Localization of low density lipoprotein receptors on plasma membrane of normal human fibroblasts and their absence in cells

Transcription

1 Proc. Natl. Acad. Sci. USA Vol. 73, No. 7, pp , July 1976 Cell Biology Localization of low density lipoprotein receptors on plasma membrane of normal human fibroblasts and their absence in cells from a familial hypercholesterolemia homozygote (low density lipoprotein-ferritin/electron microscopy/cell surface receptors/coated membrane regions/cholesterol metabolism) RICHARD G. W. ANDERSON, JOSEPH L. GOLDSTEIN, AND MICHAEL S. BROWN Departments of Cell Biology and Internal Medicine, University of Texas Health Science Center at Dallas, Dallas, Texas Communicated by E. R. Stadtman, April 21, 1976 ABSTRACT Monolayers of normal human fibroblasts were observed to bind ferritin-labeled low density li'protein (LDL-ferritin) at specific receptor sites on the cell surface membrane. When fibroblasts were incubated with LDL-ferritin at 40, more than 70% of the surface-bound ferritin cores were localized by electron microscopy to short segments of the plasma membrane where the membrane appeared indented and coated on both of its sides by a fuzzy material. These membrane segments corresponded to "coated regions" previously described in other cell types. Under the conditions of these experiments, an average of 55 LDL-ferritin particles were bound to each millimeter of plasma membrane in normal cells. In the presence of a 15-fold excess of native LDL, the number of bound ferritin cores was reduced by 75%, suggesting that the LDL-ferritin was binding to specific LDL receptor sites. Although fibroblasts from a patient with the homozygous form of familial hypercholesterolemia contained the same number of indented, coated membrane regions per millimeter of cell surface as did normal cells, no LDL-ferritin was observed to bind to the cell membrane in these mutant cells. The present ultrastructural data are consistent with previous biochemical and genetic evidence indicating that LDL exerts its regulatory action on cellular cholesterol metabolism in fibroblasts through an interaction with a specific cell surface receptor and that this receptor is defective in homozygous familial hypercholesterolemia fibroblasts. Moreover, the data suggest that the LDL receptor is localized to indented, coated regions of the plasma membrane that appear to participate in the adsorptive endocytosis of proteins. Evidence from biochemical and genetic studies indicates that low density lipoprotein (LDL), the major cholesterol-carrying protein in human plasma, is taken up by cultured human fibroblasts through a specific interaction with a receptor located on the cell's surface membrane (reviewed in refs. 1 and 2). When monolayers of normal fibroblasts are incubated with 1251-labeled LDL at 370, the lipoprotein binds with high affinity and specificity to a finite number of cell surface receptor sites, after which the receptor-bound lipoprotein is internalized by the cells and delivered to lysosomes where its protein and cholesteryl ester components are hydrolyzed (3-6). The liberated cholesterol in turn regulates two key enzymes in cellular cholesterol metabolism: (i) it suppresses the activity of 3-hydroxy-3-methylglutaryl coenzyme A reductase [mevalonate: NADP+ oxidoreductase(coa-acylating); EC ], reducing cellular cholesterol synthesis (7); and (ii) it activates an acyl- CoA:cholesterol acyltransferase (EC ), stimulating its own reesterification and storage within the cell as cholesteryl esters (8, 9). The initial binding reaction can be studied independently of the internalization step by incubating monolayers of fibroblasts with 125I-LDL at 40 so that cell surface binding occurs but the subsequent cellular uptake is prevented (3, 5). Fibroblasts from patients with the homozygous form of fa- Abbreviations: LDL, low density lipoprotein; FH, familial hypercholesterolemia. milial hypercholesterolemia (FH) lack functional LDL receptors and therefore show neither high-affinity binding at 40 (3-5) nor high-affinity receptor-mediated uptake of 125I-LDL at 370 (3-5). Because of their LDL receptor defect, these mutant cells do not respond to LDL by either suppression of the CoA reductase activity or activation of the cholesterol esterification reaction (10). Although the biochemical and genetic data have established that the regulatory action of LDL is a result of a receptormediated uptake of the lipoprotein, these techniques cannot localize the sites on the cell surface where initial binding takes place nor can they determine what mechanism the cell uses to ingest the LDL particles once they are bound to the cell membrane. To answer these questions, as well as to corroborate the biochemical data accumulated thus far, we have labeled human LDL with ferritin so that the binding and uptake of this lipoprotein can be followed with the electron microscope. The present studies show that the LDL-ferritin complex can be used to identify the LDL receptor sites on the plasma membrane in normal fibroblasts and that such receptor sites are absent in the FH homozygote cells. METHODS Cells. Cultured fibroblasts were derived from skin biopsies obtained from either normal subjects or a patient with the receptor-negative form of homozygous FH (10). Cells were grown in monolayer as previously described (10) and studied in the late logarithmic phase of growth after a 48-hr incubation in the presence of lipoprotein-deficient serum (5). Lipoproteins. Human LDL (density g/ml) and lipoprotein-deficient serum (density > g/ml) were obtained from the plasma of healthy subjects and prepared by differential ultracentrifugation (7). Coupling of LDL to Ferritin. Ferritin (Polysciences, Inc., Cat. no. 0217) was activated with a 1200 molar excess of glutaraldehyde according to the method of Kishida et al. (11). The ferritin was separated from glutaraldehyde on a Sephadex (>25 column, and the activated ferritin was mixed with purified LDL in a ferritin:ldl molar ratio of approximately 0.6 (based on molecular weights of 9 X 105 for the ferritin particle and 5 X 1i0 for the protein component of LDL). The mixture was incubated at 40 for hr on a rotating table. To separate LDL-ferritin from free ferritin and free LDL, we adjusted the density of the mixture to g/ml with solid potassium bromide; the mixture was centrifuged at 214,000 X g for 18 hr at 40 in a 1.6 X 7.6 cm tube in a Beckman type 65 ultracentrifuge rotor. After this centrifugation, the free ferritin pelleted to the bottom of the tube and the free LDL floated to the top of the tube. Immediately above the ferritin pellet was a yellow-orange zone approximately 1 cm in height that contained 2434

2 Cell Biology: Anderson et al. both LDL and ferritin. This fraction, which typically contained 20% of the initial LDL-cholesterol, was aspirated from the tube and dialyzed extensively at 40 against 0.15 M NaCl. Examination of this fraction by electron microscopy using negative staining revealed that it contained spherical LDL particles and that more than 90% of the LDL particles had been coupled with one to three ferritin cores. For purposes of quantitation, the cholesterol content of the LDL-ferritin complex in this fraction was measured by gas-liquid chromatography (12) and converted to an LDL protein concentration by dividing the sterol concentration by 1.6 (the cholesterol:protein ratio of the LDL used in these studies). Binding of LDL-Ferritin to Intact Fibroblasts. Monolayers of fibroblasts were placed in a 40 cold room for 30 min, after which the medium was removed and replaced with 2 ml of ice-cold growth medium containing 5% human lipoproteindeficient serum and LDL-ferritin at a concentration corresponding to 33,ug/ml of LDL-protein. After incubation at 40 for 2 hr, each monolayer was washed five times with ice-cold phosphate-buffered saline and fixed in the cold for 1 hr with 2% glutaraldehyde in 0.1 M sodium phosphate buffer, ph 7.3. In some experiments, cells not treated with LDL-ferritin were fixed in the cold for 1 hr with 1% tannic acid and 1% glutaraldehyde in 0.1 M sodium phosphate buffer, ph 7.3. Electron Microscopy. After fixation of the monolayers, the cells were washed in situ in the original petri dish with phosphate-buffered saline and postfixed with 2% OS04 in 0.1 M sodium phosphate buffer, ph 7.3. Cells were dehydrated through a graded series of ethanol solutions, washed several times with 100% ethanol, and embedded in Epon (13). After the plastic was hardened at 600 overnight, the 1.5-mm thick layer of plastic that covered the monolayer of cells was