Binding of low density lipoproteins to lipoprotein lipase is dependent on lipids but not on apolipoprotein B.

Lipoprotein lipase (LPL) efficiently mediates the binding of lipoprotein particles to lipoprotein receptors and to proteoglycans at cell surfaces and in the extracellular matrix. It has been proposed that LPL increases the retention of atherogenic lipoproteins in the vessel wall and mediates the uptake of lipoproteins in cells, thereby promoting lipid accumulation and plaque formation. We investigated the interaction between LPL and low density lipoproteins (LDLs) with special reference to the protein-protein interaction between LPL and apolipoprotein B (apoB). Chemical modification of lysines and arginines in apoB or mutation of its main proteoglycan binding site did not abolish the interaction of LDL with LPL as shown by surface plasmon resonance (SPR) and by experiments with THP-I macrophages. Recombinant LDL with either apoB100 or apoB48 bound with similar affinity. In contrast, partial delipidation of LDL markedly decreased binding to LPL. In cell culture experiments, phosphatidylcholine-containing liposomes competed efficiently with LDL for binding to LPL. Each LDL particle bound several (up to 15) LPL dimers as determined by SPR and by experiments with THP-I macrophages. A recombinant NH(2)-terminal fragment of apoB (apoB17) bound with low affinity to LPL as shown by SPR, but this interaction was completely abolished by partial delipidation of apoB17. We conclude that the LPL-apoB interaction is not significant in bridging LDL to cell surfaces and matrix components; the main interaction is between LPL and the LDL lipids.

The functional location of LPL is at the vascular side of endothelial cells where it is anchored through electrostatic interaction with heparan sulfate proteoglycans. LPL activity is regulated according to the nutritional state, in a tissue-specific manner, based on the needs of the tissue for fatty acids.
In addition to its lipolytic activity, LPL acts as a potent bridge between lipoproteins and cell-surface proteoglycans and lipoprotein receptors or components of the extracellular matrix (1,(3)(4)(5)(6). The presence of LPL increases by severalfold the binding of lipoproteins to cells in culture, to perfused livers, and to matrix components deposited on culture dishes by fibroblasts. This bridging ability of LPL is independent of its catalytic activity (7).
LPL is produced by macrophages and smooth muscle cells and has been found in the arterial wall in connection with atherosclerotic lesions (8 -10). Recently it has been suggested that LPL contributes to the retention of low density lipoprotein (LDL) in the vessel wall by bridging LDLs to either macrophages or the extracellular matrix (1,3,5,11,12). Specific biological responses to the retained lipoproteins lead to biochemical and cellular events that promote atherogenesis (13,14). In vitro, LDL and other lipoproteins containing apolipoprotein (apo)B bind weakly to heparin and other proteoglycans (15). It is therefore possible that molecules such as LPL, which enhances the interaction of LDL with proteoglycans, could play an important role in atherogenesis in vivo. Goldberg et al. proposed that protein-protein interaction between LPL and apoB, the sole protein moiety of LDL, is more important than the interaction between LPL and LDL lipids (16 -18). However, LPL interacts efficiently with liposomes and lipid emulsions that do not contain any apolipoprotein (19,20). We therefore investigated whether the interaction between LPL and LDL is mediated primarily by the LDL protein (apoB) or by LDL lipids. For these studies, we produced modified and recombinant LDL and a recombinant fragment of apoB (apoB17) for experiments with surface plasmon resonance (SPR) (21)(22)(23) and cultured THP-I monocyte-derived macrophages.
Human ApoB Transgenic Mice-Transgenic mice were generated with a P1 bacteriophage clone (27) that spanned the human apoB gene in which mutations had been introduced by RecA-assisted restriction endonuclease cleavage (28,29). Mice were housed in a pathogen-free barrier facility operating on a 12-h light/12-h dark cycle and were fed rodent chow containing 4.5% fat.
Chemical Modification of ApoB-To selectively modify arginines or lysines in apoB100, recombinant LDLs were incubated with acetic anhydride or cyclohexanedione, respectively, as described by Innerarity et al. (30).
