Mechanisms by which lipoprotein lipase alters cellular metabolism of lipoprotein(a), low density lipoprotein, and nascent lipoproteins. Roles for low density lipoprotein receptors and heparan sulfate proteoglycans.

We sought to investigate effects of lipoprotein lipase (LpL) on cellular catabolism of lipoproteins rich in apolipoprotein B-100. LpL increased cellular degradation of lipoprotein(a) (Lp(a)) and low density lipoprotein (LDL) by 277% +/- 3.8% and 32.5% +/- 4.1%, respectively, and cell association by 509% +/- 8.7% and 83.9% +/- 4.0%. The enhanced degradation was entirely lysosomal. Enhanced degradation of Lp(a) had at least two components, one LDL receptor-dependent and unaffected by heparitinase digestion of the cells, and the other LDL receptor-independent and heparitinase-sensitive. The effect of LpL on LDL degradation was entirely LDL receptor-independent, heparitinase-sensitive, and essentially absent from mutant Chinese hamster ovary cells that lack cell surface heparan sulfate proteoglycans. Enhanced cell association of Lp(a) and LDL was largely LDL receptor-independent and heparitinase-sensitive. The ability of LpL to reduce net secretion of apolipoprotein B-100 by HepG2 cells by enhancing cellular reuptake of nascent lipoproteins was also LDL receptor-independent and heparitinase-sensitive. None of these effects on Lp(a), LDL, or nascent lipoproteins required LpL enzymatic activity. We conclude that LpL promotes binding of apolipoprotein B-100-rich lipoproteins to cell surface heparan sulfate proteoglycans. LpL also enhanced the otherwise weak binding of Lp(a) to LDL receptors. The heparan sulfate proteoglycan pathway represents a novel catabolic mechanism that may allow substantial cellular and interstitial accumulation of cholesteryl ester-rich lipoproteins, independent of feedback inhibition by cellular sterol content.

tein that is an integral component of several plasma lipoproteins. It exists in three general forms: apoB-100, the fulllength protein of 550 kDa (2); apoB-48, a truncated version limited to the N-terminal 48% of apoB-100 (2); and apoBapo(a), a covalent complex consisting of apoB-100 linked by a disulfide bridge to apo(a), a polymorphic protein homologous to plasminogen (3,4). ApoB-100 is the major protein of VLDL and LDL, and binds to LDL receptors on cell surfaces, thereby mediating high-affinity cellular uptake of these lipoproteins (5). ApoB-48 is a major protein of chylomicrons and chylomicron remnants, but lacks the LDL receptor-binding domains (2). ApoB-apo(a) is the major protein of Lp(a) and binds weakly to LDL receptors (3), apparently because the presence of apo(a) shields or alters the receptor-binding domains of apoB-100 (6). Lp(a) also binds to cell surface receptors for plasminogen, which recognize epitopes in apo(a) (3,7,8). All of these apoB-rich lipoproteins have been linked to the development of atherosclerosis in humans (2- 4,9,10).
Of these atherogenic lipoproteins, Lp(a) is the least well understood. There are several competing hypotheses to explain its association with vascular disease. Lp(a) accumulates within the arterial wall, apparently by traversing the endothelium and binding to interstitial proteoglycans (3,ll); Lp(a) may be the predominant form of apoB in atherosclerotic lesions (11). Lp(a) is susceptible to oxidation (3) and other modifications (11) that promote cellular uptake in vitro, particularly by macrophages. Lp(a) displaces plasminogen from its binding site on endothelial cells (3,7,8) and clots (3, 11) and thereby may inhibit thrombolysis. Atherogenic effects correlate with plasma concentrations, which vary over several orders of magnitude in the normal population (4). Mechanisms underlying this extreme variability are unknown, although plasma concentrations tend to correlate inversely with the molecular weight of apo(a) isoforms (3). In general, mechanisms of assembly, secretion, and clearance from plasma remain incompletely characterized (3).
