Oxidation of low density lipoproteins greatly enhances their association with lipoprotein lipase anchored to endothelial cell matrix.

Native and oxidized low density lipoprotein retention within arterial wall endothelial cell matrix (ECM) is an early event in the pathogenesis of atherosclerosis. Previously we showed lipoprotein lipase (LPL) addition to ECM enhanced the retention of apoB-containing lipoproteins. In the present studies we examined whether the oxidation of low density lipoprotein (LDL) increases its retention by LPL-containing ECM. Except where noted, I-labeled moderately oxidized LDL (ModOxLDL) was prepared by long term storage of I-LDL. Without LPL, I-ModOxLDL matrix binding was low and nonsaturable. LPL preanchored to ECM resulted in I-ModOxLDL binding that was saturable and 20-fold greater than in the absence of LPL, with an association constant equal to 2.6 nM. Copper-oxidized LDL (Cu-OxLDL) was able to compete with I-ModOxLDL, whereas a 60-fold native LDL excess had no effect. Reconstituted apolipoprotein B from Cu-OxLDL also reduced I-ModOxLDL to LPL, whereas liposomes derived from the lipid extract of Cu-OxLDL had no effect on binding. These data suggest that the increased binding of oxidized LDL to LPL•ECM may be due to the exposure of novel apoB binding sites and not an oxidized lipid moiety. I-ModOxLDL binding was also not affected by either preincubation with a 300-fold molar excess of apoE-poor HDL or an 340-fold molar excess of Cu-OxHDL. In contrast, a 4-fold apoE-rich HDL excess (based on protein) totally inhibited I-ModOxLDL matrix retention. Positively charged peptides of polyarginine mimicked the effect of apoE-rich HDL in reducing the I-ModOxLDL retention; however, polylysine had no effect. We postulate that the oxidation of LDL may be a mechanism that enhances LDL retention by the ECM-bound LPL and that the protective effects of apoE-containing HDL may in part be due to its ability to block the retention of oxidized LDL in vivo.

I-ModOxLDL retention; however, polylysine had no effect. We postulate that the oxidation of LDL may be a mechanism that enhances LDL retention by the ECMbound LPL and that the protective effects of apoE-containing HDL may in part be due to its ability to block the retention of oxidized LDL in vivo.
The process of LDL oxidation is not very well understood, but it appears to occur within the artery wall, facilitated by cellular lipoxygenase-dependent mechanisms or by nonenzymatic pathways. The phospholipid fatty acids of LDL are believed to be oxidatively modified first (25,26), resulting in the formation of phospholipids containing a spectrum of peroxidized fatty acids. Some of these, through chemical decomposition and/or enzymic hydrolysis, are believed to generate reactive aldehydes, which covalently modify the lysine residues of apoB and cause the surface charge to become more negative (27). Due to the oxidative attack, apoB also undergoes cleavage, which exposes neoepitopes on the protein (28). Although most of the biological effects of OxLDL are thought to result from oxidized lipid products, the increased scavenger receptor uptake of OxLDL is mediated by changes in surface protein charge (3). Another important step in the process of atherogenesis is believed to be the trapping of LDL within the subendothelial space. LDL retention increases its residence time within the artery wall and therefore its susceptibility to oxidative modification. In physiological ion and salt concentrations, very little native (unmodified) LDL binds to endothelial cell-derived matrix in vitro (29). Although lowering the salt concentration or adding divalent cations can increase LDL retention to ECM (30,31), this does not appear to be physiologically relevant. However, under physiological conditions, the ECM avidly binds lipoprotein lipase (LPL), which in turn markedly enhances LDL binding (29). LPL is normally present in the artery wall and increases, due to macrophage and smooth muscle cell production, during atherogenesis (32,33). The increased binding of LDL to LPL-containing ECM is mediated by positively charged clusters of arginine and lysine domains of apoB (34). Furthermore, purified or HDL-associated apolipoprotein E (apoE) is able to efficiently displace LDL from LPL-containing ECM by virtue of its positively charged domains (35). Arginine or lysine polymers were able to mimic the LDL displacement effects of apoE (34).
