Kinetics of Membrane Internalization and Recycling During

Proc. Nati. Acad. Sci. USA Vol. 77, No. 2, pp. 1015-1019, February 1980 Cell Biology Kinetics of membrane internalization and recycling during pinocytosis in Dictyostelium discoideum (plasma membrane marker/pinosomal membrane/primary pinosome size/membrane shuttle) LUTZ THILO AND GUNTER VOGEL Max-Planck-Institut fur Biologie, 74 Tubingen, Federal Republic of Germany Communicated by George E. Palade, November 13,1979

ABSTRACT Internalization and recycling of plasma membrane internalization, the antibodies appeared in the membrane during pinocytosis in Dictyostelium discoideum was vacuolar system of the cells, and the antibody complex was analyzed quantitatively. A labeling technique was used by which [3Hlgalactose could be enzymaticallybound to and re- shown by isopycnic centrifugation to become associated with leased from the plasma membrane. Label internalized with the plasma membrane to which it had been recycled. Despite the plasma membrane was no longer accessible to enzymatic re- elegance of this method, it gave only qualitative evidence for lease and could therefore be distinguished quantitatively from membrane recycling. label remaining on the cell surface. Internalization of labeled An approach for analyzing internalization and recycling of membrane components was measured as a function of pinocytotic uptake. Direct experimental evidence for membrane re- membrane is presented in this study. It is based on a labeling cycling was obtained by demonstrating that previously inter- system by which a radioactive marker can be. enzymatically nalized label reappeared at the plasma membrane. The exper- bound to and released from the plasma membrane. Label on imental data agree with a kinetic model requiring that a shuttle internalized membrane is no longer accessible to enzymatic of membrane between two membrane compartments leads to release and, therefore, can be quantitatively distinguished from the same surface concentration of label in both. The two com- label on the cell surface. We have used this technique to partments consist of the plasma membrane and of cytoplasmic study vacuolar membranes; their relative membrane surface areas are the flow of membrane during pinocytosis in Dictyostelium 1 and 0.5, respectively. One surface area equivalent of the discoideum, which depends on endocytosis as the sole mecha- plasma membrane is internalized during a pinocytotic uptake nism of food uptake. amounting to 15% of the cell volume. At the observed rate of pinocytosis, this occurred once every 45 min. The average size of the primary pinosomes, as weighted according to their con- MATERIALS AND METHODS tribution to pinocytotic uptake, was calculated to be about 0.6 Culture Conditions. D. discoideum strain AX2 (ATCC Am. 24397) was grown axenically at 20°C as described (17). Cells Endocytosis is performed by a large variety of eukaryotic cells were harvested at a density of 2 X 106 cells per ml by centri- (1-5). During endocytosis, the plasma membrane invaginates fuging at 100 X g for 5 min and were washed in cold phosphate and encloses extracellular fluid (pinocytosis) or particles buffer (20 mM potassium phosphate, pH 6.5). Cell density was (phagocytosis) in plasma-membrane-derived vesicles. This determined with a particle counter (Coulter model DN). process leads to an extensive internalization of plasma mem- Pinocytosis Assay. Fluorescein-labeled dextran (FITC- brane varying between 1 and 20 times the total cell surface area dextran 40, Pharmacia) is a suitable fluid phase marker for per hr, depending on cell type and culture conditions (6-12). measuring pinocytotic fluid uptake (unpublished data). Uptake However, during endocytosis no reduction of the cell surface of FITC-dextran is directly proportional to its concentration area is observed. This implies that internalized membrane is in the medium (0.5-10 mg/ml) and proceeds linearly with time replaced at