A carboxyl-terminal fragment of lipoprotein lipase binds to the low density lipoprotein receptor-related protein and inhibits lipase-mediated uptake of lipoprotein in cells.

It has previously been shown that lipoprotein lipase can mediate uptake of remnant lipoprotein particles via binding to the low density lipoprotein receptor-related protein/alpha 2-macroglobulin receptor (LRP). Binding of lipoprotein lipase, and of triglyceride-rich lipoproteins associated with the lipase, to LRP depends on an intact carboxyl-terminal folding domain of the lipase (Nykjaer, A., Bengtsson-Olivecrona, G., Lookene, A., Moestrup, S. K., Petersen, C. M., Weber, W., Beisiegel, W., and Gliemann, J. (1993) J. Biol. Chem. 268, 15048-15055). Here we show that the site for binding to the receptor is within residues 380-425 of the bovine and residues 378-423 of the human lipoprotein lipase. We demonstrate that a carboxyl-terminal fragment of human lipoprotein lipase (residues 378-448), expressed as fusion protein in Escherichia coli, binds to purified and cellular LRP but not to lipoproteins. Binding of the fragment to purified LRP was blocked by heparin. In addition, the fragment inhibited the binding of lipase and the lipase-mediated binding of lipoproteins to the purified receptor. The fragment exhibited reduced binding to proteoglycan-deficient cells. Moreover, the fragment inhibited the uptake of lipoproteins in cells mediated by the lipase via binding to heparan sulfate proteoglycans and LRP. We conclude that the fragment contains the site for binding to LRP and a candidate site for interaction with heparan sulfate proteoglycans, whereas binding to lipoproteins is inefficient. The fragment can therefore inhibit the lipase-mediated lipoprotein uptake, a process that may promote the development of atherosclerosis when occurring in cells of the arterial wall.

tein receptor-related protein/a,-macroglobulin receptor (LRP) following activation by the addition of apolipoprotein E (apoE) (1)(2)(3) or lipoprotein lipase (LpL) (4)(5)(6). Uptake mediated by LRP is thought to participate in the hepatic clearance of remnant particles (1-4, 7, 8). In addition, LRP is expressed in macrophages of human atherosclerotic lesions together with the scavenger receptor, and LRP is, so far, the only lipoprotein receptor expressed in lesion smooth muscle cells (9). LpL, secreted by subsets of lesion smooth muscle cells and macrophages (lo), promotes the retention of lipoproteins in the subendothelial space (11,12). In addition, LpL has an important function in tethering lipoproteins to heparan sulfate proteoglycans at the cell surface, thereby facilitating their receptor-mediated internalization (5,6,13) (for review, see Refs. 14 and 15). These results suggest that LpL and LRP may play a role in the pathogenesis of atherosclerosis. In support of this hypothesis, LpL is reported to markedly increase the uptake of P-VLDL in arterial smooth muscle cells, whereas apoE had no effect (16). Moreover, inbred mice exhibiting increased macrophage LpL secretion develop atherosclerosis (17). Thus, LpL-and LRPmediated lipoprotein uptake may be important in the development of foam cells characteristic of atherosclerotic lesions.
