Transcytosis of Lipoprotein Lipase across Cultured Endothelial Cells Requires Both Heparan Sulfate Proteoglycans and the Very Low Density Lipoprotein Receptor*

Lipoprotein lipase (LPL), the major enzyme responsible for the hydrolysis of circulating lipoprotein triglyceride molecules, is synthesized in myocytes and adipocytes but functions while bound to heparan sulfate proteoglycans (HSPGs) on the luminal surface of vascular endothelial cells. This requires transfer of LPL from the abluminal side to the luminal side of endothelial cells. Studies were performed to investigate the mechanisms of LPL transcytosis using cultured monolayers of bovine aortic endothelial cells. We tested whether HSPGs and members of the low density lipoprotein (LDL) receptor superfamily were involved in transfer of LPL from the basolateral to the apical side of cultured endothelial cells. Heparinase/heparinitase treatment of the basolateral cell surface or addition of heparin to the basolateral medium decreased the movement of LPL. This suggested a requirement for HSPGs. To assess the role of receptors, we used either receptor-associated protein, the 39-kDa inhibitor of ligand binding to the LDL receptor-related protein and the very low density lipoprotein (VLDL) receptor, or specific receptor antibodies. Receptor-associated protein reduced125I-LPL and LPL activity transfer across the monolayers. When the basolateral surface of the cells was treated with antibodies, only anti-VLDL receptor antibodies inhibited transcytosis. Moreover, overexpression of the VLDL receptor using adenoviral-mediated gene transfer increased LPL transcytosis. Thus, movement of active LPL across endothelial cells involves both HSPGs and VLDL receptor.

tiple organs but especially in cardiac and skeletal muscle and in adipose tissue (1). This enzyme hydrolyzes the triglyceride in circulating lipoproteins such as chylomicrons and VLDL and produces free fatty acids that are used for metabolic energy or for fat storage. Endothelial cells do not synthesize LPL; rather myocytes and adipocytes produce it. Thus, it is a protein that requires transcytosis across the endothelial cell barrier, in this case from the interstitial fluid to the luminal side of the cells.
There are several possible ways that LPL could cross the endothelial barrier. Nonspecific transport of molecules across endothelial monolayers occurs either via paracellular routes between the cells or via vesicular transit through cells (2). Alternatively, a specific transcytosis pathway could exist which requires LPL to associate with a cell surface receptor and then transports LPL through the cells. This process would be analogous to that which transfers IgA across epithelial cells (3). The first step in a specific LPL transcytosis pathway would involve LPL interaction with the basolateral side of endothelial cells. LPL binds to a number of cell surface molecules including heparan sulfate proteoglycans (HSPGs) and members of the LDL receptor family (4). In bovine endothelial cells the most highly expressed of these receptors is the VLDL receptor (VLDLr) (5). A previous study suggested that HSPGs are required for LPL transcytosis (6). It is, however, unclear whether HSPGs are sufficient for transport or whether HSPGs must operate in concert with receptors. The binding of LPL to several members of the LDL receptor family leads to uptake and degradation of LPL by cells. There are no data on whether these receptors participate in transendothelial movement of LPL or other ligands.
In this report, we present data showing that LPL transcytosis across endothelial monolayers requires both HSPGs and the VLDLr. LPL transcytosis was diminished by removal of HSPGs and inhibition of receptors by RAP, a 39-kDa protein that was copurified with the LDL receptor-related protein (LRP) (7). This protein binds to members of the LDL receptor family and inhibits ligand binding and uptake by those receptors (8,9). Furthermore, antibodies against the VLDLr blocked LPL translocation and increased expression of this receptor-increased transcytosis. Thus, LPL requires both HSPGs and receptors for translocation across endothelial cells.

