Apolipoprotein E Containing High Density Lipoprotein Stimulates Endothelial Production of Heparan Sulfate Rich in Biologically Active Heparin-like Domains

Reduced heparin and heparan sulfate (HS) proteoglycans (PG) have been observed in both inflammation and atherosclerosis. Methods to increase endogenous heparin and heparan sulfate are not known. We found that incubation of endothelial cells with 500–1,000 μg/ml high density lipoprotein (HDL) increased35SO4 incorporation into PG by 1.5–2.5-fold. A major portion of this increase was in HS and was the result of increased synthesis. Total PG core proteins were not altered by HDL; however, the ratio of 35SO4 to [3H]glucosamine was increased by HDL, suggesting increased sulfation of glycosaminoglycans. In addition, HDL increased the amount of highly sulfated heparin-like HS in the subendothelial matrix. HS from HDL-treated cells bound 40 ± 5% more125I-antithrombin III (requires 3-O sulfated HS) and 49 ± 3% fewer monocytes. Moreover, the HS isolated from HDL-treated cells inhibited smooth muscle cell proliferation (by 83 ± 5%) better than control HS (56 ± 6%) and heparin (42 ± 6%). HDL isolated from apolipoprotein E (apoE)-null mice did not stimulate HS production unless apoE was added. ApoE also stimulated HS production in the absence of HDL. ApoE did not increase35SO4 incorporation in macrophages and fibroblasts, suggesting that this is an endothelial cell-specific process. Receptor-associated protein inhibited apoE-mediated stimulation of HS only at higher (20 μg/ml) doses, suggesting the involvement of a receptor-associated protein-sensitive pathway in mediating apoE actions. In summary, our data identify a novel mechanism by which apoE and apoE-containing HDL can be anti-atherogenic. Identification of specific apoE peptides that stimulate endothelial heparin/HS production may have important therapeutic applications.

Proteoglycans (PG), 1 important constituents of vascular cell membranes and extracellular matrix (1,2), consist of a core protein to which long chains of negatively charged polysaccharides termed glycosaminoglycans (GAG) are attached. The three major PG classes in the vessel wall are heparan sulfate (HS), chondroitin sulfate, and dermatan sulfate. HSPG play an important role in the regulation of various vascular functions. They bind and promote lipoprotein lipase activity, the key enzyme in the hydrolysis of triglyceride-rich lipoproteins (3). Basic fibroblast growth factor, a potent mitogen and angiogenic factor, requires the presence of cell surface HSPG or exogenous heparin to bind to its high affinity cell signaling receptor (4). In addition, HSPG potentiate the thrombin-inhibiting actions of antithrombin (5).
A reduction in arterial HS and heparin has been observed under conditions of inflammation and atherosclerosis as well as with increased age (6 -14). The age-dependent decrease in HS is more pronounced in atherosclerotic tissues than in normal tissues (10,11). An inverse correlation between the amount of cholesterol in the lesion and the concentration of HS was observed in human aortas. More importantly, this negative correlation was observed in both normal and atherosclerotic vessels. 4 -5-fold more cholesterol was found in vessels that have 50% less HS.
The negative relationship between HS and atherosclerosis is not surprising because arterial HS is known to inhibit smooth muscle cell (SMC) proliferation and promote antithrombin activity (5,(15)(16)(17). In addition, our recent studies show that HS masks subendothelial proteins such as fibronectin and prevents lipoproteins such as lipoprotein(a) and monocytes from associating with the matrix (18,19). Thus, an increase in vascular HSPG could be athero-protective. Although the decrease of HSPG in atherosclerosis has been known for several years, it is not clear how this occurs. In vitro, a reduction in subendothelial HSPG was observed when endothelial cells were exposed to moderately oxidized LDL or lysolecithin, a product of lipoprotein oxidation. This reduction in subendothelial HSPG was found to be caused by secretion, by endothelial cells, of a HSPG-degrading heparanase activity (18,19). During these studies an initial observation was made that HDL blocked the oxidized LDL-and lysolecithin-mediated decreases in HSPG (19). We now show that HDL, more specifically the apoE component of HDL, stimulates endothelial HS synthesis. MATERIALS (20)) or a commercially obtained apoE3 (Calbiochem) was used.
