The Biochemistry and physiology of the endocytic apparatus

Cells have evolved a number of specialized mechanisms to transport molecules across the plasma membrane barrier. Endocytic processes involve the invagination and internalization of membrane vesicles and their contents. Conversely, secretory events involve fusion of membrane vesicles with the plasma membrane. The net effect of such activities is the mixing of intracellular and plasma membrane components. Cells must therefore posses some type of sorting mechanism to counteract this randomization and it is believed that the endocytic pathway plays a part in sorting internalized membrane and its constituents. The present study made use of a specific tracer molecule to analyze the compartments involved in the endocytosis/recycling pathway. Transferrin and the transferrin receptor were used to study the endocytic pathway because this ligand along with its receptor traverses the entire endocytic and recycling pathway without being degraded. Transferrin was chemically coupled to horseradish peroxidase in order to specifically target peroxidase activity to the endocytic apparatus. Peroxidase can be used to alter the physical properties of compartments containing the enzyme by affecting an increase in buoyant density, and a crosslinking of luminal contents. The density shift procedure was used to demonstrate that transferrin, low density lipoprotein, and epidermal growth factor are internalized into the same endocytic compartment. A combination of the density shift and crosslinking reactions was used to determine that the endocytic pathway responsible for the accumulation of ligand does not include the Golgi apparatus. Use of the conjugate also provided a means to study the movement of unoccupied receptors. It was determined that intracellular transferrin receptors in HeLa cells are the result of the constitutive internalization of unoccupied receptors. It was further demonstrated that endocytic compartments internalized at different times remain temporally segregated. These results suggest that one way in which surface receptor number can be regulated is the by rate of internalization of unoccupied receptors. The results of the present studies provide a quantitative approach to describing the nature and constituents of the endocytic/recycling pathway. They provide strong support for the notion that the endosome is the intracellular organelle responsible for sorting different receptor/ligand complexes and raise the possibility that this organelle may also participate in the regulation of plasma membrane proteins. Further, these studies provide a method and conceptual approach for further experimentation on the role of the endosome in cellular physiology and metabolism.

THE UNIVERSITY OF UTAH GRADUATE SCHOOL SUPERVISORY COMMITTEE APPROVAL of a dissertation submitted by Richard Scott Ajioka This dissertation has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. ({t(�. )/ �)t:. Chairman: / Costa Georgopoulos Martin Rechsteiner Steven Wiley THE UNIVERSITY OF UTAH GRADUATE SCHOOL FINAL READING APPR()V AL To the Graduate Council of The Ulli\'ersity of Utah: Richard Scott Ajioka I have read the dissertation of in its final form and have found that ( I) it!'llo " rmat. citations, and bibliographic sty le are consistent and acceptable; (2) its illustrative materials including fig ure s tables. , and charts are in place; and (3) the filial manu!'Icript is sati!'lfactory to the Supervis­ ory Committee and i� ready for submi!'lsion to the Graduat e Schoo\ UalC Approved for tht" !\t�jor De pa rt m ent John M. Matsen Chllirnmn I nt'an B. Gale Dick UC<I" C11' TIlt' (in"h',lIt' Stlll"11 . Copyright © Richard Scott Ajioka 1987 All Rights Reserved ABS1RACT Cells have evolved a number of specialized mechanisms to transport molecules across the plasma membrane barrier. Endocytic processes involve the invagination and internalization of membrane vesicles and their contents. Conversely, secretory events involve fusion of membrane vesicles with the plasma membrane. The net effect of such activities is the mixing of intracellular and plasma membrane components. Cells must therefore possess some type of sorting mechanism to counteract this randomization and it is believed that the endocytic pathway plays a part in sorting internalized membrane and its constituents. The present study made use of a specific tracer molecule to analyze the compartments involved in the endocytosis/recycling pathway. Transferrin and the transferrin receptor were used to study the endocytic pathway because this ligand along with its receptor traverse the entire endocytic and recycling pathway without being degraded. Transferrin was chemically coupled to horseradish peroxidase in order to specifically target peroxidase activity to the endocytic apparatus. Peroxidase can be used to alter the physical properties of compartments containing the enzyme by effecting an increase in buoyant density, and a crosslinking of luminal contents. The density shift procedure was used to demonstrate that transferrin, low density lipoprotein, and epidermal growth factor are internalized into the same endocytic compartment. A combination of the density shift and crosslinking reactions was used to determine that the endocytic pathway responsible for the accumulation of ligand does not include the Golgi apparatus. Use of the conjugate also provided a means to study the movement of unoccupied receptors. It was determined that intracellular transferrin receptors in HeLa cells are the result of the constitutive internalization of unoccupied receptors. It was further demonstrated that endocytic compartfl1ents internalized at different times remain temporally segregated. These results suggest that one way In which surface receptor number can be regulated is the by rate of internalization of unoccupied receptors. The results of the present studies provide a quantitative approach to describing the nature and constituents of the endocytic/recycling pathway. They provide strong support for the notion that the endosome is the intracellular organelle responsible for sorting different receptor/ligand complexes and raise the possibility that this organelle may also participate in the regulation of plasma membrane proteins. Further, these studies provide a method and conceptual approach for further experimentation on the role of the endosome in cellular physiology and metabolism. v This work is dedicated first to my parents and family for their patient support throughout my graduate career. taught me how to think and much about life. gave me bad advice. To Jerry, who To Costa, who never And to Tim, who took me fishing. "Whenever you fall, pick something up," Oswald Avery CONIENfS ABSTRACT ................................................ iv LIST OF TABLES .......................................... ix LIST OF FIGURES .......................................... x ACKNOWLEDGEMENTS . . . . . . .. . .......................... xiii Chapter I. INTRODUCTION ....................................... 1 References. . . . . . . . . . . . . . . . . . . . . . . . .. ......... 27 II. CHARACTERIZATION OFENDOCYTIC COl\1PARTl\1ENTS USING THE HORSERADISH PEROXIDASEDIAMINOBENZIDINE DENSITY SHIFf TECHNIQUE ......... 39 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . 41 Results ....................................... 41 Discussion ..................................... 47 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 48 III. THE RECYCLING PATHWAY USED FOR TRANSFERRIN-MEDIATED IRON DELIVERY IN HELA CELLS DOES NOT INCLUDE THE GOLGI APP ARATUS ....... 49 Summary ..................................... Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Materials and Methods. . . . . . . . . . . . . . . . . . . . . . . .. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Discussi on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 51 53 56 70 76 N. INTRACELLULAR POOLS OF TRANSFERRIN RECEPTORS RESULT FROM CONSTITUTIVE INTERNALIZATION OF UNOCCUPIED RECEPTORS ........................... 80 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . .. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Refer~nces .................................... 81 82 84 85 V. SUMMARY .......................................... 86 APPEND IX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 102 VITA .................................................. 117 viii LIST OF TABLES Table Page Chapter II I. The DAB Reaction Affects the Detergent Solubility of Membrane Vesicle Contents ...... : . . . . . . . . . . . . . . . .. 45 II. Titration of DAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 46 III. Effect of the DAB Reaction on Selected Activities . . . . . . . 46 IV. Localization of HRP in Lysosomes Affects Lysosomal, but not Endosomal, Activities. . . . . . . . . . . . . . . . . . . . . . . . . .. 47 V. Tf and EGF are Separated Soon After Internalization . . .. 47 Chapter III I. The DAB Reaction Inhibits the Ability to Detergent Extract Functional Tf Receptors ..................... 69 LIST OF FIGURES Fi~ure Pa~e Chapter II. 1. Distribution of internalized 125I-Tf in 12% Percoll gradients. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 42 2. Distribution of membrane activities on Percoll gradients ............................... 42 3. Internalized Tf-HRP recycles normally . . . . . . . . . . . . . . 43 4. Density shift of endosomes . . . . . . . . . . . . . . . . . . . . . . .. 44 5. Internalized Tf and EGF are in the same endocytic compartment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 6. Internalized EGF, LDL, and Tf-HRP enter the same endocytic compartment . . . . . . . . . . . . . . . . . . . . . . . 44 7. Endocytic compartments do not mix during preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 45 Chapter III. 1. Enzyme distribution on Percoll gradients . . . . . . . . . . .. 57 2. Density shift of galactosyltransferase . . . . . . . . . . . . . . . . 59 3. Density shift of incorporated 3H-fucose . . . . . . . . . . . . .. 63 4. Analysis of DAB treated material . . . . . . . . . . . . . . . . . .. 67 Page Fil:ure 5. Endo H treatment of newly synthesized VSV proteins. .................................. 72 Chapter IV. 1. Models of the relationship between intracellular and surface TfR during endocytosis .................. 82 2. Evidence that peroxidase-catalyzed oxidation of DAB within endosomes inactivates TfR . . . . . . . . . . . . . . . . . .. 83 3. Relationship between occupancy of receptors by Tf-HRP and inactivation of TtR . . . . . . . . . . . . . . . . . . . . . . . . . . .. 83 4. Effect of HRP internalized by fluid-phase pinocytosis on TtR activity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 83 5. Relationship between the cellular accumulation of Tf-HRP and receptor inactivation . . . . . . . . . . . . . . . . . . . . 84 6. Evidence that endocytic vesicles formed at different times do not intenningle their contents . . . . . . . . . . . . . . . 84 Appendix 1. PMA reduces surface receptor number in HL60 cells .................................... 105 2. PMA increases the rate of internalization of unoccupied Tf receptors . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 3. Models for the structure of endocytic compartments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 4. Temporally segregated receptor-mediated and fluid phase markers do not mix within the endocytic pathway . . . . . . . . . . . . . . . . . . . . . . . . . .. 111 Xl Page Filiure 5. Temporally segregated fluid phase and receptormediated markers do not mix within the endocytic pathway ........................... 112 6. Low levels of HRP are sufficient to catalyze DABvesicle crosslinking . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 113 7. Endocytic vesicles do not fuse with compartments containing unoccupied receptors ................... 115 xii ACKNOWLEIXJEMENTS This w'ork was supported by predoctoral training grants 5T32GM07464-08 and 5T32GM07464-09 awarded from the National Institutes of Health. The author would like to acknowledge Ms. Ina Jordan and Ms. Lisa Gren for tireless technical and moral support throughout this endeavor. The members of the laboratory and fellow graduate students provided an academic and social atmosphere ideal for scientific pursuits. I would like to thank the members of my thesis committee for their enthusiastic support and guidance throughout the development of my graduate education. Dr. Steven Wiley taught me to use the computer as a scientific tool. I especially would like to thank Dr. Jerry Kaplan who was, and continues to be an endless source of creativity, intellectual thought, support, and friendship. Chapter II is reproduced from The Journal of Cell Biology, 1987, vol. 104, pp 77-85 by permission of The Rockefeller U ni versity Press. Chapter IV is published with the permission of the National Academy of Sciences, USA. CHAPTER I IN1RODUCTION 2 The plasma membrane is the interface and boundary between the cytoplasm of a cell and the outside world. This fact places major functional requirements on the organelle since any molecule which enters or leaves the cell must encounter this barrier. In order to function properly, cells must be able to obtain nutrients, export waste, maintain osmotic balance, and carry out specialized functions such as phagocytosis and secretion. All of these processes involve the plasma membrane to some degree. As a consequence, cells have evolved various mechanisms for the transfer of molecules from one side of the membrane to the other. Transport activities involving the plasma membrane range from simple diffusion to more complex processes such as phagocytosis which require ATP hydrolysis and the inclusion of large segments of membrane. Cells exhibit a number of different mechanisms for the transport of biologically relevant molecules across the plasma membrane. Facilitated diffusion and active transport are effective mechanisms for the transfer of specific molecules across membranes, but are usually restricted to simple, low molecular weight compounds. How then are other potentially important molecules able to enter cells? One way cells can accumulate molecules which are otherwise impermeable to the plasma membrane is by pinocytosis, a term coined by Lewis in 1931 derived from the Greek word meaning "to drink" (1). He observed that macrophages could engulf fluid in droplets which were taken up in a discontinuous manner. Using phase contrast microscopy, he was able to visualize vacuoles of 1-2 Jlm in diameter which formed in areas of the cell membrane active 3 in ruffling. With the advent of electron microscopy, Palade and Bennet described pinocytic vesicles of 0.05-0.2 ~m in size (2,3). A cornbination of light and electron microscopic studies led to the description of pinocytosis in nonquantitative terms (4). Macromolecules soluble in the medium would therefore be included in the pinocytic vesicle. Pinocytosis thus represents a process by which cells can accumulate substances which are not able to cross the plasma membrane barrier. There are two characteristic features of pinocytosis which allow cells to sample the environment in an unbiased manner. First, the pinocytic vesicle will include any molecule which is soluble in the media, and is therefore nonselective. Second, solutes are accumulated by cells as a direct linear function of their concentration. The employment of assayable fluid phase markers such as horseradish peroxidase and 3 H -sucrose provided a means to quantitatively measure uptake by pinocytic processes. A combination of morphological and quantitative data led to the conclusion that even though absolute rates of uptake may vary, pinocytosis is a constitutive process. For example, Steinman et al. estimated that macrophages and mouse L cells internalized an equivalent of 0.43% and 0.05% of their total cellular volume respectively, per min (5). Because pinocytic vesicles are formed by the invagination of plasma membrane, the net effect is the internalization of some fractional area of the cell surface. With time, large amounts of membrane can be accumulated by the cell. Steinman and coworkers calculated that the fractional influx in 4 surface area was 3.1 % and 0.8% per min for macrophages and L cells respectively. It can then be calculated that macrophages and L cells are able to internalize the equivalent of their cell surface area every 33 and 125 min. Similarly it has been shown that amoebae can internalize half of their surface area in 30 min (6,7). Thus, within relatively short time periods relative to division cycles, cells are capable of internalizing greater than 100% of their surface area. In phagocytosis, particulate matter is .internalized by a process of membrane invagination similar in some respects to pinocytosis. Tsan and Berlin estimated that during a 30 min period of phagocytosis of polyvinyl toluene particles, macrophages internalized approximately 50% of their surface area within the phagocytic vesicle (8). Petty et al.(9) demonstrated that as a consequence of ingesting antibody coated lipid vesicles, macrophages could internalize a quantity of membrane equal to their surface area in 30-40 min, yet the cells exhibited only a 25% loss of surface area. This loss at the cell surface exceeds the rate at which new membrane can be synthesized. If de novo synthesis were the sole source of plasma membrane, a cell engaged in phagocytosis would soon engulf itself. It is apparent that during periods of active pinocytosis or phagocytosis, cells do not lose a corresponding amount of surface membrane. Therefore, in order to maintain pinocytic and endocytic activity over long periods, cells must possess an efficient membrane recycling system. One of the earliest mentions of membrane reutilization was made by Palade who observed that following a secretory event, 5 pancreatic exocrine cells exhibited an increase in surface area which with time, returned to normal dimensions (10,11). In 1973 Heuser and Reese (12) observed that during the stimulation of frog nerve cells, some 30% of the synaptic vesicle membrane disappeared. This loss was balanced by an increase in the surface area of the plasma membrane suggesting that the synaptic vesicles had fused with the cell surface. Moreover, within 15 min, the loss of synaptic vesicle membrane was largely balanced by the appearance of membranewalled cisternae inside the terminals. Soon thereafter the cisternae disappeared and vesicles reformed. Further studies using fluid phase markers revealed that the loss of synaptic vesicle membrane was compensated for by an increase in pinocytic activity. Similar observations have been made in mast cells following degranulation (13), and bidirectional flow between secondary lysosomes and plasma membrane has been demonstrated during phagocytosis (14). The amount of membrane recycled during endocytic events has been estimated using theoretical models of vesicle fusion and correlates well with experimental values (15). Pinocytosis represents a nonselective means of transport in that any molecule which is soluble in the media will be included in the pinocytic vesicle. Random uptake enables cells to analyze the entire spectrum of molecules in the medium. Selective internalization, on the other hand, would enable cells to discriminate between useful or irrelevant and harmful molecules before they are internalized. clear that cells are able to accumulate certain molecules in a selective manner. That is, in a given amount of time, they are It is 6 capable of binding and internalizing amounts of these molecules in excess of that predicted by pinocytic processes alone. Binding of these molecules is usually mediated by specialized plasma membrane proteins generically called receptors. Receptors exhibit several characteristics which make them particularly well suited for the task of processing specific extracellular molecules. The two most significant features are specificity and high binding affinity for a particular ligand. The dissociation constants for such interactions range from 1 x 10- 7 to 1 x 10- 12 . High affinity insures capture of ligand even when its concentration in the medium is low. A third feature which will be disoussed in more detail in following sections is the ability of cells to internalize the receptor/ligand complex. This not only enables cells to accumulate large amounts of ligand against a concentration gradient but also to remove the complex from the surface of the cell. The definition of a receptor is largely operational in the sense that the binding of ligand to receptor usually mediates a cellular event such as internalization of receptor and ligand or alteration of some metabolic characteristic. Indeed, most receptors were first described not by their binding characteristics, but the cellular response generated by the addition of ligand to cells (16). The receptor for low density lipoprotein (LDL), for example, was first observed by its involvement in the regulation of cholesterol metabolism (17,18). The asialoglycoprotein receptor was discovered while studying the role of carbohydrates in regulating the serum survival time of plasma glycoproteins (19). The mannose-6- 7 phosphate receptor was identified while studying genetic disorders in which lysosomal hydrolases accumulated in extracellular fluids instead of lysosomes (20). Receptors have been functionally divided into two general categories (21). Class I receptors elicit some physiological response (e.g., cell division or alteration of cell metabolism) upon binding of ligand. Examples of this are the receptors for epidermal growth factor (EGF) and insulin which stimulate cell division and glucose transport respectively. Class II receptors are involved in accumulating ligand and examples of this type of receptor are those for Transferrin (Tf) and LDL. The ability to bind and internalize large amounts of ligand not only serves to supply the cell with required substances such as iron and cholesterol, but also to clear potentially harmful agents as is the case for the macroglobulin protease complex receptor. Even though different receptor types are genetically distinct, they share some common physical requirements for ligand binding. For example, most hormone receptors, like those for epidermal growth factor and insulin bind ligand with high affinity independent of the presence of divalent cations (22,23). Other receptors such as those for LDL and alpha-2-macroglobulin exhibit a marked decrease in binding affinity in the absence of calcium or the presence of chelators such as EDTA (24,25). In general, those receptors which require calcium to exhibit high binding affinities are also sensitive to changes in pH. That is, they display a rapid decrease in binding affinity as the pH is reduced from neutrality (7.0) and approaches 8 5.5-6.0. The fact that binding constants can be altered by the concentration of calcium or hydrogen ions may be critical for the dissociation of receptor/ligand complexes within the cell. Although there is no direct evidence that endocytic compartments transport calcium, it is known that calcium is not accumulated within this pathway. (26). There is also evidence for acidic intracellular structures Low pH and reduced calcium have been shown to have a synergistic effect on the dissociation of asialoorosomucoid from its receptor (27). The acidic nature of elements of the endocytic apparatus will be discussed in more detail in following sections. Internalization of receptor/ligand complexes serves two basic functions. First, it removes receptor/ligand complexes from the cell surface. Studies indicate that some hormone responses precede the internalization step. Endocytosis of occupied Class I receptors then provides a means to terminate the signal. Second, internalization of Class II complexes allows cells to accumulate ligand. Occupation of Class II receptors in and of itself does not appear to lead to any alteration in cell behavior or metabolic activity. Metabolic changes if they occur at all are consequences of metabolism of the ligand. An example of this is the regulation of LDL receptor synthesis by intracellular cholesterol. Accumulation of cellular iron has also been shown to regulate the expression of Tf receptors (28), That the major function of Class II receptors is ligand accumulation is demonstrated by the finding that these receptors may be reutilized. Experiments with the receptors for LDL, macroglobulin protease complex, and mannose terminal 9 glycoproteins revealed that ligand could be bound, internalized, and delivered to intracellular compartments without a loss of receptor number at the cell surface. In fact, these receptors are capable of internalizing in the steady state an amount of ligand equal to fiveto tenfold the amount of ligand bound at the surface (29-32). Maintenance of surface receptor number cannot be explained at the level of biosynthesis since the rate of synthesis is from 1 to 15% the rate of internalization (33,34). Furthermore, it was found that the rates of internalization for LDL and macroglobulin protease complex could be maintained for extended periods in the absence of protein synthesis (29,35). The possibility that receptors could be continually recruited from a cryptic intracellular pool was ruled out by the experiments of Doyle et al. (36). These investigators fused hepatocyte plasma membrane vesicles (which for asialoglycoproteins) to mouse L cells. contai~ed receptors L cells do not express receptors for asialoglycoproteins and therefore do not bind these ligands. After fusion, the mouse cells mediated several rounds of ligand internalization. Thus, continued uptake of asialoglycoproteins could not be due to recruitment from intracellular pools of receptors. Although there is evidence that some Class I receptors may be reutilized, many appear not to be recycled. This is based primarily on the observation that surface binding is reduced in cells which have been exposed to ligand. This apparent "endocytic down regulation" has been observed for the EGF receptor (37). However the possibility exists that occupancy reduces the binding affinity of 10 the receptor. From the standpoint of regulation, it is important to be able to terminate a signal once the information has been transmitted. One way this can be accomplished is to internalize the receptor/ligand complex. occupied ins~lin For example, internalization of the receptor may not be required in order to produce an effect (38). Thus, removal of the complex from the cell surface may provide a means to stop insulin-specific effects. Further, it has been shown for the human choriogonadotropin receptor that the hormonal response ceases upon removal of the receptor/ligand complex (39). Regardless of the functional significance of different receptors, the fact remains that some internalized receptors are recycled, and some are not. These complexes must therefore be processed differentially by the cell. It is not clear at what cellular level the decision to recycle is made. It could be made when the receptor/ligand complex is formed, at the internalization step, or within some intracellular compartment. Although there must be information in the primary structure