INTRODUCTION

Proteoglycans (PG)¹ contain a core protein and highly charged carbohydrate chains, glycosaminoglycans (GAG) (1, 2). The three major classes of GAG are heparan sulfate (HS), dermatan sulfate, and chondroitin sulfate. HSPG are ubiquitous components of cell membranes and the extracellular matrix (3, 4) and have important functions as structural proteins and as cell surface receptors. Moreover, portions of HS function as cofactors for enzymatic reactions. An example of this is the modulation of coagulation by heparin (5). HS have also been hypothesized to modulate cell proliferation (6, 7). In general, the specificity of the actions of HSPG are determined by the carbohydrate sequences within the GAG chains. Five- to twelve-unit oligosaccharides that modulate the actions of antithrombin (8) and bind basic fibroblast growth factor (bFGF) (9), hepatocyte growth factor (10), and lipoprotein lipase (LpL) (11, 12) have been characterized.

LpL has an unusual extracellular transport that, in part, involves its association with HSPG. LpL is synthesized primarily in adipocytes and myocytes but hydrolyzes triglycerides in circulating lipoproteins while bound to the luminal surface of endothelial cells (13, 14). Much of the physiological regulation of LpL activity occurs without changes in LpL protein (15). Examples of this include LpL regulation by feeding/fasting (16, 17) and the increase in adipose LpL by insulin (18-20). Unlike most secretory proteins, newly synthesized LpL appears to transiently reside on cell surfaces, e.g. adipocytes (21, 22). LpL is a very unstable protein in solution (23). How LpL is able to maintain its activity after its release from adipocytes and during its transport to endothelial cells is unclear.

Within cells, a number of unstable proteins are protected by being associated with a second, chaperone, molecule (24, 25). Most chaperones are proteins that function by binding specifically and non-covalently to unstable proteins preventing their misfolding and degradation. LpL, however, has no known chaperones involved in maintaining its activity and facilitating its transport.

While investigating the mechanisms for LpL release from adipocytes, we discovered that endothelial cells produce an HS-degrading heparinase-like enzyme that dissociates the LpL from cultured adipocytes. The released LpL was associated with an oligosaccharide that made it more stable in solution. In contrast to heparin, this oligosaccharide did not inhibit LpL binding to endothelial cells. Thus, the oligosaccharide serves as a chaperone for the LpL. We hypothesize that other HS oligosaccharides may serve a similar function to aid in the transport and stabilization of heparin binding proteins.

MATERIALS AND METHODS

Bovine endothelial cells were isolated and cultured as described (26). The cells (5-15 passages) were grown in minimal essential medium (MEM) containing 10% fetal bovine serum (Life Technologies, Inc.). Brown fat cells (BFC-1B) were grown on collagen (type III, Sigma)-coated six-well tissue culture plates (Falcon, Becton Dickinson, Lincoln Park, NJ) as described previously (27). The cells were cultured in Dulbecco's modified Eagle's medium containing 15 mM Hepes buffer, 100 units/ml penicillin G, 100 µg/ml streptomycin, 17 µM pantothenate, 33 µM biotin, 1% glutamine, and 10% (v/v) fetal bovine serum. Two days after seeding, the confluent cells were converted to adipocytes by incubation in differentiation medium (the same medium supplemented with 10 nM insulin and 2 nM triiodothyronine). The medium was changed every 1-2 days, and the cells were used for experiments at 14-20 days after confluence. To confirm that the cells were producing LpL, LpL activity was measured before each experiment using a glycerol triolein emulsion (28) as described previously (29).

LpL was purified from fresh bovine milk as described previously (29) following the method of Socorro et al. (30) and stored at 70 °C. Purified LpL was radioiodinated (29) using lactoperoxidase and glucose oxidase enzymes (Sigma). The reaction mixture containing purified LpL (800 µg), 20 mM glucose, 1 mCi of Na¹²⁵I (Amersham Corp.), lactoperoxidase 7.5 mg/ml, and glucose oxidase 1 mg/ml was incubated in 50 mM of Tris-HCl buffer, pH 7.4, for 5 min on ice. The mixture was then applied to a column containing 3 ml of heparin-agarose gel (Bio-Rad). The column was washed first with 20 mM Tris-HCl buffer, pH 7.4, containing 0.4 M NaCl and then with 20 mM Tris-HCl buffer, pH 7.4, containing 0.75 M NaCl. Radioiodinated LpL was then eluted with 20 mM Tris-HCl buffer, pH 7.4, containing 1.5 M NaCl and 0.1% BSA. ¹²⁵I-LpL was stored at 70 °C. A typical preparation had a specific activity of about 800-1000 cpm/ng LpL and greater than 90% of the counts were precipitable by 10% trichloroacetic acid.

