THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1986 by The American Society of Biological Chemista, InC.

Vol. 261, No. 36, Issue of December 25, pp. 17127-17133.1986 Printed in U.S.A.

Detection and Quantitation of Low Density Lipoprotein (LDL) Receptors in Human Liver by Ligand Blotting, Immunoblotting, and Radioimmunoassay LDL RECEPTOR PROTEIN CONTENT IS CORRELATED WITH PLASMA LDL CHOLESTEROL CONCENTRATION*

(Received for publication, June 25, 1986)

Anne K. Soutar$, Kristin Harders-SpengeIg, David P. Wades, and Brian L. Knights From the $Medical Research Council Lipoprotein Team, Hammersmith Hospital, London, United Kingdom and the Medizinische Poliklinik der Universitat Munchen, Miinchen, Federal Republic of Germany

Low density lipoprotein (LDL) receptor activity has been detected and identified in human liver samples by ligand blotting with biotinylated lipoproteins and by immunoblotting with a monoclonal antibody raised against the bovine adrenal LDL receptor. The molec- ular weight of the human liver LDL receptor, approx- imately 132,000 on nonreduced polyacrylamide gels, is identical to that of LDL receptors detected in normal human skin fibroblasts by the same methods. LDL receptor-dependent binding activity in human liver samples has been semi-quantitated by integrating the areas under the peaks after scanning photographs of ligand blots, and receptor protein determined by ra- dioimmunoassay with purified bovine adrenal LDL re- ceptor protein as the standard. There was a highly significant correlation between the values obtained by each method for seven different liver samples (r = 0.948). The LDL receptor protein content of liver membranes from 10 subjects as determined by radioim- munoassay was inversely related to the plasma LDL cholesterol concentration (r = -0.663, p = 0.05) but not to other plasma lipid values, including total plasma cholesterol, high density lipoprotein cholesterol, or plasma triglyceride concentrations.

It is now generally believed that the majority of plasma low density lipoprotein (LDL)’ catabolism occurs by LDL recep- tor-mediated uptake in the liver. This view is based on many observations of the uptake and clearance both by cultured cells and in whole animals of normal LDL and of chemically modified LDL (1-3). Confirmation of the role of hepatic LDL receptors in plasma LDL catabolism was obtained when transplantation of a normal liver into a subject with familial hypercholesterolemia, in whom there was previously a total lack of LDL receptor function due to a genetic defect, essen-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

5 Recipient of a grant from Wissenschaftliches Herausgeberkolle- gium der Miinchener Medizinischen Wochenschrift e.V.

The abbreviations used are: LDL, low density lipoprotein; VLDL, very low density lipoprotein; BVLDL, very low density lipoprotein with &electrophoretic mobility from a cholesterol-fed rabbit; PMSF, phenylmethylsulfonyl fluoride; CHAPS, 3-[ (3-cholamidopro- pyl)dimethylammonio]-1-propanesulfonatq b-LDL, biotin-modified human LDL; b-PVLDL, biotin-modified OVLDL; HDL, high density lipoprotein.

tialiy normalized LDL clearance (4). However, despite its implied importance in maintenance of plasma LDL concen- tration little is known of the factors that regulate or determine LDL receptor-dependent uptake in the liver in uiuo (5).

A major difficulty in studying LDL receptor activity in the liver has been the lack of suitable methods of measurement. Most assays for LDL receptor activity have depended on measurement of saturable or EDTA-sensitive binding of ‘*‘I- labeled LDL either to the surface of cultured cells (6) or to membrane fractions of tissues (7). However, with some he- patic membranes it has proved difficult to distinguish between LDL receptor-dependent and -independent binding of LDL (8,9), and the presence of other lipoprotein binding proteins in the liver has been suggested (10). We have shown recently that it is possible to identify LDL receptors with properties similar to those found in human skin fibroblasts or bovine adrenal cortex in hepatic membranes from dogs, rats, and rabbits by ligand blotting with biotin-modified lipoproteins (11) and that this method can be used to quantitate LDL receptors (12). We have also developed a radioimmunoassay for the determination of LDL receptor protein based on a double antibody sandwich technique that can be successfully employed with human skin fibroblasts (13).