of these proteins, analysis of the amino acid sequences for many receptors has revealed no consensus between recycled and nonrecycled receptors. Analysis of how receptors interact with the endocytic pathway may provide insight into how different receptor/ligand complexes are sorted by the cell. For example, the formation of the receptor/ligand complex could cause a conformational change in the receptor which might then act as the message to recycle or to be transferred to lysosomes. Recycling receptors may be internalized 11 VIa a different mechanism than nonrecycling receptors. Or the endocytic apparatus itself may determine the various fates of the complex. Many of the responses observed for ligand binding are initiated by the internalization of receptor/ligand complexes. Once ligand has bound to its receptor, the receptor/ligand complex appears to be internalized via specialized regions of the plasma membrane called coated pits. One of the first observations of this structure was made by Roth and Porter when they described bristle coated invaginations of mosquito oocytes (40). The coat was later found to be made up primarily of the 180 kd protein clathrin (41,42). In all cases where ultrastructural data are available, receptor/ligand complexes enter the cell through coated pits (43,44). The precise role of coated pits in receptor mediated endocytosis is unclear. However, one class of mutants of the LDL receptor discovered in the disease of familial hypercholesterolemia are capable of binding LDL, but do not internalize the receptor/ligand complex. When the distribution of receptors was analyzed by electron microscopy, it was found that the receptors could not cluster over coated pits (45). This result suggests that receptor internalization is dependent on the ability to cluster over coated membrane. However, clathrin-deficient mutants in yeast have been described which are both viable and capable of secretion (46). The role of clathrin and coated pits in receptor mediated endocytosis might be to limit the size of the endocytic vesicle and provide a means to concentrate receptor/ligand complexes. The ability of 12 receptor/ligand complexes to cluster prIor to internalization would greatly increase the ratio of ligand to membrane and thus reduce the amount of membrane which would have to be recycled by the cell. Recent experiments indicate that some receptors are capable of being phosphorylated. For example, the binding of EGF and insulin to their receptors results in a rapid hyperphosphorylation of the receptor (47 -49). Tumor promoting agents such as phorbol esters are known to stimulate phosphorylation of cellular proteins by activating protein kinase C (50). The receptors for EGF and Tf are among those proteins found to have increased levels of phosphorylation. A secondary effect of these agents is the redistribution of surface and internal receptors. It has thus been suggested that phosphorylation may affect the internalization or recycling rate of receptors (51,52). Although phosphorylation of receptors may reflect an altered metabolic state in the cell, it does not appear to be the signal for internalization or recycling of receptors. This is based on three observations. First, phosphorylation of the receptor alone has not been demonstrated to be sufficient for redistribution. Second, the action of phorbol esters has opposite effects in the redistribution of Tf receptors in different cell types (53,54). Third, deletion of two phosphorylation sites on the EGF receptor abolished the high affinity state of the receptor, but had no effect on its ability to be internalized via endocytosis (55). Studies on insertion mutants demonstrate that the protein- tyrosine kinase activity of the receptor can be abolished while both 13 high affinity binding and the ability to be internalized are retained (56). Ligands may have multiple binding sites for receptors, 1.e., a single ligand may be capable of binding several receptors. The net result of such multivalent ligand/receptor interactions would be the crosslinking of receptors at the cell surface. It has been suggested that the valency of receptor/ligand complexes may determine their intracellular fate (57). Evidence for this hypothesis is based primarily on two sets of observations. First, studies on the Fc receptor in macrophages suggest that this receptor recycles in the absence of ligand (58,59). If presented with antibody coated red blood cells, the receptor fails to recycle from the phagolysosome. Monovalent Fab fragments, however, bind to the receptor, are internalized, and are recycled along with the receptor (60,61). Second, incubation of cells with antibodies directed against receptors for LDL, mannose-6-phosphate, Tf, and Fc prevent recycling of these receptors and induce transfer to lysosomes (62-66). Crosslinking of receptors at the cell surface alone is not likely to be the signal for lysosomal transfer since LDL and proteins containing mannose-6phosphate are multivalent. These receptors recycle to the surface in the unoccupied state suggesting that ligand dissociates in an internal compartment. Thus, when the ligand/receptor complex dissociates, receptors would be released from the multivalent complex and free to recycle. Although the crosslinking of receptors may direct the Fc receptor along the lysosomal pathway, insulin and EGF are thought to bind via monovalent attachment to their receptors, yet both are 14 transported to lysosomes. The hypothesis of receptor crosslinking and lysosomal transfer may therefore apply to some receptors, but does not appear to be universal. Thus, neither receptor phosphorylation nor surface crosslinking alone appears to be the signal for recycling. Analysis of the interactions between receptor and ligand have not revealed a common set of rules which govern decisions for recycling. It may therefore be profitable to begin looking at the way in which the receptor/ligand complex interacts with the components of the endocytic apparatus. Experimental evidence indicates that plasma membrane differs from membrane derived from intracellular organelles both in phospholipid makeup and protein content. The continual exchange of surface 'membrane with internal membrane pools has the net effect of randomizing these constituents. The observed differences must then be maintained by some type of cellular sorting process. This could be achieved either by selectively internalizing segments of membrane which recycle through the cell by different pathways, or by sorting components from the endocytic compartment. The first possibility has been assessed by analyzing the components included in pinocytic and phagocytic vesicles. Pinocytic compartments have been studied by specifically radiolabeling proteins included in the pinocytic vesicle. The enzyme lactoperoxidase is internalized by cells by fluid phase uptake and can be used to catalyze the radioiodination of proteins exposed on the luminal face of the pinocytic compartment. U sing this approach, 15 Mellman and Galloway (67) were able to compare iodinated plasma membrane proteins to those of pinocytic vesicles. Analysis of proteins immunoprecipitated using monoclonal antibodies raised against surface antigens revealed similar patterns of radiolabeled protein on SDS polyacrylamide gels. This result led the authors to suggest that the proteins included in the pinocytic vesicle are present in the same relative concentrations as on the plasma membrane. That is, the process of pinocytosis is not selective with respect to the segment of membrane which is internalized. Analysis of plasma membrane proteins or activities included in phagocytic vesicles has yielded conflicting results. Comparisons of labeling patterns following surface or phagosome iodination revealed no difference in the spectrum of proteins labeled, suggesting that membrane proteins are randomly included in the phagocytic vesicle (59,68). Tsan and Berlin (8) analyzed cell surface transport activities for adenine and lysine during phagocytosis. They found no loss of activity suggesting the transporters were not included in the phagocytic vesicle. Buys and Kaplan (69) measured the effects of phagocytosis on receptor mediated endocytosis in macrophages. Neither the surface receptor number nor the rates of internalization or recycling was affected. These results are consistent with those of Tsan and Berlin suggesting that the phagocytic vesicle selectively excludes certain plasma membrane components. A possible explanation for this discrepancy may be that plasma membrane proteins included within the phagocytic vesicle are rapidly returned to the surface of the cell. For example, 16 Muller et al. (59) reported that within 5 min after introducing radiolabel into the phagolysosome, 81 % of the radioactivity had redistributed to the plasma membrane. It is important to note that a major difference between these studies exists in the assay systems. Specifically, iodinatable membrane proteins are compared to plasma membrane activities. The most abundant membrane proteins are likely to be the most highly labeled and changes in minor proteins may not be detectable. Analysis of membrane associated activities is more sensitive and may therefore be a more accurate way of measuring changes in plasma membrane A major limitation of analyzing labeled proteins from composition. pin.ocytic or phagocytic compartments is that it is difficult to distinguish between the existence of multiple internalization pathways. Proteins isolated from cells labeled in this manner may represent an aggregate of endocytic compartments. U sing a lactoperoxidase-asialoorosomucoid conjugate to label endosome proteins, Watts found a distinct pattern compared to surface labeled proteins (70). This result suggests that with a more specific method of labeling internalized vesicles, differences in endocytic vesicle and plasma membrane proteins can be detected. Although there may be some question as to the selectivity of membrane internalized at the cell surface, there is strong evidence for sorting within intracellular compartments. Much of the protein internalized with the phagolysosomal vesicle is degraded, but a significant portion returns to the plasma membrane (14). As described earlier, the problem associated with determining where 17 molecular sorting occurs is in identifying specific endocytic compartments. Receptor mediated endocytosis provides a useful model system in this regard. Several well characterized receptor/ligand complexes are known to be processed differently and therefore provide a suitable means to study intracellular sorting processes. Current thought suggests that intracellular sorting occurs in a nonlysosomal compartment. The evidence for a discreet internal compartment for receptor/ligand complexes derives from several areas of investigation which include morphological, kinetic, physical, and biochemical approaches. In situ binding and internalization of asialoorosomucoid (ASOR) conjugated to horseradish peroxidase was analyzed in perfused rat livers (71). Peroxidase activity was found in coated pits and coated vesicles at early time points. Within 5 min, the marker appeared in Golgi-Iysosome regions and vesicles of approximately 200 nm in diameter. Analysis of ligand conjugated to electron dense probes confirmed the peroxidase data. Most importantly, at least 15 min was required before lysosomal transfer could be detected, suggesting that the compartment between the coated vesicle and the lysosome was fairly long-lived. Immunocytochemical techniques using colloidal gold conjugates to antibody directed against the asialoglycoprotein receptor confirmed the presence of receptors within these compartments (72). Furthermore, receptors could not be detected in lysosomes indicating that the receptor/ligand complex dissociated prior to delivery of ligand to lysosomes. 18 The second body of research suggesting the existence of an internal compartment for receptor/ligand complexes is the kinetic analysis of ligand movement through cells. U sing a technique which would precipitate receptor/ligand complexes but not receptor or ligand alone, Bridges et al. (73) demonstrated that 125 1-ASO Rand the asialoglycoprotein receptor were internalized as a unit. They also determined that the unoccupied receptor recycled, and that the complex dissociated prior to the degradation of ASOR in lysosomes. Reduction of the temperature to 20 0 C prevented transfer of ligand to lysosomes despite that fact that endocytosis continued suggesting that the endocytic and lysosomal pathways could be separated (74). Subcellular fractionation studies using internalized radioligand to trace the endocytic pathway demonstrated that ligand was first found in a compartment of relatively low buoyant density which was quite distinct from the higher density of lysosomal enzyme markers (75). Further studies using EGF, LDL, and mannose terminated glycoprotein indicated that these ligands first entered a low density compartment and with time were transferred to a compartment with a buoyant density corresponding to lysosomes (76). Studies using fluorescein labeled alpha-2-macroglobulin revealed that the ligand was internalized into a compartment which rapidly became acidified (77). Further analysis showed that the compartment was devoid of acid phosphatase activity, indicating that it was nonlysosomal. The possible significance of low pH in the prelysosomal compartment was suggested by the use of agents 19 which caused an increase in intracellular pH. Weak bases such as ammonium chloride will produce this effect. Incubation of cells with ammonium chloride prevented transfer of 125I-ASOR to lysosomes and caused ligand accumulation in the prelysosomal fraction (78). Furthermore, mutants unable to acidify the endocytic compartment were also blocked in the delivery of ligand to lysosomes (79,80). Agents which raise the pH of intracellular compartments may then be used to disengage the endocytic compartment from the lysosome. The endocytic compartment has been gIven several names in recent years including endosome, compartment for uncoupling of receptor and ligand (CURL), and receptosome. For the sake of clarity, it will be referred to here as the endosome. The endosome can be functionally defined as the compartment which contains internalized receptor and ligand. Because different ligands can be found in the same endocytic vesicle, and the various receptor/ligand complexes have different eventual fates, it is believed that the endosome is the organelle responsible for sorting these complexes (81-85). Following the loss of clathrin from the coated vesicle, morphological descriptions of the itinerary for receptors and ligand become somewhat varied (86-88). One point which is generally agreed upon is that newly formed endocytic vesicles of approximately 100 nm in diameter fuse with other membranous compartments because ligand can be found in vesicles of approximately 200 nm in diameter as well as tubular structures (71,72). 20 In general there have been two basic morphological approaches used to investigate the struc~ural nature of the endocytic apparatus. Pulse chase experiments using electron dense tracers or enzymes conjugated to ligand have been used extensively to describe the endocytic/recycling pathway (71,86). More recently, the use of monoclonal antibodies has allowed the localization of specific receptors in the absence of ligand (82,83). Antibodies which recognize either the asialoglycoprotein receptor or asialofetuin can be coupled to colloidal gold particles of differing sizes. U sing these probes, Geuze and co-workers described the tubulo-vesicular network which they termed the CURL. In rat parenchymal cells, this structure is a peripheral network adjacent to the sinusoidal membrane. This network changed into "tubulo-vesicles" in the region of the Golgi. The authors suggested that this region was equivalent to the endosome defined in other systems. In other morphological studies internalized ligand was found in regions of the cell near the Golgi (89-91). This observation has led to the hypothesis that receptors recycle via the Golgi apparatus. Morphological and biochemical data suggest that some fraction of surface receptors traverse a longer circuit through the cell which may include elements of the Golgi. Hep 2 cells were analyzed using a monoclonal antibody directed against the asialoglycoprotein receptor (92). In the absence of protein synthesis significant levels of asialoglycoprotein receptor could be detected within the Golgi stacks. This suggests that these receptors reached the Golgi via a route different from the biosynthetic pathway. The authors 21 proposed that these receptors originated at the cell surface, that IS, they represent internalized receptors. Biochemical evidence suggests that internalized proteins can be exposed to enzymatic activities associated with the Golgi. Regoeczi et al. (93) determined that approximately 25% of 125I-asialoTf injected into rats was resialated with time. Human erythroleukemia cells which had been treated with neuraminidase to desialate surface proteins exhibited a time-dependent resialation of the Tf receptor (94). More recently, Fishman and Fine (95) have succeeded in isolating Golgi-derived exocytic vesicles which contain internalized Tf. This type of biochemical evidence provides a compelling argument for the proposition that surface proteins can recycle through Golgi compartments. Kinetic analysis, however, indicates that the half time for traversing the pathway which includes Golgi-specific activities is on the order of hours (95,96). Recycling times for the pathway involved in ligand accumulation, however, vary from 10 to 20 min (43,97). Resolution of this discrepancy will require a more complete characterization of the compartment(s) included in the endocytic apparatus. The biochemical nature or composition of the endosome has been difficult to assess for one major reason. That is, a definitive endosomally associated enzymatic activity has not yet been identified. An endosomal associated protease activity has been reported in rabbit alveolar macrophages (98). The degradation of 125I-mannose BSA was found as early as 6 min following internalization suggesting that it had not yet reached the lysosome. 22 Subcellular fractionation studies confirmed that this early compartment was nonlysosomal. Although the data are consistent with an endosomal protease, it is not known whether the enzyme is unique to endosomes. It has been demonstrated that the endosome maintains an acidic pH (77,99). Cell free systems have been used to study the proton pumps associated with partially purified endosomes and lysosomes (100). The two organelles displayed similarities in that the pH gradient could be dissipated by proton ionophores and weak base. Furthermore, neither was affected by vanadate or oubain nor did they require permeant anions such as Sodium or Potassium. were inhibited by N-ethylmaleimide. Both These similarities led the authors to suggest that endosomes and lysosomes used similar if not identical proton pumps to reduce intraluminal pH. However, mutant cell lines have been isolated in which the ATP-dependent acidification of endosomes is impaired (101). In vitro analysis revealed that the acidification of lysosomes was unaffected. The similarities are therefore either coincidental or more than one type of proton pump could exist in the lysosome. While low pH is probably physiologically important for endosomal function, at present, it would not be a suitable marker for the endosome. Endosomes are one of a number of intracellular compartment which exhibits a low intraluminal pH. Lysosomes were among the first acidic organelles to be characterized (102,103). More recently, elements of the Golgi and secretory vesicles have been demonstrated to contain proton pumps (104,105). Thus low 23 pH alone would not suffice to mark the endosomal compartment. It is possible that the endosomal proton pump is different from pumps associated with other organelles, and therefore could be specifically recognized by antibodies. Such antibodies do not exist at this time. D nfortunately, aside from the acidic nature of the endosome and the reported endosomally associated protease, there is little else known about the biochemical nature of this organelle. Thus, there is no clear, inherent biochemical marker for the endocytic compartment. A major approach to the purification and characterization of endosomal compartments has been to separate them based on differences in their physical characteristics. In this procedure cells are disrupted to the point where the plasma membrane is broken and the cellular contents are released. Most segments of membrane released by the disruption procedure vesicularize. The pioneering efforts of Claude, DeDuve, and Palade (106-109) demonstrated that intracellular organelles exhibit differential centrifugation characteristics and these differences could be used as a method of purification. Techniques of differential centrifugation have been modified and refined to the point where n1any different systems are currently available for subcellular fractionation. More recently a separation technique called free-flow electrophoresis has been reported (110). By this method, membrane vesicles are allowed to segregate in an electrical field according to their relative charge. Both methods have their advantages but also suffer from a major limitation. That is, by these methods alone, it cannot be determined 24 whether activities which copurify are in the same or different com partmen ts. There are two general tactics which can be used to determine whether activities which copurify are in the same compartment. First, the specific activities of the two markers can be compared at different points along the purification scheme. If they are in the same compartment, they should exhibit identical degrees of enrichment. The second approach is to specifically alter the compartment which is known to contain one of the markers. If the second marker is within the same compartment, it should undergo a similar change. Although the first method is feasible, it suffers from a major limitation. If the two markers have differential stabilities, that is, they lose activity at different rates, it would be impossible to compare relative levels of purification. The second approach is not subject to this limitation, and simply asks whether the two markers exhibit similar changes. Most importantly, if a population of vesicles can be physically altered in a specific manner, it might be possible to purify them. The approaches which have been suggested physically alter intracellular compartments have been to either increase or decrease their relative buoyant densities (111-115). There are two recently described methods which can specifically increase the buoyant density of endocytic compartments (116-118). Thus, the endosome can theoretically be purified away from other membranous organelles which have similar physical characteristics. The method of Courtoy involves placing peroxidase activity into the endosome. 25 Vesicles derived from cellular homogenates are then incubated In the presence of H202 and diaminobenzidine (DAB). The peroxidase/H202-driven oxidation reaction causes DAB to polynlerize, thus becoming trapped inside the peroxidase-containing vesicle. Deposition of the polymer within the lumen results in an increase in the buoyant density of the vesicle. Peroxidase activity can be targeted to the endocytic apparatus by chemically coupling horseradish peroxidase to the appropriate ligand. The technique of Helmy et al. (118) uses the activity of acetylcholinesterase to produce a dense lead phosphate precipitate within vesicles. The method can also be made specific for the endosome by attaching the enzyme to ligand. To date, these are probably the most direct approaches to purifying the endosome. One of the most useful strategies in studying complex systems is reduction. Specifically, this approach seeks to gain understanding of the whole by dissecting it into simpler, more basic components. This general strategy has proven to be a powerful tool in the study of biological systems at all levels. The ability to characterize the intracellular structures or activities included in the endocytic/recycling pathway is dependent on the ability to identify the constituents involved. This, in turn is contingent upon either having a marker for, or the ability to purify these constituents. The ability to increase the density of endosomal compartments in a specific manner provides a means to isolate the constituents of the endocytic apparatus. Thus, the molecules or activities involved in this pathway can be carefully examined in a quantitative manner. 26 In the studies described here, we have made use of the density shift technique of Courtoy to analyze components of the endocytic apparatus. In order to make the enzyme specific to the endosome we chemically coupled horseradish peroxidase to Tf. Because the Tf/Tf receptor complex traverses the endocytic/recycling pathway without being transferred to lysosomes, cells incubated with the conjugate can be studied under steady state binding conditions. Thus, the entire endocytic/recycling pathway would be expected to contain the conjugate which would allow us to density shift all of the integral compartments. During the course of these studies we also found that the DAB density shift procedure results in the inactivation of peroxidase containing compartments. I have used Tf-HRP and a combination of the DAB density shift and inactivation techniques to study the endocytic compartment. In the following chapters I describe the use and limitations of the DAB procedure, characterization of endocytic compartments, and how the technique can be used to test models for receptor mediated endocytosis. 27 References 1. Lewis, W.H. 1931. Pinocytosis. Bull. Johns Hopkins Hosp. 49: 17 2. Holter, H. 1959. Endocytosis. Int. Rev. Cytoi. 8: 481 3. Palade, G.E. 1956. The endoplasmic reticulum. J. Biophys. Biochem. Cytoi. 2 (Suppl.): 85 4. Bennett, H.S. 1956. The concepts of membrane flow and membrane vesiculation as mechanisms for active transport and ion pumping. J. Biophys. Biochem. Cytol. 2 (SuppI.): 99 5. Steinman, R.M., Brodie, S.E., and Z.A. Cohn. 1976. Membrane flow during pinocytosis. A stereologic analysis. J. Cell BioI. 68: 665 6. Chapman-Andresen, C. 1962. Studies on pinocytosis in Amoebae. C.R. Trav. Lab. Carlsberg. 33: 73 7. Bowers, B., and T.E. Olszewski. 1972. Pinocytosis in Acanthamoeba castellanii. J. Cell BioI. 53: 681 8. Tsan, M., and D. Berlin. 1971. Effect of phagocytosis on membrane transport of nonelectrolytes. J. Exp. Med. 134: 1016 9. Petty, H.R., Hafeman, D.G., and H.M. McConnell. 1981. Disappearance of macrophage surface folds after antibodydependent phagocytosis. J. Cell BioI. 89: 223 10. Palade, G. 1959. Functional changes in the structure of cell components. In Subcellular particles. T. Hayashi, ed. The Ronald Press Company, New York. P. 64. 11. Jamieson, J.D., and G .E. Palade. 1971. Synthesis, intracellular transport, and discharge of secretory proteins in stimulated pancreatic exocrine cells. J. Cell BioI. 50: 135 12. Heuser, J.E., and T.S. Reese. 