Unique Cellular Events Occurring during the Initial Interaction of Macrophages with Matrix-retained or Methylated Aggregated Low Density Lipoprotein (LDL)

A critical event in atherogenesis is the interaction of arterial wall macrophages with subendothelial lipoproteins. Although most studies have investigated this interaction by incubating cultured macrophages with monomeric lipoproteins dissolved in media, arterial wall macrophages encounter lipoproteins that are mostly bound to subendothelial extracellular matrix, and these lipoproteins are often aggregated or fused. Herein, we utilize a specialized cell-culture system to study the initial interaction of macrophages with aggregated low density lipoprotein (LDL) bound to extracellular matrix. The aggregated LDL remains extracellular for a relatively prolonged period of time and becomes lodged in invaginations in the surface of the macrophages. As expected, the degradation of the protein moiety of the LDL was very slow. Remarkably, however, hydrolysis of the cholesteryl ester (CE) moiety of the LDL was 3–7-fold higher than that of the protein moiety, in stark contrast to the situation with receptor-mediated endocytosis of acetyl-LDL. Similar results were obtained using another experimental system in which the degradation of aggregated LDL protein was delayed by LDL methylation rather than by retention on matrix. Additional experiments indicated the following properties of this interaction: (a) LDL-CE hydrolysis is catalyzed by lysosomal acid lipase; (b) neither scavenger receptors nor the LDL receptor appear necessary for the excess LDL-CE hydrolysis; and (c) LDL-CE hydrolysis in this system is resistant to cellular potassium depletion, which further distinguishes this process from receptor-mediated endocytosis. In summary, experimental systems specifically designed to mimic thein vivo interaction of arterial wall macrophages with subendothelial lipoproteins have demonstrated an initial period of prolonged cell-surface contact in which CE hydrolysis exceeds protein degradation.

A critical event in atherogenesis is the interaction of arterial wall macrophages with subendothelial lipoproteins. Although most studies have investigated this interaction by incubating cultured macrophages with monomeric lipoproteins dissolved in media, arterial wall macrophages encounter lipoproteins that are mostly bound to subendothelial extracellular matrix, and these lipoproteins are often aggregated or fused. Herein, we utilize a specialized cell-culture system to study the initial interaction of macrophages with aggregated low density lipoprotein (LDL) bound to extracellular matrix. The aggregated LDL remains extracellular for a relatively prolonged period of time and becomes lodged in invaginations in the surface of the macrophages. As expected, the degradation of the protein moiety of the LDL was very slow. Remarkably, however, hydrolysis of the cholesteryl ester (CE) moiety of the LDL was 3-7-fold higher than that of the protein moiety, in stark contrast to the situation with receptor-mediated endocytosis of acetyl-LDL. Similar results were obtained using another experimental system in which the degradation of aggregated LDL protein was delayed by LDL methylation rather than by retention on matrix. Additional experiments indicated the following properties of this interaction: (a) LDL-CE hydrolysis is catalyzed by lysosomal acid lipase; (b) neither scavenger receptors nor the LDL receptor appear necessary for the excess LDL-CE hydrolysis; and (c) LDL-CE hydrolysis in this system is resistant to cellular potassium depletion, which further distinguishes this process from receptor-mediated endocytosis. In summary, experimental systems specifically designed to mimic the in vivo interaction of arterial wall macrophages with subendothelial lipoproteins have demonstrated an initial period of prolonged cell-surface contact in which CE hydrolysis exceeds protein degradation.
Arterial wall macrophages are prominent features of both early and advanced atherosclerotic lesions (1)(2)(3), and there is increasing evidence that these cells play important roles in early atherogenesis (4 -6) as well as in the progression to acute clinical events (7,8). A critical event in the life span of the arterial wall macrophage is its interaction with subendothelial lipoproteins; for example, when macrophages internalize these lipoproteins, massive CE 1 accumulation, or foam cell formation, can ensue (9 -11). Most studies have attempted to investigate macrophage-lipoprotein interactions in vitro by incubating monolayers of cultured macrophages with tissue culture medium containing monomers of certain lipoproteins, such as oxidized LDL, ␤-VLDL, or acetyl-LDL (12)(13)(14). In vivo, however, lesional macrophages encounter lipoproteins that are mostly retained on a three-dimensional network of extracellular matrix (15)(16)(17)(18)(19)(20). For example, Smith et al. (15) showed that only 8% of lesional lipoproteins in human aortic fatty streaks could be released by extraction in aqueous buffer or by electrophoresis. Furthermore, matrix-retained lesional lipoproteins are often aggregated and fused (16, 20 -25). The importance of these issues in foam cell formation is demonstrated by the finding that prolonged incubation of macrophages with either aggregated LDL or matrix-retained and aggregated LDL leads to massive CE accumulation (26 -31).
