Lipid Free Apolipoprotein E Binds to the Class B Type I Scavenger Receptor I (SR-BI) and Enhances Cholesteryl Ester Uptake from Lipoproteins*

The Class B type I scavenger receptor I (SR-BI) is a physiologically relevant high density lipoprotein (HDL) receptor that can mediate selective cholesteryl ester (CE) uptake by cells. Direct interaction of apolipoprotein E (apoE) with this receptor has never been demonstrated, and its implication in CE uptake is still contro-versial. By using a human adrenal cell line (NCI-H295R), we have addressed the role of apoE in binding to SR-BI and in selective CE uptake from lipoproteins to cells. This cell line does not secrete apoE and SR-BI is its major HDL-binding protein. We can now provide evi-dence that 1) free apoE is a ligand for SR-BI, 2) apoE associated to lipids or in lipoproteins does not modulate binding or CE-selective uptake by the SR-BI pathway, and 3) the direct interaction of free apoE to SR-BI leads synthesis was measured by simultaneous incubation of cells with Na 235 SO 4 (20 (cid:3) Ci/ml) (42, 43). After incubation cells were washed six times with phosphate-buffered saline and cells were then solubilized with 1 M NaOH or were incubated with 12.5 (cid:3) g/ml trypsin in phosphate-buffered saline with 1 m M EDTA for 15 mn at 37 °C to release cell surface proteoglycans. Total cells and trypsin-released materials were counted for 35 S incorporation. The decrease by 24 (cid:3) 5% of Na 2 35 SO 4 incorpora- tion in cells and of 49 (cid:3) 11% in cell surface trypsin-released materials demonstrates that their proteoglycan content was effectively decreased during xyloside incubation.

The Class B type I scavenger receptor I, SR-BI, 1 binds HDL and mediates the selective uptake of HDL cholesteryl esters (CE) in cultured cells (1). Another property of the Class B scavenger receptor, shared with CD 36, is to bind either native (2,3) or modified lipoproteins (acetylated or oxidized (3)(4)(5)). They are also the first defined receptors to be able to specifically bind anionic but not cationic or zwitterionic liposomes (6). However CD 36 is less efficient than SR-BI in promoting CE-HDL uptake from native lipoproteins to cells (7).
There are probably different binding sites on SR-BI. HDL compete for the binding of LDL to SR-BI but LDL poorly inhibit the binding of HDL, and there is no reciprocal cross-competition between these two ligands (1,8). The study of several mutants of SR-BI also supports for the proposal that the interaction of SR-BI with HDL differs from that with LDL (9).
Apolipoproteins (apo) AI, AII, and CIII of HDL either associated with lipids or in lipid free forms can directly mediate their binding to SR-BI (10). Williams et al. (11) demonstrated that SR-BI can interact with multiple sites in apoAI and identified the Class A amphipathic ␣-helix as a recognition motif. The specific role of apoAI in the delivery of cholesterol to adrenal cells was clearly demonstrated in mice deficient either in apoAI or apoAII (12). We have demonstrated, in a human adrenal cell line, that apoAII, which binds SR-BI with high affinity, can act as an antagonist of apoAI in CE-HDL uptake (13).
ApoE is a constituent of trigyceride-rich lipoproteins and is essential for the receptor-mediated uptake of their remnants (14) and for their catabolism by pathways involving heparansulfate proteoglycans (15). ApoE deficiency in mice leads to impaired catabolism of these remnants and increased atherosclerotic lesions (for review, see Ref. 16). Direct interaction of apoE with SR-BI has never been demonstrated, and its implication in CE uptake is still controversial.
It was previously shown that HDL binds murine SR-BI (1), human SR-BI (4), and human CD 36 (2) independently of the removal of apoE by chromatography on heparin-Sepharose (for review, see Ref. 17). These results clearly demonstrated that HDL binding can occur with high affinity in the absence of apoE.