Radiolabeling of Lipoproteins for Cell Experiments-LDLs were iodinated by the iodine monochloride method as modified by Helmkamp et al. (31). Iodinated LDLs were dialyzed at 4°C against Dulbecco's phosphate-buffered saline with 0.01% (w/v) EDTA and 0.02% (w/v) NaN 3 and filtered through 0.22-m Millipore filters (MILLEX-GS). Aliquots of LDL were delipidated with 20 volumes of methanol:diethyl ether (1:1, v/v), and the radioactivity in lipid and protein moieties was determined (22). More than 98% of 125 I radioactivity in the LDL preparations was in the protein moiety. Iodination did not change the electrophoretic mobility of LDL compared with unlabeled LDL. 125 I-LDL was used for the experiments within a few days after preparation.
Liposomes-Liposomes composed of free cholesterol/phosphatidylcholine (FC/PC), PC only, or lyso-PC/PC were prepared by probe sonication. Chloroform solutions of FC and PC and a solution of lyso-PC (10 mg/ml) in chloroform:methanol (1:1) were mixed in 1:1 molar ratios and evaporated under a stream of nitrogen. After addition of 5 mM Tris-HCl, 0.15 M NaCl, 0.02% (w/v) NaN 3 , pH 7.4, the mixture was sonicated for repetitive cycles of 15 s with a 15-s break for 30 min at 4°C with an MSE Soniprep 150 equipped with a 10-mm probe. After equilibration at 37°C for 2 h, the liposomes were separated from undispersed lipid and multilamellar vesicles by low speed centrifugation (3000 rpm, 10 min). The liposomes were dialyzed overnight at 4°C against the same buffer used for the dispersion and were used immediately. Cholesterol was determined with a CHOD-PAP kit (Axiom, Bü rstadt, Germany), and phospholipids were determined with a phospholipids B kit (Wako Chemicals, Neuss, Germany).
ApoB17 Fragment-Recombinant baculovirus encoding human apoB17 (a kind gift from Dr. Alan Attie, University of Wisconsin) was produced as described by Gretch et al. (33). ApoB17 was purified from the culture medium by adsorption to an immunosorbent column prepared with rabbit polyclonal antibodies to human apoB (DAKO) coupled to CNBr-activated Sepharose 6B (Amersham Pharmacia Biotech). ApoB17 was eluted with 50 mM diethylamine, pH 12.0, neutralized with 1 M NaH 2 PO 4 , dialyzed extensively for 4 h against 0.01 M Dulbecco's phosphate-buffered saline with 0.01% EDTA, 0.02% NaN 3 , and concentrated by centrifugation through Microsep 30K (Kendro Laboratory Products, Newtown, CT). For delipidation, immunoaffinity-purified apoB17 (2 ml) was mixed with 4 ml of buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 3% sodium deoxycholate) and protease inhibitors (aprotinin (final concentration, 100 kallikrein inhibitor units/ ml), leupeptin (0.1 mM), and phenylmethylsulfonyl fluoride (1 mM)). After sonication with six 10-s pulses and one 15-s pulse, 3.0 ml of the solution was overlaid onto a 15-55% sucrose gradient in each of six SW40 ultracentrifuge tubes and spun at 35,000 rpm for 65 h at 10°C. Fractions (0.5 ml each) were unloaded from the gradient and analyzed by electrophoresis on a 10% polyacrylamide-sodium dodecylsulfate gel. Fractions containing apoB17 were pooled and used in binding experiments after dialysis. The delipidated apoB17 still contained some polar lipids, but the amount was reduced by ϳ50% as qualitatively analyzed by thin-layer chromatography. After complete delipidation with methanol:diethyl ether (1:1), apoB17 was not soluble in the buffers used for the experiments and could therefore not be studied.

LPL-mediated Binding and Uptake of LDL by THP-I Macrophages-
For the LDL binding studies, cells were cooled at 4°C for 30 min, washed twice with ice-cold medium (RPMI 1640) containing 0.2% BSA (RPMI-0.2% BSA), and incubated for 2 h at 4°C in RPMI 1640 containing 5 g/ml 125 I-LDL, 10 g/ml bovine LPL, and 4% BSA. Control wells did not contain macrophages. After incubation, the cells were washed twice with ice-cold RPMI-0.2% BSA and incubated with 100 IU of heparin/ml of RPMI-0.2% BSA for 30 min at 4°C. Heparin-releasable radioactivity was considered to represent LDL bound to the cell surface. The macrophages were washed twice with Dulbecco's phosphate-buffered saline and dissolved in 0.2 M NaOH for measurement of the 125 I-LDL remaining associated with cells (heparin-resistant LDL) and cellular protein content (22). The radioactivity in control wells was subtracted from the radioactivity of the cell-containing wells. For studies of LDL uptake, THP-I macrophages were incubated for 2 h at 37°C with the same amount of 125 I-LDL and LPL as in the binding studies. The cells were not cooled before the experiment, and all washings were performed with warm solutions.