In this paper we describe the identification of LDL recep- tors in human liver from a number of different subjects and show that LDL receptor activity in human liver membranes determined by either radioimmunoassay or ligand blotting is inversely correlated with the plasma LDL concentration.

MATERIALS AND METHODS

Preparation of Human Liver Membrane Samptes-Human liver biopsies were obtained from patients undergoing elective surgery (as described in Table I), stored in liquid N2, and transported in dry ice. The portions of liver used in these studies were macroscopically normal, but the presence of a low percentage of abnormal cells could not be totally excluded. For the preparation of the membrane fraction, the samples (4-10 g, wet weight) were finely chopped, minced, sus- pended in buffer (10 ml/g, wet weight; 20 mM Tris-HC1, pH 8.0, 0.15 M NaC1, 1 mM CaC12, and 1 mM phenylmethylsulfonyl fluoride (PMSF)) and homogenized at 4 ‘C with a Polytron homogenizer (4 x 15 s, setting 8). The homogenate was cleared of debris by centrif- ugation for 5 rnin at 500 X g, filtration through nylon mesh, and centrifugation for 15 min at 8000 X g. The membrane fraction was prepared from the 8000 X g supernatant by centrifugation for 60 min at 4 “C in a Beckman 70 Ti rotor at 40,000 rpm (g.” = 115,000). Membrane pellets were stored frozen in liquid N2.

Other Samples-Human skin fibroblasts were maintained in cul- ture and preincubated in lipoprotein-deficient medium for 48 h as described previously (14). Cells from ten 9-cm diameter dishes were washed twice with cold phosphate-buffered saline (Gibco, Europe), scraped from the dishes, and pelleted by centrifugation for 5 min at

17127

17128 Low Density Lipoprotein Receptors in Human Liver

400 X g. The cell pellet was solubilized in 200 g1 of 1.6% (w/v) Triton x-100, 1.5 mM PMSF, 0.3 mM leupeptin, and 5 M urea as described by Daniel et al. (15). LDL receptor protein from bovine adrenal cortex was purified by DEAE-cellulose chromatography and affinity chro- matography as described by Schneider et al. (16).

Blotting Procedures-For ligand blotting and immunoblotting, liver membrane pellets were solubilized in 1.0 ml of 50 mM Tris- maleate, pH 6.0, containing 2 mM CaC12, 1% (w/v) Triton X-100 and 1 mM PMSF by homogenization through a 19-gauge and a 25-gauge needle. After centrifugation for 60 min at 40,000 rpm in a Beckman 50.3 rotor at 4 "C, the protein content of the supernatant extract was determined by the method of Bensadoun and Weinstein (17) with bovine serum albumin as standard. Samples for electrophoresis were adjusted to give 1% (w/v) SDS and 10% (w/v) glycerol. Electropho- resis, transfer to nitrocellulose membranes, and ligand blotting with biotinylated lipoproteins were carried out as described previously (12). For immunohlotting, anti-receptor IgG-4B3 was purified from ascites fluid and labeled with '%I as described (18). Nitrocellulose blots were incubated with labeled antibody for 2 h (2 pg of antibody/ ml, specific activity, 7800 cpm/ng) and washed as described by Be- isiegel et al. (19). Bands were visualized by autoradiography with preflashed film and an intensifying screen (18).

Radioimmunoassay for LDL Receptor Protein-The double anti- body radioimmunoassay was carried out by the standard assay pro- cedure as described previously (13). Human liver LDL receptor, for comparison with the purified bovine adrenal LDL receptor used as standard, was partially purified by DEAE-cellulose chromatography and affinity chromatography from 77 g of human liver using the procedures described by Schneider et al. (16) with the exceptions that the membranes (5.98 g) were solubilized in 50 ml of 50 mM Tris- maleate, pH 6.0, containing 2 mM CaC12, 1 mM PMSF, and 30 mM CHAPS, and the DEAE-cellulose column was washed and eluted with buffers containing 30 mM CHAPS instead of Triton X-100 or n-octyl glucoside.

Samples of human liver membranes for immunoassay were solu- bilized as described above for the blotting procedures, and the extracts were then adsorbed onto small columns of DEAE-cellulose. The columns were washed with buffer containing 30 mM CHAPS and eluted with 0.4 ml of the same buffer containing in addition 0.5 M NaCl. The eluate was desalted on a small column of Sephadex G-25 equilibrated in the same buffer. This procedure has been described in detail previously (13).