Different results, obtained before the description of SR-BI as an HDL receptor, implicated cellular apoE in enhancing HDL binding to cells and CE uptake from HDL. Selective uptake of CE by HepG2 cells was reduced by antibodies directed against the receptor binding domain of apoE (18). The enhancing effect of apoE on CE uptake was not due to the transfer of apoE to HDL or to the LDL-receptor pathway. Inversely, Ji et al. (19) demonstrated, in cultured hepatocytes, that if apoE (in apoEenriched HDL) enhanced the uptake of HDL particles, it had little effect on the selective uptake of CE-HDL. Similar results were described by Rinninger et al. (20). More recent studies demonstrated that apoE expression by mouse adrenocortical cells increased both endocytic and selective uptake of CE-LDL but had little influence on CE uptake from HDL (21). Basal CE-selective uptake could be observed in the absence of secreted apoE but was enhanced by the apoE synthesis.
In vivo studies in mice deficient in apoE cannot determine the importance of apoE in CE-selective clearance. Different authors suggest that in mice lacking apoE, selective uptake of CE from remnant lipoproteins is not impaired, but occurs probably via SR-BI and could be facilitated by hepatic lipase (HL) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In this paper, we have examined the role of apoE, free or associated with lipids, in binding to SR-BI and in selective CE uptake from lipoproteins to cells. We used a human adrenal cell line, which naturally expresses a high level of SR-BI and where SR-BI is the major HDL-binding protein (13). These cells do not secrete apoE, and we wanted to determine whether apoE can have direct interaction with SR-BI either free or associated with lipids in reconstituted proteoliposomes or in lipoproteins. We have demonstrated that free apoE competes with high affinity with reconstituted HDL, 1-palmitoyl-2-oleoyl-L-phosphatidylcholine-apoAI (POPC-AI), in binding to cells. But when associated with phospholipids in 1,2-dipalmitoyl-L-3-phosphatidylcholine-apoE (DPPC-E), apoE fails to compete with POPC-AI. The presence of apoE in lipoproteins, naturally or by in vitro incubation with high apoE concentrations, does not modify binding and selective uptake of CE by cells in conditions where the LDL receptor pathway is inhibited. However, the addition of small amounts of apoE3 to cells, in the presence of lipoproteins devoid of apoE, leads to a 2-fold increase in CEselective uptake by cells. In these experimental conditions, apoE does not bind to lipoproteins, and we suggest that direct interaction of apoE with SR-BI, in the same site as the POPC-AI binding site, even if it is not necessary for selective uptake, could modify the SR-BI structure in cell membranes and facilitate CE uptake.
Lipoprotein Isolation and ApoE Depletion or Enrichment-LDL and high density lipoprotein fractions 2 and 3 (HDL 2 and HDL 3 ) were prepared from human plasma by sequential ultracentrifugation, respectively, at density 1.030 g/ml Ͻ d Ͻ 1.053 g/ml, 1.063 g/ml Ͻ d Ͻ 1.12 g/ml, and 1.12 g/ml Ͻ d Ͻ 1.21 g/ml (29). HDL 2 were purified on heparin-Sepharose column (30) to obtain an apoE free unretained HDL 2 fraction. Lack of apoE in HDL 3 and in apoE free unretained HDL 2 fraction was tested by ELISA (25). VLDL from cholesterol-fed rabbits or from apoE-deficient mouse serum were prepared by ultracentrifugation at density 1.006 g/ml. In some experiments, lipoproteins devoid of apoE were enriched with human recombinant apoE as described previously in a 0.6/1 (w/w) ratio (31). Briefly, concentrated lipoproteins (0.5 mg/ml) were incubated with recombinant apoE3 (0.3 mg/ml) for 1 h at 37°C and then diluted for incubation with cells. In these conditions nearly all apoE3 remains associated with lipoproteins. In the experiments indicated lipoproteins were reisolated by ultracentrifugation at the appropriate density to remove free apoE3. When the lipoproteins were labeled for binding or CE-selective uptake experiments, apoE3 enrichment was performed after labeling.
