Independent regulation of cholesterol incorporation into free apolipoprotein-mediated cellular lipid efflux in rat vascular smooth muscle cells.

Cholesterol was poorly available to free apolipoprotein (apo)A-I-mediated cellular lipid efflux from cholesterol-loaded rat vascular smooth muscle cells generating cholesterol-poorer pre-β-HDL particles than those generated from macrophages by the same reaction (Li, Q., Komaba, A., and Yokoyama, S. (1993) Biochemistry 32, 4597-4603). The factors known to induce transformation of the smooth muscle cells into a macrophage-like stage were used in order to modulate this reaction, such as human platelet-derived growth factor, macrophage colony-stimulating factor, and phorbol 12-myristate-13acetate (PMA). When the cells were stimulated by PMA following the pretreatment with platelet-derived growth factor plus macrophage colony-stimulating factor, cholesterol efflux mediated by free apoA-I increased 3-fold without changing phospholipid efflux, resulting in generation of pre-β-HDL particles more rich in cholesterol. This treatment had only a little or no effect on apparent cellular cholesterol efflux to HDL or lipid microemulsion, respectively. Overall cellular free cholesterol pool size was unaffected by the treatment, and probing by extracellular cholesterol oxidase did not detect gross change in the cellular surface cholesterol. This specific enrichment of cholesterol in the apoA-I-mediated cellular lipid efflux was reversed by protein kinase C inhibitors. Measurement of intracellular cholesterol esterification suggested that PMA induced translocation of intracellular cholesterol to a specific pool for apoA-I-mediated efflux, and a protein kinase C inhibitor reversed this effect.

When cholesterol is transported from the peripheral cells to the liver for its biological degradation, the first step of the pathway is efflux of cellular cholesterol to plasma lipoproteins. High density lipoprotein (HDL) 1 is believed to play a primary role in this step by "accepting" the cellular cholesterol. Cellular free cholesterol is exchangeable with extracellular lipoprotein surface by a nonspecific physicochemical mechanism such as its diffusion through aqueous phase or direct collision between the membranes (Rothblat and Phillips, 1982;Johnson et al., 1986;Karlin et al., 1987;Johnson et al., 1988). The net efflux could therefore be induced by a gradient of cholesterol content between cell membrane and lipoprotein surface, so that cholesterol esterification on HDL may play an important role in such a mechanism (Murphy, 1962;Glomset, 1968;Ray et al., 1980;Fielding and Fielding, 1981;Czarnecka and Yokoyama, 1995). Specific function of HDL to accommodate cholesterol may also be attributed to physicochemical properties of its surface (Stein et al., 1986;Phillips et al., 1987). On the other hand, a HDLbinding protein(s) on the cell surface is proposed to play a specific role being linked to intracellular signal transduction that results in mobilization of intracellular cholesterol to cellular surface (Graham and Oram, 1987;Slotte et al., 1987;Mendez et al., 1991;McKnight et al., 1992).
Recent development of the investigation in this field shed light for further understanding of the mechanism for cellular cholesterol efflux. Fielding and co-workers discovered that cellular cholesterol appeared in a specific HDL fraction, pre-␤-HDL, in the very early stage of its efflux and then was transferred to other lipoproteins (Castro and Fielding, 1988;Miida et al., 1990Miida et al., , 1992 and proposed that this lipoprotein fraction is the most efficient "acceptor" of cellular cholesterol. Later, we demonstrated that lipid-free apolipoproteins having multisegments of amphiphilic helix mediate the net efflux of cholesterol and phospholipid from various types of cells, generating new HDL-like particles with these cellular lipids Yokoyama, 1991, 1992;Komaba et al., 1992). The particles were consistent with pre-␤-HDL with respect to the chemical and physical properties. Other groups have also reported similar data using lipid-free apolipoproteins (Bielicki et al., 1992;Forte et al., 1993). These results may explain why cellular cholesterol appears in pre-␤-HDL fraction in the very early stage of the efflux (Castro and Fielding, 1988;Miida et al., 1990Miida et al., , 1992; pre-␤-HDL particles are newly generated with cellular lipid by apolipoproteins dissociated from HDL, rather than the particles in the medium preferably "accepting" cholesterol from cellular surface. The reaction can be carried out by a number of amphiphilic helical apolipoproteins (Hara and Yokoyama, 1991;Bielicki et al., 1992) or their model peptides (Mendez et al., 1994), and it may induce specific translocation of intracellular cholesterol that would render cholesterol readily available for the efflux (Mendez et al., 1994). The involvement of a protein binding site(s) for such apolipoproteins has not clearly been identified on cellular surface. However, many circumstantial evidences may support a potential role of such a site(s) in this reaction. Apparent K m values for apolipoprotein-mediated lipid efflux were as low (Hara and Yokoyama, 1991; as that of the LDL-receptor interaction (Ϸ10 Ϫ8 M) (Brown and Goldstein, 1986), and lower than that of the lipidapolipoprotein interaction (Ϸ10 Ϫ7 M) (Tajima et al., 1983;Yokoyama et al., 1985), indicating some type of specific interaction. Proteolytic treatment of cellular surface resulted in complete inhibition of this pathway (Li et al., 1995). Furthermore, specific lack of interaction with lipid-free apoA-I in fibroblasts from the patients with Tangier disease strongly suggested that there is a genetically defined cellular interaction site for lipid-free apolipoproteins, and that this reaction is a major source of plasma HDL (Francis et al., 1995).