RESULTS

Identification of LDL Receptors in Human Liver-Extracts of human liver membranes were subjected to electrophoresis on 7.5% polyacrylamide gels in the presence of SDS but without 2-mercaptoethanol, and the separated proteins were transferred to nitrocellulose membranes. After incubation with either biotin-modified human LDL (b-LDL) or biotin- modified rabbit P-very low density lipoprotein (b-PVLDL) and then development with biotinylated streptavidin-peroxi- dase, several bands were detected. Strong bands of activity were associated with a protein of approximately M, 132,000 and with a doublet of approximately M, 68,000. This doublet of proteins bound the biotinylated streptavidin-peroxidase complex directly on strips containing liver membrane extracts that had not been incubated with biotinylated lipoprotein ligands (Fig. lb). These streptavidin-binding proteins, pre- sumably biotin-containing enzymes, have not been observed in extracts of bovine adrenal membranes or in solubilized human skin fibroblasts (Fig. lb). The Mr 132,000 protein was indistinguishable in electrophoretic mobility from the LDL receptor protein detected in normal human skin fibroblasts by ligand blotting (Fig. lb) . The band associated with this protein was much stronger when b-PVLDL rather than b- LDL was the ligand, and thus subsequent experiments were carried out with P-VLDL. Binding of b-BVLDL to the M, 132,000 was inhibited by 10 mM EDTA and by an excess of unmodified PVLDL, but not by high density lipoprotein (HDL) (Fig. IC). Binding of b-PVLDL was greatly reduced in the presence of an excess of unmodified LDL, while binding

of b-LDL was completely inhibited by excess unmodified PVLDL (data not shown). Faint bands were sometimes de- tectable with proteins of high M, that were not well resolved in this gel system (Fig. la). There was considerable variation in the amounts of these bands relative to M, 132,000 band that were detectable with different preparations of ligand and with different preparations of the same liver sample; with some preparations, the high M, bands were not detectable. Variability in the presence of these proteins made it difficult to study them further.

Confirmation that the LDL binding protein of M, 132,000 in normal human liver is the same as the LDL receptor in human skin fibroblasts was obtained by immunoblotting with a monoclonal antibody raised against the bovine adrenal LDL receptor, anti-receptor IgG-4B3, that is known to recognize the human receptor (18). When nitrocellulose blots of solu- bilized human liver membranes or human skin fibroblasts were incubated with lZ51-labeled antibody, a single band of radioactivity was detected by autoradiography associated with a protein of M , 132,000 in each case (Fig. 2).

Quantitation of LDL Receptors in Human Liver Mem- branes-Our preliminary studies of ligand blotting of mem- brane extracts from several different liver samples suggested that there ' might be significant variation in the amount of LDL receptor present. We, therefore, attempted to quantitate LDL receptors by ligand blotting and by a double antibody radioimmunoassay in samples of liver from a number of patients undergoing surgery, as shown in Table I.

We have established previously that the intensity of the band obtained when different amounts of purified bovine adrenal membrane LDL receptor are ligand blotted with b- LDL or b-PVLDL is proportional to the amount of protein applied to the gel (12). A similar result was obtained with solubilized human liver membranes (Fig. 3), and thus it was possible to compare the LDL receptor-dependent binding activity in different samples of human liver. The membrane fraction was prepared from liver samples from eight of the subjects shown in Table I. When equal amounts of soluble membrane protein were applied to the gel, there were consid- erable differences between subjects in the amount of LDL receptor that could be detected by ligand blotting (Fig. 4). There was also variation in the amount of streptavidin bind- ing proteins that could be detected, but the ratio of LDL receptor protein to streptavidin binding protein was not the same in each sample, suggesting that the differences observed were not due to differential solubilization of membrane pro- teins. The relative amount of LDL receptor-dependent bind- ing of b-BVLDL in different liver samples could be assessed by integrating the area under the peak after scanning photo- graphs of the blots.