Cells-Bovine aortic endothelial cells were isolated and cultured as described (21). The cells (5-15 passages) were grown in minimal essential medium (MEM) containing 10% fetal bovine serum (Life Technologies). The subendothelial matrix was prepared as described previously (18)  DEAE-Cellulose Chromatography of Proteoglycans-To determine changes in PG, DEAE-cellulose chromatography was performed as described previously (21,22). Endothelial cells were labeled with [ 3 H]leucine, and labeled proteins were extracted using 10 mM Tris-HCl buffer, pH 7.4, containing 1% octyl glucoside, 1% CHAPS, 0.1 mM each EDTA and phenylmethylsulfonyl fluoride, and 1 g/ml leupeptin. Equal amounts of labeled proteins from control and HDL-treated cells were loaded on a DEAE-cellulose column previously equilibrated with HEPES buffer containing 0.15 M NaCl and 0.1% Tween 80. Stepwise elution was done with HEPES buffer containing 0.25 M and 0.5 M NaCl.
GAG Analysis-The relative incorporation of [ 3 H]glucosamine and 35 SO 4 was determined in isolated GAG. To prepare GAG chains, aliquots of purified PG were incubated with 0.10 volume of 10 N NaOH for 18 h at 26°C with constant shaking and then neutralized with 10 N HCl. Samples were dialyzed, and protein-free GAG were purified by DEAE-cellulose chromatography. Chondroitin sulfate/dermatan sulfate-GAG were degraded by treatment with chondroitin ABC lyase (0.1 unit) treatment as described previously (22). To determine the size of the GAG, gel filtration on Sepharose 6B was performed using 0.2 M NaCl as eluant as described previously (22).
To determine heparin-like sequences in the subendothelial matrix, 35 SO 4 -labeled matrix prepared from control and HDL-treated cells was incubated with 1 unit/ml of heparitinase for 2 h at 37°C. Released material (containing degraded and undegraded HS) was then passed through a PD-10 gel filtration column. Radioactivity eluting in the void volume was determined as undigested heparin-like HS. The undegraded material was further subjected to low pH nitrous acid treatment as described previously (22). Briefly, 1 ml of PD-10 void volume material was incubated with an equal volume of a mixture containing 20% butyl nitrite in 1 M HCl. The mixture was incubated for 2 h, and the digested material was gel filtered on a PD-10 column.
Labeling of Monocytes and Heparin-binding Proteins and Binding to the Subendothelial Matrix-THP-1 monocytes were labeled with [ 3 H]leucine (100 Ci/1 ϫ 10 7 cells) for 2 h at 37°C. The label was removed, and cells were washed three times with MEM-BSA and suspended in MEM-BSA. Suspended cells were added to the subendothelial matrix prepared from control or HDL-treated endothelial cells in 24-well plates (2-4 ϫ 10 5 cells/well) and incubated for 1 h at 37°C. The spontaneous release of radioactivity under these conditions was about 5%. Unbound monocytes were removed by washing four times with MEM-BSA, and bound radioactivity was extracted by incubation in 0.1 N NaOH and 0.1% SDS for 1 h.
Antithrombin and lipoprotein lipase were iodinated using the en-zymes lactoperoxidase and glucose oxidase as described previously (23). Iodinated proteins were purified by heparin-agarose chromatography, and proteins were eluted with 1.5 M NaCl. 5 g of iodinated protein was incubated with subendothelial matrix in MEM-BSA for 2 h at 37°C. Unbound protein was removed, and bound radioactivity was determined as above.
Smooth Muscle Cell Proliferation-Rat aortic SMC were kindly provided by Dr. L. Rabbani (Department of Medicine, Columbia University). SMC were grown in basal medium supplemented with growth factors (Clonetics). To determine the effects of HS on SMC proliferation, cells were plated at low density (8 ϫ 10 4 /well) and cultured for 3 days in the presence or absence of matrix HS. Matrix HS was isolated from [ 3 H]glucosamine-labeled control and HDL-treated endothelial cells. Equal amounts of glucosamine radioactivity from control and HDLtreated cells were used. On the 4th day, the cell number was counted with a hemacytometer. Net growth was determined by subtracting the final cell number from the initial cell number. The percentage of SMC growth inhibition was calculated by using the average of each triplicate with the formula % Inhibition ϭ ͩ 1 Ϫ Net growth in the presence of HS Net growth in the controls ͪ ϫ 100 (Eq. 1) In another experiment, SMC were cultured in the above conditions for 3 days, and the cells were then labeled with [ 3 H]thymidine for 6 h; radioactivity incorporated into the DNA was determined. For this experiment HS isolated from nonlabeled cells was used.