In view of this background, it occurred to us that the cellular processes involved in the initial interaction of macrophages with lipoproteins that are retained and aggregated in a threedimensional matrix may differ substantially from the processes involved in the initial interaction of macrophages with monomeric lipoproteins dissolved in tissue culture medium. In particular, the usual experimental system involves receptor-mediated endocytosis (9,32) while the situation in vivo most likely involves some form of phagocytosis (31,33) or other non-clathrin-coated pit mechanisms (34). In this light, we have established two experimental systems that attempt to mimic certain unique aspects of the early stages of encounter between macrophages and retained and aggregated lipoproteins. Using these systems, we have found that the retained and aggregated li-poproteins remain bound to the external surface of macrophages for an extended period of time and that there is a delay in the degradation of the protein moiety of the lipoproteins. Remarkably, however, CE hydrolysis proceeds at a high rate during this period and markedly exceeds the rate of protein degradation. These events, which differ from those observed with receptor-mediated endocytosis, may more accurately reflect the initial events that occur when macrophages encounter subendothelial lipoproteins in developing atherosclerotic lesions.

EXPERIMENTAL PROCEDURES
Materials-Tissue culture media and reagents were purchased from Life Technologies, Inc., tissue culture plates were from Corning, and defined fetal bovine serum was from HyClone Laboratories, Inc. (Logan, UT). Low potassium medium was made substituting the salt solution of DMEM with potassium-free buffer (142 mM NaCl, 3.6 mM CaCl 2 , 0.81 mM MgCl 2, 20 mM HEPES, pH 7.4). Lipoprotein-deficient serum was prepared by ultracentrifugation of the fetal bovine serum to obtain the d Ͼ1.21 g/ml fraction. [1,2,6, (38) was kindly provided by Dr. John Heider of Sandoz, Inc., East Hanover, NJ. Stock solutions (10 mg/ml) were prepared in dimethyl sulfoxide.
Cells-J774.A1 macrophages (from the American Type Culture Collection) (39) were maintained in spinner culture in DMEM, 10% (v/v) fetal bovine serum containing penicillin (50 units/ml), streptomycin (50 units/ml), and glutamine (2 mM). The medium was replaced with fresh medium each day. Mouse peritoneal macrophages were obtained from 25-35-g female mice that had been injected intraperitoneally with 1 ml of sterile thioglycollate broth 4 days prior to cell harvesting (31); the three mouse strains used were ICR mice, LDL receptor knockout mice on the C57BL/6 background (40), and lysosomal acid lipase knockout mice on a 129CV/CF-1 mixed background (41). On the day prior to the experiments utilizing monolayers of macrophages, the macrophages were plated at ϳ80% confluence in 22-mm wells (12-well dishes) and placed in a 37°C CO 2 tissue culture incubator. On the day of the experiments involving retained and aggregated LDL, the 1.5 ϫ 10 6 macrophages were plated in 16-mm wells (24-well dishes) on top of these retained aggregates. Human peripheral blood monocytes were isolated from normal subjects as described previously (42) and grown for 48 h in 250-ml tissue culture flasks in RPMI medium containing 30% heat-inactivated pooled human serum plus penicillin, streptomycin, and glutamine. The cells were then plated in 22-mm wells as above and induced to differentiate into macrophages by the addition of 1 ng of GM-CSF/ml of medium on days 1, 4, and 11 of culture as described previously (42); by day 14, the cells were differentiated as assessed by morphological changes (e.g. increased spreading) and increased expression of scavenger receptor activity (cf. Ref. 43).
Bovine aortic endothelial cells and smooth muscle cells were obtained as described previously (44). The cells were plated in 16-mm wells (24-well dishes) in DMEM, 10% (v/v) fetal bovine serum, containing penicillin, streptomycin, and glutamine, and allowed to grow until confluent. The day prior to the experiment, the cells were washed three times with warm PBS and then incubated with 1 ml of DMEM, 0.2% (w/v) fatty acid-free BSA per well.