The radioimmunoassay for LDL receptor protein depends on a solid phase double antibody sandwich assay with anti- receptor IgG-4B3 used to coat the wells of the plate and lZ51- labeled anti-receptor IgG-1OA2 as second antibody (13). When LDL receptor protein was assayed in directly solubilized liver membranes, the curve obtained for the amount of lZ5I-labeled antibody bound with increasing amounts of membrane protein was not parallel with that observed for purified bovine adrenal LDL receptor (Fig. 5). However, when human liver membrane LDL receptor was partially purified by chromatography on DEAE-cellulose and LDL-Sepharose, the line obtained for the amount of second antibody bound was now parallel with that for bovine adrenal receptor over a wide range of dilutions (Fig. 5). Although the bovine adrenal LDL receptor can be purified to near homogeneity by the purification procedure described, the human liver LDL receptor was still less than

Low Density Lipoprotein Receptors in Human Liver a b C

17129

O r i g i n .

205,000 -

116,000.

97,000 ‘

66,000’

45 ,000 ’

114 * ’ Ir R

1 2 1 2 3 4 5 6 1 2 3 4 5 6

FIG. 1. Ligand blotting of human liver membrane proteins with biotin-modified lipoproteins. Protein samples as described below were fractionated by gel electrophoresis (7.5% acrylamide) in the presence of SDS without 2-mercaptoethanol and transferred to nitrocellulose membranes. Blots were incubated with biotinylated @VLDL (10 pglml) or biotinylated LDL (20 pg/ml) and then developed by incubation with biotinylated streptavidin- peroxidase complexes. The molecular weight markers were myosin, M, 205,000; @-galactosidase, M, 116,000; phosphorylase b, M, 97,400; and bovine serum albumin, M, 66,000. (a ) Solubilized human liver membranes (150 pg of soluble protein/lane) incubated with b-@VLDL (lane 1 ) or b-LDL (lane 2). ( b ) Solubilized human liver membranes (150 pg of soluble protein/lane, lanes 1 and 41, solubilized normal human skin fibroblasts (115 pg of protein/lane, lanes 2 and 5), and DEAE-purified bovine adrenal LDL receptor protein (67 pg of protein/lane, lanes 3 and 6) incubated with (lanes 1-3) or without (lanes 4-6) b-PVLDL. (c) Solubilized human liver membranes (150 pg of soluble protein/lane) incubated with b-@VLDL alone (lanes 1, 3, and 5) or in the presence of unmodified PVLDL (1.0 mg/ml, lane 2), EDTA (10 mM, lune 4 ) , or unmodified apoE-free human HDL (1.0 mg/ml, lane 6) .

10% pure as estimated from SDS-polyacrylamide gel electro- phoresis. Since insufficient human material was available to attempt further purification, subsequent assays of the LDL receptor content of human liver samples were related to a given volume of the partially purified human LDL receptor. The amount of LDL receptor protein in the partially purified human sample was estimated by comparison with a purified bovine adrenal LDL receptor sample from the parallel region of the curves shown in Fig. 5.

Since some component of directly solubilized human liver membranes interfered in the immunoassay (Fig. 5 ) it was necessary to effect a partial purification of the samples. Mem- branes were solubilized, adsorbed onto small columns of DEAE-cellulose, eluted with 0.5 M NaC1, and the eluates then desalted on small columns of Sephadex G-25 as described under “Materials and Methods.” Four dilutions of each sample were assayed in duplicate, and the results from three repre- sentative samples are shown in Fig. 6. The lines obtained in the immunoassay for the liver membrane samples treated in this way were each parallel with that obtained for the affinity- purified human liver LDL receptor preparation. When 10 different samples were assayed using duplicate membrane preparations for each subject, the variation between the du- plicates was 8.9 t 1.2% (mean f S.E.; range 4.6-15.7). The mean value for the LDL receptor protein content of the 10 liver samples of the subjects shown in Table I was 20.2 t 5.1 ng/mg of membrane protein (mean t S.D.). However, there was a wide variation in the values (range 10.7-29.4). To

protein that could be assayed in different samples were related to differences in LDL receptor-dependent binding of b- PVLDL in solubilized membranes detected by ligand blotting, the two values were compared. As can be seen from Fig. 7, there was a highly significant correlation ( r = 0.948, p < 0.001), which suggested that there were no differential losses of receptor protein from the samples during the DEAE- adsorption and elution procedure necessary for the immu- noassay.