Lipoprotein Labeling-Native lipoproteins (1 mg) were labeled with [ 3 H]CE as described previously (27). The final specific activities varied between 45 and 400 dpm/ng of cholesterol. Lipoproteins (native or reconstituted) were labeled in protein moiety with [ 125 I]iodine using the iodine monochloride method (36). The final specific activities varied between 400 and 1500 dpm/ng of protein.
Binding of 125 I-Labeled Lipoproteins and Competition with Unlabeled Apolipoproteins and Lipoproteins-Cells cultured in 12-well plates were washed and preincubated for 1 h at 37°C in DMEM/F-12 serum-free medium. For direct binding, cells were incubated for 1 h at 37°C with 125 I-labeled lipoproteins without (total binding) or with a 40-fold excess of unlabeled lipoproteins (nonspecific binding). Cells were then washed and dissolved with 1 mol/liter NaOH, and cellassociated radioactivity was counted. An aliquot was used to quantify cellular proteins (37). Results were expressed as nanograms of bound or degraded proteins per milligrams of cellular proteins. Specific binding was calculated as the difference between total and nonspecific binding.
Competition studies were performed with 125 I-POPC-AI at 2 g/ml and increasing concentrations of the different unlabeled apolipoproteins or lipoproteins for 1 h at 37°C. Cells were preincubated, washed, and then counted as in direct binding studies. Results were expressed as a percentage of the binding measured without competitor.
In the experiments indicated, binding studies were performed in the presence of 50 M chlorpromazine to inhibit endocytosis by coated pit pathway (38).

Selective Uptake of [ 3 H]CE-labeled
Lipoproteins-Cells cultured in 12-well plates were washed and preincubated for 1 h at 37°C in DMEM/ F-12 serum-free medium. They were then washed and incubated for the indicated times with [ 3 H]CE-labeled lipoproteins without (total uptake) or with a 40-fold excess of unlabeled HDL (nonspecific uptake). Cells were then washed and dissolved with 1 mol/liter NaOH, and cellassociated radioactivity was counted. An aliquot was used to quantify cellular proteins (37). Results are expressed as nanograms or milligrams of cholesterol incorporated per milligram of cellular proteins. In the experiments indicated, CE-selective uptake studies were performed in the presence of 50 M chlorpromazine to inhibit endocytosis (38).
To measure the effect of direct interaction of apoE3 with cells, cells were preincubated for 1 h at 37°C in DMEM/F-12 serum-free medium with 10 g/ml (or the indicated concentrations) of apoE3. In some experiments apoE3 (100 g) was pretreated with thrombin (2 g) in 100 mM bicarbonate buffer for 24 h as described previously by Bradley et al. (39,40). The efficiency of the cleavage of recombinant apoE3 (40 kDa) was verified by the presence of two major fragments (10-kDa C-terminal fragment and 26-kDa N-terminal fragment) after SDS-PAGE in reducing conditions.
To inhibit proteoglycan synthesis, cells were preincubated for 20 h with 1 mM p-nitrophenyl-␤-D-xylopyranoside (␤-D-xyloside), a general inhibitor of proteoglycan synthesis (41), and during the lipoprotein uptake assay. The effect of ␤-D-xyloside on glycosaminoglycan synthesis was measured by simultaneous incubation of cells with Na 2 35 SO 4 (20 Ci/ml) (42,43). After incubation cells were washed six times with phosphate-buffered saline and cells were then solubilized with 1 M NaOH or were incubated with 12.5 g/ml trypsin in phosphate-buffered saline with 1 mM EDTA for 15 mn at 37°C to release cell surface proteoglycans. Total cells and trypsin-released materials were counted for 35 S incorporation. The decrease by 24 Ϯ 5% of Na 2 35 SO 4 incorporation in cells and of 49 Ϯ 11% in cell surface trypsin-released materials demonstrates that their proteoglycan content was effectively decreased during xyloside incubation.