Lipoproteins-LDL (density, 1.020 -1.063 g/ml) was isolated from fresh human plasma by preparative ultracentrifugation as described previously (45). LDL was methylated by the procedure of Weisgraber et al. (46), acetylated as described previously by Goldstein et al. (14), and labeled with DiI by the method of Pitas et al. (47,48). Native or modified forms of LDL were labeled with [ 3 H]CE by first incorporating the label into a liposome, followed by CETP-mediated transfer to HDL 3 and then finally CETP-mediated transfer from the HDL 3 to LDL (49); the specific activity was 20 -40 cpm/ng of CE. LDL was labeled with [ 3 H]sphingo-myelin exactly as described previously (50). The lipoproteins were iodinated with 125 I as follows; a solution of lipoproteins diluted to 1-2 mg/ml in 0.3 M borate buffer, pH 9.0 was placed in IODOGEN-coated tubes (Pierce), and 0.5 mCi of Na 125 I was added. After incubation for 15 min at room temperature with gentle agitation, the solution was transferred to a tube containing 10 l of 0.1 M sodium bisulfite and then dialyzed against 150 mM NaCl containing 0.3 mM EDTA, pH 7.4. The 125 I-labeled lipoproteins, which had a specific activity of 250 -400 cpm/ng protein, were used within 3 weeks of iodination. Aggregation of native or methylated LDL was induced by vortexing for 1 min at maximum setting (27), by CuSO 4 oxidation (51), or by treatment with bacterial SMase (30). The largest aggregates were removed by centrifuging at 10,000 ϫ g for 30 s.
Experimental System Involving Retained and Aggregated LDL-For the experimental system depicted in Fig. 1A (below), monolayers of endothelial or smooth muscle cells were incubated for 6 h at 37°C with DMEM, 0.2% BSA, containing 10 g of lipoprotein lipase/ml. The cells were then rinsed with PBS and incubated for 18 h with DMEM, 0.2% BSA, containing the indicated lipoproteins. Next, the wells were rinsed five times with warm PBS containing 1 mM CaCl 2 , 0.5 mM MgCl 2 , and 0.2% BSA; the last two of these rinses lasted 15 min each and were followed by a final rinse with warm PBS. Macrophages in DMEM, 0.2% BSA containing 5 g of 58035/ml (unless indicated) were then added at a density of 1.5 ϫ 10 6 cells/16-mm well.
Protein Degradation and Lipid Hydrolysis Assays-Degradation of 125 I-lipoprotein protein (apo-B100) was determined from the 125 I cpm of trichloroacetic acid-soluble, non-chloroform-extractable material (i.e. 125 I-tyrosine) in the cell-culture medium (52). The cell monolayer was dissolved in 1 ml of 0.1 N NaOH for the determination of cell-associated 125  Fluorescence Microscopy-The experimental system was set up on poly-D-lysine-coated glass coverslip-bottom dishes (48). Fluorescence images were obtained with either a Bio-Rad MRC-600 laser scanning confocal unit (Bio-Rad Microscience, Cambridge, MA) (Fig. 3) or a LSM-510 laser scanning unit (Zeiss, Oberkochen, Germany) ( Fig. 8) on an Axiovert inverted microscope using a 63ϫ, numeric aperture 1.4 Plan-Apo infinity-corrected objective (Zeiss). For Fig. 3, the illumination sources were the 488-and 514-nm lines from a 25-milliwatt argon laser for CMFDA and DiI, respectively. For CMFDA fluorescence, a 510-nm dichroic mirror and a 515-nm long pass emission filter were used, and for DiI fluorescence, a 580-nm dichroic mirror and a 580-nm long pass emission filter were used. For Fig. 8, a 1.0-milliwatt helium/ neon laser emitting at 543 nm was used, and DiI emission was collected using a 560-nm long pass filter. The images were processed with Metamorph (Universal Imaging Co) and Photoshop (Adobe) software.
Statistics-Unless indicated otherwise, results are given as means Ϯ S.D. (n ϭ 3). Absent error bars signify S.D. values smaller than the graphics symbol.
FIG. 1. Two experimental models to study the interaction of macrophages with aggregated and "retained" LDL. In the model depicted in panel A, endothelial or smooth muscle cells are the source of matrix to which aggregated LDL is pre-bound, using lipoprotein lipase as a "bridging molecule." Subsequently, macrophages are added to the system. In the protocol shown in panel B, aggregated methylated LDL is added directly to a monolayer of macrophages. The methylation delays the degradation of the aggregated LDL, thereby mimicking the delayed catabolism of matrix-retained aggregated LDL. See "Experimental Procedures" and text for further details.