Binding Studies by SPR-The binding studies were performed on a BIAcore 2000 instrument. Biotinylation and analysis of kinetics were performed as described previously (21,22,34). Briefly, avidin was covalently coupled to the dextran matrix of CM5 sensor chips. Biotinylated heparin or biotinylated LPL was bound to the avidin. LDL samples were injected into the sensor chips, and binding of LDL was recorded by monitoring changes in the refractive index close to the dextran layer. These changes are expressed in arbitrary response units (RU). The binding experiments were carried out in 10 mM Hepes, 0.15 M NaCl, pH 7.4, at 25°C.
Radiolabeling of LDL and LPL for Calibration of Binding on BIAcore Sensor Chips-Human LDLs (d ϭ 1.03-1.04 g/ml) were radiolabeled with 125 I by using IODO-GEN (Pierce). More than 90% of the radioactivity was precipitated in 10% trichloroacetic acid. The iodinated LDLs were kept at 4°C and used within 2 weeks. LPL was radiolabeled by the lactoperoxidase method and then repurified by chromatography on heparin-Sepharose (35). The protein concentrations of LDL and LPL were determined with a bicinchoninic acid protein assay (BCA, Pierce). The specific radioactivity of both preparations was about 500 cpm/ng. The molar concentration of LDL was calculated assuming a protein content of 21% and an average molecular mass of 2.3 ϫ 10 6 kDa. For the LPL dimer, a molecular mass of 110 kDa was used.
Determination of Mass/Response Relationship for LDL and LPL-Iodinated LDL and LPL were immobilized on the dextran matrix of sensor chips with the N-ethyl-NЈ-[(diethylamino)propyl]carbodiimide/ N-hydroxysuccinimide method (36). Differences in pre-and postimmobilization response levels and the bound radioactivity were measured, and bound radioactivity on the chip was measured in a ␥ counter. The mass of bound LPL and LDL was calculated from their respective specific radioactivities.

RESULTS
Binding of LDL to Heparin and LPL-To investigate, in real time, how LPL interacts with LDL we used SPR. The use of this method to study interactions of LPL and heparan/heparan sulfate proteoglycans with lipoproteins was described previ-ously in detail (21,22). LDLs were injected into sensor chip flowcells coated with heparin, LPL, or LPL-heparin. The response values (RU), corresponding to the amount of bound LDL, were then registered at steady state and used to calculate the relative affinities (21).
To assess the role of basic residues of apoB100 in the LPL-LDL interaction, we analyzed three groups of LDL by SPR. The first group consisted of human LDL in which arginines were selectively modified with acetic anhydride (Ac-LDL) and lysines with cyclohexanedione (CHD-LDL). The second group consisted of a recombinant human LDL (isolated from transgenic mice) in which arginines and lysines in the principal proteoglycan binding site (residues 3359 -3369) of apoB100 were mutated to serines and alanines, respectively (Mut-LDL) (29), to abolish the interaction between LDL and the arterial proteoglycans versican and biglycan (37). The third group consisted of human plasma LDL and recombinant wild-type LDL (controls).
Ac-LDL, CHD-LDL, and Mut-LDL did not interact with heparin alone at measurable affinity under the conditions of the experiments (Table I). Wild-type recombinant LDL bound with somewhat lower affinity than LDL purified from human plasma, consistent with earlier studies of the interaction between LDL and arterial proteoglycans (37).
The presence of LPL on heparin or proteoglycans markedly increases the affinity of LDL (6,21,22,38). Therefore, we compared the affinities of the LDL variants for LPL and LPLheparin. LPL increased the binding of all LDL variants to heparin, regardless of their own affinities for heparin (Table I). We then studied the importance of the carboxyl terminus of apoB100 in the LPL-LDL interaction. Human recombinant apoB100-containing LDL and apoB48-containing LDL had similar binding affinities for LPL (Table I). These results show that the interaction between LPL and LDL was not dependent on basic residues in apoB100 or on any specific binding site in the carboxyl-terminal half of apoB100.