Determination of 35 SO 4 Incorporation in Mouse Tissues-Control and apoE-null mice (4 weeks old, three each) were injected intraperitoneally with 100 Ci of 35 SO 4 in 100 l of saline. Mice were sacrificed after 4 h, tissues were perfused (through the left ventricle) with PBS, and liver and heart (together with the proximal aorta) were removed. Tissues were homogenized (Polytron) for 30 s in ice-cold HEPES pH 7.0 buffer containing 4 M urea, 0.5% CHAPS, 0.05 M NaCl, 1 mM each phenylmethylsulfonyl fluoride and benzamidine, and 5 g/ml leupeptin. Homogenates were centrifuged (14,000 rpm, 20 min), and the supernatants were dialyzed extensively to remove free sulfate. Aliquots of dialyzed supernatants were counted, and radioactivity was expressed per mg of tissue protein.  35 SO 4 incorporation in all three pools, secreted, cellular, and matrix, increased by 50 -100%. An increase of 43% with 250 g of HDL protein and an increase of 83% with 500 g of HDL protein was observed in matrix. HDL also increased [ 3 H]leucine incorporation into proteins by 15-20% (not shown). However, the increase in PG sulfate shown in Fig. 1 was found after normalizing for protein. These data suggest that HDL stimulated sulfate incorporation into PG. It should be noted that these changes were found using levels of HDL which are within the normal plasma concentrations of HDL protein (150 -180 mg/dl fasting plasma (24)).

HDL Stimulates Endothelial PG Production-To
HDL Increases Endothelial Cell PG Synthesis-To determine whether the HDL-mediated increase in sulfate incorporation was caused by increased synthesis, cells were labeled with 35 SO 4 for different time periods in the presence of HDL (1,000 g/ml). HDL increased sulfate incorporation by 25% in 2 h and by 37% in 6 h in cells ( Fig. 2A) and matrix (Fig. 2B). When endothelial PG were first labeled with 35 SO 4 and chased in cold medium for different time periods in the presence or absence of HDL, HDL did not alter the turnover rates of PG (not shown). These data suggest that HDL-mediated increases in 35 SO 4 incorporation were primarily the result of increased synthesis.
HDL Did Not Alter PG Core Proteins but Increased Sulfation of GAG-To determine if increased 35 35 SO 4 radioactivity was eluted at salt concentrations Ͼ 0.25 M NaCl. When equal amounts of cellular proteins were loaded on DEAE, the amount of radiolabel eluted at 0.5 M NaCl was not different (Fig. 3, 2,992 cpm in control cells versus 3,079 cpm in HDL-treated cells). Although this suggests that total core proteins were not altered, whether HDL altered core proteins of specific PG (such as perlecan and syndecans, which may not reflect in total core proteins) remains to be determined.
HDL increased endothelial GAG ([ 3 H]glucosamine incorporation into PG, Fig. 4A), by 34% in cells and by 41% in matrix. Thus, the increase in 35 SO 4 incorporation into PG was, in part, the result of increased GAG synthesis. Endothelial cells primarily synthesize HS; consistent with this Ͼ75% of total GAG was resistant to chondroitin ABC lyase (which removes chondroitin and dermatan sulfates) treatment (not shown). Enrichment in GAG could be caused by either an increased number of GAG or an increased chain length. Sizing analysis of isolated GAG by Sepharose 6B gel filtration chromatography did not show changes in the chain length of GAG isolated from control and HDL-treated cells (not shown). We then compared the increases in sulfation and GAG (expressed as sulfate cpm/ glucosamine cpm) (Fig. 4B). Although HDL increased glucosamine (Fig. 4A), the GAG were relatively more sulfated in HDLtreated cells as indicated by an increase in the sulfate: glucosamine ratio. A 2.5-fold increase in the sulfation of cellular GAG and an increase of 1.5-fold in the sulfation of matrix GAG was observed.