a corresponding rate. De novo membrane synthesis for at least 1 hr at 200C. The measured uptake rates were is too slow to account for membrane replacement, considering identical to those measured with the use of horseradish perox- the long lifetime of membrane components, on the order of idase, a well-established fluid phase marker (18). Cells were 10-100 hr (3, 5, 13). Therefore, it is generally assumed that suspended at a density of 5 X 106 cells per ml in axenic medium, internalized membrane is recycled to the plasma membrane, and FITC-dextran was added to a final concentration of 2 as was initially proposed as a result of electron microscopic mg/ml. Cells were incubated at the appropriate temperature observation (14, 15). on a rotary shaker (100 rpm). Samples of 1 ml were diluted 1:5 The methods used to demonstrate membrane internalization into ice-cold phosphate buffer to stop pinocytosis. Cells were were mostly based on electron microscopic morphological ob- collected by centrifugation at 100 X g for 5 min and resus- servations identifying endocytosis-derived internal membrane pended in 1 ml of phosphate buffer. For complete removal of structures. Because these methods used either endocytotic external FITC-dextran, the cell suspension was layered over content markers or noncovalently linked markers to unspecified and centrifuged through (200 X g, 10 min) an aqueous solution membrane components, previous results were difficult to (10 ml, 7 cm height) of 20% (wt/wt) poly(ethyleneglycol) 6000 quantify. For the same reason these methods were not suitable (Serva). Cells were washed and resuspended in 2 ml of 50 mM for yielding direct evidence for membrane recycling. The first Na2HPO4, pH 9.3. After the cells were counted, they were lysed such evidence was reported by Schneider et al. (16) for cultured with 0.1% Triton X-100 and fluorescence intensity was deter- fibroblasts: After the consecutive uptake of fluorescein-labeled mined (excitation wavelength 470 nm, emission wavelength anti-IgG by pinocytosis and anti-plasma membrane IgG by 520 nm). The pinocytosed volume was determined by comparison with a standard curve. The publication costs of this article were defrayed in part by page Isolation of Total Membrane. Cells in phosphate buffer (108 charge payment. This article must therefore be hereby marked "ad- cells per ml) were disrupted by freezing and thawing, diluted vertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact. Abbreviation: FITC-dextran, fluorescein-labeled dextran. 1015 Downloaded by guest on September 26, 2021 1016 Cell Biology: Thilo and Vogel Proc. Natl. Acad. Sci. USA 77 (1980) 1:10 in phosphate buffer, and gently sonicated in a water bath Time of incubation with galactosyltransferase, min for about 30 sec at 0C. Unbroken cells (<1%) and nuclei were I 0 20 40 60 80 removed by centrifugation at 5000 rpm for 5 min. Total i c 10 membrane was isolated from the supernatant by centrifugation at 105Xgfor60minat40C. 0 0 0 Polyacrylamide Gel Electrophoresis. Membrane proteins 8 -U were dissociated by heating at 1000C for 3 min in a solution -6 containing 1% NaDodSO4 and 0.1 M 2-mercaptoethanol. U .o0 0 Electrophoresis was performed in gradient slab gels (7-20% 0 +1 polyacrylamide, 0.1% NaDodSO4) with a discontinuous buffer V 4u c E system (19). Protein bands were stained with Coomassie brilliant :3 blue. The gels were sliced at 1-mm intervals; the fractions were 0 0Oa dissolved in Protosol (New England Nuclear) for 2 hr at 7jD u 4 0co 0C 600C. 4- ._.0 c Cell Dimensions. Cells were fixed in culture by the addition :3 of 2% glutaraldehyde for 30 min at 200C. The average cell di- 2 D0 ameter was measured by light microscopy. The circumference z of electronmicroscopically observed cell cross sections was IL J0 determined to be about 2.2 times that of a circle enclosing the 101 - IX-x--X----x------x same cross-sectional area. The surface-to-volume ratio of the II 0 2 4 6 8 cells was therefore taken as (2.2)2 - 5 times that of a sphere. 