LRP is a giant 4524-residue, two-chain receptor that binds and mediates uptake of several structurally unrelated ligands (18)(19)(20)(21). Three established ligands are bound to other molecules at the cell surface before uptake via LRP; a complex of plasminogen activator inhibitor-type 1 and urokinase-type plasminogen activator binds to the urokinase receptor (19), and LpL and apoE bind to proteoglycans (5,6,13,14). The primary binding serves to concentrate the ligands at the cell surface, thereby facilitating the interaction with the endocytic receptor LRP. Since LpL associates with lipoproteins, either in the circulation or when bound to cell surface proteoglycans, it can mediate endocytosis of triglyceride-rich lipoproteins via LRP LpL is a noncovalent homodimer of monomers each containing an amino-terminal folding domain with the catalytic site and a carboxyl-terminal folding domain (for reviews, see Refs. 14 and 15). The mediation of lipoprotein binding to LRP depends on the dimeric structure of LpL, apparently because only the dimer can bind efficiently to a lipoprotein and to LRP at the same time (5). Bovine LpL (bLpL) has 450 amino acid (aa) (4)(5)(6). gel electrophoresis; PVDF, polyvinylidene difluoride; h, human; b, bovine; CHO, Chinese hamster ovary; BisTris, 2-[bis(2-hydroxyethy1)aminol-2-(hydroxymethyl)-propane-1,3-diol; CHAPS, 3-[(3-cholamidopropyl)dimethylammoniol-1-propanesulfonic acid; Tricine, N42hydroxy-1,l-bis(hydroxymethyl)ethyllglycine; VLDL, very low density lipoprotein; FH, familial hypercholesterolemia. ~~ 31747 residues, and the overall homology is 93% between bLpL and human LpL (hLpL), which is 2 residues shorter at the amino terminus (22). Both bLpL and hLpL can mediate binding and uptake of P-VLDL in cells (4). Earlier studies established that chymotrypsin-truncated bLpL, comprising residues 1-390 (23), does not bind to LRP or P-VLDL (5). Purified chymotrypsintruncated ~L P L '~~O has full catalytic activity against soluble substrates, and the dimeric structure is maintained (23). This suggests that the lack of binding to LRP following the truncation was caused by deletion or disruption of a binding site for the receptor rather than a change in the overall structure of the lipase. We decided to identify the putative LRP binding region in the carboxyl-terminal folding domain of bLpL. In addition, we wanted to express the corresponding fragment of hLpL, reasoning that a peptide capable of binding to the receptor, but not to lipoproteins, might inhibit the LpL-mediated uptake of lipoproteins via LRP.

MATERIALS AND METHODS
Lipases, LRE and 0-VLDL-The properties of the dimeric bLpL preparation purified from milk, as well as the iodination using lactoperoxidase and glucose oxidase, have been described previously (5). Monomeric bLpL was prepared by incubation with 1 M guanidinium hydrochloride for 3 h at 20 "C (24). hLpL was purified from postheparin plasma using heparin-Sepharose chromatography (4); two separate preparations were tested, and one of them was a gift from Dr. J. D. Brunzell, University of Washington. Human hepatic lipase (hHL) was purified from postheparin plasma as described (25). In brief, fresh postheparin plasma was adjusted to 0.4 M NaCl and mixed with heparin-Sepharose for 2 h at 4 "C. Following washes of the heparin-Sepharose, elution was performed with 0.7 M NaCl in 10 m M phosphate buffer, pH 7.4. hHL is eluted under this condition whereas hLpL is retained. To remove possible traces of hLpL in the eluate, the preparation was immunoprecipitated with rabbit IgG raised against bLpL (25). Then the eluate was adjusted to 0.2 M NaCl and applied to a Sepharose column with immobilized, modified (N-desulfated and N-acetylated) heparin (26). After thorough washing with 0.2 M NaCl, hHL was eluted with 0.9 M NaC1. The hHL preparation was enzymatically active (3.5 units/ ml) and, assuming a specific activity of 400 units/mg, the concentration of hHL was around 9 pg/ml. The preparation was not entirely homogeneous on SDS-PAGE but showed two to three additional bands with lower molecular weights than hHL. To exclude the possibility that results obtained with the hHL preparation from plasma might be due to reaction of LRP with contaminating hLpL or fragments thereof, some experiments were performed with hHL isolated from the medium of the human hepatoma cell line HUH, (27). The cells were grown in Dulbec-CO'S modified Eagle's medium containing 10% fetal calf serum and 7 units of heparidml. The medium was applied to heparin-Sepharose High Trap columns (Pharmacia, Sweden), and hHL was eluted using 10 m M BisTris buffer, pH 7.4, containing 1 M NaCl and 10% glycerol, followed by precipitation by dialysis against 3.4 M ammonium sulfate. The hHL prepared from the medium of HUH, cells did not contain hLpL as judged from lack of reactivity with the murine 5D2 monoclonal antibody (courtesy of Dr. J. D. Brunzell) raised against bLpL. The 5D2 antibody reacts within residues 396-405 of hLpL (28) and does not cross-react with hHL (29). Human pancreatic lipase was a gift from Dr. B. Sternby, University of Lund.