Next we studied the effects of partial delipidation of LDL on the LPL-LDL interaction. Significant amounts of lipid were removed from human LDL with diisopropyl ether:butanol (48% of total cholesterol and 51% of triglycerides). LDLs remain soluble after removal of even more lipids without altering the LDL receptor binding activity of apoB100 (30). Size exclusion chromatography analysis of the partially delipidated LDL showed that the delipidation did not result in any oligomer formation (data not shown). The partial delipidation of LDL greatly increased their affinity for heparin but significantly lowered their affinity for LPL (Table I).
Binding and Uptake of LDL by THP-I Macrophages-To investigate the interaction of LDL with heparan sulfate proteoglycans and LPL in a more physiological setting, we performed experiments with modified and recombinant LDL in cultures of THP-I macrophages (Table II). With all types of LDL, both binding (heparin-releasable radioactivity at 4°C) and uptake (heparin-resistant association with the cells at 37°C) were low in the absence of added LPL. In this system, modified LDL bound with somewhat higher affinity than native LDL probably because of the presence of scavenger receptors (39), but the difference was not statistically significant. However, addition of LPL (10 g/ml) dramatically increased both the binding and uptake of native LDL as expected from previous studies (22,40). The binding of modified LDL also increased manyfold, indicating that LPL was the main determinant for their interaction with the cells under the conditions used. In experiments performed at 37°C, most of the LDLderived radioactivity associated with cells was heparin-resistant and probably corresponded to LDL that had been taken up by the cells. In agreement with the results of SPR studies, recombinant LDL with apoB48 and recombinant LDL with apoB100 bound to cells with similar affinity (Table III). We conclude that LDL with decreased or abolished ability to interact with proteoglycans and certain lipoprotein receptors can still to a large extent be bound to cells if LPL is present.
Competition between LDL and Liposomes for LPL-mediated Binding to Cells-To investigate the role of the LDL lipids in the interaction with LPL, macrophages were incubated with LDL in the presence of PC liposomes (Fig. 1). LPL-mediated binding of LDL to the cells was markedly reduced, demonstrating that liposomes competed efficiently with LDL for interaction with LPL. Similar results were obtained with liposomes made of PC only, a mixture of PC and lyso-PC, or PC and FC.
Stoichiometry of the LPL-LDL Interaction as Studied by SPR-LDLs contain only one copy of apoB100 per particle (41). If there were only one specific interaction site in apoB100 for LPL, a 1:1 molar ratio would be expected for the interaction between LDL and LPL. To estimate how many LPL molecules bind per LDL particle, we first determined the mass/response correlation for LPL and LDL on the BIAcore sensor chip (Fig.  2B). The correlation was linear, but the slope was 1.4 times lower for LDL than for lipid-free proteins like LPL and other previously studied proteins (42). Hence, an LDL surface density of 1 ng/mm 2 increased the response level by 715 RU (Fig.  2B). Using this value, we determined the stoichiometry of the LPL-LDL interaction by passing three different concentrations of LPL over sensor chips on which LDL had been immobilized; the response at steady state was then used to calculate the amount of bound LPL ( Fig. 2A). Saturation occurred at around 14 LPL molecules/LDL particle. At LPL concentrations Ͼ100 nM, high nonspecific binding of LPL to the dextran matrix (20 -40% of the total binding) might have affected the stoichiometric determination. At LPL concentrations Ͻ50 nM (about 5 g/ml), the association kinetics could be fitted to a single exponential function, suggesting that up to three to four LPL dimers can interact independently with LDL. The interaction was characterized by a very high association rate constant (k a ϭ 2.1 ϫ 10 6 M Ϫ1 s Ϫ1 ) and a low dissociation rate constant (k d ϭ 10 Ϫ3 -10 Ϫ4 s Ϫ1 ). Therefore, several LPL molecules bound to each LDL particle. Simultaneous interaction with many LPL molecules may explain the strong effect of LPL on binding of LDL to heparin-covered surfaces and to cells.s Stoichiometry of the LPL-LDL Interaction as Studied with THP-I Macrophages-Cells were incubated at 4°C with in- creasing concentrations of 125 I-LPL in the medium in the presence of different amounts of LPL. Binding of LDL increased when the cells were incubated with up to 10 g of LPL/ml of culture medium. At higher concentrations of LPL (20 or 50 g of LPL/ml) binding of LDL did not increase further, indicating that binding of LPL to the cells was close to saturation at 10 g of LPL/ml (Fig. 3A). Therefore, we used unlabeled LPL or 125 I-LPL in a concentration of 10 g/ml of culture medium and unlabeled LDL or 125 I-LDL at 5 g/ml of medium for the study of the stoichiometry of the LPL-LDL interaction. In the presence of LPL, 1.25 g of LDL bound per mg of cell protein (Fig. 3B). In the presence of LDL, 3.8 g of LPL bound per mg of cell protein.