EXPERIMENTAL PROCEDURES
Purification and Radioiodination of LPL-LPL was purified from unpasteurized bovine milk according to the method of Socorro et al. (10) with modifications as described by Saxena et al. (11). 300 -500 g/ml purified enzyme was stored at Ϫ70°C. Enzyme activity was assayed with a glycerol-containing triolein emulsion as described previously (11). The purified enzyme had a specific activity of 40 -50 mmol of oleic acid released/h/mg of enzyme at 37°C.
LPL was radioiodinated enzymatically with glucose oxidase and lactoperoxidase (12). Radioiodinated LPL was purified by heparin-agarose (Bio-Rad) affinity chromatography and stored at Ϫ70°C. Typical specific activity of the preparation was 1,000 -2,000 cpm/ng, and Ͼ90% of the radioactivity was precipitated with trichloroacetic acid. 125 I-LPL was purified by Sephadex G-25 gel filtration (PD-10, Amersham Pharmacia Biotech) prior to use to remove degradation products. Heatinactivated LPL was prepared by heating LPL for 1 h at 52°C.
Endothelial Cell Monolayers-Primary cultures of bovine aortic endothelial cells (BAECs) were established as reported (13) and were grown in DMEM containing 10% fetal bovine serum (Gemini Bioproducts Inc., Calabasas, CA), 1% (v/v) penicillin and streptomycin solution, and 1% (v/v) glutamine solution (both from Life Technologies, Inc.). Polarized BAEC monolayers were grown on gelatin and fibronectincoated polyethylene terephthalate 10-mm filters (pore diameter, 3.0 m) (Becton Dickinson Labware, Franklin Lakes, NJ). This allowed access to both the basolateral side of the cells adjacent to the lower chamber and the apical cell surface in contact with medium in the upper chamber (14). Approximately 4 -5 ϫ 10 4 cells were seeded onto the filters. Experiments on nonviral infected cells were conducted 5-6 days after seeding the endothelial cells. The media in the upper chamber (0.5 ml) and lower chamber (1 ml) were changed every other day. Movement of both [ 3 H]dextran (70 kDa, American Radiolabeled Chemicals, St. Louis, MO) and LDL was routinely assessed to verify that the cell monolayer was intact.

125
I-LPL Transport across Monolayers-RAP-sensitive transport of 125 I-LPL across the monolayers was studied after adding 1 g/ml radiolabeled LPL to DMEM and 1.5% BSA in the basolateral chambers. In some experiments, the basolateral side of the cells was incubated with 2.5 units/ml heparinase/heparitinase (Sekagaki America Inc., Bethesda, MD) or 5 g/ml RAP-containing medium. After 1 h, the medium was removed, and the cells were washed prior to adding 125 I-LPL. 125 I-LPL that appeared in the upper chamber was monitored over time by removing 100 l of medium from the apical side of the cells at 30, 60, 120, and 180 min. The chambers were not stirred to avoid disruption of the monolayers. All experiments were performed using triplicate chambers. At the conclusion of the experiments, the cells were washed, and 125 I-LPL associated with luminal and basolateral surfaces was released by the addition of DMEM-BSA containing 100 units/ml heparin (Elkins Sinn, Cherry Hill, NJ) at 4°C for 30 min to upper and lower chambers. Intracellular 125 I-LPL was then estimated by dissolving the heparintreated cells in 0.1 N NaOH and by measuring the radioactivity. In other experiments, heparin was added to the basolateral chamber along with the 125 I-LPL, and the appearance of 125 I-LPL protein in the upper chamber was determined. The radioactive protein was routinely immunoprecipitated with 10% trichloroacetic acid; in all experiments less than 10% of the counts in the chambers were not precipitated Transport of LPL Activity across Endothelial Cell Monolayers-To study the transport of LPL activity, 100 g/ml LPL purified from bovine milk as described earlier was added to the basolateral medium under different conditions. Because LPL rapidly loses its activity during a 37°C incubation and is less stable in more dilute solutions, this higher concentration of LPL was required to measure transport of its activity. In addition, the activity was measured at an earlier time (1-2 h). In some experiments, LPL was added together with 5 g/ml RAP or 5 units/ml heparin or after a 1-h incubation of the cells with RAP and/or VLDLr antibodies. 