The LDL receptor protein concentration in liver mem- branes was compared with the values for the concentration of lipoproteins and lipids in the plasma of the different subjects. There was no correlation between LDL receptor content of liver membranes and total plasma cholesterol, plasma triglyc- eride, or HDL cholesterol concentrations. However, there was a significant inverse correlation ( r = -0.663,~ < 0.05) between LDL receptor protein content and plasma LDL cholesterol concentration, as shown in Fig. 8.

DISCUSSION

In this paper we have demonstrated that LDL receptors of the type present in normal human skin fibroblasts can be detected in human liver membranes. The human liver LDL receptor has an apparent M, of 132,000 on nonreduced poly- acrylamide gels and can be detected by ligand blotting with biotinylated lipoproteins and by immunoblotting with a monoclonal antibody to the bovine adrenal LDL receptor. Binding of rabbit PVLDL and human LDL to this receptor

determine whether the differences in amount of LDL receptor protein on nitrocellulose blots is both saturable and EDTA

17130 Low Density Lipoprotein Receptors in Human Liver

Origin

205,000

132,000 130,000

116,000

97,000

66 ,000

45,000

29,000

1 2 3 FIG. 2. Immunoblotting of human liver, human skin fibro-

blasts, and bovine adrenal LDL receptors. Solubilized human liver membranes (150 pg of protein/lane, lane I ) , solubilized normal human skin fibroblasts (115 pg of protein/lane, lane 2), and bovine adrenal LDL receptor (67 pg of DEAE-purified protein/lane, lane 3 ) were fractionated by electrophoresis on 7.5% polyacrylamide gels in the presence of SDS and the absence of 2-mercaptoethanol and the proteins transferred to nitrocellulose. Blots were incubated with lZ5I- labeled anti-receptor IgG-4B3 (2 pglml, specific activity, 7800 cpm/ ng) and developed by autoradiography for 3 h as described under “Materials and Methods.” The molecular weight markers are as described in the legend to Fig. 1.

TABLE I Details of subiects from whom liver samDles were obtained

Plasma cholesterol”

Subject Sex Age Reason for surgery LDL

cholesterol mmollliter

H. H. M 51 Metastasis of renal ca.6 3.15 2.26 A. M. F 60 Liver cell ca. 2.35 1.63 C. M. M 66 Metastasis of rectal ca. 3.54 2.66 G. B. M 58 Metastasis of unknown 5.04 4.03

L. U. F 74 Benign liver cyst 3.41 2.45 R. 0. M 61 Solitary liver metas- 2.12 1.40

K. K. M 71 Metastasis of colon ca. 3.75 2.92 H. F. M 66 Metastasis of rectal ca. 5.17 4.08 B. K. F 41 Metastasis of sigma ca. 4.19 2.84 H. Ho. F 48 Metastasis of mam- 3.15 1.96

origin

tasis

mary ca. Determined after 18-h fast a t time of surgery.

* ca., carcinoma.

sensitive. Hoeg et al. (10) failed to detect a fibroblast-like LDL receptor in normal human liver membranes by ligand blotting with ‘251-labeled LDL, presumably due to differences in their solubilization procedures compared with those de- scribed in this paper, but observed two high M, LDL-binding proteins. We have also detected some high M, binding pro- teins, but only as a minor and variable component of total binding activity even under apparently identical conditions. Indeed, the variability made further investigations to deter- mine the nature of these proteins difficult. It is unlikely that

c

a b c d e I I I I

0 100 200 300

Extract applied to gel (pg protein)

FIG. 3. Quantitation of LDL receptor activity in human liver membranes by ligand blotting. Solubilized human liver membranes (50-250 pg of protein/lane) were fractionated by electro- phoresis on 7.5% polyacrylamide gels in the presence of SDS but without 2-mercaptoethanol and the proteins transferred to nitrocel- lulose. The blot was incubated with b-PVLDL (10 pg/ml) and devel- oped with biotinylated streptavidin-peroxidase complexes. Black and white photographs of the blots were scanned and the areas under the peaks integrated (lane a, 50 pg; lane b, 100 pg; lane c, 150 pg; lane d, 200 pg; lane e, 250 pg).