This was constantly more than was bound in the absence of LDL (2.7 g/mg of cell protein). The calculated molar ratio of LPL/LDL bound to the cells was around 15 in this experiment.
Binding of ApoB17 to LPL-To directly study the interaction between LPL and apoB, we used recombinant apoB17, which contains the first 771 amino acids of apoB100 and has a molecular mass of ϳ87 kDa (33). SPR studies showed that the affinity of apoB17 for LPL was very weak (1/1000) compared with the affinity of LDL for LPL (Table IV). However, even after purification on an anti-apoB affinity column, apoB17 con-

TABLE II
Binding and uptake of human LDL by THP-I macrophages in the presence of LPL Cells were incubated for 2 h at 4°C or at 37°C in medium containing human 125 I-LDL (5 g/ml) with or without bovine LPL (10 g/ml). Heparin-releasable (at 4°C) and heparin-resistant (at 37°C) radioactivities were analyzed at the end of the incubation period as described under "Experimental Procedures." Data are expressed as nanograms of LDL protein/milligram of cell protein and represent mean values of triplicate determinations (ϮS.D.). Similar effects were obtained with lower concentrations of LPL in the incubation medium (1 g/ml; data not shown). The heparin-resistant fraction was low at 4°C (0.6 -2.6% of the heparin-releasable fractions in the presence of LPL), demonstrating that no uptake had occurred. RecomLDL100, recombinant LDL containing apoB100.  1. Inhibition of LPL-mediated binding of LDL to THP-I cells by liposomes. THP-I monocyte-derived macrophages were incubated for 30 min at 4°C with a mixture of 125 I-labeled LDL (5 g/ml), bovine LPL (10 g/ml), and liposomes (LS) prepared from PC, FC, and lyso-PC. Heparin-releasable radioactivity representing bound LDL was determined as described under "Experimental Procedures." Data are mean values from duplicate determinations. Filled bars indicate binding in the absence of liposomes. Open bars indicate binding in the presence of PC liposomes only (135 g/ml of incubation medium in the case of nondiluted liposomes, 13.5 g/ml in the case of 1:10 diluted liposomes). Hatched bars indicate liposomes of PC and FC (95 g/ml and 40 g/ml, respectively, in the case of nondiluted liposomes). Crosshatched bars indicate liposomes of PC and lyso-PC (total concentration was 135 g/ml of medium in the case of nondiluted liposomes).

FIG. 2. Stoichiometry of binding of LPL to immobilized LDL as determined by SPR.