The Initial Interaction of Macrophages with Retained and Aggregated LDL Involves Prolonged Cell-surface Contact and LDL-CE Hydrolysis That Exceeds LDL Protein
Degradation-We initially set up the experimental system diagrammed in Fig. 1A to model the interaction of macrophages with subendothelial atherogenic lipoproteins, which in vivo are substantially aggregated and retained on subendothelial matrix, rather than simply monomeric and free in solution (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). In this system, [ 3 H]CE and 125 I-protein (apo-B100) double-labeled aggregated LDL was added to a monolayer of endothelial cells or smooth muscle cells, which served as the source of extracellular matrix. Because lipoprotein lipase has been implicated in the bridging of lipoproteins to matrix in the subendothelium (53), we added this molecule to the endothelial or smooth muscle cell monolayer prior to the addition of the aggregated LDL. After washing away non-bound LDL aggregates, macrophages were added to this system and studied over the first few hours. We have shown previously that, after 24 h, the added macrophages internalize the matrix-bound aggregates and accumulate very large amounts of intracellular CE droplets (31).
As shown in Fig. 2A, lipoprotein-125 I-apo-B100 degradation by J774 macrophages proceeded at a rate that was relatively slow compared with that reported previously for the degradation of monomeric lipoproteins in solution by these cells (see Ref. 48 and below). Remarkably, however, lipoprotein-[ 3 H]CE hydrolysis occurred at a greater rate and to a greater extent than 125 I-apo-B100 degradation over the time period examined. Although the absolute values varied somewhat among repeat experiments, the rate and extent of CE hydrolysis was always 3-7-fold greater than that of protein degradation. Similar results were obtained when mouse peritoneal macrophages were used instead of J774 macrophages (Fig. 2B) and when smooth muscle cells were used as the source of matrix instead of endothelial cells (Fig. 2C).
The morphology of this interaction was investigated by confocal fluorescence microscopy. CMFDA-labeled macrophages were incubated with matrix-bound, DiI-labeled aggregated LDL. As shown in Fig. 3 (a, c, and e), the macrophages (green fluorescence) were in intimate contact with the retained and aggregated LDL (red fluorescence). The images in panels b, d, and f were acquired after the addition of TNBS to the cells shown in panels a, c, and e, respectively; TNBS is a cellimpermeant quencher of DiI fluorescence (54,55). The finding that most of the DiI fluorescence disappeared rapidly with the addition of TNBS indicates that the retained and aggregated LDL was not fully internalized but rather nestled in deep invaginations of the cell surface of the macrophages. Importantly, these morphological events occurred at the same time points at which lipoprotein-CE hydrolysis exceeded protein degradation (see Fig. 2).
The Initial Interaction of Macrophages with Aggregated Methylated LDL-Shown in Fig. 1B is a second and somewhat simpler experimental system to study the events described above. In this system, aggregated double-labeled methylated LDL was added directly to macrophages in the absence of endothelial cells or lipase; after a short incubation, unbound lipoprotein was removed, and the cellular catabolism of the lipoprotein [ 3 H]CE and 125 I-apo-B100 was assayed during the ensuing chase period. Methylation of apo-B100 abolishes its recognition by the LDL receptor (46) and has been shown by Kruth and colleagues to retard the degradation of aggregated LDL by macrophages (56). Thus, we used methylation to mimic the relatively slow degradation of LDL protein observed with matrix-retained LDL. Indeed, when the degradation of lipoprotein-125 I-apo-B100 in the two systems was directly compared, retained and aggregated LDL (i.e. the first system) and aggregated methylated LDL were degraded slowly and similarly, while aggregated non-methylated LDL was degraded much more rapidly and extensively (Fig. 4).
Using double-labeled aggregated methylated LDL, we compared [ 3 H]CE hydrolysis with 125 I-apo-B100 degradation using three different types of macrophages-J774, mouse peritoneal, and human monocyte-derived (Fig. 5, A-C). In each case, the rate and extent of CE hydrolysis was much greater than the rate and extent of apo-B100 degradation, similar to what was found with retained and aggregated LDL (above). Interestingly, however, another lipid component of LDL, sphingomyelin, was not catabolized differently from LDL protein (Fig. 5D).