Thus, free apolipoprotein-mediated cellular cholesterol efflux seems to be a distinct pathway from nonspecific physicochemical cholesterol exchange reaction between cellular surface and lipoproteins. This may lead us to the hypothesis that HDL-mediated lipid efflux includes two distinct mechanisms. One is physicochemical lipid exchange between the cell and HDL surface in which gradient of cholesterol between the two surfaces causes the net efflux and therefore cholesterol esterification on HDL may act as its driving force. The other is generation of new pre-␤-HDL with cellular lipids, as apolipoproteins locally dissociate from HDL surface to interact with the cellular surface Yokoyama, 1991, 1992). The maximum contribution of the free apolipoprotein-mediated portion could be as much as 40% of apparent HDL-mediated efflux Francis et al., 1995).
Interestingly, vascular smooth muscle cells are very resistant to the free apolipoprotein-mediated cholesterol efflux  due to generation of cholesterol-poor pre-␤-HDL (Li et al., 1993). This was specific to lipid-free apolipoprotein-mediated cholesterol efflux, whereas little significant difference was found among cell lines in the nonspecific cholesterol efflux mediated by low density lipoprotein (LDL) or lipid microemulsions (Li et al., 1993). The low cholesterol/phospholipid ratio in the HDL-mediated lipid efflux from smooth muscle cells implied that free apoA-I-mediated mechanism is in fact involved in cellular lipid efflux to HDL (Li et al., 1993). These findings indicated that cell-specific intracellular factors are involved in regulation of cholesterol incorporation into the cellular lipid efflux mediated by apolipoprotein-cell interaction.
Vascular smooth muscle cells are known to be stimulated by certain growth factors and transformed to different stages, and such transformation may be related to its roles in development of vascular atherosclerotic lesions. The smooth muscle cells isolated from arteriosclerotic lesion were found to express the c-fms gene encoding the receptor for macrophage colony-stimulating factor (MCSF) (Inaba et al., 1992a). In vitro, plateletderived growth factor (PDGF) BB-chain homodimer stimulates the smooth muscle cells to induce c-fms gene as well as the scavenger receptor gene, as transforming the cells to a macrophage-like stage (Inaba et al., 1992b). On the other hand, in the undifferentiated macrophage cell line cell THP-1, phorbol esters are known to induce differentiation of the cells with respect to many macrophage activities, including expression of the scavenger receptor (Hara et al., 1987;Takata et al., 1989) and apolipoprotein E and lipoprotein lipase Menju et al., 1989). Hence, the in vitro transformation of the cells by these factors may modulate the cell specific factors involved in apolipoprotein-mediated lipid efflux and would provide an important model to study the intracellular mechanism for cholesterol efflux by this particular pathway.
Thus, we undertook the study of lipid-free apoA-I-mediated lipid efflux from rat vascular smooth muscle cells in comparison to the apparent cholesterol efflux to lipoproteins and lipid microemulsions under the influence of stimulation of the cells by the growth factors and phorbol esters.

EXPERIMENTAL PROCEDURES
Lipoprotein, Apolipoprotein, and Lipid Microemulsion-HDL was isolated from fresh human plasma as the fraction with a density of 1.063-1.21 g/ml in NaBr. The purity of the lipoprotein preparation was verified by electrophoresis in 0.5% agarose and by apolipoprotein composition analysis in polyacrylamide gel electrophoresis in the presence of 0.5% sodium dodecyl sulfate, in order to assure no contamination of LDL or apoB. The particular preparation used was a mixture of HDL 2 and HDL 3 , in approximately 1 to 1 protein ratio, determined by a non-denaturing gradient polyacrylamide electrophoresis. ApoA-I was further isolated from human HDL by delipidation and DEAE-cellulose chromatography in 6 M urea and dissolved in PBS buffer (10 mM sodium phosphate containing 150 mM NaCl in pH 7.4), in which the protein self-associates in equilibrium, as described previously (Yokoyama et al., 1982). All the materials were stored at 4°C as a solution in PBS buffer under argon. LDL was prepared and labeled with [1,2-3 H]cholesteryl oleate (45.4 Ci/mmol, purchased from Amersham Corp.) according to a previously described method (Nishikawa et al., 1986). The labeled LDL was either acetylated or cationized by the method previously described for the purpose of loading radiolabeled cholesterol to macrophages or vascular smooth muscle cells, respectively (Hara and Yokoyama, 1991;Komaba, et al., 1992). Lipid microemulsions having a homogeneous diameter of 26 nm were prepared from egg phosphatidylcholine and triolein as described previously (Tajima et al., 1983) for a nonspecific cellular lipid efflux . Both lipids were cosonicated in the weight ratio of 1:1, and the emulsions were isolated by ultracentrifugation and gel permeation chromatography.