(a) (b)

r=-. a -

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

FIG. 4. Comparison of LDL receptor activity in different human liver samples by ligand blotting. Nitrocellulose blots of samples of human liver membranes (150 pg of soluble membrane protein/lane) were incubated with (a ) or without (b ) b-PVLDL (10 pg/ml) and developed with biotinylated streptavidin-peroxidase com- plexes as described under “Materials and Methods.” Samples were from subjects listed in Table I as follows: lanes 1-8 in order: G. B., K.K.,H.Ho,B.K.,H.H.,A.M.,H.F.,andR.O.

these proteins are related to the LDL receptor since Hoeg et al. (10) observed them in a liver sample from a subject with familial hypercholesterolemia. It is also unlikely that these binding proteins constitute a second receptor protein specific for LDL that is unique to the liver (10) since we have observed previously that two proteins of similar molecular weights co- purify with the LDL receptor from bovine adrenal cortex

Low Density Lipoprotein Receptors in Human Liver 17131

_i Samplelwell (uII

20 200

LDL receptor proteinlwell ( n g l e - 0

FIG. 5. Detection of LDL receptor protein in solubilized human liver membranes by radioimmunoassay; comparison between purified bovine adrenal LDL receptor protein and affinity-purified human liver LDL receptor protein. Samples of DEAE-purified bovine adrenal LDL receptor protein (O), human liver LDL receptor protein purified by DEAE-cellulose chromatog- raphy and affinity chromatography (0), and human liver membranes solubilized in 50 mM Tris-maleate, pH 6.0,2 mM CaC12, 1 mM PMSF, and 30 mM CHAPS (A) were diluted as shown in the figure with 50 mM Tris-maleate, pH 6.0, containing 2 r n ~ CaClz and 30 mM CHAPS. Samples (100 pl) were incubated in wells coated with anti-receptor IgG-4B3, and bound receptor detected with ‘251-labeled anti-receptor IgG-10A2 (300,000 cpm/well). Results were corrected for background radioactivity observed in wells incubated with diluting buffer alone (605 cpm/well).

I 1 10 100

Samplelwell (ul)

FIG. 6. Immunoassay of human liver membrane LDL recep- tor protein. Samples of human liver LDL receptor protein purified by DEW-cellulose chromatography and affinity chromatography (0)

A. M., L. U., 0 , and G. B., V, from Table I) solubilized and treated and three representative human liver membrane samples (subjects

by adsorption onto DEAE-cellulose as described under “Materials and Methods,” were diluted as shown in the figure with 50 mM Tris- maleate, pH 6.0, 2 mM CaCI,, and 30 mM CHAPS. Samples (100 pl) were incubated in wells coated with anti-receptor I&-4B3 and bound receptor detected with ‘%I-labeled anti-receptor I&-1OA2 (300,000 cpm/well). Results were corrected for background radioactivity ob- served in wells incubated with diluting buffer alone (618 cpm/well).

during chromatography on DEAE-cellulose and that they bind to some extent to an LDL-Sepharose affinity column. How- ever, they are undoubtedly immunologically distinct from the

FIG. 7. Correlation between ligand blot assay and radioim- munoassay for the LDL receptor in human liver membranes. Samples of human liver membranes were assayed for lipoprotein binding activity as shown in Figs. 3 and 4 and by radioimmunoassay as shown in Fig. 6. The points represent the mean of assays of duplicate membrane preparations.

y = -0.094 x *29.7

r =-0.669 I

1

Plasma LDL cholesterol (mmolesll)

FIG. 8. Correlation between LDL receptor protein in human liver membranes determined by radioimmunoassay and the plasma LDL cholesterol concentration. Samples of human liver membranes were assayed for LDL receptor protein as shown in Fig. 6.

M , 130,000 LDL receptor in the bovine adrenal gland.2 We have also shown that it is possible to compare the LDL

receptor activity of different human liver samples both by semi-quantitative ligand blotting and by radioimmunoassay. The radioimmunoassay for the LDL receptor protein in hu- man liver membranes gives highly reproducible results with duplicate samples of liver, and there was also remarkable agreement between the relative values obtained for each sam- ple by ligand blotting and by immunoassay. It is of some interest that the extrapolated regression line in Fig. 7 would not pass through the origin, which might suggest that there is some small amount of LDL receptor protein that did not bind LDL. However, it is not strictly valid to extrapolate the data to determine the significance, if any, of the intercept value. Presumably values for those familial hypercholester-

* A. K. Soutar, D. P. Wade, and B. L. Knight, unpublished obser- vations.