Limited digestion of bLpL with chymotrypsin followed previously published procedures, and bLpLlaW was purified by gel filtration chromatography followed by affinity chromatography on heparin-Sepharose as described previously (23). For digestion with CNBr, LpL was dissolved in 70% formic acid and incubated for 24 h with 40 mg of CNBr/ ml. The lyophilized products of the cleavage reaction were dissolved in 3 ml of 0.1 M acetate buffer, pH 3.0, containing 0.05 M NaCl and 3 M urea and were applied to a heparin-Sepharose column (2 ml) equilibrated with the same buffer. Bound peptides were eluted with a linear 0.1-2.0 M NaCl gradient, dialyzed against 0.05% SDS, and lyophilized. The dried samples were dissolved in 0.06 M Tris buffer, pH 6.8, containing 0.5% SDS, 10% glycerol, 5% 2-mercaptoethanol and heated to 90 "C for 3 min to reduce disulfide bonds.
Human LRP was prepared from solubilized placental membranes by affinity chromatography using Sepharose-coupled methylamine-treated a,-macroglobulin (30). Elution was in 150 m M NaC1,5 m M EDTA, 10 m sodium phosphate, 0.6% CHAPS, pH 6.0. The 40 kDa receptor-associated protein was removed by heparin-Sepharose at pH 8.0 (30). LRP was iodinated to a specific activity of about 6 x lo1, Bq/mol as described previously (5) using chloramine T as the oxidizing agent.
Electrophoresis, Blotting Procedures, and Protein Sequence Analysis-Slot blots of LpL and fusion proteins on polyvinylidene difluoride (PVDF) membranes (Millipore Corp.) were performed using a Bio-Rad vacuum blotter (5). Electrophoresis of fragmented LpL was performed in the Tricine-SDS-PAGE system (33) without urea and with 10 m dithioerythritol. The composition of the separating gel was 15.5% acrylamide, 1% N,W-bismethyleneacrylamide. Electroblotting of resolved LpL fragments onto PVDF membranes followed previously published procedures (5). NH,-terminal sequence analysis of electroblotted fragments was made on an Applied Biosystems (Foster City, CA) 477A pulsed-liquid sequencer with an on-line phenylthiohydantoin 120A analyzer. Tricine-SDS-PAGE (18.8% acrylamide, 1.2% bisacrylamide) of fusion proteins was performed as for bLpL fragments except that 6 M urea was present.
Expression of Fusion Proteins Containing hLpL Fragments-Nucleotide sequences encoding residues 378411,378423, or 378-448 of hLpL were amplified by polymerase chain reaction from pUC18 con- as the COOH-terminal primer. The fidelity of the polymerase chain reaction-generated products was confirmed by cDNA sequencing (34). The fragments were cut with BamHI and Hind111 (Boehringer Mannheim) and subcloned into the Escherichia coli T, expression vectors (35,36) pT,H,FX (fusion protein designated F-LpL), pT,C,,H,FX (fusion protein C,,-LpL), or pT,C,MLCH,FX (fusion protein C,,MLC-LpL). H,FX refers to the hexahistidine-Factor X substrate sequence MGSH,SIEGR. C, refers to the NH,-terminal30 aa of the lambda C,, phage protein, and MLC refers to the NH,-terminal 116 aa of chicken myosin light chain. Expression and purification of the fusion proteins were performed as described (19) except that 2 m glutathione/0.2 ~l l~ oxidized glutathione was added to allow disulfide reshuffling. The eluted fusion proteins were dialyzed into the aqueous incubation buffer. The fusion proteins containing Incubations-Incubations of PVDF membranes with immobilized proteins and labeled ligands followed by autoradiography were performed as described (5) using 140 m M NaCl, 10 m Hepes, 2 m M CaCl,, 1 m M MgCl,, 1% bovine serum albumin, pH 7.8 (buffer A). Immunoblotting was performed as described (18,30) using the monoclonal 5D2 antibody. Incubations with bLpL or LRP immobilized in microtiter wells (5) were performed in 100 pl of buffer A. Hep3b hepatoma cells, normal CHO cells, CHO cells (CHO mutant 745 (37)) deficient in proteoglycans (courtesy of Dr. J. D. Esko, University of Alabama), and LDL receptordeficient fibroblasts from a patient with familial hypercholesterolemia (French-Canadian mutation, courtesy of Dr. J. Davignon, IRCM, Montreal) were incubated as described previously (4), and details are provided in the legend to Fig. 9. Cross-linking of labeled bLpL and fusion proteins to cells and autoradiography of electrophoretically resolved proteins followed previously published procedures (4). For detection of LRP immunoreactivity, we used specific polyclonal rabbit anti-LRP antibodies (38) directed against a recombinant fragment of the a-chain (aa 2500-2922) (39).