Effect of Free ApoE on Ligand Binding to SR-BI-
We previously demonstrated that reconstituted HDL, POPC-AI, bind to NCI-H295R adrenal cells by direct interaction with SR-BI, which was the major if not the unique receptor for these ligands (13). Among the tested ligands, POPC-AI had the highest af-finity for SR-BI. We wanted to know if apoE was a ligand for SR-BI. We first analyzed the competition of apoE with POPC-AI to bind to SR-BI. In Fig. 1A we can see that free apoE competes with 125 I-POPC-AI. There is no significant difference between the recombinant apoE2, apoE3, and apoE4 isoforms, but their affinity for SR-BI was slightly lower than free apoAI. In the conditions used for competition experiments, apoE did not displace 125 I-apoAI from 125 I-POPC-AI, since more than 95% of 125 I-apoAI can be reisolated in POPC-AI by ultracentrifugation at d Ͼ 1.21 g/ml (not shown).
We then studied the competition of free apoE3 with other lipoproteins, which are also ligands of SR-BI, and we showed that free apoE failed to compete efficiently with native 125 Ilipoproteins (HDL 3 , LDL, or VLDL) (Fig. 1B).
After incubation with cells in the presence of apoE, 125 I-POPC-AI and 125 I-lipoproteins were reisolated by ultracentrifugation at d ϭ 1.21 g/ml and the apoE content of the different fractions was measured by ELISA (25). For 125 I-lipoproteins, the main part of apoE was found in the bottom (Ͼ90%) but during the competition with 125 I-POPC-AI, for higher apoE concentrations, a part of apoE was found in the top, indicating a partial apoE association with lipids.
Effect of Phospholipids on ApoE Binding to SR-BI-We then reconstituted lipoproteins, DPPC-E and POPC-E complexes, to evaluate the role of lipids in the interaction of apoE with SR-BI, and we studied the competition between these complexes and POPC-AI. As shown in Fig. 2, DPPC-E poorly competes with 125 I-POPC-AI to bind to SR-BI. There is no difference between the different apoE isoforms. POPC-E3 also fails to compete with 125 I-POPC-AI. In a control experiment, competition for the binding of 125 I-POPC-AI by DPPC-AI was lower than POPC-AI but much greater than DPPC-E.
Effect of ApoE, Free or Associated to Lipids, on the Selective Uptake of Cholesteryl Esters from Lipoproteins Devoid of ApoE-We then wanted to evaluate the effect of apoE, free or associated to lipids, on the selective uptake of cholesteryl esters from HDL. To inhibit the LDL-receptor pathway, the selective uptake experiments were performed in the presence of chlorpromazine, an inhibitor of endocytosis (38). We incubated cells with free apoE3, then measured the selective CE uptake from [ 3 H]CE-lipoproteins to cells. Chlorpromazine does not modify selective uptake from apoE3 free lipoproteins (not shown). Surprisingly, we can see in Fig. 3 that the presence of free apoE3 in the culture medium increases 2-3-fold the selective CE uptake mediated by SR-BI from human [ 3 H]CE-HDL 3 (Fig. 3A) or from [ 3 H]CE-VLDL prepared from apoE-deficient mice (Fig.  3B). We verified that human plasma apoE3 has the same stimulating effect that the recombinant peptide (not shown). This effect was not obtained with free apoAI (Fig. 3, A and B), which, at higher concentrations, plays a competitive role in the selective uptake process (13). The DPPC-E preparation (Fig.  3A) fails to inhibit or enhance CE-selective uptake from HDL instead of POPC-AI, which strongly inhibits selective uptake from HDL as found previously (13).
Since apoE interaction with lipids in lipoproteins is not required to stimulate selective uptake from HDL, it was interesting to determine whether the N-terminal fragment of apoE was sufficient to elicit effect on SR-BI-mediated selective CE uptake. We incubated apoE3 with thrombin for 24 h as indi- cated under "Experimental Procedures," apoE3 was cleaved in two major fragments as verified by SDS-PAGE (Fig. 4B). Digested apoE3 failed to enhance [ 3 H]CE-HDL-selective uptake, whereas in the same experiment apoE3 increased uptake by 2.8 (Fig. 4A). These results suggest that the positive effect of apoE3 on the selective uptake of cholesteryl esters could be dependent on the direct interaction of free apoE with SR-BI. However to exclude that the effect of free apoE3 could be mediated by the interaction of apoE3 with the [ 3 H]CE-lipoproteins, after the selective uptake experiments, we reisolated lipoproteins from the cell culture medium by ultracentrifugation at d Ͼ 1.21, and we found less than 8% of apoE associated with lipoproteins isolated at the top (measured by ELISA).