removed from the petri dish. Discs of 15 mm diameter were punched out of these layers and two of them were sandwiched together with fresh Epon so that the monolayers were apposed. When the Epon had hardened, the sandwich was trimmed and sections were made along the longitudinal axis of the two layers of cells. Thus, each section contained two layers of cells separated by a thin layer of Epon, and the angle of view was perpendicular to the plane of the culture dish. Blocks were sectioned with a Sorvall MT-2B ultramicrotome, stained with uranyl acetate and lead citrate, and viewed with a Philips 300 electron microscope. Quantitation of LDL-Ferritin Binding. To quantitate the amount of LDL-ferritin bound to the surface of the cells, we cut sections along the longitudinal plane of the monolayer and placed them on 100/400 slot mesh grids according to the technique of Anderson and Brenner (14) so that the longitudinal plane was perpendicular to the 400 per inch grid bars. The sections were examined without staining. Cells that extended between two grid bars were examined to determine the number of bound ferritin cores. Both the top and bottom surfaces of the cell were counted. On the average, the distance between two grid bars was 64,um; therefore, for each grid field, about 128 Aum of cell surface was examined. RESULTS To confirm that the LDL was tightly coupled to the ferritin in the isolated LDL-ferritin fraction, we performed an experiment in which 125-Ilabeled LDL was reacted with glutaraldehyde-activated ferritin and the resulting l25i-ldl-ferritin complex was isolated as described in Methods. Treatment of this '25I-LDL-ferritin with an antibody to LDL caused the precipitation of 92% of the 125I radioactivity (Table 1). Similarly, treatment of the same material with an antibody to ferritin caused the precipitation of 86% of the 125I radioactivity. Proc. Natl. Acad. Sci. USA 73 (1976) 2435 Table 1. Precipitation of 125I-labeled LDL-ferritin complex by antibodies to LDL and to ferritin Radioactivity precipitated by Anti-LDL Total Anti-ferritin Total Compound cpm cpm cpm cpm '251LDL '2I_-LDL + ferritin 'I2I-LDL-ferritin Each immunoprecipitation assay was performed in a 400-yl plastic microfuge tube (Beckman) that contained in a final volume of 205 ul: 0.15 M NaCl; either 75 Ag of anti-ldl (IgG fraction, Cappel Laboratories) or 75,g of anti-ferritin (IgG fraction, Cappel Laboratories) as indicated; and one of the following cor unds: 10 'gg of '25I-LDL (5440 cpm); 10 fig of '25I-LDL (5440 cpm + 10,gg of ferritin; or 10 fsg of 125I-LDL that had been previo ly coupled to ferritin (6210 cpm). Each reaction mixture was - cubated for 2 hr at 37 followed by incubation for 16 hr at 4. Th immunoprecipitates were collected by centrifugation (12,000 rpm, 5 min) and washed twice with 0.15 M NaCl. The bottom of the tube containing the precipitate was cut off and its radioactivity was determined directly. Each value represents the average of duplicate determinations. In control tubes containing no antibody, less than 1% of the 125I radioactivity from each preparation was found in the pellet. On the other hand, the anti-ferritin antibody precipitated less than 7% of uncoupled 125I-LDL when it was present in a mixture with free ferritin. In addition to its immunologic activity, the LDL of the LDL-ferritin complex retained sufficient biologic activity to suppress hydroxymethylglutaryl CoA reductase activity when added to normal fibroblasts, albeit at concentrations approximately 3-fold higher than required for native LDL. When examined with the electron microscope, the membranes of fibroblasts fixed with a combination of tannic acid and glutaraldehyde displayed a typical trilaminar appearance (Fig. IA and B). At intervals along the cell surface, the external side of the plasma membrane of these cells had a fuzzy appearance due to the presence of both a prominent external glycocalyx and a corresponding fuzzy coat on the cytoplasmic side of the plasma membrane (Fig. IB). These coated regions of the membrane varied in length from 0.2 to 0.5,m. Many of the coated regions were indented to form coated pits that appeared to be in the process of pinching off to form coated endocytotic vesicles (Fig. le). In addition to these coated pits, the plasma membrane also contained