Anti-atherogenic and Anti-thrombogenic Properties of HS from HDL-treated Endothelial Cells-Several anti-atherogenic actions of HS require specific sulfate groups. Studies from this laboratory showed that subendothelial HS inhibition of monocyte binding require sulfation (18). Acceleration of antithrombin-induced proteinase inhibition by HS requires a pentasaccharide with 3-O sulfate on the internal glucosamine (2). The anti-proliferative activity of arterial HS appears to require sequences rich in 2-O sulfated uronic acid (25). Similarly, HS that contains 2-O iduronic acid has a high affinity for lipoprotein lipase (26). These sequences are generally localized to heparin-like domains in HS (resistance to heparitinase). In our experiments we found a 1.7-fold increase in these heparitinaseresistant but nitrous acid-sensitive heparin-like sequences in matrix (cpm/well in a six-well plate, 10,532 Ϯ 765 in control versus 17,986 Ϯ 1,289 in HDL-treated cells).
To determine the consequences of increased matrix HS, ma- trix was prepared from control and HDL-treated endothelial cells and incubated either with [ 3 H]leucine-labeled monocytes or 125 I-labeled antithrombin III. The number of monocytes binding to the HDL matrix was decreased by approximately 49%, suggesting that increased HSPG inhibited monocyte interactions with subendothelial matrix (Fig. 5). In contrast, antithrombin III binding to matrix was increased by 43% in HDL-treated cells compared with control matrix. These data suggest an increase in specific HS (containing 3-O sulfates).
We next tested the ability of the matrix HS to inhibit SMC proliferation, a key event in the development of atherosclerosis. SMC were cultured in the presence or absence of matrix HS isolated from control and HDL-treated endothelial cells. Cell proliferation was assessed either by counting the number of cells (Fig. 6) or by assessing [ 3 H]thymidine incorporation. Endothelial HS was normalized for glucosamine concentration before adding to SMC. Incubation of SMC with HS isolated from chlorate (an inhibitor of sulfation)-treated endothelial cells did not significantly inhibit SMC growth, suggesting the requirement for sulfated HS. Heparin, a known inhibitor of SMC proliferation, at 50 units/ml inhibited SMC proliferation by 42 Ϯ 6%. HS prepared from HDL-treated endothelial cells showed most inhibition (82 Ϯ 5%), better than heparin and control HS (56 Ϯ 6%). These data suggest that HDL treatment increased antiproliferative HS in endothelial cells.
HDL Effects Require ApoE-We next tested whether HDL effects on PG sulfation require apoA-I and/or apoE. HDL was isolated from wild type (WT), apoA-I-null, and apoE-null mice and tested for its ability to stimulate sulfation. Equal amounts of HDL-protein (100 g/ml) were used. HDL isolated from control and apoA-I-null mice increased 35 SO 4 incorporation into cellular PG. HDL from apoE-null mice, in contrast, failed to stimulate endothelial PG sulfation (Fig. 7A). Similar results were obtained when apoE HDL was removed from human HDL by heparin-Sepharose chromatography; the E-deficient HDL did not increase PG sulfation (data not shown). To determine further if this lack of stimulation was caused by the absence of apoE, apoE was added to apoE-null HDL (Fig. 7A). The addition of 5 g of apoE restored the ability of E-null HDL to stimulate PG sulfation to mouse WT HDL levels. At 10 g of apoE, E-null HDL increased 35 SO 4 incorporation by 2-fold. These data show that apoE is required to stimulate PG sulfation.
We further examined the effects of apoE in emulsions and in lipid free form. ApoE was able to stimulate endothelial HS production in both free form and DMPC emulsions (Fig. 7B). These data suggest that apoE actions do not require other HDL components. Surprisingly, however, VLDL failed to stimulate HS production despite containing similar or greater amounts of apoE (determined by SDS-polyacrylamide gel electrophoretic analysis; not shown).
We next tested whether apoE would stimulate 35 SO 4 incorporation in other cells. Unlike in endothelial cells, incubation of J774 macrophages (which do not synthesize apoE) or human skin fibroblasts with apoE (10 g/ml) did not alter PG production (Table I).