Effect of ApoE Present in Lipoproteins or in rHDL on CEselective Uptake from These Lipoproteins-Despite former re-sults, which suggest that only free apoE can have a competitive role in rHDL binding and an enhancing role in CE-lipoprotein uptake, we investigated whether the presence of apoE in lipoproteins had an influence on CE-selective uptake. For this purpose, we prepared lipoproteins naturally containing apoE and compared them with lipoproteins devoid of apoE. We compared [ 3 H]CE-selective uptake from total human HDL 2 (1.063 Ͻ d Ͻ 1.12 g/liter) and from the fraction of human HDL 2 not retained on the heparin-Sepharose column (free apoE HDL 2 ). Fig. 5 does not show any difference between these two HDL subfractions either for competition with 125 I-POPC-AI (Fig. 5A) or for direct CE-selective uptake from these lipoproteins (Fig. 5B).
We then prepared VLDL from rabbits fed a cholesterol-rich diet. These VLDL were apoE-rich, but the apoE content depends on the duration of the cholesterol-rich diet (40). We obtained three VLDL preparations with increasing apoE content. After labeling, [ 3 H]CE-selective uptake from these different VLDL preparations was measured, and no difference in CE-selective uptake was observed (Fig. 6). However, lipoproteins containing apoE and devoid of apoE are often found to be also different as regards to their lipid composition, and this could modify per se the CE-selective uptake (44,45). We then artificially prepared apoE-enriched VLDL and HDL 3 by the method described previously (31). VLDL from KO E mice and human HDL 3 were incubated at 37°C with free apoE3. In the conditions described under "Experimental Procedures" (31), nearly all the apoE3 was incorporated in the lipoproteins, and this was verified by ELISA after reisolation of lipoproteins. Before incorporating apoE into the lipoproteins, they were doubly labeled with 125 I and [ 3 H]CE. In the presence of chlorpromazine, no significant effect of apoE3 enrichment of lipoprotein, either on the binding measured by cell association of 125 Ilabeled proteins (Fig. 7A), or on [ 3 H]CE-selective uptake, could be observed for HDL 3 and VLDL (Fig. 7B). In the conditions used here, there was no cellular degradation of the protein moiety of lipoproteins (Ͻ1% of equivalent CE internalized for HDL and Ͻ5% for VLDL). These results confirm that the association of apoE to lipoproteins does not modify their interaction with SR-BI or their capacity to transfer CE to cells.
To measure direct selective uptake from discoidal rHDL (100/1/0.3, PL/ApoAI/CE or PL/ApoE3/CE), dual labeled rHDL ( 125 I-apolipoproteins and [ 3 H]CE) at 5 g/ml were incubated with cells for 2 h at 37°C in the presence of chlorpromazine. The data summarized in Table I show that apoE3 or apoAIcontaining rHDL can bind to cells but that the binding of DPPC/apoE3/CE or POPC/apoE3/CE cannot be effectively displaced by a 40-fold HDL excess as is the case for POPC/apoAI/ CE. The selective uptake of [ 3 H]CE from these rHDL was lower for apoE3-containing rHDL than for apoAI-containing rHDL and was not inhibited by a 15-fold excess of POPC-AI or a 40-fold excess of HDL. The partial inhibition of selective uptake from apoE-containing rHDL (but not from apoAI-containing rHDL) obtained with a 40-fold excess of LDL can suggest that, as this selective uptake is SR-BI-dependent, the interaction site with SR-BI of apoE-containing rHDL is the LDL site rather than the HDL or the POPC-AI site.