Loading Cells with Radiolabeled Cholesterol and Labeling Cellular Choline Phospholipid-Smooth muscle cells were prepared from rat thoracic aorta . The cells in passages 4 -9 were loaded with radiolabeled cholesterol and labeled for choline phospholipid by incubating with cationized LDL containing the [ 3 H]cholesteryl ester for 1 week and with [methyl-3 H]choline chloride (15 Ci/mmol, Amersham) for the last 24 h of this period as we described previously . The cholesterol-loaded cells were washed and incubated without lipoproteins for additional 24 h. Mouse peritoneal macrophages were obtained from male or female ICR mice by peritoneal lavage (Hara and Yokoyama, 1991). The cells were loaded with radiolabeled cholesterol and labeled for choline phospholipid by incubating with the acetylated LDL containing radiolabeled cholesteryl ester, together with [methyl-3 H]choline chloride, for 24 h as described previously (Hara and Yokoyama, 1991;Li et al., 1993).
Pretreatment of the Cells with Growth Factors-The labeled cells were pretreated for stimulation with growth factors. Both macrophages and smooth muscle cells were incubated with 10 ng/ml recombinant PDGF (Sigma) and/or 100 ng/ml MCSF (a generous gift from Dr. Nobuhiro Yamada, University of Tokyo) for 24 h (Inaba et al., 1992b). In the presence of PDGF, the smooth muscle cells were transformed into the stretched shape with multipseudopods suggesting their transformation to synthetic state. The uptake of acetyl LDL was increased in this stage by 2-4-fold (Table I), indicating expression of the scavenger receptor (Inaba et al., 1992b). This change was not obvious in the presence of MCSF alone. The cells were incubated with and without phorbol 12myristate 13-acetate (PMA) (Sigma), 160 nM, for the last 45 min of this 24-h incubation period. Sphingosine, staurosporine, and 1-(5-isoquino- linylsulfonyl)-2-methylpiperazine (H-7) (all obtained from Sigma) were also added to this 45-min period as protein kinase C inhibitors, in the attempt to reverse the effect of PMA (Hannun et al., 1986;Tamaoki et al., 1986;Hidaka et al., 1984). Sphingosine was suspended with 2 mM bovine serum albumin solution in PBS to give its concentration of 25 mM, and added to the medium to make final concentration of 5-40 M at the step of the PMA treatment (Hannun et al., 1986). Staurosporine was dissolved in dimethyl sulfoxide as 4 mM and added to the medium to make final concentration of 0.5-50 nM at the stage of the PMA treatment (Tamaoki et al., 1986). H-7 was dissolved in water as 5 mM and added to the medium to make final concentration of 12.5-100 M (Hidaka et al., 1994). Efflux of Cellular Cholesterol and Choline Phospholipid-After the pretreatment, the cells were washed with the medium for three times and then incubated with HDL, apoA-I, and lipid microemulsion in the presence of 2 mg/ml bovine serum albumin. Radioactivity of free and esterified cholesterol, phosphatidylcholine and sphingomyelin in the medium, and cells was then counted after the lipid was extracted and separated by thin layer chromatography Yokoyama, 1991, 1992). Radioactivity of each lipid in the medium in the presence of bovine serum albumin alone was less than 5% of the level induced by HDL as 50 g of protein, and this was subtracted from each data point as a background Li, et al., 1993). Specific radioactivity of cellular total cholesterol and each phospholipid was calculated based on their chemical amount, measured by an enzymatic fluorescence method and organic phosphorus assay, respectively, and used for calculation of the amount of the efflux into the medium (Hara and Yokoyama, 1991;Komaba et al., 1992;Li et al., 1993). The validity of this method was verified by quantification of cholesterol with gasliquid chromatography (Hara and Yokoyama, 1991). In some experiments, the efflux was expressed as radioactivity of each lipid in the medium standardized for cellular pool of the respective cellular lipid in each culture dish specimen  or radioactivity per cellular protein. Cellular protein of each dish was measured by the method of Lowry et al. (1951). The experimental data points were duplicated unless otherwise specified, mostly within a 10% error margin, and the values in the figures represent the averages unless otherwise specified.