The 125 I-protein degradation assay essentially measures the amount of 125 I-tyrosine that is excreted into the cell culture medium (52). If there were a substantial degree of incomplete degradation of 125 I-apo-B100, however, our assay would underestimate the extent of its degradation. To test this point, the cells were harvested at various times during the chase period and subjected to SDS-polyacrylamide gel chromatography followed by autoradiography. As shown in Fig. 6 (B-E), there was little evidence of 125 I-labeled degradation products at these chase times, and the intensity of the higher-molecular weight bands changed little over the chase period. In fact, the overall pattern looked similar to that of cells treated with chloroquine during the 20-min pulse period (Fig. 6A), which blocks the degradation of 125 I-labeled aggregated methylated LDL (see below), and to that of the original aggregated methylated 125 Iapo-B100 (i.e. not incubated with cells; data not shown). In contrast, the same method is clearly able to demonstrate the rapid degradation of 125 I-labeled native LDL by macrophages (Fig. 6, F-J). Thus, in the experimental system described above, apo-B100 degradation truly occurs at a slow rate and to a small extent.
To determine whether the differential catabolism of CE (and FC) versus protein was a peculiar property of LDL aggregated by vortexing, we studied the interaction of macrophages with LDL aggregated by completely different and more physiological means, namely oxidation and sphingomyelinase treatment (25,30,57,58). As shown in Fig. 7A, aggregated methylated LDL-CE was degraded more than LDL protein when aggregation was induced by either oxidation or by sphingomyelinase treatment. Similar results were obtained using matrix-retained and aggregated LDL (Fig. 7B).
An important property of the matrix-retained and aggregated LDL system was the slow internalization of aggregated  Fig. 2A, except the aggregated LDL was labeled with DiI and the macrophages were labeled with CMFDA. The time of the macrophage incubation was 46 min. Panels a, c, and e represent three separated images of macrophages (hollow arrow) in contact with retained and aggregated LDL (arrowhead). The same three fields are shown in panels b, d, and f, respectively, except that TNBS, a cellimpermeant quencher of DiI fluorescence, was added. The solid white arrows in these panels depict cellular invaginations occupied by extracellular (i.e. not internalized) aggregated LDL. Bar, 10 m. LDL (Fig. 3). In the aggregated/methylated LDL system, we have thus far shown slow LDL protein degradation, but it is possible that this could occur after more rapid internalization (i.e. retarded intracellular degradation). To examine this important issue, DiI-labeled methylated LDL was aggregated by either vortexing (Fig. 8, A-C) or by SMase treatment (Fig. 8, D-F) and incubated with macrophages for 20 min. The cells were then washed and incubated in medium without lipoproteins for an additional 30 min. The cells were viewed directly (Fig. 8, A and D) and, after acquiring the image, they were treated with 250 g/ml trypsin for 30 min at 37°C. After trypsin treatment, the same field of cells was visualized (Fig. 8,  B and E), although occasionally the orientation of the cells became altered (see Fig. 8E). The images clearly show that trypsin released a portion of the cell-associated aggregates, but not all. The remaining material could be inside the cell or on the cell surface but inaccessible to trypsin, e.g. due to sequestration in protected cell-surface invaginations (cf. Refs. 34 and 48). To address this point, the trypsin-treated cells were treated with cell-impermeant DiI quencher, TNBS (see Fig. 3, above). As shown in Fig. 8 (C and F), most of the trypsinresistant material was rapidly quenchable by TNBS, indicating that it was extracellular; note that a small portion of the LDL was internalized by the cell shown in Fig. 8C (arrows). Thus, similar to what was observed in the matrix-retained and aggregated LDL system (Fig. 3), aggregated methylated LDL is slowly internalized by macrophages.
Evidence That a Cell-surface Protein Other than Scavenger Receptors or the LDL Receptor Mediates the Interaction of Macrophages with Retained and Aggregated LDL-To determine if one or more macrophage cell-surface proteins were necessary for the catabolism of retained and aggregated LDL, macrophages were preincubated in the absence or presence of trypsin plus the protein synthesis inhibitor cycloheximide and then washed and incubated with soybean trypsin inhibitor. The control or trypsinized macrophages were then plated on top of 125 I-protein-and [ 3 H]CE-labeled aggregated LDL that was retained on endothelial-derived matrix. As shown in Fig. 9A, the trypsin-treated macrophages degraded substantially less 125 I-apo-B100 and lipoprotein-[ 3 H]CE compared with control macrophages. A possible candidate for a receptor involved in this interaction is SR-BI, a class B scavenger that mediates the selective uptake of CE from HDL in several cell types (59). To investigate this possibility, we determined whether a very high concentration of unlabeled oxidized LDL, a ligand for SR-BI and other class B and A scavenger receptors and a competitive inhibitor for SR-BI-mediated selective uptake (60), could compete for the interaction of macrophages with 125 I-protein-and [ 3 H]CE-labeled retained and aggregated LDL. As shown in Fig.  9A, exposure of macrophages to a vast excess of unlabeled oxidized LDL both prior to and during the incubation with labeled retained and aggregated LDL had no significant effect on apo-B100 and lipoprotein-CE degradation.