Many reports indicate that cellular catabolism of VLDL, LDL, chylomicrons, and nascent apoB-rich lipoproteins is increased by lipoprotein lipase (LpL), an enzyme that is bound to heparan sulfate proteoglycans and is involved in hydrolysis of lipoprotein triglycerides and, to a lesser extent, phospholipids (12). The reported mechanisms for LpL-enhanced lipoprotein uptake, however, are varied. LpL-induced enhancement of cellular uptake of VLDL requires hydrolysis of VLDL triglyceride (13), which then results in exposure of receptorbinding domains (14,15) and adsorption of additional apoE by the VLDL (13), causing enhanced binding to LDL receptors (13,14,16). LpL-induced enhancement of LDL uptake

LpL Enhances
Catabolism of Lp(a), LDL, and Nascent Particles 13285 has been reported to require conformational changes in the LDL receptor binding domains of apoB-100, causing enhanced binding to LDL receptors (17,18). LpL-induced enhancement of chylomicron uptake has been reported to require hydrolysis of triglyceride (19) or phospholipid (20), resulting in enhanced cell surface binding. The ligand for this enhanced binding was reported to be apoE (19); LpL bound to the surface of the chylomicrons did not directly mediate cell surface binding in these studies (19). The hypothesis that LpL acts as a bridge between chylomicrons and cell surface sites has been proposed by several authors (9,12,21,22); this hypothesis is supported by many reports that LpL binds to cell surface heparan sulfate proteoglycans (9,12,21,23,24) and to chylomicrons (12,21,22,25). One recent report has suggested that the cell surface binding site for LpL is the LDL receptor-related protein (LRP), based on similarities in molecular weight between LRP and the complex to which surface-bound LpL can be chemically cross-linked (26). LpLinduced enhancement of cellular reuptake of newly secreted apoB-rich lipoproteins is blocked by heparin (20 mg/ml) (27), which could reflect involvement of either LDL receptors (28,29) or cell surface heparan sulfate proteoglycans (21,23,24,30). There are no published articles concerning the effects of LpL on cellular uptake of Lp(a).
We therefore sought to characterize in detail the effects of LpL on cellular uptake and degradation of Lp(a), LDL, and nascent apoB-100-rich lipoproteins. We chose to focus on the roles of LDL receptors and heparan sulfate proteoglycans because of their recognized effects on cellular metabolism of apoB and LpL, respectively.

MATERIALS AND METHODS
Preparation of Reagents-LpL (EC 3.1.1.34) was purified from fresh cow's milk by our modification (27) of the Intralipid binding method of Posner et al. (31). Lp(a) and same-donor LDL were isolated from fresh human plasma by ultracentrifugation followed by affinity chromatography on lysine-Sepharose (32). All preparations of Lp(a) were of the F-isoform (molecular mass of apo(a) = 281 kDa) (33). Radioiodinations were accomplished by the iodine monochloride method (5,34).
Guanidine HCI and diethyl-p-nitrophenyl phosphate (E600) were purchased from Sigma. Guanidine-inactivated LpL (35) was prepared by overnight dialysis against 6 M guanidine HCl, 10 mM Tris-HC1, pH 8.5, followed by removal of the guanidine by dialysis against lipase buffer (phosphate-buffered saline with 30% glycerol). E6OO-inactivated LpL (36) was prepared by adjusting an enzymatically active LpL preparation to 5 mM E600, immediately followed by extensive dialysis against lipase buffer to remove unreacted E600. For experiments using these preparations, active control LpL was subjected to similar dialyses but without the inactivating chemicals. All preparations were assayed for protein mass (37) and enzymatic activity (27).
Heparitinase (EC 4.2.2.8) and 6-aminohexanoic acid were purchased from Sigma. Heparin was purchased from Fisher. The monoclonal anti-LDL receptor blocking IgG was produced from a hybridoma obtained from the American Type Culture Collection (Rockville, MD, catalogue CRL1691) (38). Cellular Uptake and Degradation of Lipoproteins-The human hepatocellular carcinoma line, HepGZ, was obtained from the American Type Culture Collection (catalogue HB8065) (39). Normal human fibroblasts (GM3468A) and LDL receptor-negative fibroblasts from a patient with homozygous familial hypercholesterolemia (FH) (GM2000) were obtained from the NIGMS Human Genetic Mutant Cell Repository (Camden, NJ) (17, 40). The wild-type Chinese hamster ovary (CHO) cell line ( K l ) and mutant lines deficient in glycosaminoglycans (745 and 677) were generously supplied by Dr. Jeffrey D. Esko of the University of Alabama (41,42). All cell types were grown to near-confluence in 35-mm wells in serum-supplemented media, then, before each experiment, grown overnight in serum-free selenium-supplemented medium (HepG2 cells) (27,29), minimal essential medium with 10% lipoprotein-deficient serum (fibroblasts) (5), or Ham's F-12 with 10% delipidated serum (CHO cells) (41,43). In one experiment, LDL receptors of normal fibroblasts were sup-pressed by adding 1.0 pg of 25-hydroxycholesterol and 20 p g cholesterol per ml of medium (minimal essential medium with lipoproteindeficient serum) (5). This addition required 2 pl of ethanol/ml of medium; the same amount of ethanol was also added to control cells.