The binding characteristics of OxLDL to ECM are largely unknown. Because the association of OxLDL to the arterial subendothelial matrix is a potentially pivotal step in the initiation and progression of atherosclerosis, we explored the requirements for its binding and potential physiologic competitors. To exclude interference in the binding studies from enzymes typically used for enzymatic oxidation of LDL (lipoxygenase/phospholipase), we utilized OxLDL prepared by long term storage or by Cu 2ϩ oxidation. The OxLDL, used in these studies, formed during long term storage in the absence of EDTA, was analyzed by the standard measures for the determination of the extent of oxidation (content of thiobarbituric acid-reactive materials, free amino group, fluorescence (excitation; 360 nm; emission: 430 nm), and migration on agarose gels). Because these particles were intermediate of native and maximally oxidized LDL for all the aforementioned assays, we used the term moderately oxidized (ModOxLDL) to describe these LDLs. We found that ModOxLDLs bound to the LPL with greater affinity than LDL, that the increased binding was due to oxidative modification of the protein moiety and not lipid, and that apoE-rich HDL and polyarginine effectively blocked the interaction between OxLDL and LPL anchored to the ECM.

MATERIALS AND METHODS
Preparation of Lipoproteins-LDL (d ϭ 1.019 -1.063 g/ml) and HDL (d ϭ 1.063-1.22 g/ml) were isolated by sequential flotation ultracentrifugation (36) of plasma obtained from normolipidemic human volunteers. LDL was radioiodinated by the iodine monochloride method of McFarlane (37) as modified by Bilheimer (38). The specific activities of the 125 I-LDL preparations ranged from 360 to 600 cpm/ng. 125 I-ModOxLDL was prepared by a modification of Berliner et al. (10). Briefly, 125 I-LDL was stored at 4°C for greater than 45 days in phosphate-buffered saline (0.137 M NaCl, 0.01 sodium phosphate, pH 7.4) in the absence of EDTA. This method consistently yielded moderately oxidized LDL, with a migration on agarose intermediate of native and maximally copper-oxidized LDL. LDL was also oxidized by incubation in the presence of 2-10 M copper sulfate (Sigma). Briefly, copper sulfate was added to 1 mg of LDL in phosphate-buffered saline, yielding a final copper concentration of 2-10 M, and incubated for 12 h at 37°C. The copper sulfate was then removed utilizing Centriflo 50 concentrating cones (Amicon, Beverly, MA), the lipoproteins were then stored in phosphate-buffered saline with 0.01% EDTA. The extent of oxidation was determined by migration on agarose gels (Helena Laboratories, Beaumont, TX), content of thiobarbituric acid-reactive material (39), loss of free amino groups (40), and fluorescence (excitation: 360 nm; emission: 430 nm) (23). The OxLDL yielded a thiobarbituric acid-reactive material value of 8 -22 nmol of malondialdehyde equivalents/mg of protein, a 15-30% decrease in free amino groups (as compared with native LDL), and a fluorescence 6 -8-fold greater than native LDL. ApoE-poor HDL was oxidized as outlined above for LDL, by incubation with 10 M copper. Extent of oxidation was monitored by fluorescence (excitation: 360 nm; emission: 430 nm) (23,41), yielding a value 6.5-fold greater than native HDL.
Lipid-free apoB from Cu-OXLDL was prepared essentially as described by Parthasarathy et al. (42). Briefly, Cu-OxLDL (2 mg of protein, 0.3 ml) was extracted using 10 volumes of iced-cold MeOH followed by an equivolume of CHCl 3 . The protein precipitate was isolated, washed, and then solubilized in octylglucoside (6 mg/ml) by vigorous vortexing for 15 min. The solvent fraction containing the lipids was dried, and large unilamelar vesicles were prepared in phosphate-buffered saline by the extrusion method of Berliner et al. (10). Phospholipid content of the vesicles was measured enzymatically (Wako Diagnostics, Richmond, VA).
Bovine Milk LPL Purification-Bovine milk LPL was purified from fresh unpasteurized milk using Affi-Gel heparin chromatography as described previously (44).