100 l of medium from the upper chamber was collected at 60 and 120 min and immediately frozen at Ϫ70°C. The samples were subsequently defrosted and assayed together. For these measurements we used a more sensitive LPL activity assay described by Hocquette et al. (15), and all assays were performed in triplicate. 20% Intralipid (Amersham Pharmacia Biotech) was diluted with an equal amount of deionized water to produce a 10% soybean oil emulsion. To incorporate the radiolabel, the emulsion was sonicated (75 W, 10 min, 50% pulse mode) with 378 Ci of [ 3 H]triolein (specific activity, 21 Ci/mmol, Amersham Pharmacia Biotech). 30 l of cell culture medium was incubated for 1 h at 25°C with 10 l of the Intralipid emulsion, 10 l of heat-inactivated serum as a source of apolipoprotein C-II, 100 l of incubation buffer (0.2 M NaCl, 0.3 M Tris-HCl, pH 8.5, 6% fatty acid-free bovine serum albumin, Sigma fraction V), 2 units of heparin/ml, and 50 l of deionized water. The reaction was terminated by the addition of 0.5 ml of deionized water and 2 ml of a solution containing isopropyl alcohol/heptane/H 2 SO 4 (48:48.3:1 v/v/v). After a 3-min centrifugation at 3,000 ϫ g 800 l of the upper phase was transferred to new tubes containing 3 ml of heptane and 1 ml of of alkaline ethanol, water, 2 M NaOH (500:475:25 v/v/v). A second centrifugation was performed as above, and the upper hydrophobic phase was replaced by 3 ml of heptane. The samples were then centrifuged again in the same manner. The radioactivity in 800 l of the remaining basic hydrophilic phase was determined using 3.5 ml of scintillation fluid (Ecoscint H, National Diagnostic, Atlanta, GA) in a model 1800 liquid scintillation counter (Beckman Instruments).
Effects of RAP and Antibodies on LPL Transport-RAP was produced as a fusion protein with glutathione S-transferase in an expression system utilizing human placental RAP cDNA (7). Antibodies against RAP (Rb80), LPL and blocking antibodies against the LDL receptor, VLDLr, and LRP were described previously (16 -19).
Adenovirus Expression of VLDLr-For expression of VLDLr in cells, BAECs were infected with adenovirus-containing human VLDL receptor (AdhVLDLr) and ␤-galactosidase-expressing adenovirus (AdLacZ) when the cells were 80 -90% confluent; experiments using these cells were conducted 24 h after infection. The barrier function of the endothelial cell monolayers was examined using trypan blue and dextran transport as described previously (6).
Ligand Blotting Analyses-Cell extracts from control, AdLacZ-, and AdhVLDLr-infected BAECs were prepared as described (20). Cells on 10-mm filters were dissolved in 0.05 ml of ice-cold solution containing 50 mM HEPES, pH 7.4, 0.5 M NaCl, 0.05% Tween 20, 1% Triton X-100, 1 mM phenylmethysulfonyl fluoride, 25 g/ml leupeptin, and 2 g/ml pepstatin. The cell extract was sheared with a 21-gauge needle and then centrifuged at 14,000 rpm for 10 min. The supernatant of each condition in triplicate was collected and pooled. For RAP ligand blotting, an aliquot (10 l) was run on 5% SDS-polyacrylamide gel electrophoresis under nonreducing conditions and electrophoretically transferred to a nitrocellulose membrane. The membrane was incubated with 25 nM RAP in phosphate-buffered saline containing 3% nonfat milk, 0.05% Tween 20, and 5 mM CaCl 2 for 1 h at 25°C (blocking buffer). The membrane was then incubated with anti-RAP IgG (Rb80, 1 g/ml) in blocking buffer for 1 h at 25°C and washed three times in phosphatebuffered saline containing 0.1% Tween 20. The membrane was then incubated with a donkey anti-rabbit IgG horseradish peroxidase conjugate (Bio-Rad) for 1 h at 25°C. After washing, the bands were visualized using the ECL kit (Amersham Pharmacia Biotech).