17132 Low Density Lipoprotein Receptors in Human Liver

olemic subjects with a defective LDL receptor protein that could be recognized by the antibody would not fall on the regression line.

There are several problems inherent in the determination of the absolute values for the LDL receptor protein content of human liver by radioimmunoassay as described in this paper. First, we do not have a purified human liver LDL receptor standard, and it is possible that differences in the affinities for the human and bovine LDL receptors of the two antibodies used in the assay may exist. However, the partially purified human liver receptor preparation gave a line in the assay that was parallel with that of the bovine adrenal recep- tor standard over at least a 10-fold dilution range. Further- more, any differences between the absolute values for the bovine and human receptors would not affect the comparison of relative amounts of receptor in different human samples. Second, because of the relatively low amount of LDL receptor present in liver homogenates compared with, for example, cultured human skin fibroblasts, the assays can only be car- ried out on membrane fractions. Studies on the binding of Iz5I-LDL to different fractions of homogenized tissue have suggested that not all specific LDL binding activity is re- covered in a crude 8,000-100,000-g membrane fraction (20), and thus we cannot estimate the absolute amount of LDL receptor present in whole liver. We must assume, in making our comparisons, that the recovery of membrane-associated LDL receptor protein is the same for each sample. The values we have obtained for the mean LDL receptor protein content of human liver membranes compare favorably with values we have obtained previously for LDL binding activity. The value obtained by radioimmunoassay of 10 samples was 20.2 f 1.6 (mean k S.E.) ng/mg of membrane protein; assuming that one receptor of protein M , 93,000 (21) binds one particle of LDL of protein M, 550,000, this is equivalent to LDL receptor- dependent binding of approximately 120 ng of LDL. This compares with a mean value of 82.3 f 29 (mean f S.E.) ng of LDL bound saturably per mg membrane protein in a mem- brane binding assay (9). We have not been able to determine the LDL receptor content of human adrenal cortex mem- branes, but bovine adrenal membranes (13) contain approxi- mately 10-fold the specific activity of LDL receptor/mg of membrane protein compared with human liver membranes. Studies on the binding of LDL to membrane fractions from different bovine tissues (20) have shown that the specific binding activity of adrenal cortex also exceeded that of liver by approximately 8-fold when expressed per mg of membrane protein.

There is a wide variation in the amount of LDL receptor protein that can be measured in membranes prepared from different liver samples, and there is a weak inverse correlation between LDL receptor content of liver membranes and the plasma LDL cholesterol concentration of an individual. There is considerable evidence to suggest that LDL receptor activity in liver is an important determinant of LDL flux, not only through clearance of LDL itself but by direct removal from plasma of VLDL remnants from which plasma LDL is derived (1). Conversely, in cultured cells the concentration of LDL in the medium can regulate LDL receptor activity (5). Thus our observed correlation is not unexpected. However, it is not possible to distinguish from these results between the possi- bilities that plasma LDL concentration is determined by hepatic LDL receptor activity or that the concentration of LDL in plasma regulates receptor activity in the liver. It must be emphasized that the samples in this study were obtained from a heterogeneous group of subjects of both sexes who

were undergoing surgery for investigation of malignant dis- ease in the liver. While the samples taken for our purposes were apparently unaffected, we cannot exclude the possibility that some abnormal cells were present, and it has been shown that malignant white blood cells possess abnormally high LDL receptor activity (22). Furthermore, in some cases the plasma cholesterol values of the subjects were lower than normal, possibly due to their state of health. In no case, however, was the general condition of the subject poor. Thus, further work is needed to determine whether the same correlation exists in a group of healthy age- and sex-matched subjects. From studies in animals (8) and in human subjects (23) it might be expected that hepatic LDL receptor activity declines with age, and because of the known effects of estrogens (24) it would also be of interest to compare hepatic LDL receptor activity in men and women. The results shown in this study suggest that such investigations are possible. Our current immunoas- say procedure requires approximately 0.5 g of liver to give reliable values, while much less material is required for the semiquantitative ligand blot assay. Thus, should it prove possible to increase the sensitivity of the immunoassay, both assays could be employed on a wide basis among subjects from whom needle biopsies were being taken for other pur- poses.