A, LPL at the indicated concentrations was injected into BIAcore flowcells containing matrix-bound LDL. The response values (RU) at steady state were used to calculate the mass of LPL bound to LDL. These values were then used to calculate the binding stoichiometry of LPL dimers (110 kDa) to LDL on a molar basis. The stoichiometries presented are mean values of data from three different surface concentrations of matrix-bound LDL (1.5, 2.5, and 3.9 ng/mm 2 ). The experiments were performed in 10 mM Hepes, 0.15 M NaCl, pH 7.4. B, the mass/response correlation for LDL compared with LPL. Different amounts of 125 I-labeled LDL or 125 I-labeled LPL were coupled to sensor chips, and the response value for each chip was determined. The bound radioactivity was measured, and the immobilized mass was calculated from the respective specific radioactivities. Relationships are shown for LDL total mass (filled circles), LDL protein mass (open circles), and LPL protein mass (squares). tained a fair amount of lipids (mostly polar lipids but also traces of triglycerides). Because LPL strongly interacts with many kinds of lipids, the associated lipids might have influenced the binding of apoB17 to LPL. Complete delipidation, however, generated an insoluble protein that could not be used for interaction studies. Therefore, a more gentle reduction of the lipid content by treatment with detergents (deoxycholate) was used. Reduction of the lipid content of apoB17 by about 50% completely abolished the detectable binding of apoB17 to LPL. Size exclusion chromatography analysis of the partially delipidated apoB17 showed that the delipidation did not result in any oligomer formation (data not shown). Furthermore, the partially delipidated apoB17 displayed normal affinity to the anti-apoB monoclonal antibodies MB19 (recognizes a Thr to Ile substitution at residue 71 in apoB) (43) and 1D1 (epitopes 474 -539 on apoB) (44) and to two monospecific polyclonal antibodies directed against peptides around residues 12 and 259 in human apoB100 (generously provided by Dr. Thomas L. Innerarity, Gladstone Institute, San Francisco). The recovery of partially delipidated apoB17 in immunoprecipitation experiments with the different anti-apoB antibodies performed in the absence of detergents were 87 Ϯ 8, 103 Ϯ 14, 93 Ϯ 21, and 107 Ϯ 17%, respectively. The recoveries of native apoB17 were 82 Ϯ 12, 93 Ϯ 22, 86 Ϯ 9, and 102 Ϯ 10%, respectively. Thus, we did not find any evidence for masked epitopes in the partially delipidated apoB17. We conclude that the interaction between apoB17 and LPL is mediated mainly, if not exclusively, by lipids. DISCUSSION LPL has been proposed to have a role in atherosclerosis, functioning as a bridge that links atherogenic LDL with the subendothelial matrix of the arterial wall. Retention of LDL in the arterial wall may be a key initiating event in atherogenesis (13). LPL can also directly mediate binding and uptake of atherogenic lipoproteins by cells in the vessel wall, thereby promoting lipid accumulation in these cells. Recent data from animal models have provided compelling evidence that LPL located in the arterial wall is proatherogenic (10,(45)(46)(47).
To investigate the role of apoB in the LPL-LDL interaction, we first tested how modifications of LDL affect its ability to interact with LPL. In contrast to earlier studies (48), our results demonstrate that chemical modification of the arginines or lysines in apoB100 did not abolish the LPL-LDL interaction. Therefore, basic residues in apoB, which are essential for normal LDL-proteoglycan interaction (49,50), are not important for the interaction with LPL. Thus, LPL must interact with another site or sites in LDL. This explains why oxidized LDL can have higher affinity for LPL than native LDL has for LPL even though the basic residues in apoB lose their positive charges during LDL oxidation (19,22,36,51,52).
Eight regions in delipidated apoB100 bind heparin (46 -48). We previously showed that one of these sequences (residues 3359 -3369) is the main proteoglycan binding site in apoB100containing LDL (34) and is also the heparin binding site in LDL in vivo. This finding is in agreement with recent data by Gaus et al. (23) showing that only one or two heparin molecules are involved in binding the LDL. However, our data imply that partial delipidation exposes several of the remaining heparin binding sequences, resulting in an enhanced binding of heparin. In the current study, partially delipidated LDL had decreased binding affinity for LPL, indicating that lipid is required for proper LPL-LDL interaction.
Because it had been suggested that apoB interacts specifically with LPL (17,18), a major emphasis in this study was to determine whether a LDL particle can bind more than one LPL dimer. In SPR experiments, the average molar ratio of LPL molecules per LDL particle at saturation is around 14. Because the saturation of LDL by LPL was achieved at relatively high LPL concentrations, one might question the physiological significance of the interaction. However, even at nanomolar concentrations of LPL, the molar ratio was clearly greater than 1. In experiments with cultured macrophages, the calculated stoichiometry in the binding of LDL to LPL on the cell surface also indicated that several (up to 15) LPL dimers could bind per LDL particle. In these experiments a consistent finding was that more LPL bound to the cells in the presence of LDL than in the absence of LDL, indicating secondary attachment of some LPL to already bound LDL.