Heparitinase (heparan sulfate lyase or heparanase) and heparinase (heparin lyase) purified from Flavobacterium heparinum was obtained from Seikagaku America Inc., Bethesda, MD.

Confluent monolayers of endothelial cells were washed with MEM and incubated with MEM, 3% BSA for 16 h. The medium (ECCM) was collected and filtered through a 0.8-µm filter to remove cell debris and used to release LpL from BFC-1 adipocytes. In other experiments lysophosphatidylcholine was added to a final concentration of 50 µM to MEM, 3% BSA and then incubated with endothelial cells for 16 h, and medium (lyso-ECCM) was collected (31).

Release of LpL activity and ¹²⁵I-LpL was studied. To assess the effects of ECCM, adipocyte monolayers were incubated with MEM, 3% BSA or ECCM at 37 °C for 1 h. LpL activity released was assayed using triolein emulsions as described previously (29). ¹²⁵I-LpL binding to BFC-1 adipocytes was carried out as described previously (32). Heparin-releasable LpL associated with the cell surface was assessed after incubating the cells in 1 ml of MEM, 3% BSA containing 10 units/ml heparin.

Purified bovine LpL and adipocyte LpL released by heparitinase or by lyso-ECCM were incubated at 37 °C for up to 1 h. Aliquots at different time intervals were assayed for LpL activity. In other experiments purified bovine LpL was first incubated with HS or heparitinase-digested HS for 10 min on ice before incubating at 37 °C.

25 µg of purified bovine LpL was incubated with 8 µg of HS (bovine intestine, M_r = 40,000) for 20 min at room temperature. The complex was then incubated with 5 units of heparitinase for 2-3 h at 30 °C. The mixture was heat-denatured at 95 °C for 20 min. Denatured proteins were removed by centrifugation at 14,000 × g for 20 min. The supernatant containing the heparitinase-digested HS (HS oligo) was used for the experiments. For transport experiments ¹²⁵I-LpL was mixed with undigested HS or HS oligo and used (see below).

An LpL binding decasaccharide (Deca) of the following composition (IdceA(2-SO₄)-GlcNSO₄(6-SO₄))₃(IdceA(2-SO₄)-GlcNSO₄)(IdceA-GlcNSO₄(6-SO₄)) was isolated from commercial heparin (Ming Han, Oakland, CA) by partial depolymerization with nitrous acid, followed by Bio-Gel P-6 and polylysine-agarose chromatography (33).

Heparanase (HSPG degrading) activity was assayed using matrix prepared from ³⁵SO₄-labeled endothelial cells. Confluent monolayers of endothelial cells were incubated with ³⁵SO₄ containing medium for 48 h. Subendothelial matrix was prepared from labeled cells as described previously (34). Labeled matrix was incubated with ECCM, lyso-ECCM, or heparinase for 2-4 h at 37 °C, and released radioactivity was determined.

Adipocyte proteoglycans were labeled by incubating the cells for 24 h in medium containing 50 µCi/ml ³⁵SO₄ (32). Following removal of the radiolabeling media, the cells were washed and incubated with medium, ECCM, or lyso-ECCM for up to 4 h at 37 °C. Released ³⁵SO₄ was counted at different time points.

Transport experiments were done using tissue culture inserts (Falcon) in 6-well plates as described previously (34). Inserts were coated with 0.1% gelatin for 30 min at room temperature followed by a mixture containing collagen (50 µg per ml) and fibronectin (10 µg per ml) for 1 h at 37 °C. Unbound proteins were removed by washing with PBS followed by MEM, 3% BSA. Endothelial cells were plated (~3 × 10⁶ cells per well), and experiments were done 2 days after seeding. Endothelial cells cultured under these conditions are polarized and form a tight barrier (34). ¹²⁵I-LpL, ¹²⁵I-LpL + HS, or ¹²⁵I-LpL + HS oligo was added to the bottom chamber in Dulbecco's modified Eagle's medium, 3% BSA. ¹²⁵I label appearing in the upper and lower media was measured after 90 min.