For preincubations and subsequent experimental incubations, cells were changed to the corresponding medium supplemented with 0.2% bovine serum albumin instead of serum or serum fractions. To block lysosomal proteases, we preincubated cells for 45 min at 37 "C in 100 p~ chloroquine and maintained this concentration after the addition of LpL and '251-lipoproteins (5). To block LDL receptors with the monoclonal anti-LDL receptor antibody CRL1691, we preincubated cells for 30 min at 37 "C in a concentration of 100 pg of IgG/ml (38). We maintained this concentration during the subsequent experimental incubations with LpL and 1251-lipoproteins. Control cells received an equal concentration of irrelevant IgG. To digest cells fully with heparitinase also requires a preincubation (23,44). We chose 90 min at 37 "C in the presence of 4.0 units/ml (one unit of heparitinase, defined by Sigma, forms 0.1 pmol of unsaturated uronic acid per h at pH 7.5 at 25 "C). Consistent with prior observations (45), our preliminary experiments indicated that cell surface HSPGs may quickly regenerate once the heparitinase is removed. Hence, we maintained the concentration of heparitinase at 4.0 units/ml during the subsequent experimental incubations of cells with LpL and '251-lipoproteins.
As before, LpL and '"I-lipoproteins were added to the cells at the same time (27). The concentrations were 5.2-8.0 pg LpL/ml (27; cf. Ref. 24) and 9.7 nM '*'I-Lp(a) and 1251-LDL. Control cells without LpL received matching volumes of lipase buffer. Unless otherwise indicated, cells were incubated with LpL and '251-lipoproteins for 4 h at 37 "C. Cell-specific degradation and cell association were assayed as described previously (5,271, and results were normalized for cellular protein masses (46), which averaged 1.06 mg (HepG2 cells), 0.200 mg (GM3468A), 0.164 mg (GMZOOO), 1.14 mg (Kl), 0.925 mg (mutant 745), and 0.692 mg (mutant 677) per 35-mm well. Degradation in wells without cells was always ~0 . 0 2 % of the total lz5I radioactivity applied to the wells (5) and was not affected by any of the experimental conditions. Variations between experiments in rates of cellular degradation of lipoproteins showed the typical 3-fold range reported by Goldstein et al. (5). All incubations with '2sII-lipoproteins were performed in triplicate.
Secretion of Nascent ApoB-HepGZ cells were cultured and pretreated as described above. Our methods to assess secretion of [3H] apoB have been reported previously (27). Cells were given LpL or lipase buffer simultaneously with [3H]leucine (120-190 Ci/mmol, Amersham Corp., 75 pCi/ml of medium, 1.0 ml of medium/35-mm well), then incubated at 37 "C for 2 h. Media were subjected to apoB immunoprecipitation using the antiserum generously supplied by Dr. Charles L. Bisgaier of Columbia University (47) and total protein precipitation using trichloroacetic and phosphotungstic acids (27). Data for apoB output by the cells are expressed as the percentage of total labeled secreted protein which was labeled apoB (i.e. 100 X [3H] apoB radioactivity divided by 3H-protein radioactivity in the media). All secretion experiments were performed in quadruplicate.
Statistics-Results are given as means t S.E. In the figures, absent error bars signify S.E. values smaller than the graphic symbols.
Statistical comparisons were performed by Student's two-tailed t test. Standard errors for the differences between means of groups with equal n were calculated as the square root of the sum of the squares of the individual S.E. values (48).

RESULTS
During a 5-h control incubation in the absence of LpL, HepG2 cells degraded 121 f 3.9 fmol of "'I-Lp(a) per mg cell protein and 457 f 2.5 fmol of same-donor '*'I-LDL/mg. Degradation in the presence of 5.3 pg LpL/ml was 277% f 3.8% higher for '*'I-Lp(a) and 32.5% f 4.1% higher for 'T-LDL (n = 3, p < 0.0005) (1). Cell association of these lipoproteins during this experiment showed proportionately larger effects (cf. Refs. 27 and 49). In the absence of LpL, 58.3 f 2.3 fmol of "'I-Lp(a)/mg and 337 f 8.7 fmol of "'I-LDL/mg became cell associated. In the presence of LpL, these figures increased by 509% -+ 8.7% and 83.9% f 4.0%, respectively ( p < 0.00005). We sought to examine these effects in detail.
Effects of LpL on Degradation of "'Z-Lipoproteins- Table  I shows that 100 p~ chloroquine entirely blocked the ability of

LpL Enhances
Catabolism of Lp(a), LDL, and Nascent Particles LpL to enhance cellular degradation of Iz5I-Lp(a) and lZ51-LDL, suggesting a lysosomal pathway, consistent with a prior report (17). Table I also shows that chloroquine did not interfere with the ability of LpL to enhance cell association of '251-lipoproteins, indicating that chloroquine had no direct effect on LpL (cf. Ref. 49). As expected, chloroquine increased cell-associated lZ5I radioactivity because of the inability of the cells to generate and release 1251-tyrosine as a breakdown product (5).