Transport of 125 I-LPL across Endothelial Cell
Monolayers-We first assessed how increasing amounts of LPL added to the basolateral side of BAECs affected the amount of LPL that crossed the monolayer. As shown in Fig. 1A, increasing 125 I-LPL in the medium on the basolateral side of the cells led to more 125 I-LPL in the upper chamber. However, when the 125 I-LPL concentration exceeded 1 g/ml, a lower percentage of the LPL was transported. For that reason, 1 g/ml was used for subsequent experiments.
To determine the specificity of LPL transport, we used unlabeled LPL to inhibit 125 I-LPL transport. The addition of 300 g/ml LPL to the lower chamber in the presence of 1 g/ml 125 I-LPL decreased the appearance of 125 I-LPL in the upper chamber to 46 Ϯ 5% of control (Fig. 1B). The amounts of 125 I-LPL within the cells and released from the apical surface are shown in Fig. 1, C and D. More 125 I-LPL was within the cells than on the cell surface. Increasing amounts of 125 I-LPL in the lower chamber led to more labeled LPL in the cells and on the apical surface. In the presence of an excess of unlabeled LPL, both the intracellular and cell surface-labeled LPL were decreased by Ͼ50%. Thus cellular uptake and transfer to the apical surface were inhibited, but not completely.
Effects of Modulation of HSPGs on LPL Transport-Although the role of HSPGs and members of the LDL receptor family as LPL-binding molecules is well documented, only limited studies (22,23) have been performed with respect to their role as transcytosis molecules. Because members of the this family of receptors often act in concert with HSPGs (22, 23), we tested whether LPL movement across monolayers requires association with HSPGs. 125 I-LPL translocation across BAEC monolayers was studied in the presence of heparin. As shown in Fig. 2, heparin at 5 units/ml decreased 125 I-LPL in the upper chamber after 3 h by Ͼ54%. Inhibition also was found with a higher dose of heparin (50 units/ml).
To test whether HSGP degradation affected LPL transfer, HSPGs on the basolateral side of the cells were removed by incubating the cells with HSPG-degrading enzymes. The results shown in Fig. 2, inset, demonstrate that heparinase/heparinitase treatment of the basolateral side of the cells reduced 125 I-LPL movement across the monolayers, a 41% inhibition. Thus, optimal LPL movement across the monolayers required its association with proteoglycans on the basolateral side of the endothelial cells. However, a large amount of the radiolabeled LPL was not affected by unlabeled LPL, heparin, and heparinase. This suggested that the radiolabeled LPL was tracing two different transport pathways only one of which was reduced by unlabeled LPL and inhibition of LPL-HSPG interaction.
Effects of RAP on LPL Transport across Endothelial Cell Monolayers-We next determined whether HSPGs are sufficient for LPL transcytosis or whether they serve as accessory molecules for transcytosis receptors. To test whether LRP, VLDLr, or other members of this family are involved in LPL transport, 1 g of 125 I-LPL was added to the lower chamber of monolayers that had been incubated for 1 h with RAP. The RAP had been added to the lower chamber in various concentrations, 0.33-10 g/ml, to produce a molar ratio of 1:1-1:30 of LPL to RAP assuming that LPL is an ϳ120-kDa dimer. In a dose-dependent manner, RAP decreased 125 I-LPL movement from the lower to the upper chamber (Fig. 3A). The highest doses 5 g/ml (shown as filled squares) and 10 g/ml (shown as open triangles) reduced the amount of 125 I-LPL Ͼ40%; 3.3 g (shown as filled circles) reduced it Ͼ30%. The maximum reduc-tion of 125 I-LPL transport with 5 g/ml was 40 -80% depending on the experiment. Moreover, the percentage of inhibition appeared greater at 2 than at 3 h; Ͼ50% inhibition was found using the two highest concentrations of RAP.