To determine if heparanase and LpL activities correlate, epididymal fat was obtained from male rats in the morning (feeding) or 16 h after their food had been removed (fasting). Epididymal fat was isolated and homogenized (polytron) in 2 volumes of phosphate/citrate buffer, pH 6.5, containing 0.02% CHAPS. Cell lysates were centrifuged at 15,000 × g for 30 min at 4 °C. Supernatants were diluted with an equal volume of buffer and assayed for HSPG degrading activity using subendothelial matrix containing ³⁵SO₄-labeled proteoglycans. To specifically assess heparanase, lysate was mixed with 100 units/ml heparin before adding to matrix.

RESULTS

To determine if endothelial cells secrete factors that can release LpL from adipocytes, ECCM was added to adipocytes containing cell surface LpL and incubated for 1 h at 37 °C. Fig. 1 shows that, compared with control medium, ECCM released 2-fold more LpL activity into the medium. Previous studies from this laboratory (31) showed that lysolecithin, a product of lipolysis, stimulated endothelial cells to produce a heparanase-like activity. We therefore tested whether media from lysolecithin-stimulated endothelial cells would release more LPL. As expected, lyso-ECCM released 3-4-fold more LpL activity than ECCM. The amount of LpL released by lyso-ECCM was similar to that released by heparinase/heparitinase (1 unit/ml each). Although lysolecithin will release LpL from adipocytes, this is prevented by inclusion of albumin in the medium (27). Since addition of lysophosphatidylcholine directly to ECCM did not affect the amount of LpL released (not shown), the releasing factor could not be the lysolecithin itself. Therefore, endothelial cells secrete LpL releasing factors, and LpL release is greater using lyso-ECCM.

We previously showed, by size analysis of the HS degradation products generated from subendothelial matrix HSPG, that lyso-ECCM contains an HSPG degrading activity similar to heparinase (31). To determine whether the heparanase-like activity in lyso-ECCM would degrade adipocyte cell surface PG, adipocyte PG were labeled with ³⁵SO₄ and incubated with control medium, ECCM or lyso-ECCM (Fig. 2). Bacterial heparitinase was used as a control. ³⁵S in control medium, representing the secreted pool of adipocyte proteoglycans, was subtracted to specifically determine the effects of conditioned media and heparitinase. Incubation with ECCM led to a small increase in the amount of labeled PG appearing in the medium. Lyso-ECCM released even more ³⁵S label into the medium, and this amount of label was identical to that released by heparitinase. These data show that the heparanase-like activity in lyso-ECCM will degrade cell surface HSPG.

Only heparanase secreted from the abluminal side of the endothelial cell will interact with adipocytes in vivo. We therefore tested whether heparanase was secreted from the basolateral side of polarized endothelial cells. Cells were grown in control or lysophosphatidylcholine containing medium for 12 h, and heparanase activity was assayed in media from top (luminal) and bottom (subendothelial) chambers (Fig. 3). In total, cells exposed to lysophosphatidylcholine had 2.5-3-fold more heparanase activity. In addition, the lyso-ECCM from the bottom chamber contained approximately 40% more activity per ml than the top chamber. When adjusted for the volume differences, the bottom chamber contained 3.5-fold more heparanase. If a similar situation exists in vivo, heparanase would be preferentially secreted into the subendothelial space.

To more directly test whether ECCM could release adipocyte LpL, we performed a coculture experiment. LpL release was studied during lipolysis on the endothelial surface, a process that produces lysolecithin (27), and with lysolecithin containing medium. Endothelial cells were grown to confluence on filters, ¹²⁵I-LpL was permitted to bind to cultured adipocytes, and filters containing endothelial cells were then placed above the adipocytes. To the endothelial cells in the top chamber, media were added containing 10 µg/ml unlabeled bovine LpL, 10 µg/ml VLDL, LpL + VLDL, or 50 µM lysolecithin. The cells were incubated for an additional 6 h, a time previously shown to result in heparanase production (31); ¹²⁵I-LpL appearing in the media of lower chamber (representing LpL release from adipocyte) and top chamber (LpL transport) was counted. Addition of unlabeled LpL to the endothelial cells did not affect the amount of ¹²⁵I-LpL released from the adipocyte cell surface or transported (Fig. 4). Addition of VLDL to endothelial cells led to a 15% increase in the amount of ¹²⁵I-LpL appearing in the top chamber. LpL + VLDL and lysolecithin increased LpL released into the media by >25% and increased transport to the upper chamber by 45 and 51%, respectively. These data suggest that lipolysis and lipolysis products stimulate endothelial cells to acquire LpL from adipocytes.