pits that lacked a prominent fuzzy coat and that exhibited a flask-shaped appearance (Fig. IA). The body of these flask-shaped pits measured about 0.1,um in diameter, and their neck region measured 0.05 rm. The flask-shaped pits, like the coated pits, appeared to be involved in either endocytosis or exocytosis. When normal fibroblasts were exposed to LDL-ferritin at an LDL protein concentration of 33 ;ig/ml at 40 for 2 hr, ferritin cores were observed to bind to the coated, indented regions of the plasma membrane identified in stained sections (Fig. 1C-E). Frequently, the LDL-ferritin tended to cluster on the sides of the invagination (Fig. 1C and E). Indented, coated regions containing LDL-ferritin were found with equal frequency on both the upper and lower surfaces of the cell. LDL-ferritin was not observed to be associated with the flask-shaped pits nor was it found inside the cell after these 40 incubations. A small proportion of the membrane-bound

3 2436 Cell Biology: Anderson et al. Proc. Natl. Acad. Sci. USA 73 (1976) 44j b~~~~~~~~~~~~4 4F, *!~~ 4.4 I~ f; ''I'm,, 'are' 1, ofw04 H C0 :7 zga: ''~~~~~~.W ~~~~~~~~~~~~~~~~~~~~ A'.2'" " FIG. 1. Electron micrographs showing coated and uncoated invaginations of the cell membrane and the localization of LDL-ferritin on the cell surface of normal fibroblasts. Monolayers of cells were grown, fixed, and embedded in plastic petri dishes as described in Methods. The cells in panels A-E were stained with uranyl acetate and lead citrate. The cells in panel F were not stained. (A) Flask-shaped invagination of the cell membrane (open arrow). Presumably, such invaginations become endocytotic vesicles (solid arrow). (B) Indented, coated region of the membrane. The arrows indicate the fuzzy coat projecting into the cytoplasm. There is a corresponding glycocalyx on the environmental surface. (C-F) fibroblasts incubated with 33 jig/ml of LDL-ferritin at 40 for 2 hr. In panel C, ferritin cores are clustered on the side of an indented, coated region (arrow). In panel D, ferritin cores are localized over a region of apparently coated membrane that is not invaginated (arrow). Panel E shows a high-magnification view of LDL-ferritin bound to the sides of an indented, coated region (arrows). Panel F shows LDL-ferritin bound to an indented, thickened membrane region in an unstained preparation of the type used for the quantitative studies. The scale on all micrographs is 1oo0 A. LDL-ferritin was found associated with uncoated regions of the cell surface. Fibroblasts from an FH homozygote were indistinguishable from normal cells in their ultrastructural characteristics. In particular, the plasma membrane of these mutant cells contained both coated regions and flask-shaped pits. However, when the homozygote cells were incubated with LDL-ferritin under conditions identical to those used for the normal cells, no ferritin cores were observed to be bound at any site on the plasma membrane. To confirm these qualitative observations of LDL-ferritin binding, we quantitated the number of ferritin cores associated with the fibroblast plasma membrane. For each experiment, normal and FH homozygote cells were incubated with LDL-ferritin either alone or in the presence of agents whose effect on the binding was to be tested. To identify the ferritin cores rapidly and unequivocally for purposes of quantitation, we examined unstained sections. Figs. if and 2A and B show that under these circumstances the ferritin cores were easily recognized. As with the stained preparations, in the unstained preparations the ferritin cores were observed to be clustered at scattered sites on the plasma membrane. At many of these sites, the membrane was indented to form pits that corresponded in size and shape to the coated pits in the stained sections. In addition, in these indented regions the vaguely visualized unstained plasma membrane usually showed a discernible thickening. By using the criteria of indentation and localized thickening, we were able to identify such regions as in-