Because previous literature has concluded that LpL increases LDL degradation by enhancing the binding of apoB to LDL receptors (17), we used three independent methods to examine the role of LDL receptors. First, we compared normal fibroblasts with mutant FH fibroblasts that lack the LDL receptor. Fig. lA shows that LpL caused large increases in "'I-Lp(a) degradation by both types of cells, but the absolute effect with mutant FH cells was smaller. Thus, we concluded that the effect of LpL on the degradation of IZ5I-Lp(a) has LDL receptor-dependent and -independent components (1). In contrast, Fig. 1B shows that the absolute increases in lZ5I- Thus, we found no evidence for an LDL receptor-dependent component of the effect of LpL on lZ5I-LDL degradation. It appeared to be entirely LDL receptor-independent. Fig. 1, C and D, shows the effects of sterol depletion with lipoprotein-deficient serum to stimulate cellular LDL receptors versus sterol loading with 25-hydroxycholesterol and cholesterol to suppress LDL receptors. Fig. 1, E and F, shows the effects of a control IgG versus an antibody (CRL1691) that specifically blocks LDL receptor-mediated uptake of lipoproteins (38). As above, the effect of LpL on lz51-Lp(a) degradation displayed LDL receptor-dependent and -independent components, whereas the effect on lz51-LDL was entirely LDL receptor-independent. Note that for competition experiments, the addition of excess unlabeled LDL is unsuitable because it not only blocks LDL receptors but also adsorbs LpL (25). The CRL1691 antibody, in contrast, is a specific anti-LDL receptor reagent that does not interact with LpL (27,38).

Role of lysosomes in the ability of LpL to increase degradation of lZ5Zlipoproteins by normal fibroblasts
Lipoprotein lipase interacts with lipoproteins both structurally, by binding to the particles (12,21,22,25), and enzymatically, by hydrolyzing core triglycerides (12,21). We sought to determine whether the structural or enzymatic action was responsible for the enhancement in Lp(a) and LDL degradation. Initially, we used denaturation with guanidine HC1 to inactivate LpL because guanidine-denatured LpL retains its ability to bind to hepatocytes (35). The specific enzymatic activity of our guanidine-inactivated LpL was 0.33% of control. Active control LpL increased 'T-LDL degradation by FH fibroblasts by 881 f 41.4 fmol/mg. An equal mass concentration of guanidine-treated LpL increased the degradation by 465 f 41.7 fmol/mg, which is 52.8% k 4.7% of the increase caused by control LpL. A 300-fold dilution of control LpL, which matched the residual enzymatic activity in the guanidine-treated LpL, had no effect on T -L D L degradation. Thus, we concluded that LpL enzymatic activity was not required. We next sought a more specific method to inactivate the hydrolytic activity of LpL, without resorting to global denaturation. Because LpL enzymatic activity depends on SerI3' (50), we used E600, a serine-specific suicide substrate (36). Fig. 2A compares the degradation of Iz5I-Lp(a) by normal fibroblasts in the presence of lipase buffer, active control LpL, E6OO-treated LpL (matched to control LpL by protein mass), or a dilution of control LpL which matched the small residual activity in the LpL-E6OO. The specific activity of LpL-E600 was 2.6% of control. With this more selective method of inactivation, there was no detectable difference between the effectiveness of control LpL and LpL-E600 in increasing the degradation of '"I-Lp(a). Dilute control LpL produced comparatively small effects.
Because LpL-induced enhancement of Lp(a) degradation has LDL receptor-dependent and -independent components, we examined the effects of our active and inactivated preparations of LpL on the degradation of Iz5I-Lp(a) in the presence of the anti-LDL receptor-blocking antibody. LDL receptorindependent degradation of Lp(a) was unaffected by enzymatic inactivation of LpL (Fig. 2B).
The effects on Iz5I-LDL degradation by FH fibroblasts were similar: control LpL and LpL-E600 were equally effective at increasing degradation, and dilute control LpL produced no effects (Fig. 2C). Therefore, we concluded that the effects of LpL on the cellular degradation of Lp(a) and LDL are predominantly the result of structural, not enzymatic, effects of the LpL molecule. This result is consistent with the observation that LpL exhibits poor enzymatic activity toward lipoproteins that lack apoC-I1 (12), such as LDL and Lp(a).
One important structural characteristic of LpL is its adher-3000 A ence to cell surface heparan sulfate proteoglycans (HSPGs) (9,12,21,23,24). To investigate whether this adherence plays a role in the ability of LpL to enhance lipoprotein degradation (I), we used three different methods. The first method involved digestion of the cell surface with heparitinase (EC 4.2.2.8), an enzyme that specifically degrades the sugar side chains of cell surface HSPGs (44). Fig. 3, A and B, shows that heparitinase partially blocked the ability of LpL to increase the cellular degradation of '"I-Lp(a) and totally blocked the effect of LpL on '"I-LDL degradation. Because LDL receptor-mediated degradation was unaffected by heparitinase (the first and third columns in Fig. 3B are indistinguishable, p > 0.5), consistent with the known absence of heparan sulfate from the LDL receptor (51), the results in Fig. 3 are consistent with Fig. 1. Enhanced degradation of Lp(a) in the presence of LpL has two components: one component is heparitinase-sensitive and therefore LDL receptor-independent; the other component is heparitinaseresistant, which is consistent with an LDL receptor-dependent effect. Enhanced degradation of LDL is entirely heparitinase-sensitive and LDL receptor-independent.