Other Methods-The medium was analyzed for the lipid efflux product by using density gradient ultracentrifugation in sucrose as described earlier (Hara and Yokoyama, 1991). Sucrose solution of density 1.30 (0.7 ml) was overlayered with the solution of density 1.10 (1.2 ml), and then the culture medium (1.2 ml) was overlayered on the top in a 3-ml quick-seal centrifuge tube for a Beckman TL100.3 rotor. After centrifugation at 99,000 rpm in a TL100 ultracentrifuge for 16 h at 4°C, 200-l fractions were collected from the bottom. For each fraction, density was measured and radioactivity of free cholesterol and phosphatidylcholine was counted after lipid extraction and thin layer chromatography. The weight of each lipid was calculated by using the specific radioactivity of the respective lipid in a cellular pool (Hara and Yokoyama, 1991;Li et al., 1993). Cellular cholesterol was oxidized by extracellular cholesterol oxidase (Sigma) in order to assess accessibility from the cellular surface, and the oxidized product, cholestone, was analyzed by thin layer chromatography after extracting the cellular lipid, as described in detail previously (Lange and Ramos, 1983;Li et al., 1993). Intracellular cholesterol esterification activity was measured for the cholesterol-loaded cells after treatment with growth factors and PMA, according to the method previously described (Francis et al., 1993). The cells were treated in the same manner as the efflux experiment for cholesterol loading with cationized LDL, the growth factor treatment, and PMA stimulation in the presence and absence of H-7. After washing the medium containing [1-14 C]oleic acid (57 mCi/mmol, Amersham), 1.5 Ci (26 nmol) was added to each culture dish. After a 1-h incubation at 37°C, the cells were washed with ice-chilled medium and lipid was extracted for the analysis by thin layer chromatography. The radioactivity in cholesteryl ester fraction was counted.

RESULTS
Lipid efflux profile was compared between rat aortic smooth muscle cells and mouse peritoneal macrophages. The efflux to lipid microemulsion represents nonspecific physicochemical pathway and apoA-I-mediated efflux represents the efflux pathway by generating new pre-␤-HDL particles with cellular lipid. The condition was chosen for the V max of the efflux rate at the 24-h incubation Yokoyama, 1991, 1992), and the results shown in Table II were consistent with our previous observation (Li et al., 1993). In nonspecific pathway (to lipid microemulsion), the rate of cholesterol efflux was much higher than that of phospholipid and the ratio of cholesterol to total phospholipid was very similar for the two types of cell, giving a weight ratio around 4. In contrast, phospholipid efflux was substantially high in the apoA-I-mediated efflux from both cells. Specific difference was observed in cholesterol efflux between smooth muscle cells and macrophages. Cholesterol was very poorly incorporated from smooth muscle cells into this pathway, making its weight ratio to phospholipid substantially lower than the efflux from macrophages, 0.24 versus 1.57. In lipid efflux to HDL, the value of this parameter was intermediate between the emulsion-mediated and apoA-I-mediated pathways.
Smooth muscle cells were exposed to the potential stimuli to induce macrophage-like functions. Such factors as PDGF, MCSF, and PMA were used in an attempt to modulate lipid efflux profile for the reasons mentioned earlier, and the results are shown in Table III. When each of the three factors was individually used, there was no effect on the cholesterol efflux from the cells either by HDL or apoA-I. Since PDGF is known to induce MCSF receptor (Inaba et al., 1992b), the two growth factors were used together, but no additional effect was found by this combination. When the cells were stimulated by PMA following the treatment with growth factors, cholesterol efflux was somewhat enhanced from the PDGF-pretreated cells but not from the MCSF-treated cells. No increase was observed in the apparent cholesterol efflux to HDL in this condition. Finally, when both growth factors were used together in the pretreatment for the PMA-stimulation, the efflux of cholesterol mediated by lipid-free apoA-I was enhanced 3-fold. The increase was only 30% with the efflux to HDL by the same treatment. In contrast to the results for the cholesterol efflux, phospholipid efflux mediated by apoA-I was not significantly affected by the treatment with PDGF/MCSF and PMA. Total cellular protein has not reached significant difference by any of these treatments (Table III). To examine potential contribution to the efflux of intracellular cholesteryl ester hydrolysis as the TABLE II Cellular specific lipid efflux profile Cellular lipid efflux was induced by human HDL (100 g of protein), apolipoprotein A-I (20 g), and lipid microemulsion (312 g of phospholipid) from rat vascular smooth muscle cells and mouse peritoneal macrophages. The cells were loaded with cholesterol as described under "Experimental Procedures." In the initial stage for the efflux reaction, cellular protein and free and esterified cholesterol were 270, 2.1, and 0.9 g/dish of the smooth muscle cell, respectively. For macrophage, cellular protein and free and esterified cholesterol were 36, 1.5, and 2.7 g, respectively. The efflux data are in ng/dish and represent average and standard error of three experiments. Cholesterol/phospholipid ratio is the weight ratio of cholesterol to [phosphatidylcholine plus sphingomyelin] in each lipid efflux. cellular reaction to the growth factors, free/total cholesterol ratio was determined in the cells during the pretreatment period (24 h). There was no significant change in this ratio (Table III). Fig. 1 shows the time course of the efflux of these lipids mediated by lipid-free apoA-I and HDL from control smooth muscle cells and those treated with PDGF/MCSF plus PMA. While free apoA-I-mediated cholesterol efflux was high from the stimulated smooth muscle cells, phospholipid efflux was essentially the same between the control and stimulated cells. The stimulation of the cell did not have a prominent effect on the lipid efflux to HDL. The enhancement of cholesterol efflux by the treatment was also demonstrated in an apoA-I dose-dependent manner (Fig. 2). Cellular cholesterol efflux was more than doubled by the PDGF/MCSF plus PMA treatment of smooth muscle cells in this particular experiment, while phospholipid efflux was almost identical between the control and the treated cells. Thus, the relative enhancement effect of the combined treatment was specific to cholesterol availability to the apoA-I-mediated cellular lipid efflux that had otherwise been poor in smooth muscle cells (Li et al., 1993).