4 FIG. 2. Indented regions of the cell membrane in unstained sections of normal and FH homozygote fibroblasts incubated with LDL-ferritin. Monolayers of cells were grown, incubated with 33 Ag/ml of LDL-ferritin at 40 for 2 hr, fixed, and embedded as described in Methods. (A and B) cells; (C) FH homozygote cells. The arrows indicate ferritin cores. The scale is 1000 A. Table 2. Cell Biology: Anderson et al. Proc. Natl. Acad. Sci. USA 73 (1976) 2437 dented, coated regions even when no LDL-ferritirl was bound (Fig. 2G). As a precaution against possible observer bias, in each quantitation experiment sections of cells from each of the treatments were coded, mixed together, and examined at random by an observer who was unaware of the code. Table 2 shows the results of these quantitative experiments. In Exps. A and B, normal cells incubated with LDL-ferritin were found to contain 49 and 61 ferritin cores per mm of surface length, respectively, while no ferritin cores were observed to be bound to the cell membrane in any of the sections prepared from the FH homozygote cells (Table 2, column b). Fixation of normal cells with 2.5% paraformaldehyde before incubation with LDL-ferritin did not affect the number or the distribution of membrane-bound ferritin cores (Exp. A). In normal cells the presence of a 15-fold excess of native LDL in the incubation mixture reduced the number of bound ferritin cores by 75%, suggesting that the native LDL competed with the LDL-ferritin for a limited number of receptor sites. Furthermore, washing the normal cells with heparin, which releases 1251-LDL from its receptor site (5), caused a dissociation of the previously bound LDL-ferritin (Exp. B). Both the normal and FH homozygote cells contained about 35 indented, thickened regions per mm of membrane surface (Table 2, column a). Inasmuch as the average length of each such region was 0.4 gm, it was calculated that the indented, thickened regions occupied only about 1.4% of the linear membrane surface, yet they contained more than 70% of the LDL-ferritin bound by the normal cells (Table 2, column c). It was also observed that under the conditions of these experiments, only about one-third of the indented, thickened regions were found to contain LDL-ferritin (data not shown). Free ferritin at 100,ug/ml did not bind to the plasma membrane (Table 2, Exp. C). DISCUSSION In the current studies, an LDL-ferritin complex was observed to bind preferentially to localized regions of the plasma membrane in human fibroblasts. In stained preparations these regions had an appearance similar to the "coated regions" or "coated pits" previously observed by other workers in other cell Quantitative analysis of the binding of LDL-ferritin to coated regions of the plasma membrane of normal and FH homozygote fibroblasts % of bound ferritin No. of ferritin Indented, Ferritin cores associated cores per ferritinthickened cores with indented, labeled indented Cell regions bound thickened region strain Treatment (no./mm) (no./mm) regions (mean ± SE) Exp. A (a) (b) (c) (d) LDL-ferritin ± 4.2 Paraformaldehyde, then LDL-ferritin ± 4.6 FH homozygote LDL-ferritin 37 0 Exp. B LDL-ferritin ± 2.8 LDL-ferritin + native LDL (495 gg/ml) ± 1.2 LDL-ferritin, then heparin ± 1.6 FH homozygote LDL-ferritin 40 0 Exp. C Ferritin FH homozygote Ferritin 31 0 Cell monolayers were prepared as described in Methods. Each monolayer was chilled for 30 min, after which it received 2 ml of growth medium containing 5% human lipoprotein-deficient serum and either LDL-ferritin (corresponding to 33,g/ml of LDL-protein) or 100 Mg/ml of ferritin. Except for the cells treated with paraformaldehyde, all incubations were for 2 hr at 4. In Exp. A, before the addition of LDLferritin one monolayer of normal cells was treated with 2.5% (vol/vol) paraformaldehyde in 2 ml of phosphate-buffered saline at 40 for 45 min, after which the monolayer was washed three times with ice-cold solution containing 50 mm Tris.HCl, ph 7.4, 0.15 M NaCl, and 2 mg/ml of bovine serum albumin. The cells were then exposed to LDL-ferritin for 2 hr at 24. In Exp. B, after exposure for 2 hr at 4 to LDL-ferritin one monolayer of normal cells was washed with sodium heparin as described (5). After the appropriate incubation, each monolayer was washed and fixed with glutaraldehyde at 40 as described in Methods. Sections of the fixed cells were prepared and the amount of LDL-ferritin was quantitated as described in Methods. A total of grid fields (corresponding to about 2 mm of cell surface membrane) was counted for each monolayer.