Endothelial Cell Surface Molecules Involved in ApoE Actions-We next determined the role of lipoprotein receptors in mediating apoE actions. Competition with lipoproteins and receptor antagonists were used (Fig. 8). LDL or VLDL neither stimulated HS production (closed bars) nor inhibited HDLmediated stimulation (open bars). Similar results were obtained with LDL receptor antibody. The 39-kDa RAP at con- Labeled PG in the cell layer and subendothelial matrix compartments were determined, and the ratio of 35 S to 3 H was determined. HDL increases the sulfate:glucosamine ratio, suggesting increased sulfation of GAG.
centrations that others have used to inhibit LDL receptorrelated protein (5-10 g) did not inhibit an HDL-mediated increase in 35 SO 4 incorporation. However, at a higher dose (20 g/ml) RAP inhibited apoE-mediated HS production by Ͼ60%. The requirement for higher doses of RAP could be caused by the longer incubation times (16 h) required for HS stimulation in the current experiments. Nevertheless, these data suggest that RAP-sensitive pathways are involved in apoE function. In different experiments heparin increased 35 (Fig. 9). However, hearts (containing part of the aorta) from apoE-null mice had ϳ40% reduction in 35 SO 4 incorporation compared with hearts from WT mice. These data suggest that apoE also stimulates sulfation of PG in vivo. Suspended cells were added to matrix prepared from control or HDLtreated endothelial cells in 24-well plates (2-4 ϫ 10 5 cells/well) and incubated for 1 h at 37°C. Unbound monocytes were removed by washing four times with MEM-BSA, and bound radioactivity was extracted by incubation in 0.1 N NaOH and 0.1% SDS for 1 h. Antithrombin was iodinated using the lactoperoxidase/glucose oxidase method and purified by heparin-agarose chromatography. 5 g of iodinated protein was incubated with matrix prepared from control and HDL-treated cells for 2 h at 37°C. Unbound protein was removed, and bound radioactivity was determined.

FIG. 6. HS isolated from HDL-treated endothelial cells is a potent inhibitor of SMC proliferation.
SMC were plated at low density (8 ϫ 10 4 /well) and cultured for 3 days in medium alone (None) or media containing 50 units/ml heparin, matrix HS isolated from 25 M chlorate-treated, control, and HDL-treated endothelial cells. On the 4th day, the cell number was determined. The percentage of SMC growth inhibition was calculated as described under "Materials and Methods. "   FIG. 7. Panel A, apoE is required for HDL stimulation of endothelial HS. HDL was isolated from wild type (C57BL/6J, denoted WT), apoEnull (E(0)), and apoA-I-null (AI(0)) mice. Confluent monolayers of endothelial cells in 24-well plates were incubated in growth medium containing [ 35 S]sulfate (50 Ci/well) and different HDLs (100 g/ml) for 16 h under culture conditions. Sulfate-labeled PG in the cell layer was determined. For add-back experiments, E(0) HDL was preincubated with purified apoE (5 and 10 g) for 30 min before adding to endothelial cells. Sulfate-labeled PG in the cell layer was determined. Values represent the mean Ϯ S.D. Panel B, effects of apoE and apoE-containing particles on 35 SO 4 incorporation. Confluent monolayers of endothelial cells in 24-well plates were incubated in 35 SO 4 -containing growth medium for 37°C for 16 h. The medium contained one of the following: purified apoE (5 g/ml) or DMPC vesicles or apoE in DMPC vesicles (E-DMPC) or VLDL (200 g/ml) or HDL (500 g/ml). Sulfate-labeled PG in cell layer were determined. Loss of endothelial HS has been postulated to lead to several pathological events, in particular to events related to atherosclerosis (15)(16)(17)(18)(19)(27)(28)(29)(30)(31). These include 1) altered endothelial permeability, 2) increased cell migration through blood vessel walls, 3) thrombin generation, 4) increased monocyte binding to the subendothelial matrix, 5) increased lipoprotein retention, 6) increased SMC proliferation, and 7) increased susceptibility to bacterial infection. HS has also been shown to inhibit matrix metalloproteinase activity (32), involved in plaque rupture (33), and to regulate the bioavailability of basic fibroblast growth factor activity (34). Agents that decrease endothelial HSPG include lipopolysaccharide and tumor necrosis factor-␣ (35), homocysteine (36), lysolecithin, and oxidized LDL (18, 19).