Aortic Endothelial Cell Culture and OxLDL Retention Studies-Porcine aortic endothelial cells were isolated and cultured to confluence, and subendothelial cell matrix was prepared as described previously (35).
OxLDL and native LDL matrix retention studies were performed in both the presence and absence of LPL. Briefly, matrix was washed four times with Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 3% bovine serum albumin (BSA) (Sigma). LPL (8 g/ml) was then added in DMEM/BSA and allowed to incubate with the ECM for 2 h at 4°C. Following the preincubation with LPL, the matrix was then washed three times with DMEM/BSA to remove any unbound LPL. Next, 125 I-ModOxLDL or 125 I-LDL was added to the ECM alone or in the presence of various unlabeled lipoproteins, liposomes, reconstituted oxidized apoB, or polypeptides and incubated for 1 h at 37°C. The medium was then removed, and the matrix was washed three times with DMEM/BSA. Finally, the retained 125 I-lipoproteins were released by the addition of 50 units of heparin (Sigma) and quantitated as described previously (29).
Analytical Methods-Lipoprotein apolipoprotein content was assessed by sodium dodecyl sulfate-polyacrylamide gradient gel electrophoresis (Novex, San Diego, CA). Protein concentrations were determined using the method of Lowry et al. (45), employing BSA as a standard. Specificity of OxLDL Binding to ECM-The specificity of ModOxLDL binding to LPL-containing ECM was examined by competition experiments utilizing native LDL, native LDL that was vortexed to promote aggregation, LDL that was oxidized with copper to varying degrees (Fig. 2), or the lipid extract or reconstituted protein from Cu-OxLDL. A 60-fold excess of native LDL or vortexed native LDL was ineffective in blocking the retention of 125 I-ModOxLDL (Fig. 3). Conversely, Cu-OxLDL was able to completely block 125 I-ModOxLDL binding to ECM (Fig. 4). Using LDL preparations oxidized to various degrees with copper as shown in Fig. 2, competition was shown to be dependent on both the concentration of Cu-OxLDL and the extent of copper oxidation. When 125 I-ModOxLDL (12.5 g/ml) was incubated in the presence of 12.5 g/ml of Cu-OxLDL or reconstituted apoB from the same Cu-OxLDL there was a 50 and 25% decrease in binding, respectively (Fig. 4B). Lipid vesicles derived from the lipid extract of the Cu-OxLDL, at concentrations up to 25 g/ml, had no effect on 125 I-ModOxLDL retention by LPL-containing ECM (Fig. 4B). A 4-fold excess of Cu-OxLDL was also able to abolish the interaction of 125 I-LDL (12.5 g/ml) with LPL bound to the ECM (Fig. 5). These studies demonstrate that the major factor influencing the enhanced binding of ModOxLDL to LPL⅐ECM is the oxidation of the apoB protein.
Role of Exchangeable Lipids in the Increased Retention of OxLDL-To further demonstrate that the lipid moiety of oxidized LDL is not responsible for its increased affinity to LPL bound to matrix, 125 I-ModOxLDL was incubated with a 16-fold excess (based on phospholipid, or approximately 300 HDL particles/LDL) of apoE-poor HDL for 24 h at 37°C to exchange the oxidized lipids and then reexamined for 125 I-ModOxLDL binding. The binding characteristics of 125 I-ModOxLDL to the LPLcontaining ECM were essentially identical whether apoE-poor HDL was present during the preincubation or added just prior to the matrix binding experiment (Fig. 6). These results suggest that an exchangeable lipid on OxLDL is not responsible for its increased affinity to LPL-containing ECM. To determine if the oxidized effects observed were lipoprotein class-specific, we next copper-oxidized apoE-poor HDL (Cu-OxHDL) and used it to compete with 125 I-ModOxLDL. Up to a 60-fold excess (based on protein, or approximately 340 HDL particles/LDL), of Cu-OxHDL had little effect on 125 I-ModOxLDL binding (Fig. 7). These experiments demonstrate that neither a lipid nor apolipoprotein component of oxidized HDL can block the 125 I-Mod-OxLDL binding, suggesting that the enhanced binding may be due to specific changes of apoB during oxidation.