10 2o24 Reversible Labeling of Plasma Membrane. [3H]Galactose Time of incubation with 3-galactosidase, hr was covalently linked to terminal N-acetylglucosamine moieties FIG. 1. Kinetics of enzymatic binding and release of [3H]Gal. The on the cell surface of D. discoideum (20-22) with galactosyl- amount of [3H]Gal bound to the cell surface is shown as a function of transferase (EC 2.4.1.22) (23, 24) from bovine milk (Sigma) incubation time in the presence (0) and absence (X) of galactosyltransferase. Labeled cells (2 min at 0C) were incubated with f3-gal- according to the reaction: actosidase. The fraction of [3H]Gal remaining bound to the cell surface is shown as a function of time: at 20'C for cells fixed with glutaral- UDPGal + GlcNAc -- UDP + N-acetyllactosamine. dehyde (0) and at 0C for unfixed cells (3). Error bars, SEM (about For labeling, cells were resuspended at 00C to a density of about 4 values). 108 cells per ml in phosphate buffer containing 10 mM MnCl2, 2.2 kiM UDP[6-3H]Gal (ammonium salt, 18.5 Ci/mmol, Am- amount of [3H]Gal accessible to /-galactosidase was determined ersham; 1 Ci = 3.7 X 1010 becquerels), and galactosyltransferase by incubating fixed cells with the enzyme for about 8-10 hr at (0.5 unit/ml) was added. The reaction was stopped by 1:10 200C. dilution in ice-cold phosphate buffer, and cells were immediately washed in the same volume until no radioactivity was RESULTS found in the supernatant (about four times). Kinetics of binding Cells were labeled at 0C to prevent internalization of label under these conditions are shown in Fig. 1. During this study, during the labeling procedure (see below). Immediately after labeling was performed at 0°C for 2 min. About 106 molecules labeling, the cells were resuspended in axenic medium and of [3H]Gal were bound per cell under the chosen conditions, divided into two portions, to one of which FITC-dextran was which yielded adequate experimental sensitivity. However, up added as a fluid phase marker; both portions were then warmed to at least 3 times higher surface concentrations of label did not to 20'C. impair pinocytotic activity as compared with unlabeled cells The first portion of cells was used to demonstrate that labeled and did not affect any of the results reported below. membrane components are internalized concomitant with [3H]Gal can be released again by hydrolysis as catalyzed by membrane internalization during pinocytosis. The results are 3-galactosidase from Streptococcus pneumoniae (EC 3.2.1.23). shown in Fig. 2A. Pinocytosis was resumed and proceeded at The enzyme was isolated as described (25) and kindly supplied a constant rate of 0.014 Ml per min per 106 cells. Simultaneously, by Rudolf Weil (Sandoz Forschungsinstitut, Vienna). Cells, the fraction of [3H]Gal accessible to hydrolysis by subsequently labeled for 2 min at 0°C as above, were fixed in 2% glutaral- added external 0-galactosidase became smaller. The decrease dehyde for 30 min at 200C. After cells were washed twice in was exponential and a steady state was reached between ac- 100 mM imidazole (pH 6.8), they were resuspended therein at cessible and inaccessible label. In order to show that the pro- a density of about 108 cells per ml and 0-galactosidase (0.5 tection of label against hydrolysis is due only to its internali- units/ml) was added. The reaction was terminated by diluting zation and to test whether the label qualifies as a stable mem- the incubation mixture 1:10 in phosphate buffer at 0°C. The brane marker, not being removed by lysosomal enzymes, we fraction of [3H]Gal released was determined by comparing the investigated the fate of [3H]Gal during pinocytosis over a period amount of radioactivity in aliquots of the cell suspension and of 90 min. During this period, no [3H]Gal was released into the in aliquots of the cell-free supernatant remaining after cen- medium. At 30-min intervals samples were taken and the cells