The second method to interfere with the adherence of LpL to HSPGs was the use of mutant CHO cells that lack various cell surface proteoglycans (41,42). We used the wild-type K1 line, the mutant 745, which lacks xylosyltransferase and therefore synthesizes no heparan sulfate and no chondroitin sulfate proteoglycans, and the mutant 677, which specifically lacks HSPGs but has an excess of chondroitin sulfate proteoglycans (41). Under our incubation conditions, LDL receptor expression in these cells was suppressed in the absence of LpL, degradation of Lp(a) and LDL was low (Table 11). The addition of LpL substantially increased the degradation of both lipoproteins by the wild-type K1 line (Table 11)    HepG2 cells were used, without (-) or with (+) LpL. In the absence of LpL, heparin had no effect on '*'I-LDL degradation ( p > 0.2 for the difference between the first and third columns). In the absence of heparin, LpL caused a large, significant increase in LDL degradation ( p < 0.005 between the two leftmost columns); in the presence of heparin, LpL slightly reduced LDL degradation (0.02 > p > 0.01 between the two rightmost columns).
ing HSPGs in the LDL receptor-independent degradative pathway.
The third method to interfere with LpL-HSPG interactions was the addition of heparin to the incubations. To inhibit binding of LDL to LDL receptors, high concentrations of heparin are required (21.5 mg/ml at 37 "C; usually, 10 mg/ml is used) (28), whereas to release LpL from cell surfaces requires far less (20.1 unit/ml; usually 10 units/ml is used, which corresponds to 67 pg of our heparin preparation/ml) (30). Fig. 4 shows that 67 pg of heparin/ml had no discernible effect on lZ5I-LDL degradation by HepG2 cells in the absence of LpL but abolished the ability of LpL to increase lZ5I-LDL degradation. Interestingly, we found that 10 units of heparin/ ml in the absence of LpL reduced lZ5I-Lp(a) degradation by HepG2 cells by 35.7% f 4.1% and that 2 units/ml reduced it by 20.0% f 1.8%. Thus, low concentrations of heparin are not necessarily a selective reagent for separating the effects of LDL receptors and HSPGs on degradation of Lp(a). Even at these relatively low concentrations of heparin, there is still a molar excess of heparin over apoB-apo(a).
We next sought to determine if binding to HSPGs and enhanced binding to LDL receptors totally explain the increase in Lp(a) degradation caused by LpL. The experiment in Table I1 did not indicate the existence of any substantial LDL receptor-independent, HSPG-independent effect of LpL on lipoprotein degradation by CHO cells (see above). To extend this result to human fibroblasts, we examined the effects of heparitinase in combination with the anti-LDL receptor antibody. In the experiment shown in Fig. 3A, LpL increased Lp(a) degradation without and with heparitinase by 1,860 f 23.2 and 626 f 22.7 fmol/mg, respectively. An additional group of cells not shown in Fig. 3A was digested with heparitinase and also received CRL1691. The incubation with heparitinase was sufficient to abolish the effect of LpL on LDL degradation (Fig. 3B), so it was presumably a complete digestion. In the absence of LpL, CRL1691 reduced Lp(a) degradation by fibroblasts to 11.2% f 0.2% of control, SO the LDL receptor blockade was nearly complete. With the combination of heparitinase and CRL1691, LpL increased the degradation of Lp(a) by 418 f 10.7 fmol/mg, which was only 22.5% f 0.6% of the increase in the no-heparitinase, no-LpL control cells. Therefore, we cannot eliminate the possibility of a small, LDL receptor-independent, HSPG-independent component of the effect of LpL on Lp(a) degradation by human fibroblasts.
One possibility is that an additional degradative pathway would be the plasminogen binding site, which can be blocked with t-aminohexanoic acid (7,8). We found that 200 mM taminohexanoic acid, which is frequently used to block plasminogen receptors (7), interfered with cellular degradation of lZ5I-LDL, suggesting nonspecific effects on the cells. 15 mM t-Aminohexanoic acid, which is sufficient to block specific binding of plasminogen to its receptor (8), had minimal effects on lZ5I-LDL degradation but no effect on the ability of LpL to enhance degradation of 1251-Lp(a) (data not shown). Thus, we have no evidence that cellular binding sites for plasminogen are involved in the effect of LpL on Lp(a) degradation. Consistent with our findings, the cellular receptor for plasminogen does not appear to mediate degradation of plasminogen (3).