Effect of ApoE-rich HDL on 125 I-ModOxLDL Binding to ECM-Previous studies from our laboratory demonstrated that apoE and apoE-rich HDL blocked LDL retention by LPL anchored to ECM. We postulated that this effect was due to the arginine-rich regions contained within apoE competing with native LDL apoB for binding (35). To determine whether apoErich HDL could similarly compete with OxLDL, 125 I-ModOx-LDL (3.125 g/ml) was coincubated with an increasing concentration of apoE-rich HDL (3.125-200 g/ml) and assessed for binding to an LPL-treated ECM (Fig. 8A). ApoE-rich HDL blocked the retention of 125 I-ModOxLDL to LPL⅐ECM in a concentration-dependent fashion, blocking greater than 90% of binding at 12.5 g/ml. This level of competition was similar to that observed for LDL in our prior studies (35). ApoE-poor HDL at 400 g/ml had no effect on retention (Fig. 8B). These results suggest that apoE competes with apoB-containing lipoproteins for binding to ECM-bound lipase, whether the lipoprotein is native or oxidatively modified. These experiments also yielded further evidence that the protein moiety of OxLDL is most likely responsible for its increased retention.
Effect of Polyarginine and Polylysine on 125 I-ModOxLDL retention-To further ascertain the role of the oxidized protein in the enhanced retention of OxLDL, we investigated the role of the positively charged polypeptides polyarginine and polylysine for their ability to block 125 I-ModOxLDL retention (Fig.  9A). Polyarginine blocked 125 I-ModOxLDL retention in a concentration-dependent manner, reducing binding by 50% at 12.5 g/ml, nearly 100% inhibition at 100 g/ml (Fig. 9B). This was similar to what we observed for competition with native LDL (34). However, polylysine at a concentration up to 100 g/ml had no effect on binding (Fig. 9A). This is in contrast to the previously demonstrated ability of polylysine to compete with native LDL for LPL bound to ECM (34). The data suggest that LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-ModOxLDL (3.125 g/ml) in DMEM/BSA alone (Control) or DMEM/BSA containing LDL (200 g/ml) or the same LDL that was previously vortexed. Plates were washed with buffer, and heparin-releasable radioactivity was determined as described in Fig. 1. Data represent mean Ϯ S.E. of an experiment performed in triplicate. the apoB arginine residues may play an important role in OxLDL retention. DISCUSSION The enhanced binding of native LDL to ECM is facilitated through an LPL bridge formed between matrix glycosaminoglycans and the lipoprotein (35). The retention of LDL did not require an enzymatically active protein, since inactivated LPL (overnight incubation at 25°C) adsorbed to a microfilter plate was able to enhance LDL retention in an LPL concentrationdependent manner. 2 However, the binding of LPL to the cell proteoglycans does require the enzyme to be active (46). We previously observed that apoE-rich HDL and apoE mimetics (polyarginine or polylysine) were effective competitors of native LDL bound to LPL⅐ECM and that this interaction did not release LPL from the matrix (34). In the subendothelium of atherosclerotic plaques epitopes of oxidized apoB have been found (47)(48)(49), suggesting either their localized subendothelium production or rapid sequestration from plasma. In this regard, in the current studies, we investigated the comparative binding affinity of native LDL and ModOxLDL to ECM alone and in the presence of LPL and the ability of apoE-rich and apoE-poor HDL and apoE mimetics to reduce this retention.
In the current studies, an experimental model of the subendothelial matrix was utilized to investigate the binding characteristics of 125 I-ModOxLDL to the ECM. We found the binding of OxLDL to ECM alone was nonsaturating and of low affinity. In contrast, LPL greatly enhanced the saturable and high affinity binding of this modified lipoprotein. Compared with native LDL, OxLDL-bound ECM anchored LPL with a 4-fold greater affinity. The increased affinity was caused exclusively by the modification of LDL apoB and not the lipid. The affinity of OxLDL binding to LPL bound to ECM was dependent on the extent of LDL oxidation. Neither native LDL, aggregated LDL, nor apoE-poor HDL could effectively compete with 125 I-ModOxLDL for ECM-bound LPL; in contrast, apoE-rich HDL efficiently blocked this interaction. Polyarginine mimicked the effects observed with apoE-rich HDL; however, polylysine was unexpectedly ineffective in competing with 125 I-ModOxLDL for ECM-bound LPL.