trifugation (500 X g, 20 sec). As shown in Fig. 1, hydrolysis were disrupted to show that at all stages the label remained occurred with a biphasic reaction rate. At 200C almost 70% of bound quantitatively to isolated membranes. Furthermore, the [3H]Gal was released during the first hour and the remainder entire label remained bound by a 0-glycosidic linkage sus- with a half-life of about 8 hr. Quantitatively, the same kinetics ceptible to hydrolysis by fl-galactosidase with the same biphasic were observed with unfixed cells that had been disrupted im- kinetics as found for whole cells. Internalization of labeled mediately after labeling by freezing and thawing. We do not membrane components depends strictly on endocytotic activity. know why a residual fraction of about 10% of the label re- The initial rate of internalization and the rate of pinocytosis at mained membrane-bound even after prolonged incubation up 15'C were both about 20% the values of 200C (data not shown). to 24 hr. At 0C the reaction rate was slightly slower and is Pinocytosis was inhibited at 0C or 20'C in the presence of 1 shown in Fig. 1 for intact, unfixed cells. During this study, the AM carbonyl cyanide m-chlorophenyl hydrazone, an uncoupler Downloaded by guest on September 26, 2021 Cell Biology: Thilo and Vogel Proc. Natl. Acad. Sci. USA 77 (1980) 1017 of oxidative phosphorylation, and, accordingly, the entire label Because internalized membrane (pinosomal membrane) could remained accessible to hydrolysis upon incubation under such not be separated from the bulk of the membrane, the profile conditions (Fig. 2A). of internalized label was obtained indirectly as explained in Fig. In order to demonstrate that internalized membrane com- 3. Within experimental accuracy, the inaccessible part of the ponents are recycled to the cell surface, it was shown that pre- label quantitatively displayed the same profile as the label on viously internalized label again becomes accessible to enzymatic the cell surface. Apparently, the composition of labeled attack. The second portion of cells was incubated in parallel to membrane components remains basically unaltered during the first, resulting in the same steady-state distribution of label. internalization. Pinocytosis was stopped by cooling to 00C. Subsequently, the steady-state distribution was disturbed by releasing about 60% of the label on the cell surface. For this, cells were subjected to DISCUSSION an intermediate treatment with 03-galactosidase at low tem- The present labeling technique has previously been used during perature. A control sample of the same cells was incubated in structural (26-28) and developmental (29) studies. Taking the absence of enzyme. Cells were resuspended in axenic me- advantage of its reversibility, it is used in this study to analyze dium containing FITC-dextran and warmed to start cellular activities. As shown in Fig. 2B, pinocytosis was resumed after a lag of about 8 min and, simultaneously, an increasing fraction of [3H]Gal again became hydrolyzable by subsequently added 3-galactosidase. The original steady-state distribution between accessible and inaccessible label was reestablished. In the absence of pinocytosis at 00C, no label reappeared at the cell surface. These results demonstrate that previously internalized label is recycled back to the cell surface concomitant with pinocytotic activity. As shown by the control experiment in Fig. 2B, the additional manipulations as such did not lead to a per- manent distortion of the previously established steady-state distribution of label. In order to serve as a representative membrane marker, all labeled membrane components must become internalized to the same extent as the membrane as a whole. Evidence for this was obtained by comparing the electrophoretic profile of labeled membrane components before and after internalization. A> 0 4- 0. ~00 45 cc

4- D~

V 4- 0 45 (.45 .0

C,0 ,0 1- L., 0 O. E0

4) 4