were iodinated as described for bLpL.  (24), was capable of binding both ligands individually a s described previously ( 5 ) . hLpL truncated with chymotrypsin and purified to yield hLpL":'"" (23) did not hind "'I-LRP or '"I-p-VLDL in agreement with previously puhlished results (5). As shown in Fig. lR, I2'II-LRP a s well as "'I-p-VLDL also bound to immohilized hLpL, and binding of the labeled receptor to hLpL was inhibited by heparin as well a s bLpL in solution (not demonstrated). We also analyzed binding to hHL due to its extensive homology with hLpL (40). Fig. 1R shows that '"I-LRP and '2'II-p-VLDL bound specifically to immobilized hHL purified from postheparin plasma and that binding of '""ILRP was inhibited by heparin and EDTAas well as bLpL in solution.

Identification of an LRP Binding Region in bLpL-
Since a slight contamination of plasma-derived hHL with hLpL or hLpL fragments could not be entirely excluded, immobilized hHL purified from the medium of human HUH, hepatoma cells was also incuhated with '"I-LRP, and the results were not distinguishable from those shown in Fig. 1R. We used immobilized human pancreatic lipase as a control, and binding of '""I-LRP was not detected (Fig. 1B ).
To identify the LRP binding region, bLpL was subjected to fragmentation procedures. Fig. 2A shows the result of partial chymotryptic cleavage followed by electrophoretic separation and electroblotting of the fragments. I2'I-LRP bound to a 13-kDa fragment identified as bLpL'"""") by amino-terminal aa sequencing, hut not to bLpL"!'2"'5", which previously has been shown to be the main carboxyl-terminal fragment produced (23). The approximately 45 kDa band containing bLpL":""' did not bind ""I-LRP. Fragmentation was next performed using CNRr since residues 379 and 425 of bLpL are methionines. Incomplete cleavage was suspected since methionine 425 is followed by a serine residue. Fig. 2R, lane 2, shows binding of '?'I-LRP to 5-and 8-kDa fragments and weak binding to a less abundant 30-kDa fragment. As shown in lane 3 , the 8-and 5-kDa fragments also bound a monoclonal antibody, 5D2, reacting within residues 398-407 of bLpL (28). None of the fragments bound '"I+-VLDL (Lana 4 ).
The mixture following cleavage with CNRr was then chromatographed on heparin-Sepharose using a NaCl gradient (Fig. 3A ), and the fragments in the consecutively eluted fractions 1-5 were resolved by electrophoresis, electroblotted, and incubated with I2"I-LRP. As shown in Fig. 3B. the 8-and 5-kDa receptor binding fragments, which were eluted a t 0.8-1.2 hl NaCI, were identified by aa sequencing as hLpL"""4'"n and bLpL:Im"m , respectively. This suggested that an LRP hinding site is within residues 380-425. The 18-20-kDa hand identified as bLpLllX-2XR (or ~LpLllX-:lll )did not bind I2'I-LRP in agreement with the result obtained with hLpL' "'". We also sequmccd the minor 30-kDa fragment capable of binding '"'I-LRP since we were unable to detect reactivity with the 5D2 antihod!, and LRP binding therefore might occur to a site not prcsmt within residues 380-425. However, the 30-kDa fragment was identified as bLpL"'d'5" (Fig. 3B). We ascribe the apparent lack of reactivity (cf: Fig. 2R) to the low abundance of the the 30 kDa band as compared with the 8-and 5kDa bands. We found that the 5D2 antibody not only inhibits lipase activity of LpL, perhaps by preventing its access to emulsified lipid substrates (41), but also inhihits the hinding of '"'I-LRP to immobilized bLpL.2 Prior binding of '"I-LRP to the weak 30 kDa band may therefore block binding of the 5D2 antibody. As shown in Fig.  3R, fragments containing peptides NH,-terminal to residue 311, as well as the fragment comprising residues 426450 of bLpL, did not bind '"'I-LRP.
The results show that the region comprising residues 380-450 of bLpL contains a binding site for LRP. Since bLpL""!'" is unable to bind to LRP even though its overall structure is maintained (23), the presence of additional LRP binding sites is highly unlikely. In addition, the results suggested that the LRP binding site is within residues 380-425 of bLpL. However, due to the qualitative nature of the blotting procedures, it was not * A . Nykjzr, unpublished ohservation. possible to assess the affinities of hLpL7""""" and hLpI,""' c'' . To address this question, and in order to obtain material for functional studies, we expressed fusion proteins containing carboxyl-terminal fragments of hLpL.