Four separate experiments suggest that the binding interaction between LPL bound to matrix and OxLDL is mediated by modified apoB rather than oxidized lipid. First, the isolated, reconstituted apoprotein moiety of Cu-OxLDL decreased the interaction of 125 I-ModOxLDL with LPL⅐ECM, whereas the lipid portion had no effect. Preincubation with excess apoEpoor HDL, to remove any oxidized lipids, did not diminish matrix binding. Oxidized apoE-poor HDL also did not compete with OxLDL for binding. Last, both apoE-rich HDL and polyarginine effectively competed with OxLDL. Prior studies (31) have also shown that lyso-PC, a component of oxidized LDL, did not interfere with the retention of LDL with LPL-containing matrix. Collectively these data suggest that it is the oxida- I-ModOxLDL in the presence or absence of various concentrations of Cu-OxLDL that were oxidized to a varying extent as described in Fig. 2. The legend indicates the concentration of copper used for oxidation. B, 125 I-ModOxLDL (12.5 g/ml) was incubated in the presence or absence of 12.5 g/ml of Cu-OxLDL or the reconstituted apoB or lipid extract from the Cu-OxLDL. Plates were washed with buffer, and heparin-releasable radioactivity was determined as described in Fig. 1. Data represent the mean Ϯ S.E. of an experiment performed in triplicate.

FIG. 5. Effect of Cu-OxLDL on 125 I-LDL binding to LPL⅐ECM.
LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-LDL (12.5 g/ml) in DMEM/BSA alone (Control) or DMEM/BSA containing Cu-OxLDL at either 50 or 100 g/ml. The Cu-OxLDL was generated as described under "Materials and Methods" in the presence of 8 M copper sulfate. Plates were washed with buffer, and heparinreleasable radioactivity was determined as described in Fig. 1. Data represent the mean Ϯ S.E. of an experiment performed in triplicate.
tion of apoB that causes the enhanced binding.
The experiments involving polyarginine and polylysine may help explain an apparent discrepancy between the present studies and our previous studies with nonoxidatively modified LDL. Native LDL modified by acetylation (modification of lysine residues) (50) and cyclohexanedione (modification of arginine residues) (51) do not block LDL binding to LPL-matrix as efficiently as native LDL (34). However, in the present study, FIG. 6. Effect of preincubation with apoE-poor HDL on 125 I-ModOxLDL binding to LPL⅐ECM. 125 I-ModOxLDL was incubated with approximately a 300fold molar excess (16-fold excess based on phospholipid) of apoE-poor HDL for 24 h at 37°C, or the same concentration of HDL was added to the 125 I-ModOxLDL just prior to being added to the ECM. LPL was bound to ECM as described in Fig. 1 and then incubated with the indicated concentrations of 125 I-ModOxLDL in DMEM/BSA alone or DMEM/BSA containing apoE-poor HDL. Plates were washed with buffer, and heparin-releasable radioactivity was determined as described in Fig. 1. Data represent the mean Ϯ S.E. of an experiment performed in triplicate.

FIG. 7. Competition of 125 I-ModOx-LDL by Cu 2؉ -oxidized apoE-poor HDL.
LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-ModOxLDL alone (Control) or in the presence of the indicated concentrations of HDL that was oxidized as described under "Materials and Methods." Plates were washed with buffer, and heparinreleasable radioactivity was determined as described in Fig. 1. Data represent the mean Ϯ S.E. of an experiment performed in triplicate.