Effects of LpL on Cell Association of '25f-Lipoproteins-As noted above, the effects of LpL on cell association were always proportionately larger than the effects on degradation, consistent with prior studies (27,49). This difference suggests substantial sequestration of lipoproteins at the cell surface, on the extracellular matrix (42, 44, 52), or, possibly, in nonlysosomal intracellular compartments (52,53). We sought to determine the mechanisms involved in the enhancement of cellular association of '251-lipoproteins. Fig. 5, A and B, shows cell association data from the experiments reported in Fig. 1, C and D. For both Lp(a) and LDL, essentially none of the effect of LpL was suppressed when LDL receptors were down-regulated by sterol-loading the fibroblasts. Fig. 6 shows that the effects of LpL on cell association do not require enzymatic activity. Note that even though E6OO-treated LpL was as effective as control LpL in promoting degradation of Lp(a) and LDL (Fig. 2) and cell association of LDL (Fig. 6C), it was less effective than control LpL in promoting cell association of Lp(a) (Fig. 6, A and B ) . Note, also, that only 10.4% f 0.7% of the effect of control LpL on cell association of Lp(a) was blocked by CRL1691, consistent with Fig. 5A, but 35.0% f 1.7% of the effect of LpL-E600 was blocked by CRL1691 (compare results in Fig.  6, A and B ) .
For both Lp(a) and LDL, most of the effect of LpL on cell association was abolished by heparitinase digestion of the cells (Fig. 7, A and B ) . In the experiment with CHO cells described in Table 11 an 83.0% k 7.3% increase in cell association, whereas with heparin, LpL increased cell association by only 15.0% f 4.7% ( p < 0.002 for the difference between these two percentages).
Effects of LpL on Net Secretion of ApoB-We showed previously that LpL substantially reduces the net secretory output of [3H]apoB from cultured HepG2 cells by enhancing cellular reuptake of newly released particles (27). We now sought to determine if this enhancement of reuptake resembles the effect of LpL on LDL uptake and degradation, which is entirely heparitinase-sensitive and LDL receptor-independent, or resembles the effect of LpL on Lp(a) uptake and degradation, which has at least two components. Fig. 8. 4 shows that the effect of LpL on net secretion of ["HIapoB is entirely unaffected by the blockade of LDL receptors by CRL1691. In the absence and presence of CRL1691, LpL reduced net secretion of apoB by the same factor. Note that in the absence of LpL, CRL1691 increased net secretion of apoB to 144% f 3.8% of control (p < 0.00005), indicating significant reuptake in the absence of added LpL (cf. Refs. 27 and 29). Fig. 8B shows that the effect of LpL on apoB secretion results from a structural, not an enzymatic, action of LpL. Inactivated LpL was nearly as effective as control LpL at reducing the net cellular output of nascent apoB, whereas dilute control LpL had no effect. Fig. 8C shows that the effect of LpL on apoB secretion is entirely abolished by heparitinase digestion of the HepG2 cells. Note that in the absence of LpL, heparitinase digestion did not increase apoB secretion, suggesting that any endogenous production by HepG2 cells of other HSPG-bound enzymes, such as hepatic lipase (54) and carboxyl ester lipase (55,56), had little effect on net apoB output. In separate experiments, we found that a low concentration of heparin (67 pg/ml) abolished the ability of LpL to affect net apoB secretion (data not shown; cf. Ref. 27). As expected from the results in Fig. 4   concentration of heparin had no effect on net apoB output in the absence of LpL.

DISCUSSION
We have shown that LpL produces large increases in lysosomal degradation of LDL and, especially, Lp(a) by cultured cells. By three independent methods (Fig. l), we have found that the increase in Lp(a) degradation has LDL receptordependent and -independent components, whereas the increase in LDL degradation is essentially independent from LDL receptors. Using two methods to inactivate LpL, we have found that neither the LDL receptor-dependent nor -independent effects on lipoprotein degradation depend on LpL enzymatic activity (see Fig. 2 and "Results"). By three independent methods (Figs. 3 and 4 and Table 11), we have found that the LDL receptor-independent effects on Lp(a) and LDL degradation require the interaction of LpL-lipoprotein complexes with cell surface heparan sulfate proteoglycans. These LDL receptor-independent, HSPG-dependent effects of LpL on lipoprotein degradation are likely to be the result of bridging by LpL between lipoproteins and HSPGs (9,12,21,22), which would not require apoB (cf. Ref. 57), or the result of enhancement of the direct binding of lipoproteins to HSPGs (3,11,52,58). We speculate that the LDL receptor-dependent effect of LpL on Lp(a) degradation is the result of exposure of previously occult receptor-binding domains on the apo(a) apoB complex; this phenomenon does not appear to occur with LDL.