Binding of Fusion Proteins Containing h L p L F m g m m t s t o Purified LRP-Fusion proteins containing h1,pL"" -Its (homolpressed in E. coli. Fig. 4 shows the purity of the preparations as evaluated by Tricine-SDS-PAGE and their reactivity with the 5D2 antibody. The fusion proteins were slot hlotted onto PVDF membranes and, as shown in Fig. 5. ""I-LRP bound to t h r fusion proteins containing hLpL"" .''I" and hLpL"7' . Little or no hinding was observed to hLpL";"'"', and no hinding could br detectrd to expression products not containing hI,pI, sequences. Hrparin (1 uniffml) abolished the binding to hLpL"" I.'' and hLpIA!7'."' in agreement with the previous observation that hrparin afinity is partially linked with the carboxyl-terminal foldingdomain (14,42). The 40-kDa a,-macroglohulin receptor-associated protein, previously reported to inhihit hinding of hLpL to LRP ( 5 1 , also inhibited the hinding of '"'I-LRP to hLpI,'"' -'I' (not shown I. The results obtained with thr fusion proteins demonstrate the presence of a LRP binding site within residues 37842.3 of hLpL in agreement with the results on the fragmrnts of wild-type bLpL. Moreover, residues within aa 41242.7 and. a s drducrd from the lack of '"'I-LRP binding to hLpI,!'" .""' ( r r Fig. 2.4 1. residues within aa 378-390 of hLpL are necessary for t h r hinding to LRP. Binding of "'I-B-VLDL was not observed to the fusion proteins (Fig. 5 ) in agreement with the results ohtained with the fragments of bLpL.
Inhihition of '?51-LRP Binding. hut ,Vat '"'I-13-V12DI, Binding. to hLpL-In order to obtain estimates of the affinities for hinding to LRP, we measured the abilities of the fusion proteins in solution to inhibit binding of ".'I-LRP to hLpL immohilizrd in microtiter wells. Fig. 6A shows that the fusion proteins contencies whereas the potency of hLpL'"' -I1' was at least 25-fold lower. The concentration of monomeric hLpL causing halfmaximal inhibition of '"'I-LRP to the immohilizrd hI,pl, was similar to that of the fusion proteins containing hI,pI,','' '"I" or was severalfold higher, possibly due to the intrraction with two sites in LRP (5) or to a conformation of each site drpendcnt on the dimeric structure. Monomeric hLpL was soluhle in the aqueous buffer only to the concentration that could inhihit binding of '"'I-LRP by ahout 50'7 ( Fig. 6A J, and exprrimcmts were performed in the presence of taurodeoxvcholate 1 mlc i to increase the soluhility (Fig. 6Bl. Surprisingly, the drtergrnt caused a 10-15-fold increase in the inhihitory potcncirs ofdimeric as well a s monomeric hLpL whereas the potency of hLpL"'""'4R remained unchanged. LpL has been shown to bind to deoxycholate (4.7). and the present results suggest that interaction of the detergent in regions within residues 1-378 of hLpL may cause an increase in its afinity for hinding to t h r receptor. These experiments show that residues 42.3448 are not important for the affinity of the fragment. The higher affinity of LpL, a s compared with the fragment, appears drpendent on its dimeric structure, and the afinity o f LpL may hr further increased by detergents such as hile acids.