1973. Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. J. Cell BioI. 57: 315 28 13. Dvorak, A.M., Schleimer, R.P., Schulman, E.S., and L.M. Lichtenstein. 1986. Lab. Invest. 54: 663 14. Muller, W.A., Steinman, R.M., and Z.A. Cohn. 1983. Membrane proteins of the vacuolar system. III. Further studies on the composition and recycling of endocytic vacuole membrane in cultured macrophages. J. Cell BioI. 96: 29 15. Duncan, R., and M.K. Pratten. 1977. Membrane economics in endocytic systems. J. Theor. BioI. 66: 727 16. Anderson, R.G.W., and J. Kaplan. 1983. Receptor-mediated endocytosis. In Current topics in cell biol0l:Y. B. Satir ed., Academic Press. P. 1 17. Goldstein, J.L., and M.S. Brown. 1973. Familial hypercholesterolemia: Identification of a defect in the regulation of 3-hydroxy-3-methylglutaryl coenzyme A reductase activity associated with overproduction of cholesterol. Proc. Nat. Acad. Sci. USA. 70: 2804 18. Goldstein, J.L., and M.S. Brown. 1977. The low-density lipoprotein pathway and its relation to atherosclerosis. Ann. Rev. Bioch. 46: 897 19. Ashwell, G., and Morell. 1974. The role of surface carbohydrates in the hepatic recognition and transport of circulating glycoproteins. Adv. in Enzym. 41: 99 20. Neufeld, E.F., Timple, W.L., and L.J. Shapiro. 1975. Inherited disorders of lysosomal metabolism. Ann. Rev. Biochem. 44: 357 21. Kaplan, J. 1981. Polypeptide-binding membrane receptors: Analysis and Classification. Science 212: 14 22. Carpenter, G., and S. Cohen. 1976. 125I-Iabeled human epidermal growth factor. Binding, internalization, and degradation in human fibroblasts. J. Cell BioI. 71: 159 29 23. Kahn, R.C., and K. Baird. 1978. The fate of insulin bound to adipocytes. J. BioI. Chern. 253: 4900 24. Basu, S.K., Goldstein, J.L., and M.S. Brown. 1978. Characterization of the low density lipoprotein receptor in membranes prepared from human fibroblasts. J. BioI. Chern. 253: 3852 25. Kaplan, J., and M.L. Nielsen. 1979. Analysis of macrophage surface receptors. J. BioI. Chern. 254: 7323 26. Mellman, I., Fuchs, R., and A. Helenius. 1986. Acidification of the endocytic and exocytic pathways. Ann. Rev. Bioch. 55: 663 27. Harford, J., Klausner, R.D., Wolkoff, A.W., Bridges, K.R., and G. Ashwell. 1984. Asialoglycoprotein catabolism by hepatocytes: Insights from perturbation of the endocytic pathway. In The molecular basis of lysosomal stora2e disorders. R.O. Brady and I.A. Barranger, eds. Academic Press. P. 149 28. Ward, J.H., Kushner, J.P., and J. Kaplan. 1982. Regulation of HeLa cell transferrin receptors. J. BioI. Chern. 257: 10317 29. Kaplan, J. 1980. Evidence for reutilization of surface receptors for alpha-macroglobulin-protease complexes in rabbit alveolar macrophages. Cell 19: 187 30. Goldstein, J.L., Basu, S.K., Brunschede, G.Y., and M.S. Brown. 1976. Release of low density lipoprotein from its cell surface receptor by sulfated glycosaminoglycans. Cell 7: 85 31. Roth, T.F., Cutting, J.A., and S.B. Atlas. 1976. Protein transport: A selective membrane mechanism. J. Supramoi. Struct. 4: 527 32. Goldstein, J.L., and M.S. Brown, 1974. Binding and degradation of low density lipoproteins by cultured human fibroblasts. J. BioI. Chern 249: 5153 33. Tanabe, T., Priler, V.E., and G. Ashwell. 1979. Subcellular membrane topology and turnover of a rat hepatic binding protein specific for asialoglycoproteins. J. BioI. Chern. 254: 1038 30 34. Brown, M.S., and J.L. Goldstein. 1975. Regulation of the activity of the low density lipoprotein receptor in human fibroblasts. Cell 6: 307 35. Brown, M.S., Anderson, R.G.W., and J.L. Goldstein. 1983. Recycling receptors: The round-trip itinerary of migrant membrane proteins. Cell 32: 663 36. Doyle, D.D, Hou, E., and R. Warren. 1979. Transfer of the hepatocyte receptor for serum asialoglycoproteins to the plasma membrane of a fibroblast. J. BioI. Chern. 254: 6853 37. Wiley, H.S., and D.D. Cunningham. 1981. A steady state model for analyzing the cellular binding, internalization and degradation of polypeptide ligands. Cell 25: 433 38. Kahn, R.C., Baird, K.L., Jarrett, D.B., and J.S. Flier. 1978. Direct demonstration that receptor crosslinking or aggregation is important in insulin action. Proc. Nat. Acad. Sci. USA 75: 4209 39. Segaloff, D.L., and M. Ascoli. 1981. Removal of the surfacebound human choriogonadotropin' results in the cessation of hormonal responses in cultured leydig tumor cells. I. BioI. Chern. 256: 11420 40. Roth, T.F., and K.R. Porter. 1964. Yolk protein uptake in the oocyte of the mosquito Aedes Ael:ypti. I. Cell BioI. 20:313 41. Pearse, B.M.F. 1975. Coated vesicles from pig brain: Purification and biochemical characterization. I. Mol. BioI. 97: 93 42. Pearse, B.M.F. 1976. Clathrin: A unique protein associated with intracellular transfer of membrane by coated vesicles. Proc. Nat. Acad. Sci. USA 73: 1255 43. Goldstein, I.L., Anderson, R.G.W., and M.S. Brown. 1979. Coated pits, coated vesicles, and receptor-mediated endocytosis. Nature 279: 679 44. Gorden, P., Carpentier, I.L., Cohen, S., and L. Drci. 1978. Epidermal growth factor: Morphological demonstration of 31 binding, internalization, and lysosomal assocIatIon In human fibroblasts. Proc. Nat. Acad. Sci. USA 75: 5025 45. Anderson, R.G.W., Goldstein, J.L., and M.S. Brown. 1977. A mutation that impairs the ability of lipoprotein receptors to localize in coated pits on the cell surface of human fibroblasts. Nature 270: 695 46. Payne, G.S., and R. Schekman. 1985. A test of clathrin function in protein secretion and cell growth. Science 230: 1009 47. Hunter, T., and J.A. Cooper. 1981. Epidermal growth factor induces rapid tyrosine phosphorylation of proteins in A431 human tumor cells. Cell 24: 741 48. Iwashita, S., and C.F. Fox. 1984. Epidermal growth fact and potent phorbol tumor promoters induce epidermal growth factor receptor phosphorylation in a similar by distinctively different manner in human epidermoid carcinoma A431 cells. J. BioI. Chern. 259: 2559 49. Kasuga, M., Karlsson, F.A., and C.R. Kahn. 1982. Insulin stimulates the phosphorylation of the 95,000-dalton subunit of its own receptor. Science 215: 185 50. Niedel, J.E., Kuhn, LJ., and G.R. Vandenberk. 1983. Phorbol diester receptor copurifies with protein kinase C. Proc. Nat. Acad. Sci. USA 80: 36 51. Beguinot, L., Hanover, J.A., Ito, S., Richert, N.D., Willingham, M.C., and 1. Pastan. 1985. Phorbol esters induce transient internalization without degradation of unoccupied epidermal growth factor receptors. Proc. Nat. Acad. Sci. USA 82: 2774 52. May, W.S., Jacobs, S., and P. Cuatrecasas. 1984. Association of phorbol ester-induced hyperphosphorylation and reversible regulation of transferrin membrane receptors in HL60 cells. Proc. Nat. Acad. Sci. USA 81: 2016 53. Klausner, R.D., Harford, J., at;Kl J. van Renswoude. 1984. Rapid internalization of the transferrin receptor in K562 cells is 32 triggered by ligand binding or treatment with a phorbol ester. Proc. Nat. Acad. Sci. USA 81: 3005 54. Buys, S.S., Keogh, E.A., and J. Kaplan. 1984. Fusion of intracellular membrane pools with cell surfaces of macrophages stimulated by phorbol esters. Cell 38: 569 55. Livneh, E., Prywes, R., Kashles, 0., Reiss, N., Sasson, I., Mory, Y., Ullrich, A., and J. Schlessinger. 1986. Reconstitution of human epidermal growth factor receptors and its deletion mutants in cultured hamster cells. J. BioI. Chern. 261: 12490 56. Prywes, R., Livneh, E., Ullrich, A., and J. Schlessinger. 1986. Mutations in the cytoplasmic domain of EGF receptor affect EGF binding and receptor internalization. EMBO Journal 5: 2179 57. Mellman, I. 1985. Mechanisms of molecular sorting in endosomes. In New insights into cell and membrane transport processes. G. Poste and S.T. Crooke eds. Plenum Press. P. 377 58. Steinman, R.M., Mellman, I.S., Muller, W.A. and Z.A. Cohn. 1983. Endocytosis and the recycling of plasma membrane. J. Cell BioI. 96: 1 59. Muller, W.A., Steinman, R.M., and Z.A. Cohn. 1980. The membrane proteins of the vacuolar system. II. Bidirectional flow between secondary lysosomes and plasma membrane. J. Cell BioI. 86: 304 60. Mellman, I., Plutner, H., and P. Ukkonen. 1984. Internalization and rapid recycling of macrophage Fc receptors tagged with monovalent antireceptor antibody: Possible role of a prelysosomal compartment. J. Cell BioI. 98: 1163 61. Schreiber, A.B., Liberman, T.A., Lax. I., Yarden, J., and J. Schlessinger. 1983. Biological role of epidermal growth factorreceptor clustering. J. BioI. Chern. 258: 846 62. Anderson, R.G.W., Brown, M.S., Beisiegel, U., and J.L. Goldstein. 1982. Surface distribution and recycling of the low density 33 lipoprotein receptors as visualized with anti-receptor antibodies. J. Cell BioI. 93: 523 63. von Figura, K., Gieselmann, V., and A. Hasilik. 1984. Antibody to mannose-6-phosphate receptor induces receptor deficiency in human fibroblasts. EMBO Journal 3: 1281 64. Neutra, M.R., Ciechanover, A., Owen, L.S., and H.F. Lodish. 1985. Intracellular transport of transferrin- and asialoorosomucoidcolloidal gold to lysosomes after receptor-mediated endocytosis. J. Histochem. Cytochem. 33: 1134 65. Hopkins, C.R., and I.S. Trowbridge. 1983. Internalization and processing of transferrin and the transferrin receptor In human carcinoma A431 cells. J. Cell BioI. 97: 508 66. Mellman, I., and H. Plutner. 1984. Internalization and degradation of macrophage Fc receptors bound to polyvalent immune complexes. J. Cell BioI. 98: 1170 67. Mellman, I., and C.J. Galloway. 1983. Selective labeling and quantitative analysis of internalized plasma membrane. Meth. Enzym. 98: 545 68. Hubbard, A.L., and Z.A. Cohn. 1975. Externally disposed plasma membrane proteins. II. Metabolic fate of iodinated polypeptides of mouse L cells. J. Cell BioI. 64: 461 69. Buys, S.S., and J. Kaplan. 1987. Effect of phagocytosis on receptor distribution and endocytic activity in macrophages. J. Cell Phys. (in press) 70. Watts, C. 1984. In situ 125I-Iabelling of endosome proteins with lactoperoxidase conjugates. EMBO Journal 3: 1965 71. Wall, D.A., Wilson, G., and A.L. Hubbard. 1980. The galactosespecific recognition system of mammalian liver: the route of ligand internalization in rat hepatocytes. Cell 21: 79 72. Geuze, HJ., Slot, J.W., Strous, GJ.A.M., Lodish, H.F., and A.L. Schwartz. 1982. Immunocytochemical localization of the 34 receptor for asialoglycoprotein in rat liver cells. J. Cell BioI. 92: 865 73. Bridges, K., Harford, J., Ashwell, G., and R.D. Klausner. 1982. Fate of receptor and ligand during endocytosis of asialoglycoproteins by isolated hepatocytes. Proc. Nat. Acad. Sci. USA 79: 350 74. Dunn, W.A., Hubbard, A.L., and N.N. Aronson. 1980. Low temperature selectively inhibits fusion between pinocytic vesicles and lysosomes during hepatocyte heterophagy of 125 1asialofetuin by the perfused rat liver. J. BioI. Chern. 255: 5971 75. van Renswoude, J., Bridges, K.R., Harford, J.B., and R.D. Klausner. 1982. Receptor-mediated endocytosis of transferrin and the uptake of Fe in K562 cells: Identification of a nonlysosomal acidic compartment. Proc. Nat. Acad. Sci. USA 79: 6186 76. Merion, M., and W.S. Sly. 1983. The role of intermediate vesicles in the adsorptive endocytosis and transport of ligand to lysosomes by human fibroblasts. J. Cell BioI. .96: 644 77. Tycko, B., and F.R. Maxfield. 1982. Rapid acidification of endocytic vesicles containing (l2-macroglobulin. Cell 28: 643 78. Harford, J., Bridges, K., Ashwell, G., and R.D. Klausner. 1983. Intracellular dissociation of receptor-bound asialoglycoproteins in cultured hepatocytes. J. BioI. Chern. 258: 3191 79. Robbins, A.R., Peng, S.S., and J.L. Marshall. 1983. Mutant Chinese hamster ovary cells pleiotropic ally defective in receptor-mediated endocytosis. J. Cell BioI. 96: 1064 80. Merion, M., Schlesinger, P., Brooks, R.M., Moehring, J.M., Moehring, T.J., and W.S. Sly. 1983. Defective acidification of endosomes in Chinese hamster ovary cell mutants "crossresistant" to toxins and viruses. Proc. Nat. Acad. Sci. USA 80: 5315 81. 35 Abrahamson, D.R., and R. Rodewald. 1981. Evidence for the sorting of endocytic vesicle contents during the receptormediated transport of IgG across the newborn rat intestine. J. Cell BioI. 91: 270 82. Geuze, H.J., Slot, J.W., and G.J.A.M. Strous. 1983. Intracellular site of asialoglycoprotein receptor-ligand uncoupling: Doublelabel immunoelectron microscopy during receptor-mediated endocytosis. Cell 32: 277 83. Geuze, H.J., Slot, J.W., Strous, G.J.A.M., Peppard, J., von Figura, K.~ Hasilik, A., and A.L. Schwartz. 1984. Intracellular receptor sorting during endocytosis: Comparative immunoelectron microscopy of multiple receptors in rat liver. Cell 37: 195 84. Wolkoff, A.W., Klausner, R.D., Ashwell, G., and J. Harford. 1984. Intracellular segregation of asialoglycoproteins and their receptor: A prelysosomal event subsequent to dissociation of the ligand-receptor complex. J. Cell BioI. 98: 375 85. Via, D.P. Willingham, M.C., Pastan, 1., Gotto, A.M., and L.C. Smith. 1982. Co-clustering and internalization of low-density lipoproteins and (X2-macroglobulin in human skin fibroblasts. Exp. Cell Res. 141: 15 86. Helenius, A., Mellman, 1., Wall, D., and A. Hubbard. 1983. Endosomes. TIBS 8: 245 87. Pastan, 1., and M.C. Willingham. 1983. Receptor-mediated endocytosis: Coated pits, receptosomes and the Golgi. TIB S 8: 250 88. Hopkins, C.R., and I.S. Trowbridge. 1983. Internalization and processing of transferrin and the transferrin receptor in human carcinoma A431 cells. J. Cell BioI. 97: 508 89. Yamashiro, D.J., Tycko, B., Pluss, S.R., and F.R. Maxfield. 1984. Segregation of transferrin to a mildly acidic (pH 6.5) paraGolgi compartment in the recycling pathway. Cell 37: 789 90. Willingham, M.C., Hanover, J.A., Dickson, R.B., and 1. Pastan. 1984. Morphologic characterization of the pathway of 36 transferrin endocytosis and recycling in human KB cells. Proc. Nat. Acad. Sci. USA 81: 175 91. Woods, J.W., Doriaux, M., and M.G. Farquhar. 1986. Transferrin receptors recycle to ill and middle as well as trans Golgi cisternae in Ig-secreting myeloma cells. J. Cell BioI. 103: 277 92. van den Bosch, R.A., Geuze, H.J., and G.J. Strouse 1986. Presence of asialoglycoprotein receptors in the Golgi complex in the absence of protein synthesis. Exp. Cell Res. 162: 231 93. Regoeczi, Eo, Chindemi, P.A., Debanne, M.T., and P.A. Charlwood. 1982. Partial resialation of human asialotransferrin type 3 in the rat. Proc. Nat. Acad. Sci. USA 79: 2226 94. Snider, M.D., and O.C. Rogers. 1985. Intracellular movement of cell surface receptors after endocytosis: Resialation of asialotransferrin receptor in human erythroleukemia cells. J. Cell BioI. 100: 826 95. Fishman, J.Bo, and R.E. Fine. 1987. A trans Golgi-derived exocytic coated vesicle can contain both newly synthesized cholinesterase and internalized transferrin. Cell 48: 157 96. Snider, M.D., and O.C. Rogers. 1986. Membrane traffic in animal cells: Cellular glycoproteins return to the site of Golgi mannosidase I. J. Cell BioI. 103: 265 97. Bleil, J.D, and M.S. Bretscher. 1982. Transferrin receptor and its recycling in HeLa cells. EMBO Journal 1: 351 98. Diment, S., and P. Stahl. 1985. Macrophage endosomes contain proteases which degrade endocytosed protein ligands. J. BioI. Chern. 260: 15311 99. Marsh, M. Bolzau, E. and A. Helenius. 1983. Penetration of Semliki Forest Virus from acidic prelysosomal vacuoles. Cell 32: 931 100. Galloway, CoJ., Dean, G.E., Marsh, M., Rudnick, G., and 1. Mellman. 1983. Acidification of macrophage and fibroblast endocytic vesicles in vitro. Proc. Nat. Acad. Sci. USA 80: 3334 37 101. Robbins, A.R., Oliver, C., Bateman, J.L., Krag, S.S., Galloway, C.J., and I. Mellman. 1984. A single mutation in Chinese hamster ovary cells impairs both Golgi and endosomal functions. J. Cell BioI. 99: 1296 102. deDuve, C. 1963. The lysosome concept. In Lysosomes Ciba foundation symposium. A.V.S. de Reuck, M.P. Cameron eds. Little, Brown and Co. P. 1 103. Mego, J.L., 1971. The effect of pH on cathepsin activities in mouse liver heterolysosomes. Bioch. J. 122: 445 104. Glickman, J., Croen, K. Kelly, S, and Q. AI-Awqati. 1983. Golgi membranes contain an electrogenic H+ pump in parallel to a chloride conductance. J. Cell BioI. 97: 1303 105. Orci, L., Ravazzola, M., Amherdt, M., Madsen, 0., Perrelet, A., Vassalli, J.D., and R.G.W. Anderson, 1986. Conversion of proinsulin to insulin occurs coordinately with acidification of maturing secretory vesicles. J. Cell BioI. 103: 2273 106. Claude, A. 1946. Fractionation of mammalian liver cells by differential centrifugation. I. Problems, methods, and preparation of extract. J. Exp. Med. 84: 51 107. DeDuve, C., Pressman, B.C., Gianetto, R., Wattiaux, R. and F. Appelmans. 1955. Tissue fractionation studies. 6. Intracellular distribution patterns of enzymes in rat-liver tissue. Biochem. J. 60: 604 107. DeDuve, C. 1963. The separation and characterization of subcellular particles. The Harvey Lectures 59: 49 108. Ehrenreich, J.H., Bergeron, J.J.M., Siekevitz, P., and G.E. Palade. 1973. Golgi fractions prepared from rat liver homogenates. I. Isolation procedure and morphological characterization. J. Cell BioI. 59: 45 109. Bergeron, J.J.M., Ehrenreich, J.H., Siekevitz, P. and G.E. Palade. 1973. Golgi fractions prepared from rat liver homogenates. II. Biochemical characterization. J. Cell BioI. 59: 73 38 110. Minnifield, N., Creek, K.E., Navas, P., and D.J. Morre. 1886. Involvement of cis and trans Golgi apparatus elements in the intracellular sorting and targeting of acid hydrolases to lysosomes. Eur. J. Cell BioI. 42: 254 111. Leskes, A., Siekevitz, P., and G .E. Palade. 1971. Differentiation of endoplasmic reticulum in hepatocytes. II. Glucose-6phosphatase in rough microsomes. J. Cell BioI. 49: 288 112. Davis, G.A., and F.E. Bloom. 1973. Subcellular particles separated through a histochemical reaction. Anal. Biochem. 51: 429 113. Ryan, J.W., and U. Smith. 1971. A rapid, simple method for isolating pinocytic vesicles and plasma membrane of lung. Biochim. et Biophys. ACTA 249: 177 114. Waldo, G.L., Doss, R.C., Perkins, J.P., and T.K. Harden. 1984. Use of a density shift method to assess beta-adrenergic receptor synthesis during recovery from catecholamine-induced downregulation in human astrocytoma cells. Mol. Pharm. 26: 424 115. Wetzel, M.G., and E.D. Korn. 1969. Phagocytosis of latex beads by Acanthamoeba castellanii (Neff). J. Cell BioI. 43: 90 116. Courtoy, P.J., Quintart, J., and P. Baudhuin. 1984. Shift of equilibrium density induced by 3,3' -diaminobenzidine cytochemistry: A new procedure for the analysis and purification of peroxidase-containing organelles. J. Cell BioI. 98: 870 117. Quintart, J., Courtoy, P.J., and P. Baudhuin. 1984. Receptormediated endocytosis in rat liver: Purification and enzymic characterization of low density organelles involved in uptake of galactose-exposing proteins. J. Cell BioI. 98: 877 118. Helmy, S. Porter-Jordan, K., Dawidowicz, E.A., Pilch, P., Schwartz, A.L., and R.E. Fine. 1986. Separation of endocytic from exocytic coated vesicles using a novel cholinesterase mediated density shift technique. Cell 44: 497 CHAPTER II CHARACTERIZATION OF ENDOCYTIC COMPARTMENTS USING THE HORSERADISH PEROXIDASE DIAMINOBENZIDINE DENSITY SHIFT TECHNIQUE 40 Characterization of Endocytic Compartments Using the Horseradish Peroxidase-Diaminobenzidine Density Shift Technique Richard S. Ajioka and Jerry Kaplan Department of Pathology, University of Utah College of Medicine, Salt Lake City, Utah 84132 Abstract. We have employed a modification of the horseradish peroxidase (HRP)-diaminobenzidine density shift technique of Courtoy et al. (J. Cell Bioi., 1984, 98:870-876) to examine the biochemical properties of the endosome. This organelle is involved in receptor recycling and the sorting of intemaJized receptor ligand complexes. Transferrin covalently bound to HRP was used to place peroxidase activity specifically within the endosome. The peroxidasecatalyzed polymerization of diaminobenzidine within these vesicles causes an increase in buoyant density. thus allowing them to be separated from other membranes. Using this technique we demonstrate that 12!l1_ the process of receptor-mediated endocytosis, newly internalized receptor-ligand complexes are found in a nonlysosomal, acidic, low density compartment called the endosome (7, 19, 20). Studies suggest that dissociation of receptor-ligand complexes occurs within the endosome and that the endosome is involved in receptor recycling (5). The endosome may also be the organelle in which the sorting of internalized molecules occurs. For example, some internalized ligands are transferred to Jysosomes while others are recycled to the cell surface. The fact that molecules may have different eventual fates raises the question of whether these receptor-ligand complexes are internalized into the same endosome. Morphological studies suggest this possibility (6) although rigorous biochemical evidence is lacking. Little is known about the biochemical characteristics of the endosome beyond the fact that it maintains an acidic pH. Similarly, it is unclear which subcellular organelle(s) constitutes the endocytic pathway. Morphological studies suggestthat the endocytic pathway is associated with, and may be part of, the Golgi system (16, 26). To resolve this issue, and to define the constituents ofthe endosome, it is necessary to purify or otherwise separate the endosome from other membrane compartments. We have modified the horseradish perox.idase (HRP)'- D URING I. Abbreviations used in this paper: DAB. 3,3'-diaminobenzldine; EGF. epidermal growth factor; HRP. horseradish permudase; TF. diferric transfernn. low density lipoprotein, Oil-epidermal growth factor, and Tf-HRP are internalized into the same endosome. We discovered that the diaminobenzidine reaction product "cross-links" the lumen of the vesicle, rendering vesicular components detergent ins01uble. Furthermore, the reaction inactivates enzymatic activities associated with the endosome. Thus, the diaminobenzidine density shift procedure has limited usefulness in studies designed to isolate endosomal constituents. Nonetheless, we have found that the inactivation of enzymatic activities is confined to those endosomes that contain peroxidase. This selectivity allows us to define endosome-specific activities. 3,3'-diaminobenzidine (DAB) density-shift procedure developed by Counoy et al. (3) to specifically increase the buoyant density of the endosome. The peroxidase-H2Ot-catalyzed oxidation of DAB within vesicles causes a dense polymer of DAB to form within the lumen which increases the buoyant density of the vesicle. Thus, peroxidase-containing vesicles can be separated from other vesicles by density gradient centrifugation. Diferric transferrin (TO covalently attached to HRP can be used to place peroxidase activity within the endocytic pathway. Tf is particularly useful in studying this pathway because it traverses the complete endocytosis-recycling route without being transferred to the lysosome. Diferric Tf binds to receptors on the cell surface and the receptor-ligand complexes are internalized via coated pits. These complexes are internalized into endosomes, where iron is released from Tf. The apo-Tf-Tf receptor complex is recycled back to the cell surface where apo-Tf dissociates from the receptor and is free to bind iron again (4, 9, 10, 15). The use of the Tf-HRP conjugate allows us to specifically mark the endocytic pathway. The density shift approach is extremely useful for determining whether different receptor-ligand complexes are internalized into the same endosome. In this paper we show that at least three different receptor-ligand complexes are internalized into the same endosome. In an attempt to purify endosomes using the density-shift procedure we have discovered a severe limitation of the perox.idase-DAB technique. The DAB polymer formed during 41 the peroxidase-H 202 reaction apparently cross-links and/ or oxidizes the luminal contents of the endosome, rendering these vesicles insoluble in detergent. The inability to extract protein from DAB-treated vesicles reduces the usefulness of this approach to purify and analyze peroxidase-containing compartments. However, the ability to specifically inactivate endosomal constituents can be used to define activities that are included within the endosome. Using this approach we demonstrate that the enzyme .leucyl-j)-naphtylamidase is highly enriched in the endosome. Materials and Methods Cells HeLa cells were grown on plastic culture dishes in MEM containing 10% newborn calf serum (Flow Laboratories. (nc .. Mclean, VA), penicillin (200 Ulml), and streptomycin (0.2 mg/ml). Cells were maintained at J7°C in a 5% C~ atmosphere and were subcultured by trypsinization. For some experiments. cells were incubated in serum-free media for 12 h before use. Preparation of WI-Tf(Fe)J and /J11-EGF Transferrin was saturated with iron (24). Tf and epidermal growth factor (EGF) were radioiodinated using Iodagen (Pierce Chemical Co., Rockford, IL) as described by Wiley and Cunningham (25). EGF and radioiodinated EGF were generous gifts from Dr. Steven Wiley. IZ'I·LDL was a kind gift from Dr. R. G. W. Anderson. Conjugation of HRP to 11 The HRP-Tf conjugate was prepared by the method of Nakane and Kawaoi (14). except that the final reaction was carried out using sodium cyanoborohydride (60 min. room temperature). The final preparation was stored in PBS in the presence of 10 mg/ml BSA at -20°e. Binding of 12'I-Tf or Tj-HRP The binding of 12'(_Tf or Tf-HRP to cells was performed as described elsewhere (2). Removal of surface bound ligand was achieved by washing cells at OCC with a citric acid-phosphate buffer (pH 3.8), containing 150 mM NaCI for 3 min followed by PBS (pH 7.2) for 3 min. This cycle was repeated three times. Subcellular Fractionation All operations were performed at O°e. Monolayers were washed with PBS and cells removed using a rubber policeman. Cell pellets were resuspended in 0.25 M STE bulfer (0.25 M sucrose in 10 mM Tris HCl [pH 7.21 and I mM EIJTA). Cells were homageniz.ed in a precooled. tight-fitting Dounce homogenizer using 25-30 strokes or until 80-90% of the cells were disrupted as monitored by phase-contrast microscopy. The homogenate was centrifuged at 800 g for 10 min. and the supernatant applied to 12% Percoll (Phannacia Fine Chemicals, Piscataway, NJ). lsoosmotic Percoll and various Percoll concentrations were prepared in sucrose according to manufacturer's recommendations. Percoll density was calculated by refractive index using a refractometer (Bausch and Lomb, Inc., Rochester. NY). Gradients were centrifuged in a Ti 75 rotor (Beckman Instruments, Inc., Palo Alto, CA) at 59.000 g.. for the specified times. Gradients were fractionated by pumping from the bottom of the centrifuge tube. Diaminobenzidine 7reatment The DAB treatment described by Counoy et al. (3) was used with minor modifications. DAB solutions were prepared at a concentration of 3 mg/ml in 0.25 M STE and the pH was carefully adjusted to 7.2 with 3.0 N NaOH. This solution was filtered through a 0.45-l1m filter (MilliporelContinental Water Systems. Asby, MA) and protected from light at O°C. The final reaction mixture contained 