LDL methylation experiments indicate that LDL receptors play a role in the degradation of non-retained aggregated LDL particles by macrophages (cf. Ref. 56 and Fig. 4) but not in the uptake of CE from these particles (Fig. 5). To determine the role of LDL receptors in the initial interaction of macrophages with retained and aggregated LDL, peritoneal macrophages from LDL receptor-null mice and from gender-and agematched wild-type mice of the same genetic background were plated on top of 125 I-protein-and [ 3 H]CE-labeled aggregated LDL retained on endothelial-derived matrix. The LDL receptor-null macrophages showed the same degree of 125 I-apo-B100 degradation and [ 3 H]CE hydrolysis as the wild-type macrophages (Fig. 9B). In summary, the data in Fig. 9 suggest that a cell-surface protein other than scavenger receptors or the LDL receptor mediates the interaction of macrophages with retained and aggregated LDL.
Evidence That Lysosomal Acid Lipase Hydrolyzes the CE from Retained or Methylated Aggregated LDL-The CE of lipoproteins internalized by receptor-mediated endocytosis are hydrolyzed by lysosomal acid lipase (LAL) (32), whereas CE hydrolysis resulting from SR-B1-mediated HDL-CE selective uptake occurs normally in fibroblasts from patients lacking LAL (61). To address this central issue in our systems, two experiments were conducted. The data in Fig. 10A show that the hydrolysis of CE derived from aggregated methylated LDL was inhibited more than 3-fold by chloroquine. While these data are consistent with lysosomal hydrolysis of the CE, chloroquine can inhibit non-lysosomal trafficking pathways, including those involved specifically in the classic HDL-CE selective uptake pathway (61,62). Therefore, we examined the fate of the CE in peritoneal macrophages from LAL null mice (41). The data in Fig. 10B show that CE hydrolysis, but not protein hydrolysis, was completely blocked when these LALnegative macrophages were incubated with retained and aggregated LDL. These data definitively prove that the CE derived from aggregated LDL is hydrolyzed by lysosomal acid lipase.

Low Potassium Medium Blocks Protein Degradation but Not CE Hydrolysis during the Interaction of Macrophages with
Aggregated Methylated LDL-LAL hydrolyzes both the CE of retained or methylated aggregated LDL and the CE of monomeric lipoproteins internalized by receptor-mediated endocytosis, yet receptor-mediated endocytosis leads to nearly equivalent degradation of the protein and CE moieties of lipoproteins (see above and Ref. 32). To further distinguish the cellular pathway leading to CE hydrolysis described in this report from that occurring during receptor-mediated endocytosis of lipoproteins, we utilized the ability of potassium depletion to partially block endocytic processes (63,64). As an example of receptormediated endocytosis, we incubated macrophages with 125 I/ [ 3 H]CE-labeled acetyl-LDL (Table I)

DISCUSSION
In vivo, macrophages in atherosclerotic lesions encounter lipoproteins that are bound to subendothelial matrix components, and these lipoproteins are often aggregated and fused (15,16,18,19,(21)(22)(23)(24)(25)65). We reasoned that certain unique cellular events occurring during this interaction might be missed by using the usual method of studying the interaction of macrophages with lipoproteins, namely incubation of macrophages with monomeric lipoproteins dissolved in tissue culture medium. This latter method focuses on receptor-mediated endocytosis, which is characterized by relatively rapid internalization of lipoproteins followed by the nearly simultaneous lysosomal degradation of both the protein and CE moieties (32,48). In contrast, we have shown herein, using specialized cellculture systems, that the uptake and of matrix-retained or methylated aggregated lipoproteins is markedly delayed, even when compared with aggregated lipoproteins that are not retained or methylated (Fig. 4). Furthermore, during the early stages of this interaction, there is a dissociation between protein degradation and CE hydrolysis. Finally, CE hydrolysis is not blocked by potassium depletion (Table I), which further distinguishes this interaction from receptor-mediated endocytosis.