5 2438 Cell Biology: Anderson et al. types (15-18). In unstained preparations these regions could be identified because they appeared indented and the membrane composing them appeared thickened. Although the total number of ferritin cores bound to the cell surface was small, several lines of evidence indicated that this binding accurately reflected the attachment of the LDL-ferritin complex to the physiologic LDL receptor. First, binding of the LDL-ferritin was competitively inhibited by native LDL. Second, the bound LDL-ferritin could be removed from the surface by treatment of the cells with heparin, an agent that releases [125I]LDL from its cell surface receptor (5). Third, membrane binding of the LDL-ferritin was not observed in FH homozygote fibroblasts, which have been shown biochemically to have a specific deficiency in cell surface LDL receptors (3-5). Based on the quantitative data in Table 2, we have estimated that under the conditions of these experiments about 4000 LDL-ferritin particles were visualized bound to each cell. From previous biochemical studies with 125I-LDL, we calculated that a maximum of about 7,500-15,000 LDL particles can bind to the surface of a single cell at 40 (3, 5). Thus, although the LDLferritin appeared to be binding to the physiologic LDL receptor, not all of the available LDL receptors were capable of binding the LDL-ferritin. In addition to its binding to specific sites on the plasma membrane, a variable amount of LDL-ferritin was observed in the current studies to adhere to the proteinaceous material that coats the surface of the plastic dish between the cells. The amount of this adherent LDL-ferritin appeared similar whether the dish contained normal or FH homozygote cells. We believe that the intercellular proteinaceous material represents the site of the "nonspecific" (i.e., nonsaturable) binding that has been observed when either normal or FH homozygote fibroblasts are incubated with 125I-LDL at 40 (3, 5, 19). Since in the current studies the nonspecifically bound LDL-ferritin could be visually distinguished from membrane-bound LDL-ferritin, it was not necessary to vigorously wash the cells to remove the former. However, in studies in which 125I-LDL binding is being measured, it is essential to wash the cells with an albumin-containing buffer in order that this non-membrane-bound 25I radioactivity is preferentially removed so that it is not confused with 125I-LDL that is bound to the physiologic cell surface receptor (reviewed in ref. 20). A question raised by the current studies is whether the functional LDL receptors are specifically localized to the coated regions of the plasma membrane, or whether they migrate to these regions after they bind LDL. Although this question has not been resolved, two observations argue somewhat against the latter possibility: (i) preferential binding to the coated regions was observed at 40, a temperature at which lateral diffusion within the plane of the membrane is retarded (21), and (ii) the localization of LDL receptors to coated regions was not affected by prior fixation of the cells with paraformaldehyde at a concentration that abolishes endocytosis of the bound LDL (20). The finding that functional LDL receptors are clustered on coated