0. Mr: 1 00°000 50,000 30,000 20,000 10,000 FIG. 2. Internalization and recycling of labeled membrane FIG. 3. Composition of labeled membrane components before components. (A) Internalization (0, at 20'C; O, at 0C) was measured and after internalization. The radioactivity profiles on NaDodSO4/ in correlation with pinocytotic uptake (v). (B) Recycling (0) (control, polyacrylamide gels were obtained by applying equal amounts (Z40 0) and pinocytotic uptake (v) were measured after the previously ,ug) ofmembrane protein from total membrane fractions. Labeled cells established steady state was disturbed by partial removal of [3H]Gal were divided into three aliquots, A, B, and C. A, total membrane was from the cell surface during an intermediate treatment with f3-gal- isolated before internalization oflabel. B, total membrane was isolated actosidase (80 min, 0C; control was incubated without enzyme). before internalization but after 60% of the label had been removed Center ordinates: Fraction of label hydrolyzed during 8 hr at 201C by fl-galactosidase (80 min, 0C). C, total membrane was isolated after with all the label made accessible by repeated disruption of cells in internalization of 38% of the label (steady state after 90 min at 20'C, the presence of f3-galactosidase (30-sec sonication every 2 hr) was measured as in Fig. 2A) and subsequent removal of 37% of total label normalized to represent 100% of label on the plasma membrane. (A) (60% of 62% on cell surface as in B). The profile of internalized label Upper ordinate: 81% A 1.0 (zero time); (B) lower ordinate: 73% A 1.0 (denoted A') can be reconstructed by using profiles B and C and the (lower than in A due to intermediate treatment with f3-galactos- relationship C = 0.38A' + 0.62B, as is shown by the dotted line in idase). comparison with profile A. Downloaded by guest on September 26, 2021 1018 Cell Biology: Thilo and Vogel Proc. Natl. Acad. Sci. USA 77 (1980) the dynamic process of membrane flow. In the particular case = 0.63, the area ratio of plasma membrane to vacuolar mem- of D. discoideum, the label was found to be close to ideal as a branes is 1/0.5, compared to 1/1 as determined by morpho- membrane marker: [3H]Gal is covalently bound to membrane logical observations of the same organism (12). Whereas the components at the cell surface. It is a nondisturbing label, represent labeling technique detects only those vacuolar mem- maining fully membrane bound during prolonged periods of branes that take part in the membrane shuttle, the previous pinocytosis, without chemical modification by intrinsic cellular result includes all morphologically detectable membranes of activities. The label is attached to many different membrane food vacuoles. Therefore, about half of the vacuolar mem- components, all of which take part in membrane flow to the branes, probably lysosome-derived membranes, do not seem same extent. It therefore seems justified to consider the label to take part in the membrane shuttle. as a representative membrane marker. The curve in Fig. 4 corresponds to vo = 0.63 X 10-6 l, about The kinetics of the redistribution of label during pinocytosis 15% of the cell volume. Therefore, at the observed rate of pi- (Fig. 2) can be analyzed in terms of membrane internalization nocytosis, the plasma membrane is internalized once every 45 and recycling by making the following assumptions. (i) Inter- min. [An estimate of this can be obtained by extrapolation to nalization and recycling of label quantitatively represent in- x = 0 by using (dx/dv),=o. ] This agrees with morphological ternalization and recycling of the membrane as a whole. (ii) observations for mouse macrophages and L cells, which indicate During pinocytosis, there is a shuttle of membrane between two that the plasma membrane is internalized once every 32 and membrane compartments consisting of the plasma membrane 111 min, respectively, during a pinocytotic uptake amounting and cytoplasmic vacuolar