Two important issues raised by this study are the mechanisms involved in the events described above and the physiological significance in terms of arterial wall macrophage biology. The mechanistic issues can be divided into cell-surface events and internalization processes. The trypsin data in Fig.  9A indicate the involvement of one or more cell-surface proteins, but we have not yet identified these molecules. Interestingly, the LDL receptor is clearly not involved in the differential hydrolysis of protein and CE (Fig. 9B), and the lack of  and D), the dishes were left in place on the heated microscope stage and washed and incubated for 30 min at 37°C with 250 g of trypsin/ml PBS. After acquiring this post-trypsin image of the same respective cells (B and E), the macrophages was exposed to TNBS-quenching (C and F) (see Fig. 3). Note that the orientation of the cell shown in panel D changed after trypsin treatment. In these images, which are projections of Z series, the DiI fluorescence is orange and the rest of the field, which was visualized by Nomarski differential interference contrast (DIC) microscopy, is shown as green (Metamorph conversion). Bar, 2 m. competition by a vast excess of oxidized LDL (Fig. 9A) suggests that class A and B scavenger receptors are also not critical (cf. Ref. 60). 2 It is important to note, however, that if the retained and aggregated LDL is also oxidized (see Fig. 8), scavenger receptors or other oxidized LDL receptors may become important (cf. Ref. 30).
Another aspect related to cell-surface events is the location of the retained and aggregated LDL in deep invaginations in the macrophage cell surface (Fig. 3). This finding is reminiscent of the structures, called STEMs (surface tubules for entry into macrophages), involved in the interaction of macrophages with ␤-VLDL (34); of the cell-surface tubules described by Kruth et al. (56) during the interaction of macrophages with vortexedaggregated LDL; and possibly of the microvilli that appear to be involved in selective lipoprotein-CE uptake by steroidogenic cells (66). Of note, ␤-VLDL in STEMs was inaccessible to antibodies (34) and to suramin (48), and aggregated methylated LDL on the surface of macrophages was only partially released by trypsin treatment (Fig. 8). These findings suggest that the lipoprotein-containing cell-surface invaginations are relatively "protected," i.e. accessible to small molecules like TNBS but not to larger molecules. This model would support our speculation that the prolonged residence of the aggregated lipoproteins in the macrophage invaginations provides the proper milieu for the metabolic events that follow.
The next stage is characterized by CE hydrolysis that exceeds protein degradation. We have not yet established whether CE hydrolysis occurs extracellularly or intracellularly. 3 On one hand, we have shown definitively that LAL hydrolyzes the CE of retained and aggregated lipoproteins, and we have also determined that there is no detectable CE hydrolase activity in the conditioned medium of J774 macrophages, even when concentrated and assayed with [ 3 H]CE-LDL substrate at acid pH. 4 On the other hand, we have shown previously using fluorescence resonance energy transfer experiments that ␤-VLDL in the cell-surface invaginations of macrophages undergoes a certain degree of disruption in situ (34). Thus, despite the unpublished data mentioned above, it is theoretically possible that LAL-mediated CE hydrolysis occurs in cell-surface invaginations via secretion of the enzyme into these areas. Clearly, further experimentation will be needed to resolve this important issue.
If CE is first internalized and then hydrolyzed, there are at least two mechanisms to explain how CE internalization could exceed protein internalization. In one scenario, aggregation and fusion processes, including those thought to be physiologic like SM hydrolysis and oxidation (25,30,57,58) (see Fig. 7), might lead to the formation of CE-rich particles that are preferentially taken up by macrophages. Such CE-rich particles would have had to have been made to a similar extent during three very different methods of aggregation (Fig. 7), and the particles would have to have the curious property of excluding LDL-sphingomyelin (see Fig. 5D). Furthermore, we prepared aggregated LDL that had its CE labeled with Bodipy and its protein labeled with Cy5. This double-labeled LDL was aggregated by either vortexing or by oxidation, and was then added to endothelial cell-derived matrix that had been preincubated with LpL. The ratio of Bodipy-CE to Cy5-protein of the retained and aggregated LDL was found to be quite uniform, with less than 2% of the labeled material showing a high Bodipy:Cy5 ratio. Nonetheless, it is possible that this technique would not be able to discern a subpopulation of particles with a relatively modest enrichment of CE, and so this possibility must still be formally considered. In this regard, it is well documented that CE-rich particles exist in atherosclerotic lesions and can lead to CE accumulation in macrophages (67)(68)(69).