These results are incompatible with the notion that apoB has only one specific LPL binding site; instead, they imply that there are many LPL binding sites on apoB or, more likely, that LPL mainly binds to the exposed lipid surface. ApoB covers 40 -60% of the surface area of LDL (53). The remaining surface should be exposed for interaction with other proteins. The hydrated radius of dimeric LPL is 4.4 nm (54). Thus, packed together with apoB, 20 -25 LPL dimers could be accommodated on a LDL particle. The finding that partially delipidated LDL had markedly decreased binding affinity for LPL strongly in- FIG. 3. Binding of LDL to THP-I macrophages and calculation of stoichiometry between LPL and LDL under near saturating conditions. Binding of 125 I-labeled LDL and LPL to THP-I monocyte-derived macrophages was studied as described in Fig. 1. In all experiments the concentration of LDL was 5 g/ml. A, inverted triangles, 1 g of LPL/ml of medium; circles, 10 g of LPL/ml; squares, 20 g of LPL/ml; and triangles, 50 g of LPL/ml. B, the concentration of LPL added was 10 g/ml of medium. The  dicates that the majority of LPL molecules interact directly with the LDL lipid. This was further supported by the results of competition assays of the LPL-LDL interaction with liposomes.
In vivo, the situation is different: LPL is bound to cell-surface heparan sulfates, and LDLs are passed over the cell surfaces by blood circulation. A similar situation was analyzed by SPR in a previous study (21). At that time, we did not have values for the mass/response relationship for LDL. Therefore the effect of the amount of LPL on the amount of bound LDL was expressed only in RU. Our previous study indicated that detectable LDL binding starts when three or four heparan sulfate-bound LPL dimers interact simultaneously with the LDL particle and that the amount of bound LDL is proportional to the concentration of LPL at the surface. The slope of this relationship indicates how much LPL is needed to form a binding site for LDL. Knowing the mass/response relationship for LDL, we can now calculate that six or seven LPL dimers were needed for optimal formation of the LDL binding site in the previous experiments, again demonstrating that several LPL molecules bind per LDL particle.
Our data seemingly contradict previous findings of Goldberg and co-workers (17,18), who have suggested that proteinprotein interaction between LPL and LDL, especially the amino-terminal region of apoB, is more important than proteinlipid interaction. Recently, these authors (55) concluded that the high affinity binding between LPL and LDL involves multiple ionic and hydrophobic interactions and that the interaction is inhibited by heparin. We decided to directly study the interaction between apoB17 and LPL by SPR and compared it with the LDL-LPL interaction. The K d values differed by a factor of 10 3 , demonstrating that the interaction between apoB17 and LPL must be relatively insignificant in the physiological setting. However, several different lipid classes were associated with the purified recombinant apoB17. Because recombinant apoB17 interacts tightly with lipids (56 -58), we studied the effect of gentle delipidation of the apoB17 preparation. This treatment abolished the detectable affinity of apoB17 for LPL, again demonstrating that the lipids are necessary for the interaction with LPL. We did not further characterize the lipid content of apoB17. The lipids must originate either from the insect cells, from the culture medium, or from both.
Taken together, our results demonstrate that LPL interacts mainly with the lipids of the LDL particle and that proteinprotein interaction with the amino-terminal region of apoB plays a minor role if any. This finding may explain how LPL facilitates the binding of oxidized LDL to proteoglycans and to cells. Even mild oxidation leads to some hydrolysis of LDL phospholipids (59). This may expose an even more attractive interface for LPL binding as was shown for the interaction of LPL with synthetic phosphatidylcholine-stabilized emulsions of triglycerides and with very low density lipoproteins (60). In line with these observations, Pentikä inen et al. (61) demonstrated recently that LPL bind with high affinity to apoB-free microemulsions reconstructed from LDL lipids. In the absence of LPL, binding of LDL to artery wall proteoglycans is mediated through an ionic interaction between residues 3359 -3369 in apoB100 and the glycosaminoglycans (37). This interaction is partly diminished by oxidation (52). The data presented here show that LPL can mediate the interaction between LDL and proteoglycans independently of the physical state of apoB and thus of the affinity between apoB and proteoglycans. This effect of LPL may under certain circumstances be important for the binding of LDL to artery wall proteoglycans and to cells of the vessel wall although the amounts of LPL at these sites may not be high enough to allow the highest affinity interaction between one LDL particle and many LPL dimers.