Thus, a decrease in HS may be a general inflammatory reaction.
The present studies show that HDL increases endothelial HS and thus could decrease the occurrence of the above mentioned events. HDL has several anti-atherogenic effects, and epidemiological studies inversely correlated HDL with atherosclerosis (37). HDL facilitates reverse cholesterol transport from peripheral tissues to liver (38). In addition, platelet-activating factor acetyl hydrolase and paraoxonase enzymes associated with HDL can metabolize and reduce the content of biologically active oxidized lipids in oxidized LDL (39,40). Our current results show yet another mechanism by which HDL can be anti-atherogenic, i.e. by increasing sulfation of endothelial HSPG.
HDL increased both glucosamine and sulfate incorporation into PG without significantly affecting the total core proteins. Although total core proteins have not changed based on the elution of [ 3 H]leucine-labeled proteins from DEAE-cellulose, it is conceivable that HDL increased specific PG core proteins which remains to be determined. The ratio of sulfate to glucosamine was increased by HDL, suggesting that the GAG were more sulfated in HDL-treated endothelial cells. The increase in sulfation does not appear to be due to apoE binding to HS and preventing its degradation because incubation of 35 SO 4 -labeled endothelial cells with HDL did not prevent degradation. Thus, HDL treatment may have increased sulfation by affecting sulfotransferases. Several enzymes involved in the sulfation of HSPG have been characterized (41)(42)(43)(44)(45). These include glucosamine N-, 3-O, and 6-O sulfotransferases and uronic acid 2-O sulfotransferase. It is conceivable that HDL increased the activities of one or more of these enzymes. Regulation of sulfation and sulfotransferases in endothelial cells is a poorly studied area. This is surprising considering the several known antiatherogenic effects of endothelial heparin and HS. Evidence has recently been put forward that a deficiency of endogenous heparin or heparin-like substances predisposes to atherosclerosis (46,47), and heparin administration has been shown to increase sulfation of endothelial HS (48).
Subendothelial HS contains substantial amounts of highly sulfated blocks (referred to as heparin-like) (49,50). These are resistant to heparitinase (heparinase III) digestion. Data in Figs. 4 and 5 support the conclusion that HDL increased sulfated HS. Antithrombin binding is largely restricted to heparin chains containing glucosamine N and 3-O sulfates (51). In our experiments antithrombin binding was increased to matrix prepared from HDL-treated cells, suggesting an increase in GlcNSO 3 (3-OSO 3 )-heparins. Second, HS prepared from HDLtreated matrix inhibited SMC proliferation, suggesting an increase in HS sequences containing 2-O uronic acids (25). An increase in 2-O iduronic acid-containing HS was also confirmed by increased binding of lipoprotein lipase (1.6-fold, not shown).
Experiments with HDL prepared from apoE and apoA-I knockout mice suggest that apoE is required for stimulation of sulfation. This was confirmed further by apoE add-back experiments and apoE emulsion experiments. Lipid-free apoE also stimulated endothelial HS production. It is, however, conceivable that apoE acquired cellular lipid during the 16-h incubation. Nevertheless, these data clearly show that apoE is required for HS stimulation. Although containing apoE, VLDL failed to stimulate HS. Whether this is because of conformational changes in VLDL apoE remains to be determined.