membranes, respectively. (iii) Within to 14 and 5.6% of the cell volume, respectively (11). each membrane compartment, a homogeneous surface con- The size of the primary pinosomes can be calculated by re- centration of label is rapidly established. Under these conditions, lating the surface area of internalized membrane to the corre- shuttling of membrane between the two compartments will sponding volume taken up by pinocytosis. With a cell diameter gradually lead to an equal surface concentration of label in both. of 20 + 5 grm (SD) and a surface-to-volume ratio about 5 times It can be shown that the fraction of label that is in the plasma that of a sphere (see Materials and Methods), the surface area membrane compartment, x (1 > x > 0), changes from its initial of the cell was estimated as (6.3 + 3) X 103 Am2. This is the value, xO, to its equilibrium value, p, as a function of the pino- surface area of membrane used by the cell to enclose a volume cytosed volume, v, as described by of 0.63 X 10-6Ml (vo). The average size of the primary pinosomes turns out to be about 0.6 gim in diameter. Because this x =P + (xo-p) exp [- -.(l j. [1] value is derived by relating the pinocytosed volume to the "membrane-consuming" step during pinocytosis, it gives the The equilibrium value, p, is the ratio of the plasma membrane size of the truly "primary" pinosome immediately after the first surface area to the total surface area of both membrane com- fusion event has closed the plasma membrane invagination. No partments. The constant vo is defined as the volume that is pi- subsequent steps are detected by the present method nor can nocytosed during the internalization of an amount of mem- anything be said about the size of the vesicles by which membrane equivalent in area to that of the plasma membrane. As brane is recycled back to the cell surface. The value of 0.6 gm is shown in Fig. 4, the data from Fig. 2A can be rearranged to represents an average of pinosome sizes weighted according represent the fraction of label on the plasma membrane, x, as to their volume contribution to pinocytotic uptake. In this re- a function of the pinocytosed volume, v. The best fit of Eq. 1 spect, previously reported size distributions of primary pino- to the experimental data then yields values for p and vo. somes as observed morphologically in Acanthamoeba (7) and Interpreted in this way the present data agree well with macrophages (11) highly overemphasize the small and most previous results based on morphological observations. With p abundant pinosomes (0.1-0.2 Mm). The reported size histograms (7, 11) can be transformed to reflect the relative volume contribution of the size groups. The volume-weighted average size 1.0 of the pinosomes then turns out to be between about 0.5 and 0.8 0 c 0 gm in both cases. For mouse macrophages and L cells, the .0 volume-to-surface ratios of primary pinosomes have actually E 0) been reported (30) and can be directly used to calculate the C 0.9 volume-weighted size average as 0.6 and 0.4 gm, respectively. c Our results are therefore in full agreement with previous a morphological observations when reassessed to yield volume- c 10 weighted average values for the pinosome size. However, the t, 0.8 implication of the rather insignificant contribution to pinocy- 0 totic uptake by the more abundant small "pinosomes" then 0 raises the question as to their biological function. The rate at which label reappears on the plasma membrane .0 during the recycling experiment (Fig. 2B) is much higher than expected from the data in the internalization curve (Figs. 2A -",- and 4). This is partially due to a net transfer of membrane to ------o the plasma membrane, as shown by the control experiment 0.6 (Fig. 2B). These cells have the same surface concentration of 0 0.2 0.4 0.6 0.8 pinocytosed volume/1 06 label in both membrane compartments. Therefore, the initial v, cells, Ml low value of the fraction of label on the plasma membrane re- FIG. 