FIG. 8. Effect of apoE-rich or apoEpoor HDL on 125 I-ModOxLDL binding to LPL⅐ECM. LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-ModOxLDL (3.125 g/ml) in DMEM/BSA containing apoE-rich (A) or apoE-poor (B) HDL at the indicated concentrations. Plates were washed with buffer, and heparin-releasable radioactivity was determined as described in Fig. 1 we show that oxidation of LDL, which has been shown to modify the lysine residues, greatly enhances its association with LPL and can completely block LDL retention. Why does OxLDL bind with higher affinity to LPL-containing ECM, and why is polylysine ineffective in blocking OxLDL retention? This apparent paradox may be explained with the process of LDL oxidation. Oxidation of apoB leads to its fragmentation (28,52,53), which may expose novel sites and also confer conformational changes. This, in turn, may result in a higher affinity binding of OxLDL to LPL⅐ECM. Oxidation also modifies the lysine residues on apoB. The lysine modification would decrease its role in the binding to LPL, which may account for the inability of polylysine to inhibit the OxLDL binding. Further experiments are necessary to delineate the precise mechanisms of these effects.
The ability of apoE-rich HDL to compete with oxidized LDL increases our understanding of the importance of this lipoprotein fraction and in particular apoE. Studies in apoE transgenic and apoE-deficient mice demonstrate the antiatherogenic properties of apoE. Mice, which are normally atherosclerosisresistant, become very prone to atherosclerosis when they are made apoE-deficient by gene targeting (54,55). Interestingly, these mice present with both circulating oxidized LDL and antibodies to epitopes of oxidized apoB (49). Recently, Shimano et al. created transgenic mice overexpressing apoE in the arterial wall (56). When these mice were fed an atherogenic diet, significant decreases in fatty streak formation as compared with nontransgenic controls was observed. Therefore, in these models, the excess of apoE within the lesions may impede the retention of OxLDL and thereby inhibit the progression of atherosclerosis.
Epitopes of oxidized apoB have been demonstrated in the extracellular matrix of lesion areas in animals and humans using antibodies to malondialdehyde OxLDL (47)(48)(49)57). Furthermore, OxLDL has been shown in the plasma of subjects with severe atherosclerosis and diabetes (28, 58 -61). Based on the studies presented here, it is possible that when circulating OxLDL infiltrates the artery wall it is captured by matrixbound LPL. Alternatively, native LDL may first bind to the matrix-bound LPL, and then be oxidized by cells within the arterial wall, thereby resulting in higher affinity binding. In this regard, it has been shown in vitro that the binding of LDL to matrix proteoglycans increases the lipoprotein's susceptibil-ity to oxidation (62). In either scenario, OxLDL may be selectively trapped within the artery wall by such interactions as those presented here. The accumulation of native and CuOx-LDL have been studied in vivo in balloon catheter-deendothelialized rabbit aorta (63). The accumulation of native and Ox-LDL in areas of regenerating endothelium were similar. However, in areas that were still de-endothelialized, OxLDL was retained 2.9-fold more than native LDL, despite its more rapid clearance from plasma. Additionally, the extent of lysine modification of ModOxLDL in these studies is sufficient to be recognized by the scavenger receptor (64). Thus, oxidation may enhance retention in the arterial wall by both increased binding to LPL and uptake through the scavenger receptor, markedly increasing its ability to promote various steps in the pathogenesis of atherosclerosis.
In conclusion, we have demonstrated that oxidized LDL has a higher affinity to LPL anchored to subendothelial matrix than native LDL and that the increased retention is due to the oxidation of apoB, not the lipids. Furthermore, unlike native LDL, it is mainly the arginine residues, not the lysines, that play an important role in this retention. And finally, apoE-rich HDL has been shown to be as effective in blocking the retention of OxLDL, strengthening its role as an antiatherogenic agent.
FIG. 9. Effect of polyarginine or polylysine on 125 I-ModOxLDL binding to LPL⅐ECM. A, LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-ModOxLDL (3.125 g/ml) in DMEM/BSA alone (Control) or DMEM/BSA containing heparin (50 units/ml), polylysine, or polyarginine at the indicated concentrations. B, displacement curve 125 I-ModOxLDL by polyarginine. LPL was bound to ECM as described in Fig. 1 and then incubated with 125 I-ModOxLDL (3.125 g/ml) in DMEM/BSA containing polyarginine at the indicated concentrations. Plates were washed with buffer, and heparin-releasable radioactivity was determined as described in Fig. 1. Data represent the mean Ϯ S.E. of an experiment performed in triplicate.