0.45 rng/ml DAB and 0.003% H2O;t. Reactions were performed by adding the DAB solution to vesicles and incubating at room temperature for 15 min. H2~ (0.3%) was then added and the mixture incubated for 15 min. Reaction mixtures were gently rocked in plastic snap· cap tubes covered with alummum foil. Some DAB-endo'iomal mactlvation studies were carried out on crude membrane preparatlons obtained by centrifuging the 800 g supernatant over a sucrose step gradient conmting of the supernatant underlayered with l3% sucrose (Wt/W(j in [0 mM Trl'; HC!. pH 7.2, I mM EDTA (TE buffer) and finally with 35% 'iucrose In the same bulfer. These gradients were centrifuged at 27.000 g.v tiJr 40 min and membranes were collected at the 13-35% sucrose Intertace. Enzyme Analyses Hexosaminidase (EC 3.2.1.30). galactosyltransferase IEC 2.4.1.38), Jnd leucyl-~·naphtylamidase (EC 1499) were assayed as described by Lamb et al. (12). Because Percoll interferes with colorimetric and absorbance readings, the smallest possible sample volumes were used in these reactions (usually 50 IJ.I). In some cases, enzyme analysis was performed on samples that had been detergent solubilized and cleared of Percoll by centrifugation (see soluble receptor assays). Soluble Receptor Assays The method of Lamb et al. (l2) was used for quantifying soluble Tf receptors. Membranes were solubilized by adding) % Triton X·100 to a final concentration of 0.1 % and BSA (20 mg/ml) to I mg/m1. Percoll was removed from detergent Iysates by centrifugation of samples at 105.000 g ... for 60 min over a cushion of 35% (wtlwt) sucrose in TE bulfer. Protein Determination in the Presence ofPercoll Protein determinations were made by the method of Vincent and Nadeau (21) using BSA (Fraction V, Sigma Chemical Co.. St. Louis. MO) as a protein standard. Results Subcellular Fractionation We first evaluated the separation of subcellular organelles on Percoll gradients. Unlike most internalized ligands that are directed to the lysosome, Tf and its receptor are cycled back to the cell surface. Incubation of cells at rl°C in the presence of I2~I·Tf results in distribution of the radiolabeled ligand between the cell surface and the endocytic pathway. Surface bound ligand can be selectively removed by washing cells at O°C with isotonic citric acid-phosphate buffer (pH 3.8) alternating with PBS (see Materials and Methods), leaving internalized ligand as the only source of cell-associated radioactivity. Greater than 95 % of surface-bound ligand can be removed using this procedure (data not shown). Endosomes were marked by internalized 125I_Tf and separated from other subcellular organelles by applying cellular homogenates to Percoll gradients. The distribution of internalized 12sI-Tf on a 12 % Percoll gradient centrifuged at 59,000 gay for Z7 min is shown in Fig. 1 a. The homogenate was either bottom loaded in 12.5% (wt/vol) sucrose, top loaded in 9% Percoll, or brought to 12 % Percoll and mixed throughout the centrifuge tube before centrifugation. All of these procedures gave rise to similar gradients of Percoll as measured by refractive index. Under these conditions internalized ligand was distributed in the gradient with a bimodal distribution. For simplicity, we refer to the more dense peak as peak A, and the less dense peak as peak B. If endosomes were in eqUilibrium by buoyant density, the gradient profiles should be the same for all methods of loading. We found instead that the distribution of radioactivity differed depending on the method of loading. The bimodal distribution of the endosomal marker, however, was the same for each method of loading, with the peaks of radioactivity occurring in the same positions on the gradients. The acid wash procedure re- 42 16 A 12 1.03 -; a I 5 9 13 17 21 FRACTION NUMBER Figu~ I. Distribution of internalized 1251_Tf in 12% Percoll gradients. (A) Cells were incubated at 11°C for 30 min in media containing 5 x lQ-9 M 125(_Tf. Surface-bound ligand was removed at O°C and the cells homogenized. Homogenates were either top loaded in 9% PercoU, bottom loaded in 12.5% sucrose. or brought to 12 % Percoll and mixed throughout the Percoll before centrifugation. Samples were applied to 12 % Percoll and gradients were formed by centrifuging at 59,000 g.. for Z1 min. Gradients were collected by pumping from the bottom of the centrifuge tube. Bottom load (open circles); top load (open triangles); mixed (solid circles); density (solid squa~s). (B) Cell homogenates were prepared as described above and applied to 12 % PercoU. Samples were applied either to the top or the bottom of the tube and centrifuged at the same g-force as above. but for a period of 40 min. Plasma membrane was labeled by incubating cells at O°C for 60 min with 1l!I(_ Tf. Homogenates of these cells were applied to the bottom of the tube before centrifugation. 11°C label. bottom load (open circles); 11°C label. top load (solid triangles); O°C label. bottom load (open a bimodal endosome distribution, was also effective in separating other organelles. Several marker enzymes were used to localize lysosomes and trons-Golgi elements across the gradient (Fig. 2, a and b). The most notable feature of these gradients was that lysosomes occurred in the most dense region of the gradient and were well separated from both endosome peaks. The distribution of the lysosomal enzyme hexosaminidase was similar regardless of whether gradients were top- or bottom-loaded (data not shown). The tronsGolgi marker galactosyltransferase (17) coincided with endosome peak A, and plasma membrane showed a similar distribution compared with endosome peak B. The enzyme leucyl-~-naphtylamidase had the same distribution as endosomes, supporting the suggestion that this enzyme represents an endosomal marker in HeLa cells (12). Even though endosomal markers do not reach density equilibrium under these centrifugation conditions, a reasonable degree of separation among various subcellular organelles could be achieved. Preparation of1f-HRP Conjugates Internalized wI-Tf in peak A and the Golgi enzyme galactosyltransferase exhibited a similar distribution on 12 % Percoli (Fig. 2 b). Similarly, a plasma membrane marker (vesicles from cells labeled at O°C) had a similar centrifugation pattern to endosome peak B. These activities might merely triangles). moved >85 % of the surface-bound ligand (data not shown). so neither peak represented I2jI-Tf bound to plasma membrane receptors. These results suggest that there are two physically different populations of endosomes that also differ in some characteristic besides density. Pulse-chase experiments using 125I-Tf to label endosomes and chase times ranging from 1 to 15 min resulted in the same relative distributions between peaks A and B (data not shown). This suggests that within this time frame radioligand is not being processed from one compartment to the other. Because of the relatively short centrifugation times used in these experiments, initial separation of membrane vesicles may be due to differences in size. Increasing the centrifugation time to bring the gradients to equilibrium produced a single peak of radioactivity for both top- and bottom-loaded gradients (Fig. I b). Although endosomes appeared to reach density equilibrium by 40 min, resolution between cellular compartments was poor. Radiolabeled plasma membrane vesicles (surface-bound radioactivity) could not be separated from endosomes (internalized radioactivity) under these conditions. Centrifugation in 12% Percoll for Z7 min. which yielded 24 ~ u B !'... I 18 ~[IUlJ·Tf o Galaclosyllransflrasf! O· Label jr--..!........ \/~ \ I h • ...J ~ 12 o ~\ I- ~ 6 \o. ° fH:., \ rll. ' l .'t.,~ 0'..... I/.....!:.'\ o~o·C>-o-o.o-o-.,.. ...- ...·,. ,- , 0.... t:: i ,~'6 ' . _ ... 0 ... i i 5 9 13 17 FRACTION NUMBER 21 Figure 2. Distribution of membrane activities on Percoll gradients. (A) Cells were incubated with '2~I-Tf for 30 min at 11°C. Cells were placed at O°C, surface-bound ligand was removed, and the cells were homogenized. The 800 g supernatant from the homogenate was applied to the top of 12 % Percoll and centrifuged as described in Materials and Methods. Radioactivity as well as enzyme activity was assayed across the gradient. (B) Cellular homogenates and Percoll gradients were prepared as in A. Cells incubated with radioligand at O°C were used to label the plasma membrane. Enzymatic activities were measured as described in Materials and Methods. 43 have similar sedimentation properties or they might actually be in the same compartment. To distinguish between these possibilities we used a modification of the density shift technique developed by Courtoy et a1. (3). This technique permits placement of HRP into specific intracellular compartments by covalently coupling the enzyme to an appropriate ligand. HRP-containing vesicles can then be reacted with H2~ and DAB. The HRP-H 20 Z oxidation reaction causes DAB to polymerize into a dense complex within the vesicle. Thus, any vesicle containing HRP will increase its buoyant density under these conditions.-HRP was conjugated with Tf in order to specifically increase the density of the Tfcontaining compartment 'and to compare the sedimentation properties of the more dense compartment with markers for other organelles. The protein conjugation procedure of Nakane and Kawaoi (14) was used to couple HRP to Tf. The conjugate was prepared using an initial 3:1 molar ratio of HRP/Tf. The final ratio based on protein concentration and enzyme activity was rv1:l.5. This value represents an approximation since the conjugation procedure affected peroxidase activity (using o-dianisidine as a substrate [18]) causing the reaction rate to slow with time (data not shown). Further analysis revealed that neither H2~ nor substrate was limiting during the reaction. We conclude that one of the steps used in preparing the HRP for conjugation alters enzymatic activity. We have not pursued this issue further since the enzyme retained enough activity for use. Analysis of the conjugated material by column chromatography and SDS polyacrylamide gel electrophoresis revealed products of various molecular weights. High molecular weight material was characterized by altered recycling kinetics. Presumably this material represents multimers of Tf-HRP-Tf. Conjugates with molecular weights below 200,000 were used for most experiments. Electrophoresis data suggested that a significant proportion of this material consisted of one to three molecules ofHRP per molecule of Tf. Less than 5 % of the protein migrated in a region corresponding to unconjugated HRP. Unconjugated HRP would be internalized by fluid phase pinocytosis. As demonstrated elsewhere (I), at the concentration used in this study, the amount of HRP taken up by pinocytosis is insignificant and has no measurable effect on these experiments. The most critical test of the conjugate was to demonstrate that it not only bound to the Tf receptor, but participated in the normal Tf cycle. Cell-associated HRP activity was drastically reduced when cells were incubated with the conjugate and excess Tf (Fig. 3 a). Furthermore the rate of loss of cellassociated peroxidase activity was similar to that for the radiolabeled ligand (Fig. 3 b). This result indicated that peroxidase accumulated during incubation at "!'rc participated in the normal Tf recycling pathway. The kinetics of loss of peroxidase activity suggest a process of exocytosis rather than degradation (t'I! for degradation of HRP in HeLa cells is 14 h). Peroxidase activity was recovered in the chase media, suggesting that the conjugate was not degraded during incubation with cells. Density Shifting of Endosomes The follOWing experiments were performed to confirm that the buoyant density of endosomes could be affected by the peroxidase reaction product and that the Tf-HRP conjugate and 1"~I-Tf were internalized into the same compartment. Figure 3. Internalized Tf-HRP A 100 recycles nonnally. (A) CeJJs were incubated with 12~I-Tf (2 x 10-9 M) and Tf-HRP (5 x D Tf-HRP 10-9 M) at J7°C for 40 min in ~ 50 the presence or absence of ex;: 25 cess unlabeled Tf (1.25 x c 10-' M). After incubation at H' 40' oTfIF.12 40'CHASE .TflF.1 2 J7°C, cultu~s we~ placed at ~ c 'CHUE 00C and surface ligand was~­ g <.I'> B moved. Half the cultu~s we~ ~ 100 ~tumed to J7°C and incu• '20I-TI ...I bated for 40 min in the p~s­ ;;:: 75 o Tf ·HRP ence of 1.25 x 10·' M Tf. At ;,e 50 this time, all cuJtu~s we~ harvested and cells lysed with 25 0.1 % Triton X-loo. Cell-associated radioactivity (closed o 5 10 15 2025 30 35 40 bars) and peroxidase activity TIME OF CHASE (open bars) we~ then measured. The graph ~p~sents the percentage of maximum activity. (B) Cell cultu~s we~ incubated for 40 min at J7 c C in the p~sence of Tf-HRP and I~J-Tf. Surface-bound ligand was ~moved at O°C and cells ~turned to J7°C in prewanned media containing excess Tf (1.25 x IO·~ M) for various times. At specified times cells we~ ~tumed to O°C, surface-bound ligand removed, and lysed in 0.1 % Triton X-loo. Cell-associated radioactivity (solid circles) and peroxidase activity (open circles) we~ then measured. 75 • '20I-TI (..) Q (..) Cells were incubated at rrc in the presence of both l2'I-Tf and Tf-HRP. Surface-bound ligand was removed at O°C, cells were homogenized, and the homogenate applied to 12 % PercolI gradients (Fig. 4 a). Peak A was collected and incubated with DAB in the presence or absence of H2~. This sample was then applied to a 17% Percoll gradient. There was a significant increase in the buoyant density of l2'I-Tf-containing vesicles incubated with H2~ and DAB (Fig. 4 b). The buoyant density of vesicles from cells not incubated with Tf-HRP was unaffected by the DAB treatment, and TfHRP-containing vesicles did not increase their density if H2~ was left out of the reaction mixture (Fig. 4 c). If vesicles containing Tf-HRP were mixed with vesicles containing only 12~I-Tf, there was no increase in density of the 12sl_ Tf-containing vesicles after DAB treatment (data not shown). These data demonstrate that only vesicles that contained peroxidase activity exhibited an increase in density. Similar results were obtained using vesicles isolated from peak B. We have observed shifts ranging from 60 to 90% of internalized radioligand. This variability may represent damage to endosomes during preparation. Internalization of Different Receptor-Ligand Complexes Morphological studies have demonstrated that ligands bound to different receptors enter the endosome very soon after internalization and that different ligand receptor complexes initially utilize the same endocytic pathway (6). In addition, biochemical studies have shown that different ligands exhibit similar subcellular distributions, suggesting that they are localized in the same compartment (13). We used the density shift technique to determine whether different Jigands were in the same or different compartments. CeJls were incubated 44 A 12.0 eTf o HRP-Tf 9.0 9 ~\", I 6.0 3.0 II :t 8 e o ...J 9 13 17 21 If::, j, ',\ 5'~,,~ . ,,f; 18.1 Q.. ....« 5 C 24.0 u 19 ;5~ I i 15 are in the same endocytic compartment Cells were incubated at noc for 30 min in the presence of l2SI-Tf. I3II-EGF was added for 10 min and the cells placed at O°C. Surface ligand was removed and cells were homogenized. Homogenates were applied to 12 % Percoll. centrifuged as described in Fig. 2, and radioactivity was detennined (A). Material from peak A was pooled and divided into aliquots. One aliquot was in· cubated with DAB in the presence of H202 and centrifuged in '11 % Percoll (B). The other aliquot was incubated with DAB in the absence of H2~ and centrifuged as in B (C). ~:: ~;\ ~~ .t Figure 5. Internalized Tf and EGF 12.3 ....>(/) ~ 6.5 Z IJJ at C C 24.0 e Control (. DAB) o 18.0 12.1 6.3 0.4 iii i i • 1 5 9 13 17 21 FRACTION NUMSER HRP-Tf (-DAB) ,1\ iJ' :~ 5 9 i 13 17 21 FRACTION NUMBER Figure 4. Density shift of endosomes. (A) Cells were incubated at TJOC with 12sl_Tf (3 x 10-9 M) in the presence or absence of TfHRP (4 x 10-9 M). Cells were placed at O°C, surface-bound ligand was removed, and cells were homogenized. Homogenates were applied to 12 % Percoll and centrifuged at 59,000 g.. for '11 min, the gradients were fractionated, and radioactivity detennined. (Solid circles) Control cells; (open circles) cells incubated with Tf-HRP. (B) Fractions from the more dense peak (peak A) were pooled and incubated with DAB and H20 2 as described in Materials and Methods. This sample was applied to '11 % Percoll and centrifuged as described above. The symbols represent samples obtained from control cells (closed circles) or cells incubated with Tf-HRP (open circles), (Solid triangles) Density, (C) Samples obtained as described above were treated in the absence of DAB, applied to '11% Percoll, centrifuged, and the distribution of radioactivity detennined. (Solid circles) Control vesicles; (open circles) Tf-HRP vesicles. gradients. The distribution of 1251 and 131 I was bimodal and essentially identical (Fig. 5 a). Each peak was reacted separately with DAB and H 20 2 , and applied to a Z7 % Percol1 gradient. The distributions of both Tf and EGF were affected by the DAB reaction, indicating that the two ligands were internalized into the same compartment (Fig. 5 b). A control experiment where H 2O:2 was left out of the reaction was also performed (Fig. 5 c). The peak of 131 1 is slightly more dense than the peak of 1231. It is possible that this material represents EGF in vesicles that have begun fusion with secondary lysosomes but have not separated totally from Tf. The time-dependent transition of internalized ligand from low density vesicles to vesicles with a higher buoyant density has been observed before (13). The same general protocol was used to define the intracellular localization of Jow density lipoprotein. In this experiment cells were incubated for 10 min at rrc with 1231_ LDL, 13II-EGF, and Tf-HRP. Surface-bound ligand was removed at O°C as previously described and the cells were homogenized. Endosomal fractions were isolated from 12 % Figure 6. Internalized EGF. 260 A 19.5 13,0 (l) '!l 6.5 2 Q. (,) ..J « ~ 26.0 I t- Iff 19,5 I~O to steady-state binding at J7 C with Tf-HRP and 12sI-Tf. 131 1EGF was then added to the incubation mixture for 10 min. Surface-bound ligand was removed at ooe, the cells were homogenized, and the homogenate applied to 12 % Percoll Q 65 2 4 6 8 FRACTION NUMBER 10 LDL, and Tf-HRP enter the same endocytic compartment. Cells were incubated for 30 min at noc in the presence of Tf-HRP (5 x 10-9 M). IlsI· LDL and '3II·EGF were added to the cultures for 10 min and the cells cooled to O°c. Surface ligand was removed and cells were homogenized. Samples were applied to 12 % Percoll and centrifuged at 59,000 g., for '11 min. Peak A was pooled and incubated with DAB in the absence of H20 2 (A), or the presence of H~O~ (B). The samples were applied to 35 % Percoll and cen· trifuged at 59,000 gO' for Z7 min. 45 Percoll gradients and reacted with DAB-H 20 2 • The radioactivity profiles of 35 % Percoll gradients indicate that low density lipoprotein was internalized into the same compartment as both EGF and Tf (Fig. 6). Slightly higher Percoll concentrations were used in this experiment to insure that we could differentiate density shifted material from that which might have accumulated in lysosomes. The use of two separate isotopes allowed us to further determine the specificity of the density shift procedure. Cells were incubated in the presence_of uIJ-EGF alone or 1251_Tf plus Tf-HRP. Vesicles isolated from each culture were mixed, reacted with DAB-H2~ and applied to Z7% Percoll gradients. Under these conditions some Tf-containing vesicles exhibited a shift in density while EGF-containing vesicles did not (Fig. 7 a). Cells incubated with !3IJ-EGF were mixed and homogenized with cells that had been incubated with '2~I-Tf and Tf-HRP. Peak A was isolated and subjected to the DAB reaction. Again, there was a change in the distribution of I2~I-Tf without a concomitant change in the radioactivity profile of l31I-EGF (Fig. 7 b). Normally >80% of the endosomal marker shifts after the DAB treatment. In these experiments only 1'\J50% of the label was found in the high density region of the gradient. The degree of density shift is somewhat variable. This experiment represents one of the lower degrees of shift. It is also possible that the manipulations involved inactivated some of the peroxidase. Finally, cells were incubated with 12~I-Tf for 10 min at J1°C, shifted to O°C, and incubated with Tf-HRP for 60 min. Vesicles obtained from homogenates of these cells showed no density shift when subjected to the DAB treatment (data not shown). These data indicate that only vesicles that contain Tf-HRP can be density shifted and neither the homogenization of cells nor the various reaction conditions induced vesicle fusion or a mixing of vesicular contents. Limitations oltlle Density Shift Procedure One of the original goals of this study was to use the density shifting technique to isolate and purify endosomal populaFigu~ 7. Endocytic compartments do not mix during preparation. (A) Cells were incu18 ('1'11-£6' ~ bated with I~I-Tf and Tf-HRP 12 / for 30 min at ~oC. An equival"\~ lent set of cells was incubated 6 'I \~ 2 with I3II-EGF for 10 min. ~, Q. (,) Surface ligand was removed at ..J O°C and cells were homoge<C nized. Homogenares were ap.B plied to 12 % Percoll and centrifuged as described in Fig. 2. Peak A from each gradient was pooled and the two pools mixed. This mixture was then incubated with DAB-H20z and centrifuged in Z1 % Per~'i~C I 5 9 13 17 21 colI. (Solid circles) IZ5I_Tf; FRACTION NUMBER (open circles) I3II-EGF. (B) Cells were prepared as described in A. Cells from the two cultures were combined and homogenized together. The homogenate was incubated with DABHzOz and applied to Z1% Percoll. The figure illustrates the radioactivity profiles from the gradient. (Solid circles) I~I-Tf; (open circles) 13II-EGF. 24 A tions. Density shifted endosomal fractions from Z7 % Percoll gradients were analyzed on SDS-polyacrylamide gels. The results from such gels were inconclusive because little or no protein could be detected by Coomassie Brilliant Blue staining. To determine whether the density-shifted fractions actually contained significant amounts of protein we assayed trichloroacetic acid precipitable radioactivity from cells that had been labeled with [lsS]methionine. We were able to detect 3-4% of the total acid precipitable 35S in the densityshifted region of the gradient. Most of the J~S-Iabel in these samples was found in a precipitate that formed when samples were boiled in preparation for electrophoresis. Although Percoll will precipitate under these conditions, we determined that protein was not being trapped by the Percoll precipitate by adding Inl-Tf to Percoll and boiling using the same conditions used for electrophoresis samples. We next analyzed the ability of detergent to solubilize '2'I-Tf from shifted and nonshifted vesicles. Table I illustrates the ability of Triton X-1OO to release internalized '2sI_Tf into a high speed supernatant after incubation of vesicles with DABH2~. Similar results were obtained if the vesicles were extracted with SDS (data not shown). The data suggest that the peroxidase-DAB reaction cross-links the luminal contents of the vesicles such that 12'1 -Tf cannot be detergent solubilized from the complex. To determine whether a shift in density could be obtained without cross·linking. the amount of DAB used in the reaction was titrated. Cells were incubated to steady state at J1°C in the presence of I2~I -Tf and Tf-HRP. Surface ligand was removed at O°C, cells homogenized, and the endosomal fractions pooled from 12 % Percoll gradients. Vesicles were reacted with concentrations of DAB ranging from 1.9 to 450 j.l.glml. The amount of radio labeled ligand shifted to the lower half of the gradient was compared with the amount of radioactivity that was insoluble in detergent (Table m. The effect for both density shifting and detergent solubility were concentration dependent. However, a concentration of DAB (17 IJ8Iml) that was still effective in producing a density shift caused a major proportion of the radiolabel to become de· tergent insoluble. Thus, the level of reaction sufficient to • [IUI)-Tf 0 0 ;~:I':~ Table I. The DAB Reaction Affects the Detergent Solubility of Membrane Vesicle Contents Sample Detergent-extractable m(·Tf % Control (-DAB) 70 Control (+DAB) 69.8 ±2 ±2 ±1 Tf-HRP (-DAB) Tf-HRP (+DAB) 68 14 ± 3 ControllTf-HRP mix 49 ± 2 Cells were incubated at 37°C with usI-Tf (1 x 10-8 M) for 40 min in the presence or ab!lence (control) of Tf-HRP (5 x 10-' M). The cells were placed at O°C, surface ligand was removed, and the cells homogenized. Vesicles were obtained by centrifulation over a step gradient of sucro!le (see Materials and Methods). DAB reactions were carried out using control and Tf-HRP vesicles alone, or with an equal mixture of control and Tf-HRP vesicles. The DAB reaction within vesicles was stopped by the addition of Triton X ·100 (final concentration 0.1 "). The sample was layered over a cushion of 35" sucrose in TE buffer and centrifuged for 60 min at 100,000 8... The distribution of ra· dioactivity was determined in the top layer and in the sucrose layer (including pellet). The data in the table represent the percentage oftotaJ radioactivity from each sample remaining in the top layer. 