Analogous experiments (not demonstrated) wrre prrformed using ""I-P-VLDL. The results showed that whereas IO nv hLpL in solution inhibited hinding of the labeled lipoprotein to immobilized hLpL half-maximally, no inhihition was ohtained when using 2 pv fusion protein containing h1,pL'" -I". This result shows, together with the lack of hinding of ' ' ! " I -~-~' L I~I~ to fragment immobilized onto PVDF mrmhranrs. that  lowed by electrohlotting and incuhation with the 5D2 antihody. Designations of fusion proteins: F, the hexahistidine-Factor X substrate sequence H,FX; C,,, 30 NH2terminal residues of the A C,, protein linked to H,FX; C,,MLC, C,, linked to the 116 NH,-terminal residues of chicken myosin light chain. Left, Coomassie staining; right, binding of the 5D2 antibody to the electrohlotted proteins. Inhihition of LpL-mediated Binding of P-VLDL to Purified LRP-Since the aim was to use the fragment for inhibition of LpL-mediated binding of lipoprotein, we first analyzed whether VLDL to LRP immobilized in microtiter wells. As shown in Fig.  7A, the presence of up to 4 p~ fragment in the incubation medium did not cause significant binding of the labeled lipoprotein to the immobilized receptor whereas bLpL mediated the binding of I2'II-P-VLDL to the receptor as reported previously (5, 6). The binding of '"I-p-VLDL was maximal at about 10 nM bLpL and half-maximal at about 3 nM bLpL, in close agreement with previously reported results using "'I-labeled normal human VLDL (6). As shown in Fig. 7B, the binding of '2511-p-VLDL mediated by 4 nM bLpL was essentially blocked by the fragment at concentrations of 1-2 p~, which will cause occupation of a large fraction of the immobilized LRP molecules, whereas the expression product not containing hLpL sequences had no effect. Thus, the fragment can bind to LRP and inhibit the LpLmediated binding to the receptor, apparently because it binds poorly to lipoproteins.
Binding of LpL"'"A"JH and Inhihition of LpL-mediated p-VLDL Binding and Uptake in Cells-We used the following cell types: Hep3b hepatoma cells since previous experiments with hepatic cells have shown cross-linking of "'I-hLpL to a large membrane protein compatible with LRP as well as bLpLmediated o-VLDL binding and uptake (4); CHO cells since a mutated form not expressing proteoglycans is available (CHO mutant 745 (37)); and fibroblasts from a patient with homozygous familial hypercholesterolemia ( F H ) since they lack functional LDL receptors. We first used chemical cross-linking, followed by SDS-PAGE and electroblotting, to elucidate the nature of binding of bLpL and fragment to Hep3b cells.
The followingexperiments wrrr drsignrd to analvzr thr rffcct of unlnhtled fusion protcins containing hLpI,"" -I t -on the LpLmediated binding and uptakr of [3-VLI>I, in crlls. In initial cxperiments (not shown) we mrasurrd thr hrparin-rrlrasablr binding at 4 "C of the '251-lahelrd fragment to Iirp3h crlls and normal CHO cells and found a 75-8.5f-; inhihition hv 1 p11 unlabeled fragmrnt as wrll as hy 200 n\c hLpI,. This rrsult confirmed the similar nature of the binding of hLpL and thr fragment to the cell surfacrs. Othrr rxprrimrnts showrd that thr fusion proteins containing hLpL"" -'Ir and hLpL!" w ( 2 p\ll did not mediate binding of ".'l-fi-VLDL in HrpSh crlls and normal CHO cells, in agreement with the rrsults ohtained with purified LRP. Ry contrast, as drmonstrntrd in Fig.   9, clpprvpnnri, hLpL"'H44* inhibited the LpL-medintrd hrparin-rrlrasahlr hinding of "'I-p-VLDL in normal CHO crlls. and thc sprcificity of the hinding is shown by the inhihition in thr prrsencc of unlabeled o-VLDL. Since heparin-relrasnhlr binding may occur hoth to proteoglycans and to LRP, we prrformctl parallrl cxperiments (not shown) in proteoglycnn-dcficirnt CHO rrlls. and the results showed a 56-77T reduction in hinding of Iahrled fusion proteins containing hLpL!" .'" in a p r r m e n t with prrvious experiments using hLpL (44,. Thrse rcsults arc compatible with binding of the fragment to protroglvcans and partial competition with LpL at this levrl in addition to t h r comprtition for binding to LRP. The LpL-mrdiatrd hinding of '~'*'I-\~-VLI)I, was in different experimrnts rrducrd hy 3 f " K in t h r protroglycan-deficient mutant CHO crlls a s comparrd with thr normal CHO cells. The inhihition of LpL-mediatrd "7'I-[3-VLDL hinding by the fragment was difficult to assess in the mutant CWO crlls due to the low level of the signal. In one experiment with sufficient cell-associated radioactivity in the mutant cells, we found that the fragment (1 p~) inhibited the LpL-mediated 1251-P-VLDL binding by 60%.