4. Redistribution of [3H]Gal between the plasma membrane flects a relative reduction of the cell surface area that occurred and vacuolar membranes due to membrane shuttle during pinocytosis. during cooling. After warming, this effect is reversed before The fraction of label on the plasma membrane, x (data from Fig. 2A, upper center ordinate), is plotted as a function of pinocytosed volume, pinocytosis is resumed. However, even after the onset of pi- v (Fig. 2A, lower part). A curve as described by Eq. 1 is fitted to the nocytotic activity, in the recycling experiment, membrane flow experimental data as shown, with p = 0.63 and vo = 0.63 X 10-6 Al. is about 3-4 times faster than before, compared to the rate of Downloaded by guest on September 26, 2021 Cell Biology: Thilo and Vogel Proc. Natl. Acad. Sci. USA 77 (1980) 1019

pinocytotic uptake. This suggests that during the recovery phase 13. Morr6, D. J., Kartenbeck, J. & Franke, W. W. (1979) Biochim. the cells pinocytose by means of smaller (0.15 gim) micropi- Biophys. Acta 559, 71-152. nocytotic vesicles compared to the normally occcurring ma- 14. Palade, G. E. (1956) J. Biophys. Biochem. Cytol. Suppl. 2, 85- cropinosomes (0.6 jum). 98. 15. Palade, G. E. (1975) Science 189, 347-358. We acknowledge the expert technical assistance of Mrs. Renate Kaus, 16. Schneider, Y. J., Tulkens, P. & Trouet, A. (1977) Biochem. Soc. Roswithe Steinhart, and Rosemarie Herrmann. This investigation Trans. 5, 1164-1167. benefitted from electron microscopic observations kindly contributed 17. Watts, D. J. & Ashworth, J. M. (1970) Biochem. J. 119, 171- by Dr. Heinz Schwarz. We thank Dr. Barbara Wallenfels for bringing 174. the availability of the labeling technique to our attention and Dr. Rudolf Weil for kindly supplying us with the necessary amounts of 18. Steinman, R. M., Silver, J. M. & Cohn, Z. A. (1974) J. Cell Biol. i3-galactosidase. We thank Dr. Peter Overath and Dr. Keith Wright 63, 949-969. for many helpful suggestions. This research was supported by Fond 19. Laemmli, U. K. (1970) Nature (London) 227,680-685. der Chemischen Industrie. 20. Wilhelms, 0. H., Luderitz, O., Westphal, 0. & Gerisch, G. (1974) Eur. J. Biochem. 48, 89-101. 1. Holter, H. (1965) Symp. Soc. Gen. Microbiol. 15,89-114. 21. Gilkes, N. R. & Weeks, G. (1977) Biochim. Biophys. Acta 464, 2. Bennett, H. S. (1969) in Handbook of Molecular Cytology, ed. 142-156. Lima-De-Faria, A. (North-Holland, Amsterdam), pp. 1294- 22. Hoffman, S. & McMahon, D. (1978) J. Biol. Chem. 253, 278- 1319. 287. 3. Silverstein, S. C., Steinman, R. M. & Cohn, Z. A. (1977) Annu. 23. Morrison, J. F. & Ebner, K. E. (1971) J. Biol. Chem. 246, Rev. Biochem. 46,669-722. 3977-3q84. 4. Holtzman, E., Schacher, S., Evans, J. & Teichberg, S. (1977) Cell Surf. Rev. 4, 165-246. 24. Khatra, B. S., Herries, D. G. & Brew, K. (1974) Eur. J. Biochem. 5. Edelson, P. J. & Cohn, Z. A. (1978) Cell Surf. Rev. 5, 387-405. 44, 537-560. 6. Weisman, R. A. & Korn, E. D. (1967) Biochemistry 6, 485- 25. Hughes, R. C. & Jeanloz, R. W. (1964) Biochemistry 3, 1535- 497. 1543. 7. Bowers, B. & Olszewski, T. E. (1972) J. Cell Biol. 53, 681-694. 26. Schenkel-Brunner, H. (1973) Eur. J. Biochem. 33,30-35. 8. Stockem, W. (1973) Z. Zellforsch. 136, 433-446. 27. Schindler, M., Mirelman, D. & Schwarz, U. (1976) Eur. J. Bio- 9. Githens, S., III & Karnovsky, M. L (1973) J. Cell Biol. 58, chem. 71, 131-134. 536-548. 28. Shaper, J. H. & Stryer, L. (1977) J. Supramol. Struct. 6, 291- 10. Hubbard, A. L. & Cohn, Z. A. (1975) J. Cell Biol. 64, 461- 299. 479. 29. Wallenfels, B. (1979) Proc. Natl. Acad. Sci. USA 76, 3223- 11. Steinman, R. M., Brodie, S. E. & Cohn, Z. A. (1976) J. Cell Biol. 3227. 68,665-687. 30. Steinman, R. M., Silver, J. M. & Cohn, Z. A. (1978) in Transport 12. Ryter, A. & De Chastellier, C. (1977) J. Cell Biol. 75, 200- of Macromolecules in Cellular Systems, ed. Silverstein, S. C. 217. (Dahlem Konferenzen, Berlin), pp. 167-180. Downloaded by guest on September 26, 2021