46 Table II. Titration of DAB Concentration DAB (l1g/ml final) Density shifted Insoluble 4SO ISO 66 ± 2.4 80 ± 2.9 65 80 17 ±3 ±3 ± 60 ± 5 51 81 ± 3 78 ± 4 10 30 70 o 1.9 5.5 ±5 ±2 16 37 ±7 ±9 12 ± 5 13 ± 4 Cells were incubated at 37°C in the presence of '2'I-Tf (2 x 10-'1 M) and Tf-HRP (5 x 10-9 M) until steady state binding was reaA:hed. Surface-bound ligand wo removed at O°C and cells were harvested and homogenized. The low speed supernatant was applied to a 12% Percoll gradient and cenuifulled a described in Materials aDd Methods. Endosome peaks A and B were isolated from fractionated gradients and incubated with the indicated concentrations of DAB. The DABtreated material wo divided in half and either cenuifulled on 27% Percollilradients or detergent extraeted with 1% SDS. The data in the table represent either the percent of tocal radioactivity shifted to the lower half of the 27% Percoll gradient or the percent of radioactivity that could be pelleted after detergent extraction. produce an increase in buoyant density also prevents the contents of the vesicle from being released by detergent extraction. Titrations of Tf-HRP yielded essentially the same result. i.e., concentrations of the conjugate in the media corresponding to receptor occupancies as low as 6% caused endosomal l2'I-Tf to become detergent insoluble after DAB treatment (1). Inactivation Occurs Only within Peroxidase-contDining Endosomes The fact that the cross-linking is confined to vesicles containing peroxidase was shown by mixing vesicles containing l2'I-Tf and Tf-HRP with control vesicles before the addition of DAB-H2~. The resultant detergent-soluble radioactivity is approximately equal to half of the sum of control and TfHRP values and indicates that only the contents of the peroxidase-containing vesicle become resistant to detergent solubilization (1ible I [Mix]). Furthermore, if Tf-HRP vesicles were lysed by detergent before DAB treatment, all of the radioactivity remained in the high-speed supernatant (see below). To determine whether other intracellular organelles were affected by the DAB-H2~ reaction, we assayed both Tf receptor activity and selected enzymatic activities (1ible ill). Tf receptor activity (measured in a detergent extract of cell membranes) was significantly diminished after treatment with DAB. This measure is probably an overestimate of endosomal receptor activity since there is some background binding due to receptors from plasma membrane. The other activity that diminished after this treatment was leucyl-pnaphtylamidase, a putative endosomal marker in this cell type (U). The lysosomal enzyme hexosaminidase was unaffected by DAB treatment. While these results suggest that the peroxidase-DAB inactivation is limited to endosomes, the possibility exists that naphtylamidase and the Tf receptor are sensitive to oxidation or the DAB reaction product. To test this possibility, vesicles were lysed before reaction with DAB-H2~. This procedure resulted in a negligible loss of Tf receptor activity. Although some naphtylamidase activity was lost by detergent lysis. lysis of Tf-HRP vesicles before DAB treatment resulted in little loss of activity. Thus, neither Tf receptors nor naphtylamidase are selectively sensitive to the DAB treatment, and the inactivation observed is confined to intact vesicles. This observation supports the suggestion that in HeLa cells leucyl-p-naphtylamidase is concentrated in the endosome. To test the hypothesis that the DAB reaction will inactivate luminal contents of any vesicle that contains peroxidase. we performed the following experiment. HeLa cells were incubated at 'J7°C in media containing 2 mglml HRP for U-1S h. Under these conditions, peroxidase should accumulate in lysosomes. Cells were washed extensively with cold PBS and incubated in media without HRP at 'J7°C for 60 min. Homogenates of cells were obtained and membrane vesicles collected. DAB reactions were performed either on intact vesicles or vesicles that had been lysed with detergent. After DAB treatment, detergent was added to all samples and hexosaminidase, as well as soluble Tf receptor activity, were assayed. The results of this experiment are shown in 1ible IV. Lysosomal enzyme activity was significantly diminished while endosomal Tf receptor activity was unaffected. These results confirm the hypothesis that the DAB reaction will abrogate activities associated with the luminal side of compartments containing peroxidase. Tf-HRP and EGF Are Sorted into Different Compartments The DAB procedure was used to determine the length of time that Tf and EGF are found in the same compartment. We have already shown that the two ligands are internalized into the same endocytic compartment. If the Tf-HRP conjugate truly behaves as Tf, we predict that the conjugate and EGF would eventually sort into different compartments. To test this prediction the following experiment was performed. Table III. Effect of the DAB Reaction on Selected Activities Activity Control Control (lysed) ~ ~ ~ ~ Tf receptor Hexosaminidase NaphtyJamidase 95 ± 3 96 ± 4 95 ± 5 100 92 ± I 53 ± 5 51 ± 2 98 ± I 9 ± 3 91 ± 5 99 ± 1 38 ± 3 Tf-HRP Tf-HRP (lysed) Cells were treated in a manner similar to those in Table 1. The enzyme activities of hexosaminidase and leucyl-l3-naphtylamidase were then assayed. The ability of detergent-solubilized transferrin receptors to bind ligand was also analyzed. To detennine if the peroxidase-DAB reaction itself would affect activities. vesicles were lysed with detergent (0.1 'l Triton X-loo) before the reaction (lysed). Addition of detergent did not inhibit the DAB reaction. 47 Table /Y. Localization of HRP in Lysosomes Affects Lysosomal, but Not Endosomal, Activities Activity Control Hexosaminidase Tf receptor 98 98 Control (lysed) HRP % % ±3 95 ± 1 30 ±2 97 % ±3 HRP (lysed) % ±3 74 ±5 95 ± 3 97 ± 2 Cells were incubated at 37°C with HRP (2 mglmJ) for 12-18 h. Cells were washed extensively with PBS at O°C and incubated for 60 min at 37°C in media without HRP. Cells were homogenized, vesicles prepared as described elsewhere, and the vesicles either lysed by the addition of Triton X-loo to a final concentration of 0.1 % or left intact. These-preparations were then reacted with DAB-H 20!. The DAB reaction within vesicles was stopped by adding Triton X-loo to all ~Ies 10 yield a final concentration of 0.2%. Hexosaminidase and soluble transferrin receptor activity were measured in these Iysates. Data are presented as the percent of maximum (control) activity. Cells were incubated at J7°C in the presence or absence of Tf-HRP until steady state binding was reached. Cultures were then pulsed with 12~I-EGF and 131I-Tf for 3 min, placed at O°C, and washed free of unbound ligand. Cultures were then returned to noc in the presence or absence of Tf-HRP. At various times cultures were placed at O°C, surface bound ligand removed, and homogenized. Endosome peak A was isolated from 12 % Percoll gradients and treated with DABH2O:!. Detergent extractable 12~I and 131 1 were then measured (Thble V). The data are presented as maximum detergentextractable radioactivity from control samples. The results indicate that within the 3-min pulse of radioligand, some separation ofTf and EGF had already occurred. This separation continued until a maximum was reached at rv20 min. However, the degree of lysosomal transfer of EGF was less than expected based on studies with fibroblasts (13). Recycling of 12~I-EGF 'WOUld result in a cellular distribution similar to that of I3II-Tf and Tf-HRP. This recycling would manifest itself by the inability to detergent extract 100% of the radiolabeled EGF. Independent experiments demonstrate that in HeLa cells a significant amount of EGF is capable of recycling (data not shown). Recycling of EGF has been observed in at least one other cell type (11). This approach can therefore be used not only to observe the separation of ligands, but also to demonstrate ligand recycling. Discussion Morphological approaches used to define the endocytic com- Table V. Tfand EGF Are Separated Soon after Internalization Maximum detergent-extractable radioactivity Time of chase (min) o 20 10 % % % 1'~l-EGF 67 ± I l'II_Tf 55 ± 2 79 53 ± ± 5 1 85 ± 2 57 ± 4 40 % % 90 ± 3 56 ± 1 90 ±3 58 ±2 Cells were. Incubated in the presence or absence of Tf·HRP (I x 10-- M) at 37°C for 20 min. Cultures were then pulsed with IHI·EGF (25 ng/ml) and 1'11_Tf (5 x 10' Ml for 3 min. and cells were washed free of unbound ligand wuh Ice cold PBS. Cultures were then returned to 37°C for the indicated times. Surface-bound ligand was removed at O°C, and cells were harvested and homogenized .•md the 800 R supernatant applied to 12% Percoll and cen· tTlfuged. Peak A wa~ removed from gradients and incubated with DAB-H~01' SDS wa~ added toO.2'7r. and radioactivity released to the high speed supernatant w a~ measured. partment have been restricted by the lack of suitable cytochemical markers and the absence of absolute landmarks. Morphological data are difficult to quantify and suffer from artifacts associated with fixation and embedding. Subcellular fractionation techniques have also been used to characterize the endocytic compartment. Although activities associated with various cellular organelles can be quantified. fractionation techniques cannot distinguish between activities that are truly confined to the same compartment from those that merely have similar physical properties. We have modified a technique developed by Courtoy et at (3) to specifically increase the buoyant density of the endosome. A conjugate of Tf-HRP was used to direct peroxidase activity to the endocytic compartment. Control experiments validated that the conjugate behaved as Tf. The peroxidasecatalyzed oxidation of DAB within intact vesicles resulted in a dense polymer that was restricted to those vesicles that contained peroxidase. We demonstrated that the reaction is limited to endosomes and that mixing of vesicular contents does not occur. Thus, only material that is in the same compartment with internalized Tf-HRP will exhibit an increase in buoyant density. Percoll gradients were used to fractionate cellular homogenates. As reported by other investigators. the endocytic compartment was found in the low density region of the gradients (7, 20). The endosome was defined by internalized 12~I-Tf and two distinct peaks of internalized ligand were found. The differences in centrifugation properties between the two peaks was not due solely to density differences, since different methods of loading gradients yielded different ratios between these peaks. Some separation on the gradients may be due to size and not density. We have not pursued this possibility further. We ,used the density shift technique to determine whether different receptor-ligand complexes were internalized into the same compartment. Experiments using cells that had been incubated simultaneously with '2sI-low density lipoprotein. I3II-EGF, and Tf-HRP revealed a concomitant increase in buoyant density for all three ligands after treatment with DAB. This result indicates that al1 three ligands are internalized into the same endocytic vesicle. The finding that internalized receptor-ligand complexes that have different eventual fates are internalized into the same compartment adds compelling evidence to the notion that the endosome is responsible for sorting these complexes (5, 13). The density shift technique can also be used to measure the rate at which different receptor-ligand complexes leave the endosome. Receptor-ligand complexes that become separated from the endocytic apparatus (e.g., those entering lysosomes) will no longer be affected by the DAB-H202 reaction. Although useful for demonstrating whether different internalized complexes are in the same compartment, the density shift technique suffers from a major limitation. Material that has been shifted cannot be easily analyzed. Our results suggest that oxidation of DAB within vesicles causes crosslinking of the luminal contents. The DAB molecule contains four reactive amino grQups. Under oxidizing conditions these groups would not be expected to limit their interactions solely to other DAB molecules. Thus, the DAB polymer is likely to include many molecules associated with the endosome. The inability to extract protein from peroxidaseDAB-treated vesicles, even after boiling in 1% SDS under 48 reducing conditions. supports the idea of cross-linking. Although the technique is limited by this fact, it remains useful for analysis of the contents of the endosomal compartment. For example, one can use a method of subtraction to analyze the membrane components in density shifted vs. nonshifted gradients. An alternative method for density shifting was recently reported (8). This technique utilizes the ability to place acetylcholinesterase activity into endocytic compartments. Reaction with the modified Karnovsky-Roots incubation medium results in a dense copper- and iron-containing precipitate which increases the buoyant density of the compartment. This chemical reaction appears not to cross-link luminal contents and thus may represent a feasible approach to purifying endosomal compartments. Perhaps the most powerful use of this technique is the ability to distinguish what proteins or activities are contained within the endosomal compartment by determining specific losses after DAB treatment. For example, the rate at which Tf and EGF become separated was measured by the ability to detergent extract '2sI-EGF after incubation with DAB. Activity for the enzyme leucyl-~-naphtylamidase was lost after DAB treatment, whereas the lysosomal enzyme hexosaminidase was unaltered in its activity. This result indicates that in this cell type naphtylamidase has a definite endosomal association, and the lysosome is not included in the Tf-endocytosis-recycling pathway. We have also used the peroxidase-DAB technique to inactivate endosomal contents and demonstrate that unoccupied Tf receptors are internalized in HeLa cells (1). These types of studies can be easily expanded to localize activities associated with other organelles. For example. when peroxidase was placed specifically in the lysosome. only lysosomal enzyme activity was reduced after reaction with DAB-H2~' These approaches can be combined to determine whether newly synthesized lysosomal enzymes are targeted to Iysosomes via an endocytic mechanism or if fluid phase uptake is dependent upon receptor-mediated endocytosis. We are currently carrying out experiments of this nature. The authors ~Id like to acknowledge Dr. 1. P. Kushner for advice in p~paring this manuscript. We would also like to thank Ms. Ina Jordan for her skilled technical assistance. This work was supported by grants HL269220S and HL2S92203 from the National Institutes of Health (NIH). R. S. Ajioka was supported in part by an NIH traineeship (ST32GMU7464-09). Received for publication S May 1986, and in ~vised form 22 September 1986. I. Ajioka. R. S .. and J. Kaplan. 1986. Intracellular pools of transferrin receptors ~sult from constitutive internalization of unoccupied receptors. Proc. Natl. Acad. Sci. USA. 83:6445-6449 2. Buys. S .• E. A. Keogh. andJ. Kaplan. 1984. Fusionofintracellularmem- brane pools with cell 'iurfaces of macrophagt!', ,umulated b:. phorholl!,tcr, Jnd calCIum IOnophores. Cell. 38569-57fl. 3. Courtoy. P. J .. 1. QUlntart. and P Baudhull1. 19114. Shift llf .:quillhnum density II1duced by 3.3' -diaminobenzu.line cylOcheml;,try' J. new prpc~dure tor the analysis Jnd purificallon of pero~ldase·containIng organelles. I Cell 8/111 98:870-876 4. Dautry-Varsat. A. A. Ciechanover. and H. Lodi"h. t'J83. pH and Ihe recycling of transferrin during receptor·medlated endocytosi,. PmI' Vall Acad. Sci. USA. 80:2258-2262. 5. Geuze. H. L J. w. Slot. G. 1. A. M. Strous. H. Lodlsh. and A. Sch""artz. 1983. Intracellular site of asialoglycoprotetn receptor-ligand uncoupltng: double-label immunoelectron microscopy during receptor·mediated endocytoSIs Ceil. 32:277-287. 6. Geuze. H. 1.. 1. W. Siol. G. J. A. M. Strous.l. Peppard. K. yon Figura. A. Hasihk. and A, Schwartz. 1984. Intracellular receptor sorting dunng en· docytosis: comparative immunoelectron microscopy of multiple receptors in rat liver. Cell. 37: 195-204. 7. Harford. J•• K. Bridges, G. Ashwell. and R. D. Klausner. 1983. Intracel· lular dissociation of receptor-bound asialoglycoproteins in cultured hepatocytes. J. Bioi. Chem. 258:3191-3197. 8. Helmy. S .. K. Porter-Iordan, E. A. Dawidowicz. P. Pilch. A. L. Schwartz. and R. E. Fine. 1986. Separation of endocytic from exocytic coated vesicles using a novel cholinesterase mediated density shift technique. Ceil. 44:497-506. 9. Karin. M .. and B. Mintz. 1981. Receptor-mediated endocytosis of trans· ferrin in developmentally totipotent mouse teratocarcinoma cells. J. Bioi. Chem. 256:3245-3252. 10. Klausner. R. D., G. Ashwell. J. van Renswoude. J. B. Harford. and K. R. Bridges. 1983. Binding of apotransferrin to K562 cells: explanation of the transferrin cycle. Proc. Natl. Acad. Sci. USA. 80:2263-2266. 11. Korc. M .. and B. E. Magun. 1985. Recycling ofepidermaJ growth factor in a human panc~atic carcinoma cell line. Proc. Natl. A cad. Sci. USA. 82:6172-6175. 12. lamb. J. E.• F. Ray, J. H. Ward. J. P. Kushner. and J. Kaplan. 1983. Internalization and subcellular localizaCion of transferrin and transferrm receptors in Heu cells. J. Bioi. Chem. 258:8751-8158. 13. Merion. M.• and W. Sly. 1983. 'The role of intermediate vesicles in the adsorptive endocytosis and transport of ligand to lysosome by human fibroblasts. J. Cell Bioi. 96:644-650. 14. Nakane. P .• and A. Kawaoi. 1974. Peroxidase-labeled antibody: a new method of conjugation. J. Histochem. CytocMm. 22: 1084-1091. 15. Octave, J. N .• J. Y. Schneider, R. R. Crichton. and A. Trouet. 1981. Transferrin uptake by cultured rat embryo fibroblasts. Eur. J. Biochem. 115: 611-618. 16. Pastan. L. and M. C. Willingham. 1983. Receptor-mediated endocytosis: coated pits. receptosomes and the Golgi. Trends Biochem. Sci. 8:250-254. 17. Roth. J.. and E. O. Berger. 1982. Immunocytochemical localization of galactosyltransferase in Heu cells: codislribulion with thiamine pyrophosphatase in trans-Golgi cisternae. J. Cell Bioi. 93:223-229. 18. Steinman. R. M .. and Z. A. Cohn. 1972. The interaction of soluble horseradish peroxidase with mouse peritoneal macrophages in vitro. J. Cell BioI. 55:186-193. 19. Tycko, B .• and F. R. Maxfield. 1982. Rapid acidification of endocytic vesicles containing az-macroglobulin. Cell. 28:643-651. 20. van Renswoude. J.• K. R. Bridges, J. B. Harford. and R. D. Klausner. 1982. Receptor-mediated endocytosis of transferrin and the uptake of Fe in KS62 cells: identification of a nonlysosomal acidic compartment. Proc. Natl. Acad. Sci. USA. 19:6186-6190. 21. Vincent. R.o and D. Nadeau. 1983. A micromethod for the quantitauon of cellular proteins in Percoll with the Coomassie Brilliant Blue dye-binding assay. Anal. Biochem. 135:355-362. 22. Deleted in press. 23. Deleted in press. 24. Ward. J. H., J. P. Kushner. and J. Kaplan. 1982. Regulation of HeLa cell transferrin receptors. J. Bioi. Chem. 257:10317-10323. 25. Wiley. H. S .. and D. D. Cunningham. 1982. The endocytic rate constant. J. Bioi. CMm. 257:4222-4229. 26. Willingham. M. C., J. A. Hanover. R. B. Dickson. and I. Pastan. 1984. Morphologic characterization of the pathway of transferrin endocytosis and recycling in human KB cells. PrOf. Natl. A cad. Sci. USA. 81: 175-179, CHAPTER III THE RECYCLING PATHWAY USED FOR TRANSFERRIN-:MEDIATED IRON DELIVERY IN HELA CELLS DOES NOT INCLUDE THE GOLGIAPPARATIJS 50 Summary HeLa cells incubated with a conjugate of transferrin-horseradish peroxidase accumulate peroxidase within the endocytic pathway. Incubation of peroxidase containing membrane vesicles with diaminobenzidine and H202 results in an increase in the buoyant density of the vesicles (11) and a "crosslinking" of their contents (12). The latter effect prevents the ability of detergent to solubilize luminal or membrane components. We utilized the effects of the transferrin-directed peroxidase/diaminobenzidine reaction to determine if the Golgi apparatus is part of the transferrin recycling pathway which delivers iron to the cell. Cells were incubated to steady state with transferrin-horseradish peroxidase and the effect of the diaminobenzidine/H20 2 procedure on endosomal and Golgiassociated activities was examined. Under conditions where endosomal properties were affected by diaminobenzidine treatment, three Golgi markers were unaffected. These markers were the activity of uridyl diphospho-galactosyltransferase, the buoyant density of the compartment containing 3H-fucose incorporated into acid insoluble material following a brief pulse, and the ability to detergent extract newly synthesized G-protein in cells infected with Vesicular Stomatitis Virus. These results indicate that the Golgi apparatus is not involved in the recycling of the transferrin receptor during the time frame of receptor mediated endocytosis. 51 Introduction Receptor mediated endocytosis involves the internalization of receptor/ligand complexes through coated pits. Once internalized, the complexes are found in a low density, acidic, nonlysosomal compartment termed the endosome (1,2). Within the endosome a sorting process occurs which ultimately leads to the routing of receptor/ligand complexes either to the lysosome or back to the cell surface. Neither the specific biochemical identities of the elements involved in the endocytic apparatus nor the number of compartments involved in the sorting process is known. Electron microscopic studies have suggested that internalized receptor/ligand complexes pass through a structure which appears to be part of the Golgi network (3-5). Van den Bosch et al. demonstrated that in the absence of protein synthesis significant levels of the asialoglycoprotein receptor could be detected within the Golgi stacks of hepatocytes. (6). The concentration was higher than predicted by measurements of receptor biosynthesis suggesting that these were in ternalized receptors. Additional evidence suggesting involvement of the Golgi network in the process of receptor/ligand recycling is the finding that plasma membrane Tf receptors which have been enzymatically deglycosylated at the cell surface may be reglycosylated within the cell (7). These investigators also showed that transferrin (Tf) receptors synthesized during a chemical block of Golgi mannosidase I activity returned to the compartment containing mannosidase I 52 following release of the block (8). Studies examining the subcellular distribution of surface labeled sialoglycoconjugates suggested that they were internalized and recycled through prelysosomal, lysosomal, Golgi, and Golgi- related compartments (9). It has also been demonstrated that internalized Tf can be found in Golgiderived exocytic coated vesicles (10). Taken together these studies provide compelling evidence that surface molecules may be recycled through the Golgi apparatus. However, the Tf cycle involved in iron accumulation reaches steady state within minutes, whereas the reglycosylation of surface transferrin receptors and the detection of transferrin in exocytic vesicles require hours. We have taken a different approach to determine if the Golgi apparatus is part of the receptor recycling pathway. A modification of the peroxidase/diaminobenzidine (DAB) density shift procedure developed by Courtoy et al. was used to affect the physical nature of the compartment(s) involved in receptor recycling (11). Tf was chemically coupled to horseradish peroxidase (HRP) which enabled us to place peroxidase activity specifically within the endocytic compartment. Incubation of peroxidase-containing membrane vesicles with DAB in the presence of H202 results in the polymerization of DAB within vesicles (11). Such polymers increase the buoyant density of the vesicle allowing it to be separated from other membrane vesicles by density gradient centrifugation. A secondary effect of the DAB polymerization is that luminal components of the vesicles become "crosslinked" or copolymerized with the DAB and cannot be solubilized with detergent. The results 53 of experiments using the peroxidase/DAB technique suggest that within the normal time frame of transferrin recycling, Golgi compartments are not involved. Materials and Methods HeLa cells were grown on plastic culture dishes in MEM containing 10% newborn calf serum (Flow Laboratories, McLean, VA), penicillin (200 U/ml), and streptomycin (0.2 mg/ml). Cells were maintained at 37 0 C in a 5% C02 atmosphere and were subcultured by trypsinization. For experiments, cells were washed with PBS and incubated with Hanks MEM buffered with HEPES (5 mg/ml), pH 7.2, and 10% newborn calf serum. Preparation of Li~ands Iron saturated Tf (27) was radioiodinated using IodoGen (Pierce Chemical Co., Rockford, IL) as described by Wiley and Cunningham (28). Tf was conjugated to HRP by the method of Nakane and Kawaoi (29) as described elsewhere (14). Bindin~ of 125I-Tf and Tf-HRP The binding of 125I-Tf or Tf-HRP to cells was performed as described elsewhere (30). Surface bound ligand was removed by washing cells at 00 C with citric acid phosphate buffer (pH 3.8) containing 150 mM NaCI and 1 mM EDTA for 3 min, followed by PBS 54 (pH 7.2) with 1 mM EDT A for 3 min. This cycle was repeated three times. Subcellular Fractionation All procedures were carried out at 0° C. Cell monolayers were removed with a rubber policeman and cell pellets resuspended in 0.25 M STE buffer (0.25 M sucrose, 10 mM Tris HCI pH 7.2, 1 mM EDTA). Cells were homogenized in a precooled tight-fitting Dounce homogenizer using 25-30 up and down strokes or until 80-90% of the cells were disrupted as monitored by phase-contrast microscopy. The 800 x g supernatant was applied to 12% Percoll (Pharmacia Fine Chemicals, Piscataway, NJ) and centrifuged for 27 min in a Ti75 rotor (Beckman Instruments Inc., Palo Alto, CA) at 59,000 x g. Gradients were fractionated by pumping from the bottom of the centrifuge tube. For some experiments membranes were collected by centrifuging the 800 x g supernatant over a step gradient of sucrose in TE buffer (10 mM Tris HCI pH 7.2, 1 mM EDTA) consisting of 13% (w/w) and 35% sucrose. Gradients were centrifuged for 60 min at 27,000 x g and membranes collected at the 13-35% sucrose interface. Diaminobenzidine Treatment and Density Shift The DAB treatment described by Courtoy et al. (11) was used with minor modifications as described elsewhere (12). DAB treated material was either centrifuged over 23% Percoll (27 min, 59,000 x 55 g), or detergent extracted and centrifuged over a cushion of 35% (w/w) sucrose (105,000 x g, 60 min). Soluble Receptor Assay The ability of detergent solubilized Tf receptors to bind 125 1-Tf was assayed according to the procedure of Lamb et al. (13). Virus Infections and Radiolabelin~ VSV (Indiana) was added to cells at 0 0 (multiplicity of infection=25: 1) and allowed to adsorb for 20 min. Media without serum was added and cultures were incubated at 37 0 for 90 min. Fetal calf serum was added to 6% and cultures were further incubated for 3.5-4.5 hours. Infected cells were labeled with met (Trans label, Amersham etc.) at 200 6% fetal calf serum. ~Ci/ml 35S- in media containing Virus was a generous gift from Dr. Richard Lloyd. Antibody Treatments and Electrophoresis Radiolabeled virus was incubated with rabbit anti-VSV antiserum followed by IgSorb (Enzyme Center). Pellets were washed and resuspended in electrophoresis loading buffer. VSV antisera was a gift from Dr. John Portis. Anti- Western blots were incubated with rabbit anti-Tf receptor antisera followed by 125 1_ Protein A. Anti-Tf receptor antisera was a gift from Dr. Paul Seligman. Protein was electrophoresed in 7.5% SDS polyacrylamide gels under reducing conditions. 56 3H-Fucose Labelinfi Cells were pulse labeled with 3H-Fucose (200 JlCi/ml) following the procedure of Atkinson (16). Enzyme Treatment and Analyses Galactosyltransferase (EC 2.3.1.38) was assayed as described by Lamb et al. (13), Samples were treated with endoglycosidase H (Sigma Chemical Co., St. Louis, MO) according to the procedure of Balch and Keller (31), Results Effect of the Density Shift Procedure on Galactosyltransferase Activity Internalized 125 1-Tf was resolved into two distinct peaks of radioactivity on 12% self-forming Percoll gradients (12). Analysis of enzyme activities across such gradients indicated that the Golgi enzyme galactosyltransferase cosedimented with the more dense peak of 125 1_ Tf (Peak A, Fig. 1). Similar results have been reported in subfractionation studies using sucrose gradients (13). These results suggest either that the two molecules are within the same compartment or that they are in separate compartments with similar centrifugation properties. To distinguish between these alternatives, the density shift procedure was used to alter the buoyant density of the Tf containing compartment (endosome). 