Role of Proteoglycans on the Stimulation of Selective Uptake from HDL by Free ApoE-Results from independent studies have shown that apoE can interact with proteoglycans (15,46) and/or with HL to facilitate interaction with receptors (19,22), and it was not excluded that, in NCI-H295R cells, this mecha-nism could account for a part of the free apoE3 effect. To eliminate that possibility, we cultured cells with ␤-D-xyloside, a general inhibitor of proteoglycan synthesis (41). We verified as indicated under "Experimental Procedures," that xyloside incubation decreased by 50% cell surface expression of proteoglycans. Fig. 8 shows that the activator effect of free apoE on CE-selective uptake is not suppressed in the presence of xyloside, and so this effect is probably independent of proteoglycan interaction.

DISCUSSION
In vitro and in vivo experiments have demonstrated that SR-BI is a physiologically relevant HDL receptor that can mediate selective CE uptake in the liver and in steroidogenic tissues (for review, see Refs. 47 and 48). Apolipoprotein E is a ligand of lipoprotein receptors from the LDL-receptor family (15,49), but was not shown as a specific ligand of scavenger receptors and especially of SR-BI. However many papers have reported a role of apoE either in facilitating CE-selective uptake by cells or selective clearance of CE-HDL in vivo by the liver (50).
In this paper we have demonstrated in a human adrenal cell line NCI-H295R, which does not secrete apoE, that 1) free apoE competes with POPC-AI in binding to SR-BI, 2) apoE associated with lipids or in lipoproteins does not modify CE-selective uptake, but 3) strikingly, free apoE3 can interact with SR-BI and that this interaction leads to an increase in CE-selective uptake from VLDL and HDL.
We have shown that lipid free apoE (independently of the isoform E2, E3, or E4) competes with POPC-AI in binding to NCI-H295R cells. As SR-BI is the major cellular receptor for POPC-AI in these cells (13), this is a demonstration of the direct interaction of lipid free apoE with SR-BI in cells. In the competition with POPC-AI, high apoE concentrations induced apoE association with lipids, but we further demonstrated that the association of apoE to phospholipids highly decreased the competition between apoE and POPC-AI. We obtained direct binding of POPC/apoE3/CE and DPPC/apoE3/CE to NCI-H295R cells, but this binding was not displaced by POPC-AI or by a large HDL excess but partly by LDL. It is therefore FIG. 5. Role of apoE present in HDL 2 for binding or for cholesteryl ester transfer to cells. For binding experiments, increasing concentrations of total HDL 2 (q) (1.063 g/ml Ͻ d Ͻ 1.21 g/ml) or of HDL 2 nonretained on heparin-Sepharose column (E) (apoE free HDL 2 ) were incubated with NCI-H295R cells for 1 h at 37°C in the presence of 2 g/ml 125 I-POPC-AI (A). Selective uptake (B) from total [ 3 H]CE HDL 2 (q) or from [ 3 H]CE apoE free HDL 2 (E) was measured after a 4-h incubation at 37°C with cells. Experiments were performed in the presence of chlorpromazin (50 M) to inhibit the endocytic pathway. After incubation, cells were washed, and cell-associated radioactivity was measured after NaOH digestion as indicated under "Experimental Procedures." Results are mean Ϯ S.D. of two independent experiments performed in triplicate.
FIG. 6. Role of apoE present in VLDL for selective cholesteryl ester transfer to cells. VLDL were prepared from rabbits fed for 0, 2, or 6 weeks on a cholesterol-rich diet. ApoE content of VLDL was estimated by SDS-PAGE. The increasing content of apoE was indicated as low (f), medium (OE), or high (q). Selective uptake from these different [ 3 H]CE VLDL was measured after 4-h incubation at 37°C with cells. Experiments were performed in the presence of 50 M chlorpromazine to inhibit the endocytic pathway. After incubation, cells were washed, and cell-associated radioactivity was measured after NaOH digestion as indicated under "Experimental Procedures." Results are mean Ϯ S.D. of two independent experiments performed in triplicate.