LpL also produced large increases in cell association of apoB-100-rich lipoproteins, consistent with a prior report (27). Effects on cell association were largely blocked by heparitinase (Fig. 7) or a low concentration of heparin (see "Results"), indicating a predominant, but not exclusive, role for HSPGs. Similarly, in the CHO mutants, we found residual effects of LpL on cell association (see "Results") but nearly no residual effects of LpL on degradation (Table 11), suggesting that HSPGs mediated all of the increased degradation and much of the association but that some association was mediated by non-HSPG molecules (cf. Ref. 41). The existence of more than one binding site for LpL-lipoprotein complexes is further supported by the equal effects of control LpL and LpL-E600 on Lp(a) degradation (Fig. 2) but unequal effects on Lp(a) cell association (Fig. 6); presumably, LpL-E600 complexed with Lp(a) bound normally to sites targeted for degradation but subnormally to sites not targeted for degradation. Also, effects of LpL on cell association of Lp(a) and LDL were proportionately larger than effects on degradation. Efficiency of degradation of cell-bound '251-labeled LpL has been reported to vary considerably among normal cell types (49), perhaps because of differences in cell surface HSPGs or different ratios of binding sites targeted and not targeted to lysosomes.
The effects of LpL on net secretion of [3H]apoB operated by the same mechanisms demonstrated for LDL degradation: the effect is entirely LDL receptor-independent, HSPG-dependent, and does not require LpL enzymatic activity. These findings allow interpretation of an apparent conflict in the literature on familial combined hyperlipidemia, a common genetic disorder characterized by hepatic oversecretion of apoB-100 (59). We have hypothesized that hepatic oversecretion of apoB may arise in heterozygotes for LpL deficiency as a result of reduced hepatic reuptake of apoB-rich nascent lipoproteins released into the space of Disse (27). One group has reported that heterozygous deficiency of LpL mass and activity is indeed associated with the familial combined hyperlipidemia phenotype (60,61), yet an extensive report of a single large kindred with LpL deficiency showed familial hypertriglyceridemia, not familial combined hyperlipidemia, in obligate heterozygotes (62). The LpL mutation in this large kindred, however, allows for the secretion of enzymatically inactive LpL that still appears to bind to cell surface heparan sulfate proteoglycans (63). Based on our current results (Fig.  8B), we hypothesize that this particular LpL mutation could still augment hepatic reuptake of nascent apoB. Thus, the heterozygous phenotype in this kindred reflects solely the deficiency in triglyceride hydrolysis. In contrast, apoB oversecretion should arise in heterozygotes for LpL mutations that substantially reduce the synthesis and secretion of protein mass, or produce a protein that cannot bind to lipoproteins or HSPGs.
We wish to reconcile our findings on LpL-induced enhancement of LDL degradation with those of Aviram et d . (17,18). In their studies, LpL was reported to digest LDL triglycerides, LpL Enhances Catabolism of Lp(a), LDL, and Nascent Particles 13291 resulting in apoB conformational changes, and it was proposed that binding to the LDL receptor was enhanced. Analysis of their data by Woolf plot, however, revealed an increase in maximal binding (BmeX) (17), which suggests the involvement of additional binding sites. Other data from their studies consistent with LDL receptor-independent binding include the demonstration of an LpL-mediated increase in the degradation of lz5I-LDL by LDL receptor-negative fibroblasts (Fig. 7B in Ref. 17); the ability of LDL in the presence of LpL to cause cellular accumulation of sterol (17), which should have suppressed LDL receptors (5); and a decrease in "active" lysines (pK = 8.9) on apoB following treatment of LDL with lipase (18). A decrease in active lysines on apoB is associated with decreased, not increased, LDL receptor affinity (64). We conclude that their data are, in fact, consistent with a substantial LDL receptor-independent effect of LpL on LDL degradation.
We also wish to discuss the relationship between our findings and those of Beisiegel et al. (26), in which LpL-induced enhancement of cellular binding of chylomicrons at 0-4 "C was reported to involve binding of LpL to LRP. Because LRP at 37 "C is targeted to lysosomes (65), their results should be reconciled with our findings about lipoprotein degradation. There are three possibilities. First, it is possible that both their effects and our LDL receptor-independent effects are mediated through LRP. Because cell surface binding of LpL (12,23,24) (71), are clearly distinct from LRP. Note that if the LDL receptor-independent processes we describe involve LRP, they would allow LRP, through LpL, to participate in cellular uptake and degradation of apoB-100-rich lipoproteins that have little or no apoE.
Second, it is possible that the effects described by Beisiegel et al. and the effects we describe involve binding of LpLlipoprotein complexes to HSPGs that are not LRP. Their identification of the LpL binding site as LRP was based solely on its high molecular weight from cross-linking experiments (26). Distinct high molecular weight HSPGs have, in fact, been reported (42,52). Alternatively, each LpL molecule may bind simultaneously to several cell surface HSPGs of lower molecular weight. Because chylomicrons and chylomicron remnants adsorb LpL in vivo (22,25), we hypothesize that the HSPGs that we have described may serve as receptors for these particles. It should be straightforward to examine the effects of LpL on cellular catabolism of chylomicrons and remnants when binding to HSPGs is blocked.