We next tested whether the RAP-and heparin-mediated inhibition were additive. As shown in Fig. 3B, when heparin was added to the RAP-treated cells the inhibition of LPL transport was similar to that found using only RAP. The addition of excess unlabeled LPL also did not lead to more inhibition of transport than that found with RAP alone. This suggests that the three interventions, RAP, excess unlabeled LPL, and heparin, were inhibiting the same pathway.
To assess whether our interventions affected movement of a control molecule, 1 g/ml dextran (molecular weight 70,000) was added to the basolateral side of the cells, and its movement over the ensuing 3 h was assessed. Dextran movement to the upper chamber is shown in Fig. 3B; this was not affected by heparin or the addition of RAP.
Movement of LPL from the apical to the basolateral side of the monolayers was also studied in the presence and absence of RAP and using cells that were treated with heparinase/heparitinase. LPL movement was much greater in this direction (Fig. 3D, compare the total transport across the cells with that in Fig. 3A). As had been found for movement in the opposite direction, both RAP and removal of glycosaminoglycan chains from HSPGs decreased the amount of 125 I-LPL appearing in the medium on the basolateral side of the cells.
Effects of RAP and Heparin on Cell Surface and Intracellular 125 I-LPL-The amount of radioactive LPL within the cells and on the apical surface was determined after a 3-h incubation Ϯ RAP (Fig. 4). RAP reduced the amount of heparin-releasable Effects of Heparin and RAP on LPL Activity Transport across Endothelial Cell Monolayers-To investigate the effect of RAP and heparin on the appearance of LPL activity on the apical side of the endothelial cell monolayer, 100 g/ml purified LPL was added to the lower chamber in the absence or presence of 5 units/ml heparin, 5 g/ml RAP, or to monolayers in which the basolateral side of the cells was treated with 5 g/ml RAP. As shown in Fig. 5  (data not shown). Therefore, RAP and heparin markedly decreased the amount of LPL activity transferring from the lower to the upper chamber. This effect of RAP and heparin on LPL activity transcytosis was much greater than that on 125 I-LPL transport.
Transcytosis of Heat-inactivated LPL-One possible reason for the greater effect of RAP on LPL activity than 125 I-LPL was that inactive LPL protein was transported across the monolayers by a non-RAP-inhibited process. It should be noted that during these experiments, some iodinated LPL would have been converted to inactive monomer, and the data using this tracer would assess both active and inactive LPL. To determine whether inactivated LPL was transported in a manner similar to that of active dimeric LPL, heat-inactivated 125 I-LPL was studied. These preparations have been characterized previously and consist primarily of inactive LPL that elutes from heparin at a lower salt concentration and is thought to be monomeric (23). Our preparation was assessed by SDS-polyacrylamide gel electrophoresis and consisted primarily of an ϳ55-kDa protein; however, ϳ 80% of this preparation eluted from heparin affinity gel with 0.5 M NaCl-containing buffer. This is in contrast to nonheated LPL, in which Ͻ20% of the 125 I-LPL eluted at this salt concentration. As shown in Fig. 6, heat-inactivated 125 I-LPL (denoted Inactive LPL) was transferred from the lower to the upper chamber nearly twice as fast as nonheated 125 I-LPL (denoted Active LPL). Moreover, this transfer was unaffected by either the addition of RAP along with the LPL, or preincubation of the endothelial cells with RAP. Therefore inactive 125 I-LPL crosses the monolayer at a greater rate via a non-RAP-sensitive pathway.