To effectively function on the luminal side of the endothelium in vivo, adipocyte LpL, following its release, must maintain its activity during its transport to endothelium. We, therefore, tested whether lyso-ECCM released LpL activity is stable. Adipocytes were incubated with either lyso-ECCM or heparitinase for 1 h at 37 °C. Media containing released adipocyte LpL were collected and then further incubated at 37 °C for up to 1 h. LpL activity was determined at different time intervals. Purified bovine LpL was used as a control. Purified LpL lost 60% of its activity in 15 min and >95% of its activity in 60 min when incubated in PBS (Fig. 5A). In contrast, lyso-ECCM released LpL lost very little (<2%) activity even after 60 min. Heparitinase released LpL was also stable and lost only 20-25% activity in 30 and 60 min.

We hypothesized that LpL, both heparitinase released and endothelial heparanase released, was able to maintain its activity because it was associated with a small piece(s) of HS oligosaccharide. In our experiments, we observed that when LpL was released by lyso-ECCM from ³⁵SO₄-labeled HSPG, a fragment of ³⁵S-labeled GAG could be copurified with LpL on heparin-Sepharose chromatography (not shown). To confirm that HS oligo, resulting from heparanase treatment, stabilizes LpL, LpL was mixed with undigested HS or HS oligo (generated as described under "Material and Methods") and incubated at 37 °C for 1 h. LpL even in 1% albumin-containing buffers lost approximately 65% of its activity in 60 min (Fig. 5B). As expected, in the presence of HS LpL retained >80% of its initial activity after 60 min. In addition, the smaller HS oligo was just as effective. These results further suggest that small HS oligo associated with LpL can stabilize LpL from inactivation.

We next tested whether LpL associated with HS oligos can be transported across the endothelium and interact with the luminal side of the endothelium. Because lyso-ECCM applied to adipocytes contains secreted proteoglycans and other adipocyte factors that might interfere with LpL transport in addition to released LpL itself, we used LpL·HS oligo complex, as generated above, for these experiments. ¹²⁵I-LpL, ¹²⁵I-LpL-HS, or ¹²⁵I-LpL-HS oligo were added to the bottom chamber of endothelial cell monolayers and incubated for 90 min at 37 °C. The amount of LpL that appeared in the upper chamber medium after 90 min was slightly less with HS compared with controls (~18%) and slightly more with HS oligo (128% of controls, not shown). HS also decreased the amount of ¹²⁵I-LpL associating with the cell surface after 90 min of incubation by approximately 26% (Fig. 6). HS oligo increased the amount of LpL associating with the cell surface by more than 2-fold. Thus, these data show that HS oligo not only stabilizes LpL activity but also promotes LpL transport across and association with endothelial cells.

The size of the HS oligo is not known. The above experiments suggest that, despite the well known fact that heparin dissociates LpL from endothelial cells, small HS-derived oligos do not prevent LpL binding to endothelial cells. To directly test this, we used a previously described HS decasaccharide (Deca, approximate molecular mass of 2800 Da) that has a high affinity for LpL (11). We now tested if Deca also facilitates LpL binding to endothelial cells and stabilizes its activity (Fig. 7). Unfractionated low molecular weight heparin (molecular mass of 3000 Da) was used as a control. ¹²⁵I-LpL was mixed with different concentrations of heparin or Deca (normalized for the amount of uronic acid) and incubated with endothelial cells, and cell surface binding was assessed. LpL binding was inhibited in a dose-dependent manner by heparin and at 100 ng of uronic acid concentration, heparin inhibited LpL binding by >75%. In contrast, Deca even at higher concentrations of uronic acid (up to 1000 ng) did not inhibit LpL binding. In addition, Deca also stabilized LpL activity (inset). Therefore, HS-oligo containing as few as 10 saccharides can stabilize LpL without inhibiting LpL association with endothelial cells.