To confirm generation of new pre-␤-HDL by this reaction and incorporation of more cholesterol into this particle by the activation of smooth muscle cells, the culture medium was analyzed by density gradient ultracentrifugation. Fig. 3 demonstrates that the newly formed HDL in the medium was more cholesterol-rich when the cells were stimulated. Thus, more cholesterol molecules (relatively to phospholipid) were incorporated into the pre-␤-HDL particles generated by lipid-free apolipoproteins when the smooth muscle cells were treated with PDGF/MCSF plus PMA.
In order to observe whether this specific increase of cholesterol incorporation into apoA-I-mediated cellular lipid efflux pathway is related to gross change in cellular cholesterol distribution, surface cholesterol of smooth muscle cells was probed by extracellular cholesterol oxidase. Fig. 4 illustrates the percentage of oxidized cellular cholesterol in the control and activated conditions. There was no difference in cellular surface cholesterol observed by this procedure either in the initial rate of the oxidation or the end point of the reaction between the control cells and the cells after the combined treatment.    (160 nM) on the efflux of cholesterol and phosphatidylcholine from rat vascular smooth muscle cells Cholesterol-loaded cells were pretreated with PDGF and/or MCSF and then stimulated by PMA as described in the text. Before the treatment with growth factors, cellular protein/dish was 248.4 Ϯ 37.6 g and cellular free cholesterol was 51.2 Ϯ 1.5% of total cellular cholesterol 3.9 Ϯ 0.8 g/dish. The efflux of lipids was observed in the presence of apoA-I (10 g) or HDL (100 g of protein) for 24 h. The lipid efflux value is percentage of cellular pool of respective lipid, and cellular protein is g/dish, representing mean Ϯ S.E. of four data points. FC/TC, free/total cholesterol ratio (%) in the cell in radioactivity before the efflux mediators were added. Statistical significance for the difference from the control experiments: *, p Ͻ 0.01; **, p Ͻ 0.001.

FIG. 1. Time course of lipid efflux from smooth muscle cells induced by apoA-I (top) and by HDL (bottom). Circles, control;
triangles, treatment with PDGF and MCSF plus PMA. Left panels, cholesterol efflux. Right panels, phosphatidylcholine efflux. The cells were labeled for cholesterol and choline phospholipid as described in the text, treated with growth factors and PMA, and the lipid efflux was monitored up to 24 h in the presence of 10 g of apoA-I or 100 g of HDL as protein in the 1-ml medium. Cellular protein per dish was 310.5 Ϯ 19.4 g. Radioactivity of cellular free cholesterol per dish was 20842 Ϯ 2840 dpm, and cellular free cholesterol was 48.5 Ϯ 2.2% of total cellular cholesterol in radioactivity. Radioactivity of cellular phosphatidylcholine was 6516 Ϯ 668 cpm/dish. sphingosine in the medium, while it did not have any effect on the nonspecific cholesterol efflux. Fig. 5 further demonstrates the effect of protein kinase C inhibitors on the enhancement of apoA-I-mediated cellular lipid efflux from smooth muscle cells. Both sphingosine and staurosporine suppressed the cholesterol efflux in a dose-dependent manner (left panel). The maximum inhibition by sphingosine and staurosporine was 81% and 47% of the increment of cholesterol efflux by PMA, respectively. In contrast, staurosporine showed no significant effect on phosphatidylcholine efflux mediated by apoA-I (right panel). A specific inhibitor of cyclic nucleotide-dependent protein kinase C, H-7, was used in a similar manner (Fig. 6). The increase of apoA-I-mediated cholesterol efflux by PMA from the growth factors-treated cells was almost completely reversed, while the effect on phospholipid efflux was negligible (left panel). The same compound demonstrated no effect on nonspecific lipid efflux from the similarly activated smooth muscle cells to lipid microemulsions (right panel).