regions of the plasma membrane is in keeping with the postulated role of these receptors in facilitating the cellular uptake of LDL (1, 2). A number of studies over the past 15 years have led to the suggestion that the coated region of the plasma Proc. Natl. Acad. Sci. USA 73 (1976) membrane is a specialized site on the cell surface where adsorptive endocytosis of proteins and colloidal particles is initiated (15-18). In agreement with these earlier suggestions, we have observed that the LDL-ferritin that binds to the coated regions of the human fibroblast membrane is subsequently ingested by the cell through the transformation of a coated pit into a coated vesicle when the cells are warmed to 370 (ref. 22, and manuscript in preparation). The current data thus lend support to the concept that the coated region of the plasma membrane may be a site that is specialized for the receptormediated binding and uptake of macromolecules (23). We thank Gloria Y. Brunschede and Margaret Wintersole for excellent technical assistance. This work was supported by grants from the National Institutes of Health (HL 16024, GM 19258, and GM 21698) and the American Heart Association. J.L.G. is a recipient of a Research Career Development Award (GM 70,277) from the National Institute of General Medical Sciences. M.S.B. is an Established Investigator of the American Heart Association. 1. Brown, M. S. & Goldstein, J. L. (1976) Science, 191, Goldstein, J. L. & Brown, M. S. (1976) in Current Topics in Cellular Regulation, eds. Horecker, B. L. & Stadtman, E. R. (Academic Press, New York), Vol. 11, in press. 3. Brown, M. S. & Goldstein, J. L. (1974) Proc. Natl. Acad. Sd. USA 71, Goldstein, J. L. & Brown, M. S. (1974) J. Biol. Chem. 249, Goldstein, J. L., Basu, S. K., Brunschede, G. Y. & Brown, M. S. (1976) Cell 5, Brown, M. S., Dana, S. E. & Goldstein, J. L. (1975) Proc. Natl. Acad. Sci. USA 72, Brown, M. S., Dana, S. E. & Goldstein, J. L. (1974) J. Biol. Chem. 249, Goldstein, J. L., Dana, S. E. & Brown, M. S. (1974) Proc. Natl. Acad. Sci. USA 71, Brown, M. S., Dana, S. E. & Goldstein, J. L. (1975) J. Biol. Chem. 250, Goldstein, J. L., Dana, S. E., Brunschede, G. Y. & Brown, M. S. (1975) Proc. Natl. Acad. Sci. USA 72, Kishida, Y., Olsen, B. R., Berg, R. A. & Prockop, D. J. (1975) J. Cell Biol. 64, Brown, M. S., Faust, J. R. & Goldstein, J. L. (1975) J. ClIn. Invest. 55, Lucky, A. W., Mahoney, J. J., Barnett, R. J. & Rosenberg, L. E. (1975) Exp. Cell Res. 92, Anderson, R. G. W. & Brenner, R. M. (1971) Stain Technol. 46, Fawcett, D. W. (1965) J. Histochem. Cytochem. 13, Roth, T. F. & Porter, K. R. (1964) J. Cell Biol. 20, Friend, D. S. & Farquhar, M. G. (1967) J. Cell Biol. 35, Bennett, H. S. (1969) in Handbook of Molecular Cytology, ed., Lima-de Faria, A. (North Holland Publishing Co., Amsterdam), pp Stein, O., Weinstein, D. B., Stein, Y. & Steinberg, D. (1976) Proc. Natl. Aced. Sci. USA 73, Brown, M. S., Ho, Y. K. & Goldstein, J. L. (1976) Ann. N.Y. Acad. Sci., in press. 21. Frye, L. D. & Edidin, M. (1970) J. Cell Sci. 7, Anderson, R. G. W., Goldstein, J. L. & Brown, M. S. (1975) J. Cell Biol. 67, 10a (abstract). 23. Korn, E. D. (1975) in Biochemistry of Cell Walls and Membranes, ed. Fox, C. F. (MTP International Review of Science, London), Vol. 2, pp