Our hypothesis is that lipolysis on the endothelial surface leads to production of endothelial cell heparanase and that this enzyme, in turn, permits more LpL to transit from adipocytes to its stable binding site on the endothelial cells. As a first experiment to test this hypothesis, we assessed whether heparanase activity in adipose tissues is modulated in vivo in a manner consistent with this hypothesis. Epididymal fat was isolated from feeding and overnight fasting rats. Tissue was homogenized and tested for LpL activity and HSPG degrading activity. Since crude homogenates may contain other activities (such as proteases) that can degrade HSPG, we used heparin inhibition to specifically assess heparanase activity. Total adipose HSPG degrading activity was approximately 1.6-fold higher in feeding compared with fasting rat fat pads (Fig. 8). The heparin-inhibitable HSPG degrading activity (difference between open bars and closed bars, representing heparanase activity) was 2.1-fold higher in feeding rat adipose. A similar increase in LpL activity was also found after feeding (inset). These data show that adipose heparanase activity is modulated, and greater activity is found when LpL activity is increased.

DISCUSSION

One special aspect of the metabolism of LpL is that its actions require extensive extracellular transport. Unlike the traditional secretory protein, newly synthesized LpL appears to reside, at least transiently, on the surface of its cell of synthesis. Immunohistochemical studies (35) and experiments using cultured adipocytes (36, 37) showed that LpL is associated with the cell surface. More recently, experiments in LpL overexpressing Chinese hamster ovary cells demonstrated that cells that had a reduced amount of cell surface HSPG secreted more LpL into the media and had less LpL on the cell surface (38). Therefore, lack of HSPG prevents LpL association with the cell surface. One way adipocytes can lose cell surface HSPG in vivo is via the actions of heparanase. Our data show that endothelial cells can produce a heparanase that releases LpL from adipocytes and suggest that LpL's transit to the endothelial cells is initiated by the actions of this enzyme.

The greatest amount of LpL release was found using lyso-ECCM. Although lysophosphatidylcholine is often thought of as a product of lipoprotein oxidation, it is also a lipolysis product. Aside from its actions as a triglyceride hydrolase, LpL is a phospholipase with a predilection for fatty acids in the S_n1 position (39). Thus if our in vitro observations mimic in vivo physiology, the initial lipolysis of triglyceride-rich lipoproteins on the adipocyte capillary lumen should stimulate the production of the endothelial cell heparanase. This should, in turn, promote the release of more LpL from the adipocyte surface allowing the continuation of the lipolysis reaction. This new LpL is required because some of the endothelial LpL is released during this process (40). Our data with rat adipose is consistent with this hypothesis. More LpL and more heparanase activities were found in epididymal fat obtained from feeding than fasting rats. This further suggests, but does not prove, that LpL and heparanase activities are mutually regulated.

Several lines of evidence suggested that the releasing activity was a heparanase and not a protease. It released LpL in its active form, and the released LpL was stable over time. Concomitant with LpL release adipocyte cell surface PG were also released. The HSPG degrading activity was 1) inhibited by heparin and suramin, heparanase inhibitors (31), and 2) bound to a heparin affinity column and eluted at 0.75 M NaCl.² In addition, using ³⁵SO₄-labeled subendothelial matrix as a substrate we have shown that the degradation products released by lyso-ECCM were similar in size to those released by heparitinase (31).

LpL released by endothelial heparanase or heparitinase was able to maintain its activity even after separation from cells. This was most likely due to LpL association with a small piece of HS oligo. That HS oligo stabilizes LpL was confirmed using purified heparitinase-digested HS and Deca. We hypothesize that Deca is a protected fragment of heparanase digestion of adipocyte HSPG. The regions of adipocyte cell surface HSPG that are occupied by LpL are not readily accessible to heparanase, and therefore, unoccupied regions of HSPG may be preferentially cleaved leading to the release of LpL·Deca complexes. This piece of oligosaccharide can prevent LpL from denaturation during its course of travel from adipocyte to endothelial cell surfaces. For this reason, the LpL binding HS oligo is an "extracellular chaperone" for LpL.