Intracellular cholesterol esterification activity was determined in the same stages of activation of cholesterol-loaded smooth muscle cells by the growth factors and PMA as used for lipid efflux experiments (Fig. 7). Incorporation of radiolabeled oleate into cholesteryl ester was not significantly changed by the cellular treatment with both growth factors. Adding PMA for short periods (45 min) significantly decreased the reaction, suggesting the decrease of cholesterol pool available for this reaction. The presence of H-7 completely reversed the effect of PMA. Thus, the effect of PMA and H-7 on intracellular cholesterol esterification was reciprocal to the effect of these compounds on apoA-I-mediated cholesterol efflux from the growth factor-treated vascular smooth muscle cells.
Finally, the experiment was performed for the treatment of mouse peritoneal macrophages with PDGF and MCSF, individually or in combination. Cholesterol efflux from the control cells was already high, and no significant effect of any of these treatments was observed on cholesterol and phospholipid efflux either mediated by HDL or free apoA-I (data not shown). Further incubation with phorbol ester (PMA) of the pretreated cells with the growth factors, which led to a specific effect on smooth muscle cells as described above, did not show any further differential effect on the lipid efflux, either among the different pretreatment groups with the growth factors or between the efflux mediated by apoA-I and HDL (Table V). DISCUSSION After short periods of incubation, PMA stimulated the cholesterol efflux by free apoA-I from the rat vascular smooth muscle cells pretreated with growth factors. The effect was Smooth muscle cells were activated by PDGF, MCSF, and PMA as described in the text. ApoA-I, 10 g/medium, induced cellular lipid efflux for 24 h. The medium was then analyzed in sucrose density gradient ultracentrifugation as described in the text. Specific radioactivity of cholesterol and phosphatidylcholine was obtained for respective cellular lipid pool and used for calculation of weight amount of cholesterol and phosphatidylcholine in each density fraction of the medium. Cellular protein was 375.2 Ϯ 31.9 g/dish, and total cellular cholesterol was 5.45 Ϯ 0.55 g. Free cholesterol was 72.3 Ϯ 0.5% of total cellular cholesterol. The data are expressed as weight percentage of each lipid in each fraction to total phosphatidylcholine in the medium. Circles, cholesterol; triangles, phosphatidylcholine; solid line without symbol, density.

FIG. 4. Oxidation of cholesterol in smooth muscle cells by extracellular cholesterol oxidase.
Circles, control; triangles, treatment with PDGF and MCSF plus PMA. Rat vascular smooth muscle cells were loaded with radiolabeled cholesterol as described in the text. After pretreatment with the growth factors and PMA, cells were incubated with cholesterol oxidase (3 IU/ml) for various periods of time and the radioactivity of cellular free cholesterol and cholestone was determined by thin layer chromatography in order to measure percent oxidation of the cellular free cholesterol. Radioactivity of cellular free cholesterol was 2655 Ϯ 550 cpm/dish, and the ratio of intracellular free cholesterol to total cellular cholesterol was 52.4 Ϯ 3.4% % in radioactivity. Cellular protein was 342.4 Ϯ 12.4 g/dish.

TABLE IV
The effect of the growth factors and PMA on cholesterol efflux to lipid microemulsion Smooth muscle cells were activated by PDGF, MCSF, and PMA as described in the text. The activated cells were also treated with 40 M sphingosine. ApoA-I (10 g) or lipid microemulsions (as 156 g of phospholipid) was added to the 1-ml medium and cholesterol efflux was measured after 24 h. The data was in percentage of total cellular free cholesterol pool. Average protein per dish was 513.5 Ϯ 66.5 g, average cellular total cholesterol was 9.31 Ϯ 0.74 g/dish, and cellular free cholesterol was 54.2 Ϯ 2.3% of total cellular cholesterol. *, p Ͻ 0.01.

Control
Activated Sphingosine ApoA-I 2.24 Ϯ 0.25 3.25 Ϯ 0.24* 2.42 Ϯ 0.05 Emulsion 6.32 Ϯ 0.43 5.83 Ϯ 0.30 6.53 Ϯ 0.48 achieved, but marginally by the pretreatment with PDGF alone and more prominently when the cells were pretreated with PDGF and MCSF. Without the stimulation, cholesterol was poorly incorporated into the apoA-I-mediated lipid efflux, generating cholesterol-poor pre-␤-HDL in the medium. Stimulation of the cells by PDGF and MCSF plus PMA enhanced the apoA-I-mediated cholesterol efflux without changing phospholipid efflux rate, generating pre-␤-HDL more rich in cholesterol. Interestingly, no effect was found on a nonspecific physicochemical lipid efflux pathway to lipid microemulsions by the same treatment of the cells. The effect of the treatment on cholesterol efflux to HDL was only partial. The observation was consistent with the view that the stimulation is specific to cholesterol incorporation to the apoA-I-mediated cellular lipid efflux, while the interaction of apoA-I with the cell itself is unaffected, since phosphatidylcholine efflux was not influenced by these treatment and the same amount of pre-␤-HDL seemed to be produced regardless of the treatment. The partial effect on the HDL-mediated cholesterol efflux may reflect contribution of the apoA-I-mediated mechanism to the overall cellular lipid efflux to HDL. At least three protein kinase C inhibitors reversed the stimulating effect of PMA on the apoA-I-mediated cholesterol efflux from the vascular smooth muscle cells pretreated with PDGF and MCSF, again without changing the cellular phospholipid efflux by free apoA-I. The inhibitors had no effect on the nonspecific cellular lipid efflux to lipid microemulsions. In contrast, intracellular cholesterol esterification was suppressed by the same PMA treatment in the pretreated vascular smooth muscle cells with PDGF and MCSF, and this effect was also reversed by the protein kinase C inhibitor. However, overall cellular free cholesterol pool size was unaffected by these cellular treatment and probing of cellular surface cholesterol by extracellular cholesterol oxidase did not demonstrate any significant difference between the control cells and the stimulated cells. The data thus indicated that intracellular cholesterol is translocated to the specific pool readily available to apoA-Imediated lipid efflux by the short term incubation of the pretreated cells with PMA, and this was inhibited by the protein kinase C inhibitors. This effect was, however, not to an extent that influences cellular surface lipid composition reflecting in nonspecific physicochemical cellular lipid efflux and probing surface cholesterol by cholesterol oxidase.
