteolysis of apoA-I was about 30% of the total apoA-I in HDL 3 , whereas the proportion of apoA-I in the pre-␤ fraction of HDL 3 only ranged from 4 to 14%, proteolysis of ␣-HDL must also have occurred, a conclusion concurring with our previous observations (44). Notably, such proteolysis was without effect on the acceptor function of the ␣-HDL. In line with this observation we have found that the apoA-I molecules on the surface of the globular ␣-migrating HDL remain biologically active also in terms of HDL remodeling. Indeed, chymase cleavage of apoA-I in the ␣-migrating HDL does not abolish its ability to generate PLTP-dependent pre-␤-migrating particles (39) or to undergo cholesterol esterification by lecithin-cholesterol acyltransferase (44).
Taken together, the results obtained by proteolyzing HDL 3 with the neutral protease chymase, which is considered to be relevant in atherogenesis (45), also provide some general insight into the nature of HDL as cholesterol acceptors. Although it is widely accepted that HDL 3 mainly promotes cellular cholesterol efflux by a process of passive diffusion that can be facilitated by SR-BI (8), some evidence has also suggested that HDL 3 may act via ABCA1 (14,(17)(18)(19). However, the mode of the action has remained unclear. We propose here that this interaction is caused by induction of high affinity efflux of cholesterol resulting from a lipid-poor apoA-I pre-␤migrating fraction that specifically interacts with ABCA1. This conclusion is supported by our observations that: 1) efflux of both cholesterol and phospholipid to HDL 3 increases upon ABCA1 stimulation; 2) chymase eliminates this component of efflux by degrading the pre-␤-migrating particles in HDL 3 ; 3) enrichment of HDL 3 with pre-␤ particles by PLTP treatment enhances their ability to promote ABCA1mediated efflux; and 4) ABCA1-mediated efflux to HDL 3 is inhibited by probucol, an antagonist of lipid-poor cholesterol acceptors, as well as by brefeldin and gliburide, which are inhibitors of ABCA1.
The present observations are of potential pathophysiological relevance. The intimal fluid is rich in pre-␤-HDL (46 -48) and lipid-poor apolipoproteins, notably apoA-I and apoA-IV, both of which induce ABCA1-mediated cholesterol efflux (30,49) and are sensitive to degradation by mast cell chymase (7,20). In fact, chymase treatment of the intimal fluid reduces its ability to promote a high affinity component of efflux from human monocyte-derived macrophage foam cells (25). Moreover, the effects of chymase described here have been observed using a concentration of the enzyme which is within the estimated physiological range in the intima (24). These observations suggest that chymase secretion by activated mast cells present in the arterial intima (50) may reduce the ability of intimal fluid to clear an excess of cellular cholesterol via ABCA1, so contributing to foam cell formation in the atherosclerosis-prone areas of the arterial wall.