Uptake of '251-p-VLDL was measured as the radioactivity associated with cells, and not released by heparin, following incubation a t 37 "C. As shown in Fig. 9, middle panel, hLpL378-448 inhibited the bLpL-mediated uptake in Hep3b cells in agreement with the inhibition of the LpL-mediated binding. Moreover, the fragment inhibited the incorporation of [14C]01eate into cholesteryl esters induced by P-VLDL and LpL in FH fibroblasts (Fig. 9, lower panel). This incorporation is a n indirect measure of lipoprotein uptake by assessment of the acyl-CoA:cholesterol acyltransferase activity known to be induced by cholesterol uptake (45), and the result shows that the inhibitory effect of the fragment in cells is not dependent on the presence of functional LDL receptors. DISCUSSION The present results show that residues 378-423 in the carboxyl-terminal folding domain of hLpL contain a site for binding to LRP. This conclusion is based both on the identification of the homologous LRP binding fragments of bLpL and on the binding of hLpL fragments expressed as fusion proteins as summarized in Fig. 1OA. We propose that this region contains the only site for binding to the receptor for the following reasons. First, ~LPL"~" did not bind to LRP, indicating that regions NH,-terminal to residue 380 of bLpL (residue 378 of hLpL) do not interact with the receptor. Second, the apparent affinity of hLpL378-423 was identical with that of hLpL378448, indicating the absence of accessory sites within residues 424-.
. 448 of hLpL. Third, the concentration required to inhibit binding of labeled LRP to immobilized bLpL half-maximally was similar for the fragments and for monomeric bLpL in aqueous buffer. Finally, the binding of lZ5I-LRP to immobilized bLpL was completely inhibited by the fusion proteins containing hLpL37&448 or hLpL378423, demonstrating that only sites inhibitable by the fragment could cause binding of bLpL to the receptor. However, the inhibitory potency of dimeric bLpL was higher than that of monomeric bLpL and the fragments. In addition, we observed that the addition of taurodeoxycholate caused a n increase in the affinity of the full-length bLpL (monomeric or dimeric) but not of the fragment. The mechanism of this effect remains to be elucidated. When taken together, the results indicate that full length dimeric LpL is required for optimal LRP binding affinity, even though the region comprising residues 378-423 of the hLpL monomer is likely to contain the only site interacting directly with the receptor. After completion of this study, Williams et al. (46) reported that the carboxyl-terminal folding domain of hLpL (residues 313-448) binds to LRP and mediates binding of normal human VLDL particles to the purified receptor immobilized in microtiter wells. The result concerning location of the LRP binding site is in agreement with the present results as well as the previous observation that binding of bLpL to LRP is dependent on an intact carboxyl-terminal folding domain (5). In contrast to the present results on hLpL378-448, hLpL313448 bound to li- poproteins and was capahle of mediating their hinding to LRP (461. However, whereas hinding of "'I-laheled human VLDL ( 6 ) and rahhit P-VLDI, (present results) to purified LRP was mediated hy hLpL in the concentration range of 1-10 nal hLpL, approximately 1000-fold higher concentrations of hLpL":' .'I" (1-6 p a l ) were required to cause efficient hinding of tho lipoproteins to the receptor (46). The l o w potency of hLp1,""' 'I' in terms of mediating binding of lipoprotein to LRP ma.v he due to a low aflinity of this fragment for hinding to the lipoproteins (46). Since we have previously reported poor binding of "!-' Ilaheled rahhit P-VLDL to hLpL":""' ( 5 , and of "'I-hl .p L1-:""' to r a t chylomicrons, it seems likely that regions hoth in the NH,terminal folding domain (14, 46) and in the carhoxyl-terminal folding domain are required for efficient hinding of lipoprotein to LpI,. In the present work, thr ahsence of binding of that essent.ially hlocked the hinding of hLpL to LRP in thc. solid phase assay, provided the hasis for its inhihition of LpL-mediated hinding of P-VLDL. In addition to the LRP hinding site, it is likely that hLpL:"" 'I"" has a heparin hinding site since heparin hlockrd the hinding to LRP and since LRP does not hind heparin (30). This conclusion is supported hv the observation that hindingofthe fragment was reduced in proteoglycan-deficient crlls, and it is in agreement with previous ohservations on chimeras of hLpL and rat H L ( 4 2 ) as well as predictions from three-dimensional models of hLpL (47). I t seems likely that part of the inhihitory effect of tho fragment on the LpL-mediated hinding of [j-VI,nI, in cells may he due to comprtition for hinding of Lpl, to proteoglycans. This