In the previous few years, a concept of a specific and biologically regulated cellular cholesterol efflux has been gradually developed by several research groups including ourselves. The efflux is mediated by lipid-free helical apolipoproteins and the reaction results in generation of pre-␤-HDL with the apolipoprotein and cellular phospholipid and cholesterol Yokoyama, 1991, 1992;Bielicki et al., 1992;Forte et al., 1993). This efflux reaction is distinguishable from nonspecific physicochemical lipid exchange between cell and lipoprotein in many aspects. The reaction has substantially lower K m (in the order of 10 Ϫ8 M) (Hara and Yokoyama, 1991;  The measurement was quadruplicated. Asterisk indicates significant difference form all others by p Ͻ 0.001. Cellular protein was 341 Ϯ 9 g/dish, cellular cholesterol was 6.2 Ϯ 1.0 g/dish, and free cholesterol was 53.8 Ϯ 2.8% of total cellular cholesterol. Komaba et al., 1992;Mendez et al., 1994) than K d for apolipoprotein-lipid association (Tajima et al., 1983;Yokoyama et al., 1985) and K m for cellular lipid efflux to HDL . A large amount of phospholipid accompanies cholesterol efflux, while phospholipid efflux to lipoproteins is much lower than that of cholesterol (Hara and Yokoyama, 1991;Forte et al., 1993;Li et al., 1993). Extracellular cholesterol esterification by lecithin:cholesterol acyltransferase takes place on the pre-␤-HDL generated by apoA-I (Forte et al., 1995;Czarnecka and Yokoyama, 1995), but it does not have an impact on the rate of lipid efflux by this pathway , while it causes net efflux of cholesterol to HDL by reducing the influx of cholesterol and by shifting equilibrium distribution of cholesterol if the exchange rate is rapid . Proteolytic treatment of cellular surface causes selective inhibition of the apolipoproteinmediated lipid efflux (Li et al., 1995). Specific defect of this pathway has been identified in the fibroblasts from the patients with Tangier disease (Francis et al., 1995) as well as in normal human erythrocytes (Li et al., 1995). The two distinct mechanisms were also differentiated by the lipid efflux profiles from vascular smooth muscle cells. Although there is no essential difference in nonspecific pathway between the smooth muscle cells and mouse macrophages, the former cells are resistant to net cholesterol efflux by free apolipoproteins, not because the cells are less reactive to free apolipoproteins but because cholesterol is poorly available to this reaction Li et al., 1993).
In such a context, we undertook the attempt to modulate lipid-free apolipoprotein-mediated cellular lipid efflux by modification of cellular parameters. Vascular smooth muscle cells were chosen as a model because of their unique behavior in the apolipoprotein-mediated lipid efflux and because of their peculiar reaction to a numbers of factors leading to their various differential stages (Inaba et al., 1992a(Inaba et al., , 1992bShimada et al., 1992). They are known to be induced for macrophage-like functions, including expression of the genes encoding the scavenger receptor and the MCSF receptor in certain conditions such as stimulation by PDGF BB (Inaba et al., 1992a(Inaba et al., , 1992b. MCSF is known to enhance the activity of macrophage (Nakoinz and Ralph, 1988;Hume et al., 1988;Ishibashi et al., 1990;Shimano et al., 1990;Mori et al., 1991;Inaba et al., 1993;Ishii et al., 1994), and PMA is used for inducing differentiation of macrophage cell line cells such as THP-1 cells with respect to a number of macrophage-like functions, including expression of the scavenger receptor (Hara et al., 1987;Takata et al., 1989) and secretion of lipoprotein lipase and apolipoprotein E Menju et al., 1989).