Our data in Fig. 6 show that the HS oligo, in addition to stabilizing LpL activity, also promoted LpL transport across the endothelial cell barrier. We were, however, surprised by the observation that HS oligo + LpL complex was able to associate with endothelial cells better than control LpL. We also found that the LpL-binding Deca stabilized LpL, and at different concentrations (up to 1000 ng of uronic acid) did not inhibit LpL binding to endothelial cells. Thus, a similar size oligosaccharide may be released from adipocyte HSPG by heparanase. This oligo permits LpL to cross the endothelial barrier, whereas larger undigested HS impedes this process. The stabilized LpL that arrives on the luminal endothelial surface is then able to bind to the cells, whereas native LpL that has been inactivated during this process does not associate with these cells.

We tested this hypothesis using a coculture system (Fig. 4) and showed that even under the disadvantageous conditions used, LpL release from adipocytes was promoted when endothelial cells were stimulated with lysolecithin or with LpL-mediated lipolysis of VLDL. To maximize our ability to measure the LpL we used radioiodinated proteins. Although the magnitude of the increase in LpL release from the adipocytes (the amount found in the lower chamber) was only 25-30% greater using the stimulated endothelial cells, this was found in a system in which the endothelial cells and adipocytes were widely separated and contained 2.5 ml of intervening culture media. In vivo, these cells are almost juxtaposed and the secreted heparanase would not be diluted to a similar extent. Nonetheless, the amount of LpL in the upper chamber was increased almost 50%. These data are consistent with those in Fig. 6 that show more LpL-oligo on the upper (luminal) side of polarized endothelial cells after addition of the LpL to the lower chamber. It should be noted that a similar increase in LpL associated with the endothelial surface was not found in the coculture experiment shown in Fig. 4. This is presumably because any LpL associated with the cell surface endothelial cells was released into the medium by lysolecithin and other lipolysis products (32).

Why did the oligosaccharide not inhibit LpL binding to endothelial cells? This may be due to one or more of the following. 1) The HS oligo kept LpL in the right conformation to interact with endothelial cell surface HSPG. The HS oligo must be neither too large nor too charged to inhibit LpL binding to HSPG. A similar situation exists for FGF. FGF has a higher affinity for a hexasaccharide; however, a longer 14-unit oligosaccharide is required for its release from cells (41). 2) The HS oligo dissociated from LpL during the course of transport and before LpL interaction with endothelial HSPG. 3) Several heparin binding domains on LpL have been described (42-44). Therefore, some of these HS binding domains of LpL were available for interaction with endothelial surface HSPG even in the presence of the oligosaccharide. This, we believe, is the reason why low molecular weight heparin inhibited LpL binding and Deca did not. Heparin, although similar in size, contains a heterogeneous population of saccharides with different sequences. Some of these may bind, albeit nonspecifically, to different regions of LpL preventing its interaction with other proteins or HSPG. 4) HS oligo·LpL complexes associated with non-HSPG LpL binding proteins. One such candidate protein on endothelial cells that can bind LpL with high affinity is the recently described 116-kDa amino-terminal fragment of apoB (45, 46). Preliminary observations showed that Deca did not inhibit LPL binding to apoB (not shown). 5) HS oligo·LpL complexes interacted with endothelial cells via the HS oligo. Several endothelial cell surface molecules bind to glycosaminoglycans (47). They include cell surface adhesion molecules such as selectins, integrins, platelet endothelial cell adhesion molecules, and cadherins.

There may be other unstable proteins for which HS oligosaccharides serve as chaperones. Although bFGF and other cytokines have high affinity protein receptors, their biological effects are potentiated by complexation with HSPG or exogenous heparin (48, 49). It is hypothesized that HS keeps the bFGF in a conformation required for binding to its receptor. Similarly HS modulate the activities of antithrombin, heparin cofactor II, protease nexin-1, and interferon- (5, 50-53). Our data, however, are unique in showing that a fragment of HS can facilitate the intercellular movement of a protein. LpL movement from the adipocyte to the endothelial cell surface appears, at least in part, to regulate LpL activity in vivo. We hypothesize that in vivo HS oligosaccharide facilitates this process by maintaining LpL conformation and activity. Thus, a fragment of HS derived from the actions of endothelial heparanase may be a critical factor in the delivery of energy to cells.