57 A 25 II \ 20 15 10 5 o o 2 4 6 8 10 12 14 16 18 20 Fraction # Figure 1. Enzyme distribution on Percoll gradients The 800 x g supernatant from cellular homogenates was centrifuged on 12% Percoll as described in Experimental Procedures and three different activities compared. Endosomes were marked by internalized 125 1-Tf (circles). The compartment containing galactosyltransferase was assayed by the ability of different fractions to incorporate 14C-UDP galactose into acid insoluble material (closed triangles). The location of pulse-labeled 3H-fucose was determined by measuring TCA precipitable 3H in the various fractions (open triangles). The data were normalized to the total activity across the gradient. 58 We then determined whether galactosyltransferase, or other activities associated with the Golgi, were similarly affected. Cells were incubated at 37 0 C in the presence of Tf-HRP (5 x 10- 9 M) and trace amounts of 1251-Tf (1 x 10- 9 M) until steady state binding was achieved (30 min). The cells were placed at 00 C, surface bound ligand was removed, and the cells were harvested and homogenized. The lowspeed supernatant was applied to 12% Percoll and centrifuged as described in Experimental Procedures. Peak A, which contains both the higher density peak of 125 1-Tf activity and the peak of activity for galactosyltransferase, was incubated with DAB in the presence or absence of H202. After 30 min, the incubation mixture was applied to 23% Percoll gradients and centrifuged at 57,000 x g for 27 min. Fractions from the gradients were collected and assayed for radioactivity (Fig. 2A) and galactosyltransferase activity (Fig. 2B). Under conditions in which the buoyant density of 125 1_ Tf was markedly increased, the density of the Golgi marker was unaffected. In addition to increasing the buoyant density of membrane vesicles, the DAB reaction product copolymerizes or crosslinks the luminal contents of the vesicle (14). This renders activities associated with DAB treated vesicles insoluble in detergents. Under conditions where the endosomal marker (Tf receptor activity) was lost following treatment with DAB, galactosyltransferase activity was unaffected (data not shown). These results indicate that Tf-HRP and galactosyltransferase are in separate compartments. 59 A ~ j-\ _ C Control 20 p~ Tf-HRP .l!l 10 I \ _ ~ \ ~ a a ;Ei:~.:~:.,.'·"--." ~IJ. c_c·c_c-c a .o I B •.•.•.g=~:g i i 8 4 i i 16 20 12 C Control ~ 20 • .os: "';:::; Tf-HRP co 10 .- {:. ?f!. jr::J .1 __ ,[jrEj:C 0 Figure 2. tr\ a 0 « e.5J-~'C. C·C • • 0 'b_ .)g~ . ~.~ i i i 4 .'g i 8 12 Fraction # i 16 ... I 20 Density shift of galactosy ltransferase (A). Endosomes were marked by internalized 125I-Tf. Fractions containing the peak of galactosyltransferase activity were pooled from homogenates of cells incubated in the presence (closed squares) or absence (open squares) of Tf-HRP and centrifuged on 12% Percoll. This material was subjected to the DAB procedure and centrifuged on 23% Percoll. The data represent the percentage of total radioactivity measured across the gradient. (B). The ability to incorporate 14C-UDP galactose into TCA precipitable material was compared in equivalent 23% Percoll gradients. Control vesicles (open squares), vesicles from cells incubated in the presence of Tf-HRP (closed squares). The data are presented as the percentage of total enzymatic activity as measured by the incorporation of 14C-UDP galactose into acid precipitable material. 60 In these experiments all of the galactosyltransferase activity present in the initial cell homogenate was recovered following subfractionation. The possibility exists, however, that there is a Golgi galactosyltransferase activity in the recycling pathway which has been inactivated by our homogenization conditions or was not detectable using our substrate or reaction conditions. To exclude this possibility we assessed the effect of the DAB shift procedure on two other Golgi-specific markers. Density Shiftin~ of 3H-fucose Fucose is found as a terminal carbohydrate and is added to membrane or secretory proteins by a fucosyltransferase located in the trans Golgi region (15-18). Brief exposure of cells to 3H-fucose results in the incorporation of radioactivity into protein or lipid located in the most terminal portion of the Golgi. Following short periods of incubation with 3H-fucose, radioactivity would be expected to concentrate in the terminal portion of the Golgi apparatus and could serve as a marker for this region of the Golgi. To determine the utility of 3H-fucose as a marker of the terminal portion of the Golgi apparatus, cells which had been incubated for 15 min at 37 0 C with 3H-fucose were homogenized and the homogenate applied to a 12% Percoll gradient and centrifuged for 27 min at 57,000 x g. The majority of trichloroacetic acid precipitable radioactivity was found in the same density region as galactosyltransferase activity (Fig. 3A). Less than 10% of the total acid insoluble radioactivity was found in areas coincident with 61 plasma membrane markers (fractions 7-9). This result suggests that only a minor fraction of the incorporated label had reached the plasma membrane during the 15 min incubations. In order to demonstrate that the labeled components were eventually transferred to the plasma membrane, cells were incubated with the labeled precursor for 15 min, and then in its absence for 120 min. Acid precipitable radioactivity was now found in the regions of the gradient corresponding to Golgi (fractions 2-4) and plasma membrane (Fractions 7-9, Fig. 3A). These results are consistent with the results of published studies showing a vectoral transfer of fucose-containing material from Golgi to plasma membrane. We next determined whether the compartment which contained radioactive fucose after the short exposure also contained internalized Tf. Cells were incubated at 37 0 C in the presence of Tf-HRP (5 x 10-9 M) until steady state binding was achieved. 3H- fucose (45 flCi/ml) was added and cells were incubated for an additional 15 min. In a parallel set of experiments, cells were incubated with Tf-HRP and 125I-Tf (5 x 10-9 M) in order to "label" endosomes. Surface ligand was removed at 0 0 C, cells were harvested, homogenized, and the lowspeed supernatant applied to 12% Percoll gradients. Fractions corresponding to endosome Peak A were collected from both gradients and samples were incubated with DAB in the presence or absence of H202 and then applied to 23% Percoll gradients. Gradients were fractionated and the distribution of radioactivity determined. Under conditions where almost all of the 125I-Tf had been shifted to the dense region of the 62 Figure 3. Density shift of incorporated 3H-fucose (A). 800 x g supernatants from cells pulsed for 15 min with 3Hfucose (open squares) or pulsed for 15 min and chased for 120 min (closed squares) were applied to 12% Perc 011 , and centrifuged as des~ribed in Experimental Procedures. The data represent the percentage of total acid insoluble radioactivity from fractionated gradients. (B). Endosomes were marked by internalized 125I-Tf. Fractions containing the peak of acid precipitable 3H-fucose were treated as in Figure 2A. Control vesicles (open squares), Tf-HRP vesicles (closed squares). The data represent the percentage of total radioactivity measured across each gradient. (C). Following DAB treatment, vesicles were centrifuged on 23% Percoll as in Figure 2B and TCA precipitable 3H was measured in each fraction. Control vesicles (open squares), Tf-HRP vesicles (closed squares). The data represent the percentage of total acid insoluble radioactivity from fractionated gradients. 63 ~ A 15 0 CL 15 min pulse, 120 min chase (J) ::0 :::s "0 (J) 10 [] r::: • "C ·0 <C 0 15 min pulse /\\ •., () ., •.• O~ ~ol °li ... \ 5 tU ...... 0 / '. .r· [] J~ 0 0 0 0 "[] ,0\ • ~ o Control • Tf-HRP 0 10 . • ~ 0 0 30 () (J) :::s (5 ,/ \ 20 16 20 Fraction # 0 Control • Tf-HRP [] / ....'f 0 I r::: I 12 .~a "C 10 I· [] tU ...... 0 • l- ~ 0 \ 12 (J) "0 <C 20 [] [] 8 4 C CL ::0 . 16 "" / 0 ••-.......... ' .[J'O 0_0 .0 ~.O'O·O:O .•.•,. , .•• 0 o. '0I .. 0-0i '.' 0 ~ I 0 iI 0'0 CL J- ,0'0'0 Fraction # () tU ...... 0 O.[J·O r I 12 8 B 20 \ / [J • 4 • 0 •.•/ ~[]-11- _ O·O •[]:r;J:[]:~1 • • -"0•0j • 0" i 0 4 8 12 Fraction # o ,0.0 o,g·.·· 16 20 64 gradient (Fig. 3B), the majority of acid precipitable 3H-fucose remained in the low density region of the gradient (Fig. 3C). trace radioactivity was detected at the higher density. Only This trace radioactivity may represent proteins which had reached the plasma membrane during the 15 min incubation period. The fact that most of the radiolabeled material did not change density while the endosomal marker did, again suggests that the compartment containing fucosyltransferase does not contain internalized Tf. Analysis of Newly Synthesized Protein It is possible that the Golgi markers analyzed may be contained in portions of the Golgi complex physically distinct from a part which might recycle Tf. To examine this possibility we used a more general marker for the Golgi complex, namely newly synthesized protein. No single protein is produced in high enough levels to mark Golgi function but when cells are infected with vesicular stomatitis virus (VSV), the viral G protein is glycosylated by processing through the Golgi (19). Studies have taken advantage of the presence of the VSV G protein in the Golgi to define the sequence of glycoprotein maturation and to develop in vitro assays of Golgi function (20). Another advantage of studying VSV infected cells is that late in infection, host protein synthesis is inhibited, and the only proteins synthesized are of viral origin. Infected cells are still capable of endocytosis but at a slightly reduced level (data not shown). 65 To determine if internalized Tf-HRP could be found in compartments containing newly synthesized VSV proteins, cells were infected with VSV at a multiplicity of 25: 1 and incubated at At 6.5 hours postinfection Tf- HRP (5 x 10- 9 M) was added to 37 0 C. half of the cultures. After a further 20 min incubation at 37 0 C the cells were incubated with 35S-met (with or without Tf- HRP ) for 10 min. The cells were washed and incubated in 35 S-met free media in the presence and absence of Tf-HRP (5 x 10- 9 M) for an additional 25 min. The cells were then placed on ice, surface ligand removed (Experimental Procedures), and the cells were then homogenized. The lowspeed supernatant was applied to a 12% Percoll gradient and centrifuged as described in Experimental Procedures. Endosome Peak A (as marked by a parallel culture labeled with 125 1_ Tf) which contains the peak of galactosyltransferase activity wa.s pooled from the fractionated gradient. This material was incubated in the presence or absence of DAB/H20 2 and extracted in 0.2% Triton X100. The detergent treated sample was underlayered with 35% sucrose and centrifuged for 60 min at 105,000 x g. The solubilized material (top) was then incubated with rabbit antiserum directed against VSV proteins. Immune complexes were precipitated with IgSorb, solubilized with sample buffer and electrophoresed on 10% SDS polyacrylamide gels. Autoradiographic analysis of the gels revealed that there was no significant loss of G protein in the detergent solubilized material regardless of whether the samples had been exposed to DAB/H202 or not (Fig. 4A). In order to determine that the DAB reaction had in fact been effective in 66 Figure 4. Analysis of DAB treated material (A). VSV infected cells were incubated in the presence or absence of Tf-HRP for 30 min, pulsed with 35S-met for 10 nlin, and chased in the presence or absence of Tf-HRP for 25 min. Endosome Peak A was pooled from 12% Percoll gradients, treated with DAB, detergent extracted, and the extract incubated with antiserum to VSV. Immune complexes were precipitated using IgSorb and the precipitated protein electrophoresed on a 7.5 % SDS polyacrylamide gel which was prepared for autoradiography. The graph represents densitometric tracings of the radioactive band corresponding to VSV G protein. Error bars represent standard deviation from four different gels. Values were normalized to the most dense band. (B). Membrane was collected from cells incubated in the presence or absence of Tf-HRP by centrifuging homogenates over a step gradient of sucrose (see Experimental Procedures). Vesicles were treated with DAB and extracted with 0.2% Triton X-IOO. Extracts were electrophoresed in 7.5% SDS polyacrylamide gels and transferred to nitrocellulose paper by electroblotting. The nitrocellulose paper was probed first with rabbit anti-transferrin receptor antisera followed by 125I-protein A. Values were normalized to the control and errors represent standard deviation (n=2). "'.""" .......... "'~\'"~. .,~" ~ i.'t:::: '; J,' " - f,' • 100 ~ ,,' i""~~ . "' .. '\ +'_ 'f.... "','_l A 80 S :J S .,-j GO X rtl ~ 40 20 o +DAB -DAB +DAB Control -DAB Tf-HRP B +DAB -DAB Control +DAB -DAB Tf-HRP 67 68 crosslinking the endosome, Tf receptors were also measured In the detergent extracted material using a soluble receptor assay. A major reduction in the ability to bind 125I-Tf was found in samples exposed to DAB/H202 as compared to control samples (no DAB reaction, Table 1). The reduction in Tf receptors detected by the soluble binding assay was confirmed by a different protocol using Western blots. Cells infected with VSV for 6.5 hours were exposed to TF-HRP for 30 min. The cells were homogenized and the supernatant from a low speed centrifugation was applied to a sucrose step gradient. Membranes collected from the 13-35% sucrose interface were incubated in the presence or absence of the complete DAB reaction mixture. Triton X-IOO was added to 0.2% and the sample centrifuged over a cushion of 35% sucrose as described in Experimental Procedures. The detergent extracted material was electrophoresed on 10% SDS polyacrylamide gels and electroblotted onto nitrocellulose paper. The paper was probed with rabbit anti-Tf receptor antiserum followed by 125 1-protein A. In samples exposed to the complete reaction mix there was a marked decrease in Tf receptors as detected by the ability to bind Tf receptor with antibody (Fig. 4B). Thus using two different assays we could demonstrate an alteration in the content of the endosomes containing Tf-HRP yet no alteration in newly synthesized G-protein, suggesting that these molecules are in different compartments. To demonstrate that the 25 min chase period was long enough to allow transfer of labeled G protein from the endoplasmic reticulum 69 Table 1. The DAB Reaction Inhibits the Ability to Detergent Extract Functional Tf Receptors. -DAB Control 98.5 +/- 0.87 Tf-HRP 88 +/- 12 +DAB 97.43 +/- 1.02 41.8 +/- 2.5 VSV infected cells were incubated in the presence or absence of TfHRP for 30 min. Endosome Peak A was isolated from 12% Percoll gradients, treated with DAB, and extracted with detergent. This material was assayed for the ability of the solubilized transferrin receptor to bind 125I-Tf. Values were normalized to maximum binding (n=3 ). 70 to the Golgi apparatus (Le., during the period in which cells were incubated with Tf-HRP, G protein would reach the Golgi) we analyzed the sensitivity of G protein to endoglycosidase H (Endo H). This enzyme is specific for high mannose carbohydrate and will not cleave glycoproteins once they have been modified in the Golgi apparatus (20). Within 20 min after pulse labeling, most G protein was found to be endo H insensitive, and by 30 min, all of it appeared to be insensitive (Fig. 5). The 25 min chase period was thus sufficient to "chase" the newly synthesized G protein out of the endoplasmic reticulum and into the Golgi. Discussion Previous morphologic and biochemical studies have strongly suggested that recycling of plasma membrane components involves the Golgi apparatus. A parameter common to all of these studies IS the time required to demonstrate the phenomena described. Measurements of the T 1/2 for reglycosylation of neuraminidase treated membrane proteins is between 2-3 hours. The half time measured for the return of the transferrin receptor to mannosidasecontaining compartments was greater than 6 hours (8). However, the complete endocytic cycle which affects iron uptake in the case of Tf, or ligand accumulation in the case of other receptors, is complete within 10-20 min or less (22,23). This discrepancy, which was pointed out by Snider and Rogers (8), led us to examine whether the endocytic pathway involves the Golgi apparatus. 71 Figure 5. Endo H treatment of newly synthesized VSV proteins VSV infected cells were pulsed for 10 min with 35 S met and the label was chased for various periods of time. Monolayers were scraped, cells pelleted, and lysed in 0.1 % Triton X-lOO. Lysates were incubated with rabbit anti-VSV antisera followed by IgSorb. Pellets were solubilized and prepared for treatment with Endo H as described in Experimental Procedures. Samples were electrophoresed in 7.5% SDS polyacrylamide gels and prepared for autoradiography. The figure shows the area of the gel corresponding to VSV G protein. 72 o M o N + o r-I + o + -~ ·rl 8 ::c (il 0 U~; rr< ,', ,.".., ,,I r '" c ~ ~ ~ ~~-,' 73 We have analyzed three Golgi-associated activities in an attempt to determine if the rapid recycling pathway involved in ligand uptake includes compartments of the Golgi. Using a conjugate of Tf- HRP we were able to target peroxidase activity to the endocytic/recycling pathway. Treatment of vesicles obtained from cells incubated in the presence of Tf-HRP with DAB and H202 resulted in an increase in buoyant density of the peroxidase containing compartnlent and an inability to detergent extract endosomal associated activities (12,14). We were thus able to determine, in a quantitative manner, whether biochemical markers of the Golgi apparatus were affected by this reaction. Our studies, using three different markers for the Golgi apparatus, do not provide evidence that Tf passes through Golgi compartments. The morphological and biochemic~l data of others, suggesting that endocytic markers traverse the Golgi stacks, can be reconciled with our results. The essential difference between the two can be explained by the time course of the experiments. In our experiments we used a time scale appropriate only for the rapid recycling pathway (i.e., a time scale commensurate with the Tf cycle). The longer term experiments of others suggest the existence of two separable pathways for receptor recycling with different kinetics. A number of other studies provide evidence supporting the existence of two pathways for receptor recycling. Oka and Weigel demonstrated two kinetically different compartments for the dissociation of internalized receptor/ligand complexes using the 74 asialoglycoprotein receptor (24). Morphological studies using conjugates of Tf-HRP have been used to describe a peripheral "short circuit n and a juxtanuclear, longer circuit for the intracellular routing of Tf (25). Stein and Sussman have described a monensin sensitive and monensin resistant pathway for the recycling of the transferrin receptor (26). These authors felt that the monensin sensitive pathway might represent the recycling of transferrin receptors through the Golgi. Fishman and Fine demonstrated that between 0 and 40 min of incubation, transferrin could not be found in coated vesicles isolated from the biosynthetic pathway in perfused rat liver cells (10). When the incubation time was extended to 1-2 hours, however, transferrin could be detected in coated vesicles of the biosynthetic system suggesting that the ligand could recycle via both short and long pathways. Why should two pathways for recycling of plasma membrane proteins exist? functions. We speculate that the two pathways serve different One pathway is selective and the other nonselective. The selective pathway is used for rapid recycling of internalized surface proteins and may be used for ligand accumulation and receptor mediated endocytosis. The nonselective pathway may involve all, or most, elements of the plasma membrane and is involved with compartments of the Golgi. The nonselective pathway may be involved in urepair" or degradation of surface proteins (Le., a "proof reading" mechanism). This hypothesis is consistent with the re- glycosylation of surface proteins in Golgi stacks. Another function of this nonselective pathway may be the maintenance of cell surface 75 area. Membrane added to cells as a result of secretory activity must be internalized and this may occur by this longer circuit. Studies in which Tf has been demonstrated in the Golgi stacks by morphological approaches have employed secretory cell types (e.g., hepatic parenchymal cells or myeloma cells) which would be expected to exhibit a high rate of compensatory or fluid phase pinocytosis. In our study we used a cell type that exhibits an extremely low rate of fluid phase pinocytosis. This difference may explain why HeLa cells incubated in the presence of Tf-HRP for 12 hours demonstrated no decrease in galactosyltransferase activity after treatment with DAB (unpublished results). While these studies support the view that two different pathways for recycling exist, our results do not define whether the two pathways are completely separable. There may be a common internalization step but the two pathways separate within the endocytic apparatus. endocytic events. Alternatively there may be separable Experiments are in progress to distinguish between these two possibilities. 76 References 1. van Renswoude, J., Bridges, K.R., Harford, J.B., and R.D. Klausner. 1982. Receptor-mediated endocytosis of transferrin and the uptake of Fe in K562 cells: identification of a nonlysosomal acidic compartment. Proc. Nati. Acad. Sci. USA 79: 6186 2. Tycko, B., and F.R. Maxfield. 1982. Rapid acidification of endocytic vesicles containing alpha 2-macroglobulin. Cell 28: 643 3. Yamashiro, D.J., Tycko, B., Fluss, S.R., and F.R. Maxfield. 1984. Segregation of transferrin to a mildly acidic (pH 6.5) paraGoigi compartment in the recycling pathway. Cell 37: 789 4. Willingham, M.C., Hanover, J .A., Dickson, R.B., and I. Pastan. 1984. Morphologic characterization of the pathway of transferrin endocytosis and recycling in human KB cells. Proc. Nati. Acad. Sci. USA 81: 175 5. Woods, J.W., Doriaux, M., and M.G. Farquhar. 1986. Transferrin receptors recycle to cis and middle as well as trans cisternae in Ig-secreting myeloma cells. J. Cell BioI. 103: 277 6. van den Bosch, R.A., Geuze, HJ., and GJ.A.M. Strouse 1986. Presence of asialoglycoprotein receptors in the Golgi complex in the absence of protein synthesis. Expti. Cell Res. 162: 231 7. Snider, M.C., and O.C. Rogers. 1985. Intracellular movement of cell surface receptors after endocytosis: resialation of asialotransferrin receptor in human erythroleukemia cells. J. Cell BioI. 100: 826 8. Snider, M.C., and O.C. Rogers. 1986. Membrane traffic in animal cells: cellular glycoproteins return to the site of Golgi mannosidase I. J. Cell BioI. 103: 265 9. Fishman, J.B., and J.S. Cook. 1986. The sequential transfer of internalized, cell surface sialoglycoconjugates through the 77 lysosomes and Golgi complex in HeLa cells. J. BioI. Chern. 261: 11896 10. Fishman, J.B., and R.E. Fine. 1987. A trans Golgi-derived exocytic coated vesicle can contain both newly synthesized cholinesterase and internalized transferrin. Cell 48: 157 11. Courtoy, P.J., Quintart, J., and P. Baudhuin. 1984. Shift of equilibrium density induced by 3,3'-diaminobenzidine cytochemistry: a new procedure for the analysis and purification of peroxidase-containing organelles. J. Cell BioI. 98: 870 12. Ajioka, R.S., and J. Kaplan. 1987. Characterization of endocytic compartments using the horseradish peroxidase/ diaminobenzidine density shift technique. J. Cell BioI. 104: 77 13. Lamb, J.E., Ray, F., Ward, J.H., Kushner, J.P., and J. Kaplan. 1983. Internalization and subcellular localization of transferrin and transferrin receptors in HeLa cells. J. BioI. Chern. 258: 8751 14. Ajioka, R.S., and J. Kaplan. 1986. Intracellular pools of transferrin receptors result from constitutive internalization of unoccupied receptors. Proc. NatI. Acad. Sci. USA 83: 6445 15. Atkinson, P.H., and D.F. Summers. 1971. Purification and properties of HeLa cell plasma membranes. J. BioI. Chern. 246: 5162 16. Atkinson, P.H. 1975. Synthesis and assembly of HeLa cell plasma membrane glycoproteins and proteins. J. BioI. Chern. 250: 2123 17. Haddad, A., Smith, M.D., Herscovics, A., Nadler, N.J., and C.P. Leblond. 1971. Radioautographic study of in vivo and in vitro incorporation of fucose- 3 H into thyroglobulin by rat thyroid follicular cells. J. Cell BioI. 49: 856 18. Goldberg, D.E., and S. Kornfeld. 1983. Evidence for extensive subcellular organization of asparagine-linked oligosaccharide 78 processing and lysosomal enzyme processing. J. BioI. Chern. 258: 3159 19. Zilberstein, H., Snider, M.D., and H.F. Lodish. 1981. Synthesis and assembly of the vesicular stomatitis virus glycoprotein. Cold Spring Harbor Symposia on Quantitative Biology 46: 785 20. Balch, W.E:, Dunphy, W.G., Braell, W.A., and J.E. Rothman. 1984. Reconstitution of the transport of protein between successive compartments of the Golgi measured by the coupled incorporation of N-acetylglucoseamine. Cell 39: 405. 21. Robbins, P.W., Hubbard, S.C., Turco, S.J., and D.F. Wirth. 1977. Proposal for a common oligosaccharide intermediate in the synthesis of membrane glycoproteins. Cell 12: 893 22. Bleil, J.D., and M.S. Bretscher. 1982. Transferrin receptor and its recycling in HeLa cells. EMBO J. 1: 351 23. Goldstein, J.L., Anderson, R.G.W., and M.S. Brown. 1979. Coated pits, coated vesicles, and receptor-mediated endocytosis. Nature 279: 679 24. Oka, J.A., and P.H. Weigel. 1983. Recycling of the asialoglycoprotein receptor in isolated rat hepatocytes. J. BioI. Chern. 258: 10253 25. Hopkins, C.R. 1983. Intracellular routing of transferrin and transferrin receptors in epidermoid carcinoma A431 cells. Cell 35: 321 26. Stein, B.S., and H.H. Sussman. 1986. Demonstration of two distinct transferrin receptor recycling pathways and transferrin-independent receptor internalization in K562 cells. J. BioI. Chern. 261: 10319 27. Ward, J.H., Kushner, J.P., and J. Kaplan. 1982. Regulation of HeLa cell transferrin receptors. J. BioI. Chern. 257: 10317 28. Wiley, H.S., and D.D. Cunningham. 1982. The endocytic rate constant. J. BioI. Chern. 257: 4222 29. 79 Nakane, P., and A. Kawaoi. 1974. Peroxidase-labeled antibody: a new method of conjugation. J. Histochem. Cytochem. 22: 1084 30. Buys, S.S., Keogh, E.A., and J. Kaplan. 1984. Fusion of intracellular membrane pools with cell surfaces stimulated by phorbol esters and calcium ionophores. Cell 38: 569 31. Balch, W.E., and D.S. Keller. 1986. ATP-coupled transport of vesicular stomatitis virus G protein. J. BioI. Chern. 