Acknoukdgments-We are grateful to Dr. S. Gavigan for mainte- nance of cultured fibroblasts, and to Dilip Patel and Saro Niththyan- anthan for assistance with the radioimmunoassay. We are indebted to Prof. H. Denecke and the surgeons of Chirurgische Klinik, Gros- shadern, and to Mechthild Haberkamp for help in the collection of liver samples. K. H.-S. is grateful to Herr Prof. Dr. N. Zollner for his continued support and encouragement.

REFERENCES 1. Goldstein, J. L., Kita, T., and Brown, M. S. (1983) N. Engl. J.

Med. 309,288-295 2. Carew, T. E., Pittman, R. C., and Steinberg, D. (1982) J. Bwl.

Chem. 257,8001-8008 3. Dietschy, J. M., and Spady, D. K. (1986) in Receptor-medioted

Uptake in the Liuer (Greten, H., Winder, E., and Beisiegel, U., eds) pp. 56-65, Springer-Verlag, Heidelberg

4. Bilheimer, D. W., Goldstein, J. L., Gmndy, S. M., Starzl, T. E., and Brown, M. S. (1984) N. Engl. J . Med. 3 1 1 , 1658-1664

5. Goldstein, J. L., and Brown, M. S. (1984) J. Lipid Res. 25,1450- 1461

6. Goldstein, J. L., Basu, S. K., Brunschede, G. Y., and Brown, M. S. (1976) Cell 7,85-95

7. Basu, S. K., Goldstein, J. L., and Brown, M. S. (1978) J. Bid. Chem. 253,3852-3856

8. Mahley, R. W., Hui, D. Y., Innerarity, T. L., and Weisgraber, K. H. (1981) J. Clin. Znuest. 68, 1197-1206

9. Harders-Spengel, K., Wood, C. B., Thompson, G. R., Myant, N. B., and Soutar, A. K. (1982) Proc. Natl. Acud. Sci. U. S. A. 79 ,

10. Hoeg, J. M., Demosky, S. J., Lackner, K. J., Osborne, J. C., Oliver, C., and Brewer, H. B. (1986) Biochim. Biophys. Acta

11. Wade, D. P., Knight, B. L., and Soutar, A. K. (1986) Eur. J.

12. Wade, D. P., Knight, B. L., and Soutar, A. K. (1985) Biochem J.

13. Knight, B. L., Preyer, S., and Soutar, A. K. (1986) Biochem. J.

14. Knight, B. L., and Soutar, A. K. (1982) Biochem. J. 202 , 145-

15. Daniel, T. O., Schneider, W. J., Goldstein, J. L., and Brown, M.

16. Schneider, W. J., Beisiegel, U., Goldstein, J. L., and Brown, M.

17. Bensadoun, A., and Weinstein, D. (1976) Anal. Biochem. 70,

6355-6359

876.13-21

Bwchem. 169,333-340

229,785-790

238 ,405410

152

S. (1983) J. Biol. Chem. 2 5 8 , 4606-4611

S. (1982) J. Bid. Chem. 257, 2664-2673

241-250

Low Density Lipoprotein Receptors in Human Liver 17133

18. Soutar, A. K., and Knight, B. L. (1986) Biochem. J. 2 3 3 , 683- W., and Schneider, W. J. (1985) Ann. Reu. Cell Biol. 1, 1-39 22. Peterson, C., Vitols, S., Rudling, M., Blongren, H., Edsmyr, S.,

19. Beisiegel, U., Schneider, W. J., Brown, M. S., and Goldstein, J. and Skoog, L. (1985) Med. Oncol. Tumor Pharmacotherapy 2 ,

20. Kovanen, P. T., Basu, S. K., Goldstein, J. L., and Brown, M. S. 23. Miller, N. E. (1984) Lancet i, 263-268

21. Goldstein, J. L., Brown, M. S., Anderson, R. G. W., Russel, D. Chem. 254,11367-11373

690

L. (1982) J. Biol. Chem. 2 5 7 , 13150-13156 143-147

(1979) Endocrinology 104,610-616 24. Kovanen, P. T., Brown, M. S., and Goldstein, J. L. (1979) J. Biol.