Stimulation of the smooth muscle cells by these factors brought us very interesting results as summarized above. None of the individual factor alone influenced lipid efflux profile either to HDL or by lipid-free apoA-I. When the two growth factors were used in combinations, additional PMA caused prominent specific stimulation of cholesterol availability for lipid-free apoA-I-mediated lipid efflux while phospholipid efflux in this pathway was unchanged. Accordingly, this resulted in generating pre-␤-HDL particles with higher cholesterol/ phospholipids ratio than the particles generated with the nonstimulated cells. More interestingly, the enhancement of cholesterol efflux by the apoA-I-mediated by PMA was reversed by protein kinase C inhibitors. These effects were not observed in a nonspecific lipid efflux pathway. Thus, the apolipoproteinmediated cellular lipid efflux pathway is well distinguished from a physicochemical pathway.
The stimulation of cholesterol efflux was independent of interaction of the apolipoproteins with the cellular surface because phospholipid efflux was intact. Thus, incorporation of cholesterol molecule into the free apolipoprotein-mediated cellular lipid efflux pathway is regulated independently of the cell-apolipoprotein interaction and of generation of pre-␤-HDL. From the fact that the stimulation was observed by the short term incubation with PMA and reversed by the protein kinase C inhibitors, it is fair to conclude that cholesterol enrichment in this pathway is regulated by a mechanism that involves protein kinase C at least in vascular smooth muscle cells. Thus, the mechanism of this increase is more specific than overall activation of lipid metabolism reported in the cells of dividing stage (Fielding et al., 1982). In fact, the possibility of protein kinase C involvement in cellular cholesterol efflux has been suggested in a few other previous works although in a more general manner (Theret et al., 1990;Mendez et al., 1991;Bernard et al., 1992;Voyno-Yasenetskaya et al., 1993). Our data including intracellular cholesterol esterification suggested that translocation of cholesterol from the pool available for the esterification to the specific pool readily available for the apolipoproteinmediated efflux is induced by PMA and prevented by the protein kinase C inhibitor. This view may be consistent with the hypothesis proposed by Mendez et al. (1991). A more specific approach is required for further characterization of the reaction including phosphorylation of specific proteins.
As discussed above, such specific translocation of cholesterol may result in general increase of cholesterol in the cell surface. However, the attempt of probing surface cholesterol by extracellular cholesterol oxidase failed to demonstrate any difference between the control and treated cells. Thus, overall surface cholesterol may not be a regulating factor for the cholesterol availability to apolipoprotein-mediated cellular lipid efflux unless this method may be inappropriate to probe the change of surface cholesterol correctly (Brasaemle and Attie, 1990;Liscum and Dahl, 1992). This may also be consistent with the unaffected nonspecific cellular lipid efflux from the stimulated smooth muscle cells. It should be noted that free/total cholesterol ratio in the cell was not changed by the growth factor treatment, suggesting no increase in overall intracellular free cholesterol pool by such a treatment. Thus, a concept of specific cholesterol pool available to the apolipoprotein-mediated efflux may have to be introduced to interpret these results, and the regulation would be strictly limited for such a specific cholesterol pool. This specific pool should be distinguishable from the cellular cholesterol pool that undergoes nonspecific exchange with lipoprotein surface.
Stimulation of mouse macrophages by human PDGF, MCSF, and PMA, or combinations of these factors, did not demonstrate The cholesterol-loaded macrophages (75.2 Ϯ 7.2 g of protein/dish, free cholesterol was 58.1 Ϯ 1.8% of total cellular cholesterol (7.8 Ϯ 1.7 g)) were pretreated with PDGF (10 ng) and/or MCSF (100 ng) and then stimulated by PMA (160 nM) as described in the text. The efflux of cholesterol was observed for 24 h, in the presence of apoA-I (10 g) and HDL (100 g of protein). The values are percentage of cellular free cholesterol pool representing mean Ϯ S.E. of four data. specific effect on lipid efflux mediated either by lipid-free apoA-I or HDL. This may not be consistent with the report that MCSF stimulated HDL-mediated cholesterol efflux from human monocyte-derived macrophage Inoue et al., 1992). Although the general effect of human MCSF was demonstrated on mouse macrophage (Hume et al., 1988;Ishibashi et al., 1990), the effect may not be entirely similar between the species. Mouse macrophages are generally considered very active in comparison to other species so that they may already be in an activated stage and further activation may not be necessary to achieve the maximum cholesterol efflux level. The results of this study implicate the importance of specific cellular factors in regulation of cholesterol efflux by the lipidfree apolipoprotein-mediated pathway. It seems to be particularly important to investigate its regulation by intracellular trafficking of cholesterol in relation to such a factor as manipulation of cellular protein phosphorylation in order to understand the entire process. Since apolipoprotein-cell interaction seems to be a major source of plasma HDL (Francis et al., 1995), it is increasingly important to study the cellular factors involved in this reaction system.