Effects of Antibodies to LPL and Receptors-Antibodies (IgG, 30 g) against LPL or members of the LDL receptor family were added to the lower chamber prior to the addition of 125 I-LPL. These amounts of antibodies were sufficient to inhibit the respective receptors (24) or to bind to each LPL molecule. As shown in Fig. 7A, anti-VLDLr and anti-LPL antibodies inhibited transcytosis by ϳ50%; these data are shown in the open squares and open inverted triangles, respectively. In contrast, antibodies to the LDL receptor (open circles) and LRP (filled triangles) had no effect. The same antibody treatment was used to assess the role of the VLDLr in transport of active LPL (Fig.  7B). Anti-VLDLr antibodies decreased LPL movement to the apical side of the cells by Ͼ80%. This inhibition was comparable to that found with RAP; RAP and anti-VLDLr antibody did not have an additive effect on inhibition of transport. Therefore, inhibition of the VLDLr decreased LPL transport across the cultured endothelial cells.
VLDLr Overexpression Increases LPL Transport across Endothelial Cell Monolayers-We next tested whether more VLDLr expression increases LPL transcytosis. BAECs were infected with either AdhVLDLr or AdLacZ, and the expressed VLDLr was examined by RAP ligand blotting of membrane extracts (Fig. 8A). In control and infected cells a strong band for a protein of M r ϳ120,000 which corresponded to the VLDLr was found. The identity of this band was confirmed using anti-VLDLr antibodies. Only AdhVLDLr-infected cells had an intensely staining human VLDLr band, seen in Fig. 8A, lane 3. BAECs express a lower molecular weight form of the VLDLr than human cells (5). Two other less intense high molecular weight bands were observed. The second band reacted with anti-LRP antibodies; the highest molecular weight protein was, presumably, megalin.
As expected, VLDLr expression increased LPL transport. A time course of 125 I-LPL transcytosis across control and infected BAECs is shown in Fig. 8B. Cells infected with AdhVLDLr are VLDLr overexpression increased 125 I-LPL transport Ͼ 60% in BAECs after 180 min. RAP completely blocked the increased transcytosis due to AdhVLDLr infection and reduced the transport to below that of control cells. These data confirm that LPL transport across endothelial cells can be mediated by the VLDLr. Fig. 8, C and D, shows the effects of VLDLr overexpression on apical cell surface and intracellular 125 I-LPL, respectively. BAECs infected with AdhVLDLr had more intracellular and more cell surface 125 I-LPL than control cells; 125 I-LPL on the apical surface was approximately doubled, and that inside the cells was increased Ͼ60%. This effect was completely blocked by RAP. Thus, VLDLr overexpression increased cellular uptake and transfer of LPL across the monolayer. DISCUSSION Our experiments demonstrate that LPL transport from the basolateral to the apical side of cultured endothelial cell monolayers is modulated by two processes: LPL binding to HSPGs and interaction with the VLDLr. Our previous (6) and current experiments confirm the participation of HSPGs in the LPL transcytosis process. As has been shown for endocytosis, our experiments suggest that the role of HSPG binding is to concentrate the LPL on the cell surface and increase the efficiency of its binding to cell surface receptors.
Our data suggested that receptors, especially the VLDLr, were involved in LPL movement across the monolayers. 1) Incubation of the monolayers with RAP decreased the transfer of radioactive LPL from the abluminal to luminal side of the cells. 2) Transfer of LPL activity to the apical side of the cells was inhibited to an even greater degree by RAP. 3) Antibodies against the VLDLr decreased LPL transfer. 4) Overexpression of the VLDLr in endothelial cells increased LPL movement.