possible that the binding of POPC/apoE3/CE reflects interactions with other receptors or with binding sites on SR-BI independent of apoAI or HDL. Competition curves with POPC-AI are the best indicators of the differential interaction of lipid free apoE and DPPC-E or POPC-E with SR-BI but on a particular site (the POPC-AI binding site). The competition of DPPC-E or POPC-E is low but not negligible and is coherent with the difference in binding affinity to SR-BI published by different authors (35,51) between POPC-AI (about 4 g/ml) and POPC-E (about 35 g/ml). Thuahnai et al. (35) showed a direct interaction between reconstituted lipoproteins (POPC/ CE/apoE or DPPC/CE/apoE, 100/0.3/1) and SR-BI-transfected cells, with these lipoproteins transferring their CE to cells. We also observed transfers of CE to NCI-H295R cells from reconstituted POPC/apoE/CE, but the selective uptake from these reconstituted HDL is very much lower than from native lipoproteins, certainly because of the low availability of CE in these lipoproteins. ApoAI in these rHDL is more efficient in promoting CE uptake than apoE. In the same way we demonstrated that the presence of apoE in lipoproteins does not modify their interaction with cells and does not impair the selective uptake of CE from these lipoproteins to cells in presence of chlorpromazine, which inhibits the LDL receptor pathway. These results were obtained either with native lipoproteins (HDL 2 , VLDL) or with lipoproteins devoid of apoE and further in vitro enriched with apoE3 (VLDL from mice lacking apoE gene or human HDL 3 ). This was previously described for HDL 3 by different authors (1,4), but we extended these results to VLDL. The low affinity for SR-BI of apoE associated with lipids does not modify the affinity for this receptor of apoEcontaining lipoproteins, which can interact through other apolipoproteins present at their surface and/or by their lipid moiety.
The most interesting result is the capacity of free apoE to increase the selective CE uptake from different lipoproteins without any effect on their binding capacity. In our experiments free apoE3, at low concentrations (from 2 to 10 g/ml), is allowed to interact with cells for 1 h before adding lipoproteins. Then apoE is maintained in the cell medium during CE-selective uptake experiments. However, simultaneous addition of lipoproteins and apoE3 led to nearly the same stimulation of CE-selective uptake (not shown). ApoE incorporation in lipoproteins cannot occur in cell medium when free apoE is added to labeled lipoproteins because of the very low concentration of the lipoproteins (Ͻ0.1 mg/ml). Our leading hypothesis of different binding sites for lipoproteins and free apolipoproteins is coherent with the described "nonreciprocal cross-competition" between LDL and HDL for SR-BI binding (1) or the dissociation of LDL and HDL binding activity of murine SR-BI by retrovirus library-based activity dissection (9), with these two papers showing that the multiple ligands of SR-BI can bind to different sites on the extracellular loop of the protein. Direct interaction of apoE with SR-BI, without playing a role in the binding of CE-rich lipoproteins, could therefore modify the structure of the complexes and enhance CEselective uptake mechanism. An important feature is that a constant apoE3 concentration (10 g/ml) has the same stimulating effect on CE uptake from lipoproteins independently of their own concentration (up to 100 g/ml). This strengthens the idea that the apoE effect does not imply any interaction of apoE with lipoproteins but rather a direct interaction with SR-BI. This direct interaction implies the whole apoE peptide, since  thrombin fragmentation of apoE impairs its stimulating effect on CE-selective uptake. It is possible that cell membrane lipids or hydrophobic interactions with SR-BI could be involved, but this hypothesis needs further investigation. The hypothesis that the SR-BI-mediated removal of HDL lipids by the liver appears to involve both apoE and HL as ligands or co-receptors has previously been suggested (for review, see Ref. 52), and so we tested the possibility of an interaction of apoE with other cellular components that could interact directly with SR-BI or with lipoproteins bound to SR-BI to facilitate CE-selective uptake. Proteoglycans and proteoglycanbound lipases could be good candidates to bind apoE for this function as co-receptor. They play a major role in apoE interaction with LDL-receptor-related protein and in lipoprotein metabolism (53), and their implication was suggested in HDL interactions with cells (18 -20). We wanted to know whether the presence of proteoglycans was important in mediating the apoE role in selective uptake, and we cultured cells with ␤-Dxyloside, a general inhibitor of proteogycan synthesis (41,54). In cells cultured in the presence of ␤-D-xyloside, apoE maintains a high activating effect on CE-selective uptake, indicating that proteoglycans, and indirectly lipases, do not play a major role in that apoE effect. However our results do not exclude the participation of another cellular protein in a complex interaction with SR-BI.