The other possible explanation is selective CE uptake, i.e. "extraction" of CE either from extracellular lipoproteins (above) or from lipoproteins recycling through endocytic compartments. Although the classic selective uptake pathway involves HDL interacting with SR-B1 (59), other studies have shown that cells can selectively internalize LDL-CE (49,62). In addition, as mentioned above, the microvilli thought to be involved in selective uptake by steroidogenic cells (66) have similarities with the cell-surface invaginations described here. Because selective uptake involves CE but not phospholipids, 5 our finding that CE hydrolysis but not SM hydrolysis exceeds protein degradation (Fig. 5) is consistent with a selective uptake process. One aspect of our system that is different from HDL-CE selective uptake by fibroblasts is the involvement of lysosomal acid lipase in lipoprotein-CE hydrolysis (Fig. 10B and Ref. 61). If indeed selective uptake is the mechanism responsible for our 2 We would have liked to have supported the conclusion drawn from the data in Fig. 9A by using macrophages from SR-BI-deficient mice (79). However, neither these mice nor their peritoneal macrophages were available for our use during the course of this study. 3 Prior to our finding out that protease resistance was not a reliable assay for internalization (Fig. 8), we conducted a pulse-chase experiment in which macrophages were incubated with aggregated methylated LDL doubly labeled with [ 3 H]cholesteryl ether and 125 I-protein to determine if the nonhydrolyzable cholesteryl ether was internalized to a greater extent than LDL protein. Indeed, we found that trypsinresistant [ 3 H]cholesteryl ether was 2-fold greater than trypsin-resistant 125 I-protein at chase times of 25 and 95 min. Given the data in Fig. 8, however, we feel that it is difficult to conclude definitively from this experiment that aggregated LDL-CE is internalized at a faster rate than LDL protein. 4 G. Kuriakose and I. Tabas, unpublished data. findings, the difference may be related to the nature of the lipoproteins (e.g. their large size), the cell type, and/or the apparent lack of involvement of SR-B1 (above). The resistance of CE hydrolysis to potassium depletion in our system (Table I) deserves comment. In general, potassium depletion preferentially inhibits clathrin-mediated endocytosis (63). While several investigators have found that this treatment does not inhibit non-coated vesicular uptake, such as occurs during fluid-phase endocytosis or internalization of ␤-adrenergic receptors (70, 71), Carpentier et al. (64) found that cholera toxin and horseradish peroxidase internalization, which occur via non-coated invaginations, was inhibited by potassium depletion. Of potential relevance to this report, Koval et al. (72) reported that phagocytosis of large (i.e. 2-3m) IgG-opsonized polystyrene beads was relatively resistance to potassium depletion. The effect of potassium depletion on SR-B1-mediated selective uptake of HDL-CE has not been reported.
How might the initial events described in this report relate to the biology of the arterial wall macrophage? In terms of CE loading, we envision a two-phase process. In the first few hours, ϳ20% of the matrix-retained and aggregated LDL-cholesterol provided to the macrophages appears to originate from the unique events described herein (Fig. 2); in the aggregated methylated LDL system, the percentage of cholesterol delivered during this phase is larger (Fig. 5 versus Fig. 2). Subsequently, large pieces of the aggregates are internalized, 6 probably by a process that resembles phagocytosis (31). In keeping with our focus on the initial events occurring during the interaction of macrophages with retained or methylated aggregated LDL, we have not yet determined the metabolic fate of the cholesterol derived from the first phase (e.g. incorporation into cellular membranes, efflux, or esterification). In this regard, Stangl et al. (62) have shown that LDL stimulates both cholesterol esterification and cholesterol efflux when incubated with LDL receptor-negative Chinese hamster ovary cells with supraphysiologic amounts of SR-B1. Moreover, we have previously demonstrated that cholesterol esterification is markedly activated in macrophages plated on retained and aggregated LDL for 24 h. Further studies will be needed, however, to determine if the FC delivered from the early pathway contributes to the subsequent stimulation of cholesterol esterification. Even if the initial phase does not directly affect cholesterol esterification, it may influence subsequent metabolic events by modifying the composition of the extracellularly retained and aggregated particles, for example by depletion of CE relative to protein and phospholipid.
Finally, the events described herein may have other effects on macrophage biology that could be relevant to atherogenesis. For example, the interaction of macrophages with retained and aggregated LDL may represent a modified form of "frustrated phagocytosis," which refers to a process whereby phagocytic cells interact tightly with a surface that cannot be engulfed and internalized (33). In this case of the process described here, the engaged material is eventually phagocytosed (31), but only after an initial period of non-internalization (Figs. 3 and 8). Frustrated phagocytosis per se is associated with a variety of cellular events, including release of lysosomal enzymes, reactive oxygen species, and proteoglycans (73)(74)(75)(76); redistribution of clathrin and reorganization of the Golgi (77,78); and changes in cytosolic free calcium (33). Therefore, it will be interesting to determine if similar events occur during the interaction of macrophages with retained and aggregated LDL.