Excess of unlabeled LPL markedly reduced 125 I-LPL transport across BAEC monolayers. However, we noted that the level of competition did not exceed 50%. Furthermore, when heparin, RAP, or VLDLr antibodies were used to block 125 I-LPL transcytosis, the effect was no greater than 60% in most of the experiments. Important for the interpretation of these experiments is the observation that purified LPL rapidly converts to an inactive monomeric form during a 37°C incubation (26). For that reason, our experiments using 125 I-LPL assessed the transfer of both active and inactive LPL. Although this complicated both the experiments and interpretation, we were still able to define cellular requirements needed for this process. Two types of experiments clarified this situation. Studies were performed using heat-inactivated 125 I-LPL, and others assayed movement of LPL activity. Heat-inactivated LPL transferred across the monolayer more rapidly, and its transport was not decreased by RAP. In contrast, LPL activity transcytosis was almost completely blocked by heparin, RAP, or VLDLr antibodies. Therefore, monomeric and active dimeric LPL differed in their transport. Active LPL transcytosis occurs via a HSPGrequiring and RAP-sensitive pathway, whereas presumably smaller inactive 125 I-LPL monomers transfer more quickly through nonspecific paracellular or transcellular routes.
Both HSPGs and the VLDLr modulated intracellular and cell surface LPL. Much more 125 I-LPL was found inside than on the apical surface of the cells. This was not surprising because endothelial cells internalize and recycle LPL (25), and some of the intracellular LPL may have been destined for recycling. All of the interventions that decreased the amount of 125 I-LPL in the upper chamber medium also decreased apical cell surface and intracellular 125 I-LPL: heparin, heparinase, excess unlabeled LPL, RAP, and anti-VLDLr antibodies. Therefore, movement of LPL to both the surface and into the medium is likely to be via the same process. We postulate that internalized LPL translocates across the cell only if it is targeted via the initial VLDLr uptake.
Endothelial cells, unlike other cells, degrade very little internalized LPL (25). A similar observation was noted by Friedman et al. (27) and Argraves et al. (20). We tested the BAECs used in these experiments and found that like porcine cells (32) Ͻ5% of cell surface 125 I-LPL was degraded by the cells. Although the VLDLr mediates LPL degradation in other cells, receptors in this family do not always lead to degradation of internalized ligands. It has been shown recently that hepatic uptake of apoE leads to recycling of this apoprotein (28). Like LPL, apoE is a strong heparin-binding protein that also interacts with LRP. Thus, it is likely that the cell type determines whether these receptors participate in protein transcytosis, retroendocytosis, or degradation. In agreement with this hypothesis, it has been reported recently that gp330 is responsible for transcytosis of thyroglobulin by thyroid cells grown on filters (29) even though this receptor usually leads to lysosomal degradation of ligands.
IL-8 is another protein that binds to both proteoglycans and cellular receptors (30,31). Unlike LPL, this protein enters the bloodstream via the postcapillary venules and then circulates in the blood. A histological study (32) showed that injection of heparanase blocked IL-8 transcytosis, i.e. the IL-8 remained in the injected tissues presumably because it was unable to bind to proteoglycans, a step required for interaction with the IL-8 receptor. Our current studies suggest that a similar process is required for LPL. Although our observations in cultured BAECs implicate HSPGs and VLDLr as components of a LPL transport system one must question whether these observations are of physiologic importance for LPL actions. Muscle and adipose tissue are the two most important sites of LPL-medi- FIG. 7. Effects of anti-receptor antibodies on LPL transport. Panel A, antibody effects on 125 I-LPL transport. Endothelial cell monolayers were incubated with different antibodies (30 g/ml) added to the medium on the basolateral side for 1 h at 37°C after which 1 g/ml 125 I-LPL was added, and 125 I-LPL movement to the upper chamber was determined. Panel B, antibody effects on LPL activity transport. Endothelial cell monolayers were incubated Ϯ 30 g/ml anti-VLDLr antibody, 5 g/ml RAP, or anti-VLDLr antibody and RAP together. After 1 h, 100 g/ml purified bovine LPL was added to the lower chamber, and the cells were incubated at 37°C for 1 h. LPL activity in the upper chamber was determined. Values represent the average Ϯ S.D. of two experiments performed in triplicate. ated hydrolysis of lipoprotein triglyceride. These are also sites of abundant VLDLr expression (17). This receptor is highly expressed in endothelial cells (33) and, therefore, is in the right tissues and cells to transport LPL. Moreover, the VLDLr is modulated by feeding and fasting, and its expression parallels changes in LPL activity in those tissues (34).