Third, it is possible that LpL-chylomicron complexes bind to LRP, but complexes of LpL with apoB-100-rich lipoproteins bind to heparan sulfate proteoglycans that are not LRP. This is the least likely possibility, given that LpL in the absence of lipoproteins binds to a similar site as do LpLchylomicron complexes and that heparin inhibits binding of LpL.apoB-100 ( Fig. 4 and Ref. 27) and LpL.apoB-48 complexes (26).
The physiologic relevance of our findings requires that LpL, lipoproteins, and HSPGs encounter each other at appropriate concentrations. The concentrations of LpL required to enhance LDL receptor-independent cellular catabolism of lipoproteins can be estimated from the published Kd for the binding of LpL to avian adipocytes (16 nM, i.e. about 1 pg/ ml) (24) and from our published dose-response curve for LpLmediated reuptake (half-maximal effect was at about 1 pg of LpL/ml) (27). The concentration dependence of the LDL receptor-dependent effect of LpL on Lp(a) remains to be determined; because it is largely a structural effect, we assume that the number of LpL and Lp(a) molecules will have to be similar. Thus, LpL concentrations in fasting, preheparin plasma (8.0-25.0 ng/ml) (72) may be too low for either the LDL receptor-dependent or -independent effects. The effects of LpL may be more important in postprandial plasma: the LpL concentration approximately doubles with fat feeding (73), and most of it is adsorbed to large, apoB-rich lipoproteins (22,25). The effects of LpL may be especially important in confined tissue compartments, such as the space of Disse (27), where LpL is abundant postprandially (74, 75); endothelial surfaces (12,21); and the arterial wall (211, including atherosclerotic lesions (10,(76)(77)(78)(79). Aortic uptake and accumulation of cholesteryl ester correlate positively with aortic content of LpL (lo), and parenteral administration of sulfated mucopolysaccharides reduces the development of atherosclerosis in experimental animals (9,lO). Other proteoglycan-bound molecules, such as hepatic lipase (54) and carboxyl ester lipase (55), might similarly influence the local metabolism of lipoproteins. Enzymatically inactive forms of LpL (72) may also play a role.
The existence of LpL-mediated pathways for particle uptake which are independent of LDL receptors has several implications. First, the pathways are not regulated by cellular sterol content (Figs. 1, C and D, and 5, A and B); thus, they could cholesterol-load cells without feedback inhibition, much as scavenger receptors do (80). These LpL-mediated pathways may be particularly relevant to smooth muscle cells (77), which form foam cells in atherosclerotic lesions, despite a relative paucity of scavenger receptors (80). Also, lipid debris (81) and aggregated lipoproteins (82) may be more avidly taken up by arterial macrophages and smooth muscle cells in the presence of LpL. Second, the trapping of lipoproteins without lysosomal degradation may promote other processes, such as oxidation or lipolysis. Because many proteoglycans are secreted (42, 52), not cell surface bound, LpL may contribute to the substantial interstitial trapping of these lipoproteins in vivo (3,11,58). LpL might also participate in LDL receptor-independent transport of Lp(a) across endothelium (3). Third, the LpL-heparan sulfate pathway may explain our previous observation that hepatic uptake of intravenously administered phospholipid liposomes does not require normal expression of LDL receptors (83). These injected liposomes function as synthetic, antiatherogenic lipoproteins, acquiring apoE and tissue cholesterol before their eventual removal by the liver (83). Liposomes adsorb LpL (26,84), and a nonenzymatic action of LpL enhances cellular uptake of liposomes in vitro (57). Fourth, hepatic uptake, or reuptake, of Lp(a), LDL, nascent apoB-100-rich lipoproteins, chylomicrons, and remnants may depend, in part, on LpL. Fibric acids induce LpL (27); these drugs are among the few known compounds that can decrease serum concentrations of Lp(a) (85,86).
Overall, our work suggests that LpL may serve a dual function, depending on its location. In the arterial wall, it would be atherogenic, by promoting accumulation of apoBrich lipoproteins (9, lo), independent of feedback inhibition by cellular sterol content. In the liver, it would be antiatherogenic, by enhancing reuptake and thereby decreasing net secretory output of nascent apoB-rich lipoproteins (27), including, perhaps, Lp(a). In the liver, LpL may also promote clearance of mature plasma lipoproteins (22) and might en-hance reverse cholesterol transport (83). Control of LpLmediated uptake and reuptake of lipoproteins and control of LpL movement from peripheral tissues to the liver may be key regulatory steps in atherogenesis. Because these effects are structural, not enzymatic, LpL could be regarded as an apolipoprotein.