What is the in vivo relevance of the stimulating effect of apoE on CE-selective uptake? Arai et al. (24) showed that apoEdeficient mice had a decreased clearance of CE-HDL, which was not further impaired by attenuation of the SR-BI gene. Free apoE does not exist in plasma in substantial amount, but the potential physiologic relevance is probably in the capacity of most tissues to secrete apoE and especially liver and adrenal cells. ApoE is not secreted by NCI-H295R cells in the culture medium, but in vivo a cell surface "blanket" of apoE was detected on rat adrenocortical cells (55). This apoE could play a major role in speeding up CE-selective uptake by SR-BI in adrenals. In vitro expression of apoE was induced in a murine adrenocortical cell line (21), although its effect on enhancing CE-selective uptake was only shown on LDL and not on HDL. The discrepancy with our in vitro results can arise from differences in species or from SR-BI differential interaction with exogenous or endogenous apoE, and the same authors in further studies demonstrated the implication of proteoglycans in the effect of apoE on CE-LDL selective uptake by these cells (54). In the liver, another tissue with a high level of CEselective uptake, it has been demonstrated that hepatocytes highly secrete apoE, which is recaptured by LDL-receptorrelated protein, facilitating the endocytosis of triglyceride-rich lipoproteins (15). This apoE could also interact directly with SR-BI or with another cellular protein to enhance SR-BI mediated CE-selective uptake. Therefore, in steroidogenic tissues and in the liver it can be postulated that apoE contributes to a high level of CE-selective uptake, and when SR-BI is deficient it can be speculated that overproduction of apoE contributes to the formation of large apoE-rich HDL observed in SR-BI-deficient mice (50).
Here we have clearly demonstrated the role of apoE in CEselective uptake by the SR-BI-mediated pathway. But apoE is secreted in macrophages, and its secretion is correlated with enhanced cholesterol efflux (56,57). In a recent paper, von Eckarstein et al. (58) showed that the ATP-binding cassette protein-A1 (ABC-A1) transporter (which modulates cholesterol efflux via apoAI) also controls apoE intracellular transport and secretion. Stimulation of apoE secretion was also obtained by apoAI (59). Do these stimulations of apoE secretion have a stimulating effect on CE-selective uptake dependent on SR-BI, or does apoE secreted by macrophages have a preferential effect on cholesterol efflux dependent or not on SR-BI? The positive antiatherogenic effect of apoE expression in macrophages has already been demonstrated (60,61) and even sometimes discussed (62). However the mechanism is still unclear, and the balance between apoE-mediated CE uptake stimulation and cholesterol efflux remains to be investigated and especially in macrophages. FIG. 8. Lack of effect of xyloside, an inhibitor of proteoglycan synthesis, on the activation of CE-selective uptake by direct interaction of apoE with cells. NCI-H295R cells were incubated for 24 h with 1 mM xyloside. 1 h before lipoprotein addition, 10 g/ml apoE3 was added to cells, then cells were incubated for 1 or 4 h with [ 3 H]CE-HDL 3 at 30 g/ml (7.5 g/ml cholesterol) to measure CE-selective uptake. Xyloside and apoE3 were maintained during selective uptake measurement. Cells incubated without xyloside and/or without apoE are used as controls. Results expressed in ng of CE internalized per mg of cellular proteins are mean Ϯ S.D. of two independent experiments performed in triplicate.