Although apoE is known to play a key role in lipoprotein clearance, recent studies from several groups indicate that its anti-atherogenic effects go beyond its role in remnant clearance. Shimano  Control and apoE-null mice (4 weeks old, three each) were injected intraperitoneally with 100 Ci of 35 SO 4 in 100 l of saline. Mice were sacrificed after 4 h, tissues were perfused with phosphate-buffered salineS, and liver and heart (together with proximal aorta) were removed. Tissues were homogenized (Polytron) for 30 s in ice-cold HEPES (pH 7.0) buffer containing 4 M urea, 0.5% CHAPS, 0.05 M NaCl, 1 mM each phenylmethylsulfonyl fluoride and benzamidine, and 5 g/ml leupeptin. Homogenates were centrifuged (14,000 rpm, 20 min), and the supernatants were dialyzed extensively to remove free sulfate. Aliquots of dialyzed supernatants were counted, and radioactivity was expressed per mg of tissue protein. without significant changes in plasma lipoproteins. Recently, Fazio et al. (54), by transplanting apoE-null macrophages into normal C57BL6 mice, increased atherosclerosis. How apoE protects the vessel wall from accumulating lipoproteins is not clear. This may in part be the result of its ability to facilitate removal of cholesterol from the cells of arterial walls (37).
Other studies have also suggested possible anti-atherogenic roles for apoE-HDL, including inhibition of lipase-mediated LDL retention (55) and inhibition of platelet aggregation through the nitric oxide pathway (56). Studies from humans as well as mice suggest that low apoE as well as apoE-HDL are important risk factors for vascular disease (57)(58)(59)(60). Our observation that apoE-HDL increases sulfation of HS offers an alternative explanation for the anti-atherogenic effects of vascular/macrophage apoE. Atherosclerotic vessels have decreased HS, and we showed that removal of subendothelial HSPG resulted in a 2-10-fold increase in the binding of atherogenic lipoproteins such as lipoprotein(a) and monocyte-macrophages (18,19). Consistent with this, our present data show that increasing HSPG by HDL treatment decreased the number of monocytes binding to the subendothelial matrix. In addition, an apoE-HDL-mediated increase in vascular HSPG can prevent SMC proliferation in the subendothelial space.
The mechanism of apoE-HDL-mediated stimulation of sulfation is not clear. ApoE is known to affect cell signaling by increasing cAMP and cGMP levels (56). Preliminary experiments showed that 3-isobutyl-1-methylxanthine, a phosphodiesterase inhibitor (1-5 mM), did not inhibit apoE actions (not shown). It appears that the apoE-mediated effects are cellspecific as no increase in 35 SO 4 incorporation was observed in either macrophages or fibroblasts. Data from WT and apoEnull mice further support this. Hearts (with proximal aorta) but not livers showed a significant decrease in 35 SO 4 incorporation in apoE-null mice. This may be caused by differences in the tissue metabolism of lipoproteins or endothelial cell heterogeneity (61). However, further experiments are needed to prove that the HS decrease is a direct effect of apoE deficiency. It is conceivable that vascular cell-specific surface molecules mediate apoE actions. ApoE has several known receptor molecules, including LDL receptor, LDL-receptor related protein, VLDL receptor, and proteoglycans. Endothelial cells are not known to express LDL receptor-related protein (62). Our data showed that LDL receptor antibody, excess LDL or VLDL did not inhibit apoE-mediated HS stimulation. RAP, a LDL receptor family antagonist, inhibited apoE actions at high doses. VLDL receptor, which is abundant on endothelial cells, appears to be one potential candidate for mediating apoE actions (62). It is also conceivable that RAP, by direct interaction with cell surface HSPG, inhibited apoE binding (63). However, RAP binding to HSPG is controversial (64). Heparin also inhibited apoE actions, suggesting that HSPG, either directly or indirectly facilitate binding to RAP-sensitive receptors and are involved in the mediation of apoE actions. Experiments with apoE-null mice also showed decreased 35 SO 4 incorporation only in heart but not liver. Although this is surprising considering the fact that liver is an endothelial cell-rich organ, one possible reason for this is the lack of VLDL receptors or other potential RAP-sensitive receptors.
In summary, our data show that apoE HDL enhances sulfation of endothelial PG and suggest a novel mechanism by which apoE can be anti-atherogenic. These effects were seen within the physiological concentrations of HDL and within the levels of apoE found in HDL (24). Our data, in addition, offer an alternative explanation for the anti-atherogenic effects of macrophage apoE. After synthesis and secretion from macrophages, apoE or apoE particles can act locally and stimulate endothelial HSPG. Increased HSPG can inhibit subsequent accumulation of lipoproteins and monocytes and inhibit subendothelial SMC proliferation. Thus, identification of specific apoE peptides that can stimulate endothelial cell heparin production may have important therapeutic application.