The observation that RAP inhibits the transcytosis of LPL to its site of action is consistent with and explains several in vivo observations. Overexpression of RAP in mice consistently increased circulating triglyceride levels; increased numbers of remnant lipoproteins were found only in LDL receptor knockout mice (35). RAP-induced hypertriglyceridemia was associated with an increase in the size of circulating triglyceridecontaining lipoproteins (36). In mice lacking LRP in the liver, RAP overexpression reduced postheparin LPL activity (21). Therefore, RAP has an action other than inhibiting chylomicron remnant uptake via liver LRP. In experiments not shown we have found that RAP neither inhibits LPL binding to the luminal side of endothelial cells nor inhibits LPL-mediated hydrolysis of emulsion or VLDL triglyceride, an observation also made by others (21). We propose that inhibition of LPL transcytosis is the cause of RAP-induced hypertriglyceridemia.
One seemingly inconsistent piece of data is the lack of an abnormal lipoprotein phenotype in VLDLr knockout mice (37). These mice, however, are less obese than their littermates. Of note, mice not expressing LPL in fat also have a normal lipoprotein phenotype, but when crossed with ob/ob mice they are less obese (38). One possible explanation for the lack of hypertriglyceridemia in the VLDLr knockout mice is that LPL transcytosis is not rate-limiting. It should be noted that postheparin plasma LPL activity in mice is ϳ5 times greater than in humans (39), therefore a fraction of normal LPL activity may be sufficient for normal lipolysis. In support of this hypothesis is a recent report showing that mice deficient in both VLDLr and the LDL receptor develop hypertriglyceridemia when they are placed on a high fat diet (40). Thus, when mice have greater lipoprotein flux through the circulation lipolysis may be limited by VLDLr deficiency.
In previous studies from this laboratory we studied LPL interaction with the surface of endothelial cells. With a number of colleagues, we reported that LPL binds to a 200-kDa transmembrane heparan sulfate proteoglycan on the luminal surface of endothelial cells (41). This cell surface proteoglycan is syndecan 4. LPL binding is mediated by the glycosaminoglycan chains on the proteoglycans, and a highly sulfated decasaccharide is required for optimal LPL binding (42). We also reported that LPL bound to a 116-kDa protein that was not a proteoglycan (43). A protein was isolated and sequenced and was an amino-terminal fragment of apolipoprotein B (44); this portion of apoB associates with LPL and heparin (45). The VLDLr has a molecular mass of 120 kDa (5). It is likely that our attempts to identify LPL-binding proteins by ligand blotting and LPL affinity chromatography missed the VLDLr because of its size similarity to the more abundant apoB fragment.
In summary, we have shown that LPL transcytosis across endothelial cells is inhibited by reducing LPL interaction with HSPGs and by blocking the VLDLr. Overexpression of the VLDLr increases this process. In addition, a second pathway that allows for movement of inactive LPL molecules also exists and is not sensitive to RAP or heparin. These data suggest that HSPGs participate with members of the LDL receptor family to internalize and transfer ligands from the interstitial space to the circulation. This process may be important for the extracellular transport pathway of active LPL. Similar receptormediated pathways are likely to be involved in the physiological functioning of a number of other proteins that must transfer from parenchymal cells to the circulation. Acknowledgment-We are indebted to Dr. Joachim Herz who suggested a number of the preliminary experiments from which these studies originated and who shared unpublished data on the in vivo effects of RAP on LPL and lipoprotein metabolism.