261: 14690 CHAPTER IV INTRACELLULAR POOLS OF TRANSFERRIN RECEPTORS RESULT FROM CONSTITUTIVE INTERNALIZATION OF UNOCCUPIED RECEPTORS 81 Intracellular pools of transferrin receptors result from constitutive internalization of unoccupied receptors (endosome / peroxjdase) RICHARD S. AnOKA AND JERRY KAPLAN Department of Pathology, University of Utah. College of Medicine. Salt Lake City. UT 84132 Communicated by Stanley Cohen, May 12, 1986 ABSTRACT In HeLa cells the majority of transferrin {TO receptors are found within the endocytic apparatus, with only 20 % of receptors exposed at the cell surface. Receptor distribution is unaltered by the presence or absence of Tf. The mechanism responsible for the cellular distribution of receptors was explored by selectively inactivating receptors within the endocytic apparatus. This was accomplished by employing Tf-horseradisb peroxidase conjugates. Peroxidase-catalyzed oxidation of diaminobenzidine within an endosome destroys Tf receptor activity. Using such conj ugates, we bave demonstrated tbat the majority of internal Tf receptors could be inacti· vated when <6.0% of receptors were occupied by the conjugate at steady state. This result indicates that occupied and unoccupied receptors are in the same compartment. Furthermore, horseradisb peroxidase that was internalized by fluid-phase plDoeytosis inactivated intraceUuiar Tf receptors in the absence of Tf; this lDdieates that the presence of internal receptors is Upad independent. Following exposure of ceUs to the conjugate, receptor inactivation was proportional to the percentage of the endocytic cycle traversed hy the conjugate-tbat is, the rate of Ugand accumulation was the same as the rate of endosomal Tf receptor inaetiva&ion. Wben the Tf-horseradisb peroxidase conjugate and lUI_labeled Tf were internalized simultaneously, both Ugands were found in the same compartment. However, if the two ligands were administered as separate pulses and the period between pulses was as short as 1 min, the ligands remained separate within the cell. TOlether these results demonstrate that the intraceUular pool of Tf receptors reflects the constitutive internalization of unoccupied Tf receptors, which. once internalized, remain segregated. tant because it may account for the cellular distribution of receptors. For many receptors involved in ligand accumulation, the bulk of the cellular content of receptors is intracellular. For example, in HeLa cells only 20% of the transferrin r:eceptors (TfR) are on the cell surface, with 60% residing in the endosome (4) (the remaining 20% are presumably within the biosynthetic pathway). Similar observations have been made for insulin receptors in 3T3 cells, asialoglycoprotein receptors in hepatocytes, and mannose terminal glycoprotein receptors in macrophages (5-7). The presence of this pool of internal receptors may be explained by two general models for receptor internalization (Fig. 1). The pool of internal receptors may represent receptors in membrane vesicles that are in transit through the cell (model I)-that is. unoccupied receptors are continually being internalized-or the endosome may exist as a unique subcellular organelle that acts as a central pool of TfR (model II). We have taken advantage of the observation that horseradish peroxidase (HRP)-catalyzed polymerization of diaminobenzidine (DAB) within an endosome effectively crosslinks and inactivates luminal components, including unoccupied TfR. Using this technique. we present studies designed to discriminate between the two models. Our data indicate that intracellular unoccupied TfR exist within the Tf recycling pathway and are internalized in the absence of ligand. Furthermore. the endosome does not appear to be a steady-state organelle in the sense that it continually fuses and buds endocytic and exocytic vesicles. The data are consistent with the endosome being a transient organelle that remains physically and temporally segregated from other endosomes and internal unoccupied TfR are a result of continual internalization. Our understanding of membrane and receptor recycling has increased greatly in the past few years. Of panicular importance was the discovery that receptor-ligand complexes are internalized into an acidic nonlysosomal companment referred to as the endosome. Within this compartment. receptor-ligand complexes may dissociate, allowing for different eventual fates-that is, some molecules are transferred to lysosomes while others are recycled to the cell surface. Some morphological and biochemical features of the endosome have been defined. but imponant questions concerning its physiology and relationship to receptor internalization remain unanswered. Although it is clear that ligand-receptor complexes are internalized, it is uncenain whether unoccupied receptors are internalized. Many of the techniques employed to determine whether unoccupied receptors are internalized. such as antibodies. cell-surface iodination. and trypsin treatment. may themselves penurb membrane physiology, making the results of such experiments difficult to interpret 11-3}. An understanding of unoccupied receptor movement is impor- Cell Cultures. HeLa cells were grown on plastic tissue culture dishes in Eagle's modified minimal essential medium (MEM) containing 10% newborn calf serum (Flow Laboratories), penicillin (200 units/ml), and streptomycin (0.2 mg/ml), at 37°C in a 5% CO 2 in air atmosphere. Cells were subcultured by trypsinization. Binding experiments were carried out in room air using Hanks' MEM buffered with 10 mM Hepes (pH 7.2). This medium was supplemented with 10% newborn calf serum and 4 mg of bovine serum albumin per ml. Preparation of !lSI-Labeled Tf{Feh (I%SI_ TO. Human Tf <Calbiochem) was saturated with iron (8) and radioiodinated using Iodo-Gen (Pierce) as described by Wiley and Cunningham (9). Conjugation of HRP to Te. Tf-HRP conjugates were prepared according to the method of Nakane and Kawaoi (10), except NaCNBH3 was substituted for NaBH4 in the reduction step. The final conjugate contained HRP and Tf at a The publication costs of thi" article ..-ere defrayed in part by page charge payment. ThiS article must therefore be hereby marked "advertisement" In accordance with III L' .S.c. ~1734 snlel~ to Indicate thi~ fact. diarninobenzidine. MATERIALS AND METHODS Abbreviations: Tf. difernc transfernn; TfR. transferrin receptorfsl; l:~I·Tf. 1:'I·labeled TfIFe):: HRP. horseradish peroxidase; DAB. 82 Model 11 FIG. 1. Models of the relationship between intracellular and surface TiR during endocytosis. Modell represents a possible model of the endocytic process whereby endosomal contents arc processed throuJh the cell in a temporally segregated manner. Model II illustrates how internalized receptor-ligand complexes might be segregated in a central, common organelle. molar ratio of 1:1.5. It should be noted that this ratio is an average since multiple molecular weight species are produced by the coqjugation procedure. The concentration ofTf in the conjugate was determined by its ability to compete with 1251-Tf for receptor occupancy. Similar results were obtained if the concentration ofTfin the conjugate was determined by equilibrium binding at O°C or under steady-state binding conditions at 37c C. The shape of the dilution curve for the conjugate was similar to that of Tf(Feh, indicating that the affinity of Tf for the receptor was not significantly altered by the conjugate procedure. Receptor occupancy was calculated by the formula: occupancy = ligand/(ligand + Kd) (11), where the Kd = 2 X 10- 8 M (8). Furthermore, the behavior of the conjugate was indistinguishable from Tf (data not shown). This finding is consistent with those of others who have used conjugates of Tf-HRP (12, 13). The binding to cells of 1251_Tf or the Tf-HRP conjugate was performed as described elsewhere (14). Surface-bound ligand was removed by washing cells at OCC with citric acid/phosphate buffer (pH 3.8) containing 5 mM EDTA and ISO mM NaCI for 3 min, followed by washing with phosphate-buffered saline (PBS) for 3 min, and the cycle was repeated twice. Subcellular Fractionation. Cells were removed from the monolayer by use of a rubber policeman and homogenized as described by Lamb et al. (4). Homogenates were centrifuged at 800 x g for 10 min, and the supernatant was applied to 15% Percoll (Pharmacia). Isoosmotic Percoll was prepared and diluted as per manufacturer's recommendations. Endosomes were typically collected from IS%; Percoll gradients centrifuged in a Beckman Ti 75 rotor centrifuged at 59,000 x g (average) for 40 min. Fractions containing endosomes were pooled (the endosomal fraction was identified by assaying gradients for membrane-bound 1~51_Tf). Endosomal fractions were incubated with DAB and H20 z using the procedure of Courtoy et al. (15) with only minor modifications. DAB solutions were prepared at a concentration of 3 mg/ml in 0.25 M STE buffer 10.25 M sucrose/lO mM Tris'HCL pH 7.4/1 mM EDTAI and carefully adjusted to pH 7.2 with NaOH. This solution was flltered through a 0.45-""m Millipore filter and protected from light at O°C. The final reaction mixture contained 0.45 mg of DAB per ml and 0.003% H20 2 • Reaction mixtures were gently rocked in plastic snap-cap tubes covered with aluminum foil at room temperature for 30 min. Soluble TfR Assay. Endosome-enriched fractions were collected from 15% Percoll gradients that had been centrifuged 40 min at 59,000 x g (average). The vesicles were treated with DAB/H 20 2 as described. Triton X-100 (1.0%) and bovine serum albumin (20 mg/ml) were added to the membranes, yielding final concentrations of 0.2% and 1 mg/ml, respectively. The sample was then underlaid with a solution of 35% (wt/wt) sucrose containing 10 mM Tris·HCI (pH 7.2) and 1 mM EDT A. The preparation was centrifuged in a Beckman Ti 75 rotor at 104.000 x g (average) for 60 min. As a result of this procedure, Perc 011 and DAB-containing endosomes were found in the pellet, whereas unreacted endosomal contents remained above the sucrose cushion. Samples of this supernatant were assayed for TfR activity as described by Lamb et al. (4). Protein determinations were performed according to the procedure of Vincent and Nadeau (16) with bovine serum albumin as a standard. RESULTS Peroxklue Reaction Product WltbbI EndOlODles Ioac:dvatel TIR. Incubation of endosomes containing the Tf-HRP conjugate with H20 2 and DAB in vitro results in a significant loss of Tf binding activity (Fig. 2A). Not only is there a loss of functional activity but there is a loss of immunological reactivity as weU (data not shown). After DAB treatment, the ability to detect TfR by either monoclonal (B3/25) or polyclonal antibodies was lost. Additionally, after DAB reaction, luminal contents were unable to be released by detergent. As demonstrated in Fig. lB. DAB treatment results in a quantitative reduction in the ability of Triton X-l00 (or sodium dodecyl sulfate) to release 125I-Tfinto a high-speed supernatant. Peroxidase-catalyzed receptor inactivation is restricted to those vesicles that contain luminal HRP. Reconstruction experiments employing mixtures of endosomes obtained from cells incubated with Tf-HRP and control cells demonstrate that receptor inactivation is additive (see Fig. 2 A and B far right columns). Further, the addition of free HRP to vesicles did not result in receptor inactivation (data not shown). Finally, the ability of DAB polymerization to render 125 1_ Tf insoluble was restricted to those vesicles that contain Tf-HRP. These results demonstrate that receptor inactivation or crosslinking of endosomal components does not affect vesicles that do not contain Tf-HRP. Unoccupied and Occupied Receptors Are in the Same IntraceUular Compartment. A previous study indicated that Tf-TtR complexes were internalized as a unit (4). However. we were unable to determine whether occupied and unoccupied receptors were in the same vesicle. To answer this question we incubated cells to steady state with various concentrations of Tf-HRP and assayed receptor binding activity after exposure of isolated endosomes to DAB IH 20 2 • As demonstrated in Fig. 3, occupancy of as little as 6.0% of cellular receptors with Tf-HRP resulted in the inactivation of ==75% of cellular receptors. Receptor occupancies were determined according to data obtained by equilibrium binding studies (see Materials and Methods), Since 20-40% of the cellular content of receptors are not in the recycling pathway (within the biosynthetic pathway or on the cell surface). this level of receptor inactivation probably includes all receptors within the endocytic apparatus. This result suggests that occupied and unoccupied receptors are in the same intracellular compartment. 83 - 100 '"A ~ :E u .9 75 U 50 <IS Q. r...; ~ '2 C 0 r...; 25 · r+- r-- 100 ~ :E ·. · ·. - 75 u r...; 50 Q. ~ ri- '2 C cr...; .. ~ u<IS ... c 25 t" ~ 0 10 20 30 40 .50 60 9( TfR occupancy FIG. 3. Relationship between occupancy of receptors by Tf-HRP and inactivation of TfR. Cells were incubated at 37°C in medium containing different concentrations of Tf-HRP until steady-state levels of binding were reached (30 min). Cultures were washed and surface-bound ligand was removed at O°C. The cells were homogenized, and endosomal fractions were collected from 15% Percoll gradients and incubated with DAB/H 20 2 • TfR activity was measured in the detergent-extracted supernatant. Data are presented as a percentage of control binding (mean ± SEM). See text for determination of receptor occupancy. 75 e fr 50 1j:: 25 .. Tf-HRP + DAB .. Tf-HRP Control ConlTol MI'lUre - DAB + DAB - DAB .. DA8 FIG. 2. Evidence that peroxidase-catalyzed oxidation of DAB within endosomes inactivates TfR. HeLa cells were incubated to steady-state binding at 37°C in the presence or absence of Tf-HRP (0.01 ~M). 12'I-Tf (2 nM) was added as an endosomal marker. Cells were washed. stripped of surface-bound ligand at O·C, and homogenized. The homogenate was applied to 15% Percoll gradients and en do somal fractions were pooled. Endosomes were either incubated with DAB/H 20 2 or DAB without H 20 2 as a control. The incubations were stopped by the addition of Triton X-100 to 0.1%. bovine serum albumin was added to a concentration of 1 mgl ml, and the lysate was centrifuged over a cushion of sucrose. Data are presented as mean (±SEM. where indicated). (A) Receptor activity remaining in the high-speed supernatant as measured by the ability to bind 12'I-Tf. (B) Ability of detergent to release internalized Il'I-Tfinto the high-speed supernatant. Fluid-Phase HRP Can Inactivate Internal TfR. Does the presence of occupied and unoccupied receptors in the same compartment reflect a normal physiological event? It is possible that the presence of intracellular unoccupied receptors is a consequence of the formation of a surface receptor-ligand complex-that is, occupancy of even a minor fraction of cell-surface receptors could trigger internalization of unoccupied TfR. To address this possibility, we examined whether HRP that has entered the endocytic apparatus by fluid-phase pinocytosis can inactivate internal TfR. Previous studies demonstrated that in HeLa cells HRP is taken up by fluid-phase pinocytosis (17). Cells were incubated with different concentrations ofHRP for 60 min at 37"C. washed. and homogenized. Endosomal fractions were isolated and incubated with H 20 2 and DAB. As demonstrated in Fig. 4. internalized HRP can inactivate TfR in a concentrationdependent manner. Identical results were obtained in the absence of calf serum. eliminating the possibility that TfR were internalized as a result of binding bovine Tf. The amount of peroxidase activity required to inactivate 50% of TfR is > lOOO-fold higher than that required using Tf-HRP. This observation is consistent with the difference in cellular accumulation between fluid- and receptor-mediated endocytosis. Control experiments ha ve revealed no effect of HRP on Tf binding. internalization. or distribution of receptors be- tween the surface and internal pool (data not shown). The inactivation of TfR by HRP internalized by fluid-phase pinocytosis was also independent of the presence of Tf. These results sugaest that the presence of internal TfR was not ligand induced. Inactivation or Unoccupied Recepton Is Time Dependent. As discussed in the Introduction and outlined in Fig. 1, two different models can explain how unoccupied and occupied receptors might exist in the same intracellular compartment. These models lead to different predictions regarding the rate of receptor inactivation. If occupied receptors enter or fuse with a vesicle containing a large number of unoccupied receptors. then inactivation of a majority of unoccupied receptors should occur even after a short pulse of Tf-HRP (modellI). Alternatively, if the first model were correct-that 100 ;;.., .~ ~ 7' E. :.. ..... "() ~ ~5 '-' o 10 HI 1- I f l HKP. nl!! ml FIG. 4. Effect of HRP internalized by fluid-phase pinocytosis on TfR activity. Cultures were incubated at 37°C for 60 min in medium containing the specified concentrations of HRP in the presence or absence of Tf 02.5 /-lMl. Cells were washed ex.tensively with cold PBS at O°C. harvested. and homogenized. Samples of the homogenate were applied to 159( Percoll gradients and endosomal fractions were collected. This material was incubated with DAB/H 20 2 and treated with detergent. and Tf binding activity was determined in the high-speed supernatant. TfR binding activity was measured and is presented as a percentage of control binding values Imean SEMI. 84 is, endocytic vesicles remain distinct during transit through the cell-the rate of endosomal TfR inactivation would be similar to the rate of ligand accumulation. Thus, the Oldy unoccupied TfR inactivated would be those internalized along with Tf-HRP and the amount ofTfR inactivated would reflect the proportion of the Tf cycle traversed by the conjugate. We tested the models by incubating cells for various times at 27°C in the presence of Tf-HRP and trace amounts of mI-Tf. The lower temperature (27°C) was employed to lengthen the transit time ofTfthrough the recycling pathway (20 min) and thus increase temporal resolution. Surface ligand was removed at O"C, cells were homogenized, endosomes were allowed. to react with DAB/H 20 2, and TfR activity was assared. Fig. 5 demonstrates the accumulation of intracellular 25 1_ Tf compared to the inacti vation of endosomal TfR. The data indicate a linear relationship between I2!iI-Tfaccumulation and inactivation of intracellular receptors. We observed similar results when the experiment was carried out in serum-free medium or performed at 37°C (data not shown). This result suggests that there is no central repository of TfR. Rather, each endocytic vesicle contains unoccupied TfR and inactivation of endosomal TfR results only when each vesicle also contains the Tf-HRP coqjugate. This observation suggests that unoccupied TfR and Tf-HRP must be internalized in the same vesicle and that fusion of vesicles containing occupied receptors with vesicles containing unoccupied receptors does not occur. A second experimental approach was used to demonstrate that internalized endosomes do not fuse with each other. We took advantage of the observation that DAB polymerization renders endosomal contents (i.e., mI-Tf) detergent insoluble to demonstrate that the contents of different endosomes remain separate. If cells were incubated for 5 min with both Tf-HRP and l2!iI-Tf, then 80% of the l2SI-Tf was detergent insoluble after DAB treatment. However, if the cells were incubated sequentially with the two ligands with as little as 1 min separating the pulses, essentially all of the endosomal 1251_Tf could be released by detergent (Fig. 6). Control experiments revealed that neither preincubation of cells with Tf nor the acid stripping procedure that was employed to ____ 0 100 r 21 7~ '-' :'0 ~5 ~ 0 T~m • f! i , i i i i ~ 10 15 ~O ~5 30 i 35 i 40 Incorporation. mlO FIG. 5. Relationship between the cellular accumulation of Tf-HRP and receptor inactivation. Cells were incubated at 27°C for various lengths of time in medium containing !251_Tf and Tf-HRP. Cultures were stripped of surface-bound ligand at O°C and the cells were homogenized. Endosome-enriched fractions were collected from a 15'7c Percoll gradient and treated with DAB/H,Ol' Receptor binding activity released into the detergent-extracted high-speed supernatant was measured .•. Percentage of radiolabel accumulated within cells relative to the maximum level of incorporation: :. loss of receptor binding activity normalized to the maximum loss of binding activit}. This takes into account that a percentage of Tf binding activity !20-30'7c I is not in the endocytic apparatus and would never be affected by thi~ procedure. Values (mean SEMI from four ,eparate e.xperimenls are presented. 100 E Q,. (J IU :i5 75 ~ c:... 50 :::l ;;Il ~ !j ":) .-:: :-(J ~;... ~~ · ~ · · · · · * r- - Control () Pre ~ • r- [ o w FIG. 6. Evidence that endocytic vesicles formed at different times do not intermingle their contents. Cells were incubated at 2~C with !l'I-Tf (0.01 ~) for 5 min in the presence or absence of Tf-HRP. The cultures incubated in the presence of Tf-HRP were then harvested. Cultures incubated in the absence of Tf-HRP were placed at O°C and surface-bound ligand was removed. The cultures were then returned to 27°C by the addition of warm medium and incubated for the specified times before being exposed to Tf-HRP for 5 min. Surface-bound ligand was then removed at O°C. Cells were homogenized and endosomes collected from a 15% Percoll gradient were incubated with DAB/H 20 2 • The amount of detergent-releasable radioactivity was then determined. To demonstrate that the procedure used to strip surface-bound ligand did not affect endocyti<: activity, cells were incubated with Tf for 5 min at 27°C. Surfacebound ligand was removed and the cells were incubated for 15 min at 27°C prior to as-min exposure to mI·Tfand Tf-HRP. The samples were then treated as described and are denoted as "0 Pre." remove surface-bound ligand affected subsequent internalization of Tf (see "0 Pre," Fig. 6). These results provide further evidence that individual endosomes formed at different times do not fuse with one another. DISCUSSION In this study we have taken advantage of the ability to specifically inactivate endosomal components to probe endosomal physiology. We were particularly interested in the mechanism(s) that affects the distribution ofTfR within HeLa celis, a culture line in which ==80% of the cellular TfR content is located intracellularly (4). We examined the movement of unoccupied receptors by use of HRP·medjated inactivation of endosomal contents. Incubation of cells with HRP did not affect receptor distribution, 125I_Tf binding, or l2.51_ Tf internalization. Yet HRP internalized by means of fluid-phase pinocytosis was able to inactivate a significant fraction of internal TfR, indicating that the presence of internal receptors was independent of ligand. Watts (18) was able to iodinate endosomal TfR in cultured hepatoma cells using an asiaioglycoprotein-Iactoperoxidase conjugate that had been internalized by the asialoglycoprotein receptor and concluded from this result that unoccupied TfR were internalized into the same vesicle as the asiaioglycoprotein receptor. Our results extend these studies by ruling out the possibility that binding of ligand to one receptor induces the internalization of an unoccupied second receptor. Since we obtained identical results in the presence or absence of bovine serum. the possibility that internalization of bovine Tf by HeLa cell receptors induced receptor internalization can be excluded. The degree of TfR inactivation depends on the proportion of the endocytic cycle traversed by the Tf-HRP conjugate. This result suggests that there is no central common receptor repository. This point is verified by the observation that 85 endocytic vesicles formed at different times do not fuse with each other. These results suggest that the "pool" of TfR does not represent receptors located in a defined subcellular organelle. Rather the pool is the summation of receptors within endocytic vesicles throughout the endocytic pathway. Morphological and biochemical studies have demonstrated that though Tf and other ligands may be internalized in the same vesicle, the contents of the vesicles at some stage diverge. Tf recycles to the surlace while most other ligands are directed to the lysosome. The results presented here verify those observations for the fluid-phase marker HRP. The amount of TfR inactivated by HRP internalized by fluid-phase pinocytosis is less than that of HRP internalized by means of the TfR. The result is consistent with the view that there is a population of endocytic vesicles that contain TfR but not the fluid-phase marker. The observation that endocytic vesicles formed at different times do not fuse is of interest in the context of morphological studies of the endocytic pathway. Electron microscopic examination of cells internalizing Tf reveals that ligand is initially found in coated pits and once internalized is found in structures of larger diameter (19. 20). Our data indicate that endocytic vesicles do not fuse with those formed at earlier or later times. * It should be noted that the model depicted in Fig. 1 is not meant to imply that the endocytic pathway consists of discrete vesicles. It offers a simple mechanism to explain how internalized receptors can remain segregated. Receptor-ligand complexes moving in a linear manner through a tubular structure would yield the same result since homogenization of membranes may form vesicles. If endocytic vesicles do not fuse with other endocytic vesicles. then with what do they fuse? It has been suggested that endocytic vesicles fuse with structures in the trans-Golgi region. There are repons that state that part of the Tf cycle involves the Golgi apparatus (12, 21). However. under conditions whereby two endosomal activities (TfR and leucyl-/3-napthalamidase) are greatly inhibited by the Tf-HRP/DAB procedure, the trans-Golgi enzyme galactosyltransferase is unaffected (unpublished). We interpret this result to suggest that if the Golgi apparatus is included in the endocytic pathway, it is only a minor pan of the pathway. Our results indicate that in HeLa celJs unoccupied receptors are internalized. However. in other cell types unoccupied receptors are either not being internalized or are being internalized at rates that are a fraction of the internalization rate of the occupied receptor. For example. in HL60 cells or rabbit (22,23) and human U.K .. unpublished) reticulocytes, the distribution of TfR between surlace and internal pools is radically altered upon addition of ligand. In the absence of ligand, the bulk of cellular receptors are on the surlace. *Endocytic vesicles may fuse with endocytic vesicles formed at the same time. At the temperature of these experiments the Tf cycle takes 20 min. Our data indicate that vesicles do not fuse with others formed 1 min later. Thus. the only vesicles capable of fusing are those that are formed in 1 min. which represents only 5'k of the endocytic cycle. whereas in the presence of ligand most of the receptors are intracellular. Additionally. within a given cell type, the distribution of unoccupied receptors may be subject to alteration by a variety of different agents. including mitogenic agents (24). ionophores (14). temperature (25), and hypoosmolar shock (unpublished). For example. macrophage tumor cells respond to phorbol esters or the calcium ionophore A23187 by a redistribution of receptors, resulting in a new state in which there is a 2- to 3-fold increase in surlace receptor number (10). We suggest that the new steady state may reflect a decrease in the internalization rate of unoccupied receptors. We thank Drs. Many Rechsteiner and Steven Wiley for their helpful suggestions in the preparation of this manuscript. We also acknowledge the technical help of Ms. Ina Jordan and the word processing skills of Ms. Chris Hall. This work was supponed by grants from the National Institutes of Health (HL2592203 and HL2692205). R.S.A. was supponed in pan by a National Institutes of Health traineeship (5T32GM07464-08). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. Klausner. R. D., Harford, J. & VanRenswoude, J. (1984) Proc. Natl. Acad. Sci. USA 81, 3005-3009. Watts, C. (1985) J. Cell Bioi. 100,633-637. Bleil. J. D. & Bretscher. M. S. (1982) EMBO J. 1,351-355. Lamb, J. E., Ray. F .. Ward, J. H., Kushner, J. P. & Kaplan. J. (1983) J. Bioi. Chem. 158, 8751-8758. Deutsch, P. J .. Rosen, 0: M. & Rubin, C. S. (1982) J. BioI. Chem. 257,5350-5358. Tanabe. J., Pdcer, W. E., Jr .. & Ashwell. G. (1979) Proc. Natl. Acad. Sci. USA 76, 1038-1043. Stahl, P .. Schlesinger. P. H .• Sigardson. E., Rodman. J. S. &: Lee. Y. C. (1980) Cell 19, 207-215. Ward,J. H., Kushner,J. P.&:Kaplan.J.(1982)J.Biol.Chem. 257,10317-10323. Wiley, H. S. &: Cummingham, D. D. (1982)J. Bioi. Chem. 257, 4222-4229. Nakane. P. K. &: Kawaoi, A. (1974) J. Histochem. Cytochem. 11, 1084-1091. J.,angmuir, l. (1918) J. Am. Chem. Soc. 40, 1361-1374. Willingham. M. c., Hanover, J. A., Dickson, R. B. &: Pastan, I. (1984) Proc. Natl. Acad. Sci. USA 81, 175-179. Hopkins. C. R. (1983) Cell 35, 321-330. Buys, S .• Keogh, E. A. & Kaplan, J. (1984) Cell 38, 569-576. Counoy, P. J., Quintan. J. & Baudhuin. P. (1984)J. Cell Bioi. 98, 870-876. Vincent. R. &: Nadeau. D. (1983) Anal. Biochem. 135,355-362. Kaplan. J. (1976) Nature (London) 263, 596-597. Walls. C. (1984' EMBO J. 3, 1965-1970. Wall, D. A., Wilson. G. & Hubbard, A. L. (1980) Cell 21, 79-93. Geuze,H. J..Slot,J. W.&:Strous,G. J. A. M.(1983)CeIl32, 277-287. Snider. M. D. & Rogergs. O. C. (1985) J. Cell Bioi. 100, 826-834. May. W. S., Jacobs. S. &: Cuatrecasas. P. (1984) Proc. Natl. Acad. Sci. USA 81, 2016-2020. McArdle. H. J. & Morgan. E. H. (1984) J. Bioi. Chem. 259, 1629-1643. Wiley. H. S. & Kaplan. J. (1986) Proc. Natl. Acad. Sci. USA 81, 7456-7460. Weigel. P. H. & Oka. 1. A. (1983) J. Bioi. Chem. 258, 5089-5094. CHAPTER V SUMMARY 87 These studies were designed to investigate the constituents of the endocytic pathway and the properties of organelles involved in that pathway. Specifically, transferrin and its receptor were used to probe the nature of the endocytic apparatus. It is within the endocytic pathway that different receptor/ligand complexes are directed either to lysosomes, or recycled to the cell surface. The physical and biochemical nature of the endosome are poorly understood and the number of compartments involved in the sorting process are unknown. Questions also exist as to whether there are more than one endocytic pathway. The endosome was studied by use of a technique which alters its physical nature to the point where it can be separated from other intracellular organelles. Peroxidase confined within a membrane vesicle can be reacted with diaminobenzidine (DAB), and hydrogen peroxide (H202). The oxidizing atmosphere generated within the vesicle by the peroxidase and H202 causes the deposition of a dense polymer within the vesicle lumen which increases the buoyant density of the vesicle. Peroxidase was directed to the endosome by chemically coupling horseradish peroxidase (HRP) to transferrin (Tf). Unlike other ligands Tf is not directed to lysosomes and remains bound to its receptor throughout the entire endocytosis/recycling pathway. Thus, under steady state binding conditions at 37 0 C, this pathway would be exclusively labeled with Tf-HRP. Vesicles obtained from cells incubated in the presence of Tf-HRP can be reacted in vitro with DAB and H202 resulting in an increase in their 88 buoyant density providing a means to purify them away fronl contaminating vesicle populations. A secondary effect of the peroxidaselDAB reaction is the "crosslinking" of intraluminal components. This crosslinking renders the endosomal contents insoluble in detergent. Experiments in which control vesicles were mixed with Tf-HRP vesicles confirmed that the reaction was confined only to those vesicles which contained Tf-HRP. The inability of detergent to constituent from DAB reacted material was determined by three approaches. First, radiolabeled tracer molecules within endosomes were not released by detergent. Second, enzymatic activities within such compartments were inactivated. Third, antibody could no longer detect antigens in vesicle extracts. A combination of the density shift technique and the loss of detergent extractable activities was used to probe the physiology and biochemical makeup of the endocytic compartment. As a prerequisite, it was necessary to determine that the Tf-HRP conjugate behaved identically to the native ligand. We demonstrated the conjugate could compete with Tf for binding to the receptor, exhibited indistinguishable binding affinity, and recycling kinetics compared to Tf alone. Most importantly, following treatment with DAB and H202, Tf-HRP vesicles exhibited a significant increase in buoyant density which was selective only for those vesicles which contained conjugate. U sing this conjugate we demonstrated that different receptor/ligand complexes were internalized into the same 89 endocytic compartment. Once internalized, receptor/ligand complexes are known to have different fates. For example, EGF, along with its receptor, is degraded in lysosomes while LDL is directed to the lysosome and its receptor recycled. In the case of Tf, iron is released from diferric Tf within the cell, and the apoTf, Tf receptor complex is recycled to the cell surface. Morphological evidence suggests that different receptor/ligand complexes are present within the same coated pit (1,2). The density shift technique was used to provide biochemical evidence as to whether Tf, EGF, and LDL were internalized into the same endocytic compartment. Under conditions where Tf-HRP caused a major increase in the buoyant density of endosomes, there was a concomitant increase in density in the compartment containing internalized, radiolabeled Tf, EGF, and LDL. We could thus demonstrate in a quantitative manner that different receptor/ligand complexes are internalized into the same endosomal compartment. These results are consistent with the hypothesis that the endosome is responsible for sorting internalized receptor/ligand complexes. The crosslinking effect of the peroxidase DAB reaction rendered this method of little use for the isolation of endosomal constituents. The inability to detergent extract this material, however, can be used to determine activities associated with the endosome. For example, we determined that the enzyme leucyl-betanaphtylamidase was enriched in endosomes in HeLa cells by demonstrating that the Tf-HRP/DAB treatment inactivated most of the cellular activity of this enzyme. Peroxidase activity could be 90 placed in lysosomes by allowing cells to internalize unconjugated HRP by pinocytosis and much of the activity of the lysosomal enzyme hexoseaminidase was also lost following DAB treatment. Thus organelle-specific constituents can be analyzed as long as peroxidase can be specifically placed within the lumen of the organelle being studied. Conversely, the peroxidase/DAB technique can also be used to isolate membrane fractions which do not contain peroxidase activity. The fact that the peroxidase/DAB procedure can be used in a quantitative manner makes it useful for other analyses. The amount of time required for apoTf to be recycled to the cell surface and the time required for the delivery of EGF to the lysosome are known. The amount of time, however, that the two ligands spend In the same endocytic compartment is unknown. Morphological studies have led only to rough estimates, of this value (3). Use of the Tf-HRP conjugate allowed us to determine that within 10 min, 85% of internalized EGF had separated from the Tf-containing compartment. Since EGF takes approximately 20 min to reach the lysosome, there must be a fairly long lived compartment between the endosome and the lysosome. Questions exist as to what other intracellular organelles may be included in the endocytosis/recycling pathway. It has been proposed that Tf recycles through Golgi compartments (3,4). This hypothesis was tested using Tf-HRP and three different markers for the Golgi. In all cases tested, under conditions in which major alterations in endosomal activities occurred, no similar effect was 91 observed for Golgi activities. It was therefore concluded that the Golgi apparatus is not included in the endocytosis/recycling pathway at least within the time frame of the Tf iron delivery cycle. There is strong biochemical evidence that transferrin recycles through compartments which contain enzymatic activities associated with Golgi compartments. In concert, the biochemical, morphological, and kinetic data suggest the existence of two recycling pathways. Our data suggest that the shorter pathway involved in ligand delivery does not include the Golgi apparatus. The apparent longer pathway probably includes Golgi compartments and may be part of a "proofreading" mechanism for membrane proteins. We were able to study the movement of unoccupied receptors by incubating cells with various concentrations of Tf-HRP. By this method, Tf receptors could be analyzed under steady state binding conditions where a low percentage of receptors were occupied with Tf-HRP. This in turn provided a means to test different models for endocytosis. The majority of Tf receptors in HeLa cells are intracellular. We demonstrated that with as little as 6% of the receptors occupied at steady state by Tf-HRP, nearly all the intracellular receptors could be inactivated following treatment with DAB. Those unaffected by the procedure could be accounted for by receptors present within the biosynthetic pathway. Occupied and unoccupied Tf receptors must therefore be present in the same endocytic compartment. Two models can account for this finding. The first model states that only occupied receptors are internalized, 92 and the endocytic vesicle fuses with an intracellular compartment which contains unoccupied receptors. The second model proposes that both occupied and unoccupied receptors are internalized together, and the intracellular pool of Tf receptors represents individual endocytic compartments in transit through the cell. The models lead to specific predictions regarding the rate at which intracellular receptors can be inactivated using Tf-HRP and DAB. Experiments designed to test these mode~s demonstrated that the rate at which internal receptors were inactivated by the DAB procedure was proportional to the fraction of the endocytic cycle traversed by the conjugate, suggesting that the second model is correct. The second model also predicts that endocytic vesicles remain temporally segregated. That is, endosomes formed at one point in time do not intermingle their contents with those formed at other times. Experimental results supported this prediction. The physical structure of the endosome is unknown. It has been described morphologically as being both vesicular and tubular in nature. Receptor/ligand complexes internalized at different times might remain segregated because they are contained within different vesicles which do not mix with each other. On the other hand, these complexes may be present within a tubular structure in which lateral movement is restricted. These experiments cannot differentiate between the two possibilities. Experiments presented In the Appendix will directly address this question. Current evidence indicates that endocytic processes which incl ude pinocytosis as well as receptor-mediated endocytosis 93 involve an internal, intermediate compartment distinct from lysosomes. This organelle has been termed the endosome and is responsible for sorting membrane constituents internalized during various endocytic processes. This dissertation has focused on the biochemical composition of compartments involved in receptor mediated endocytosis. As integral membrane proteins, receptors not only provide a sensitive and specific means to trace the endocytic pathway, but also a way to monitor membrane turnover and dynamics. Transferrin and its receptor were chosen to follow the endocytic/recycling pathway since both receptor and ligand are are recycled to the cell surface following endocytosis. An understanding of the physiology and biochemistry of the endosome should therefore lead to a better concept of the mechanisms involved in the regulation of membrane proteins. The ability to specifically label different receptor/ligand complexes with various tracer molecules has enabled investigators to describe the endocytic apparatus in kinetic and morphological terms. The major deficiency of these studies has been the inability to dissect the endocytic pathway in a quantitative manner. The primary requirement for this type of analysis is a method by which the endosome can be isolated or otherwise made physically distinct from other intracellular organelles. The Tf-HRP/DAB density shift and crosslinking reactions are particularly well suited for this since they are endosome-specific can be interpreted in a quantitative manner. The remainder of this section will therefore suggest specific questions which can be addressed using this technology. 94 Different membrane molecules differ in their distributions between the surface and internal compartments. For example, In human fibroblasts greater than 90% of the cell's LDL receptors are found on the surface, whereas most of the Tf receptors are internal. When exposed to ligand, both receptors recycle and both are internalized into the same endocytic compartment. This observation raises the question of how the cell maintains these different distributions when all other aspects of the receptors are the same. There are two general models which could account for this observation. First, it is possible that only occupied LDL receptors are internalized. Second, unoccupied LDL receptors may be internalized along with unoccupied Tf receptors, but the LDL receptor may segregate into an extremely rapid recycling vesicle. The most straightforward way to test these models is to determine whether unoccupied LDL receptors are internalized. This can be achieved by incubating cells with Tf- HRP in the absence of LDL, isolating endosomes, treating with DAB, and determining whether LDL receptors are affected. If receptors are not affected we could conclude that the LDL receptor is not internalized in the unoccupied state. The positive control experiment would be to perform the same experiment in the presence of LDL. Under these conditions LDL and its receptor would be in the same endocytic vesicle as TfHRP. Treatment of endosomes isolated from these cells with DAB would then result in a loss of LDL receptor activity or the ability of an anti-LDL receptor antibody to detect the receptor in detergent extracts. 95 The possible existence of a selective and nonselective pathway for the recycling of transferrin was discussed in a previous chapter. Aside from analyzing morphological or biochemical markers along the recycling route, it is difficult to test this hypothesis without a method of segregating the pathways. Stein and Sussman reported that a monensin could be used to demonstrate two different endocytic pathways in the erythroleukemia cell line K562 (5). They observed a 50% decrease in surface transferrin receptors following treatment of cells with the drug. These receptors were found to be inside the cells and unavailable to ligand. however, appeared to recycle normally. The remaining receptors, The authors concluded that the receptors lost from the cell surface become trapped within intracellular compartments which are separate from the monensin insensitive recycling pool. This hypothesis can be tested directly with the DAB technique. If there are two pathways which can be segregated by the administration of monensin, then only a fraction of the internal pool of transferrin receptors would be available for inactivation by Tf-HRP and DAB. Conversely, if monensin merely causes a redistribution of surface receptors within a single recycling pathway, all of the internal receptors would be inactivated by DAB treatment. One of the more direct questions which can be answered by the use of an endosome-specific probe is simply what activities are associated with the endosome. This question is important to address for the following reasons. First, it can be used to determine if other intracellular organelles are included in the recycling of 96 receptor/ligand complexes as discussed in a previous chapter. Second, there are a number of proteins or activities which appear to be recruited to the cell surface from intracellular storage sites. These include the glucose transporter In adipocytes, the Mac-I, LF A1 adhesion molecules of monocytes, and the T-cell receptor in lymphocytes (6-8). A number of different agents can be used to alter the distribution of receptors between the cell surface and the endosome (9,10). Phorbol esters increase the number of surface transferrin receptors in macrophages (9) and also increase the level of expression of LFA-1 in monocytes (7). It is possible, and indeed probable that molecules other than receptors reside within endosomal compartments. Endosomal components may thus represent a general reservoir of physiologically active molecules which, under the appropriate stimulation, may be recruited to the plasma membrane. The Tf-HRP/DAB inactivation reaction can be used to directly test this hypothesis. Antibodies are available for the glucose transporter, Mac-1 LF A-1, and the T-cell receptor and may be used to probe for these activities. If the ability of the antibody to distinguish antigen is lost following DAB treatment, it would indicate that these activities were associated with the endosome. Thus, any protein for which there is an antibody or assay can be tested in this manner. Application of this technology would be important in the development of an in vitro system to measure vesicle fission. Significant progress has been made in the understanding of Golgi processing due to the establishment of an in vitro assay for the 97 fusion of Golgi compartments (11). The major advantage of the peroxidase/DAB system is that one can measure the separation, or fission of endocytic compartments. It is known, for example, that internalized transferrin and EGF have different intracellular fates. Transferrin and its receptor are recycled· whereas EGF and its receptors are transferred to lysosomes. We have analyzed the kinetics of separation using Tf-HRP and radio labeled EGF in intact cells. Thus the "window" in which to expect the separation is already known. The assay is reasonably straightforward. That is, endosomes from cells incubated with Tf-HRP and radiolabeled EGF will be mixed with cellular extracts and the ability of DAB to render the EGF detergent insoluble will be measured. Successful fission of compartments will be exhibited by the ability of detergent to release radioligand following treatment with DAB. The converse of this approach can be accomplished either by conjugating HRP to a ligand which is transferred to lysosomes, or by using HRP internalized in the fluid phase. This provides an opportunity to confirm results obtained by the Tf-HRP approach. of these types of experiments is twofold. The significance First, development of an in vitro system for the separation of endocytic markers will allow for the determination of cytosolic components required for fission. Second, if there is a distinct compartment between the endosome and the lysosome, this approach should facilitate its identification and possible isolation. Other questions regarding the role of endocytic processes in cell metabolism can be addressed by selectively inactivating the 98 receptor-mediated endocytic pathway. It has been suggested that cell motility is accomplished via polar endocytosis/exocytosis activity (12). The model suggests that membrane is inserted at the leading edge of a motile cell by exocytic processes, and retrieved at the trailing edge by endocytosis. This is in part based on the observation that receptors exhibit a polar distribution in these cells. The role of receptor-mediated endocytosis in pinocytosis or motility can be directly tested if the endocytic apparatus can be inhibited in live cells. This might be accomplished through the use of the peroxidase/DAB inactivation of endocytic compartments. different approaches could be used to accomplish this. Two The endocytic/recycling pathway could be filled with peroxidase using Tf-HRP. Membrane movement would be stopped by placing cells at 0° C, at which time they would be incubated with DAB and H202, causing specific inactivation of the endocytic apparatus. The possibility exists that DAB is not permeable enough to traverse both the plasma membrane and and endosomal membrane. A possible solution to this problem would be to incubate cells with both Tf-HRP and DAB. Thus DAB would be included in the fluid phase of the endocytic vesicle and transfer across membranes would not be required. Alternatively, glucose oxidase would be chemically coupled to transferrin. Cells would then be incubated in the combined presence of Tf-glucose oxidase, HRP, and DAB. These cells would then be incubated at 00 C with media containing glucose. The hydrogen peroxide generated by the glucose oxidase-glucose reaction would drive the polymerization of DAB in the presence of 99 HRP. Because Tf-glucose oxidase would be present only within the endocytic apparatus, DAB inactivation would be restricted to those compartments. Cells would then be returned to 37 0 C where it could be determined if either pinocytosis or motility were functional. 100 References 1. Carpentier, J.L., Gorden, P., Anderson, R.G~W., Goldstein, J.L., Brown, M.S., Cohen, S., and L. DrcL 1982. Co-localization of 125 1_ epidermal growth factor and ferritin-low density lipoprotein in coated pits: A quantitative electron microscopic study in normal and mutant human fibroblasts. J. Cell BioI. 95: 73 2. Geuze, HJ., Slot, J.W., Strous, GJ.A.M., Peppard, J., von Figura, K., Hasilik, A., and A. Schwartz. 1984. Intracellular receptor sorting during endocytosis: Comparative immunoelectron microscopy of multiple receptors in rat liver. Cell 37: 195 3. Willingham, M.C., Hanover, J.A., Dickson, R.B., and I. Pastan. 1984. Morphologic characterization of the pathway of transferrin endocytosis and recycling in human KB cells. Proc. Nat. Acad. Sci. USA 81: 175 4. Fishman, J.B., and R.E. Fine. 1987 A trans Golgi-derived exocytic coated vesicle can contain both newly synthesized cholinesterase and internalized transferrin. Cell 48: 157 5. Stein, B.S., and H.H. Sussman. 1986. Demonstration of two distinct transferrin receptor recycling pathways and transferrin-independent receptor internalization in K562 cells. J. BioI. Chern. 261: 10319 6. Suzuki, K., and T. Kono. 1980. Evidence that insulin causes translocation of glucose transport activity to the plasma membrane from an intracellular storage site. Proc. Nat. Acad. Sci. USA 77: 2542 7. Springer, T.A., and D.C. Anderson. 1986. The importance of the Mac-I, LFA-l glycoprotein family in monocyte and granulocyte adherence, chemotaxis, and migration into inflammatory sites: insights from an experiment of nature. Ciba-Found-Symp. 118: 102 8. Tse, D.B., AI-Haideri, M., Pernis, B., Cantor, C.R., and C.Y. Wang. 1986. Intracellular accumulation of T-cell receptor complex molecules in a human T- cell line. Science 234: 748 101 . 9. Buys, S.S., Keogh, E.A., and J. Kaplan. 1984. Fusion of intracellular membrane pools with cell surfaces stimulated by phorbol esters and calcium ionophores. Cell 38: 569 10. Wiley, H.S., and J. Kaplan. 1984. Epidermal growth factor rapidly redistributes pools of transferrin receptors. Proc. Nat. Acad. Sci. USA 81: 7456 11. Balch, W.E., Dunphy, W.G., Braell, W.A., and J.E. Rothman. 1984. Reconstruction of the transport of protein between successive compartments of the Golgi measured by the coupled incorporation of N -acetylglucoseamine. Cell 39: 405 12. Bretscher, M.S. 1984. Endocytosis: Relation to capping and cell locomotion. Science 224: 681 APPENDIX 103 Different cell types exhibit dissimilar distributions of receptors between the surface and internal compartments. For example, HeLa cells express only 20% of their Tf receptors on the surface, whereas in the human leukemia cell line HL60 greater than 90% the Tf receptors are found on the cell surface (1,2). One way in which cells can regulate the number of surface receptors is to vary the rate at which unoccupied receptors are internalized. This is true for transferrin receptors in HeLa cells, but other cell types may regulate receptors by a different mechanism. Surface Tf receptor number can be reduced by the addition of ligand (3), or by incubating certain cell types in the presence of phorbol esters (2). In both cases, this loss was shown to be due to the production of an internal pool of receptors at the expense of surface receptors. The difference, however, between the two is that in the the first case, loss of surface receptors is ligand-induced, and response to phorbol esters is independent of ligand. In order to determine whether these responses are due to the similar, or different mechanisms, it IS important to account for the movement of unoccupied receptors. The internal receptor pool generated by phorbol esters could be a result of an increase in the rate of unoccupied receptor internalization, or by the wholesale internalization of a fraction of receptors which are retained in a cryptic internal pool. The first possibility predicts that the internal pool remains part of the normal transferrin recycling pathway. converse prediction. The second supposition makes the That is, the internal pool becomes segregated from the endocytosis/recycling pathway. These possibilities were 104 tested using Tf-HRP and the DAB inactivation procedure. If the first possibility were true, then under conditions of low receptor occupancy with Tf- HRP, at steady state binding conditions, a majority of internal receptors could be inactivated following treatment of endosomes with DAB. The second proposal predicts that the internal receptors would remain insensitive to this inactivation even with high receptor occupancies. The first set of experiments performed were designed to confirm surface receptor depletion in response to incubation with phorbol ester. HL60 cells were incubated at 37 0 C in the presence of phorbol myristate acetate (PMA) (0.1 ug/ml) for various times at which the temperature was reduced to 00 C and surface binding determined (Figure 1). Significant reduction was observed by 30 min. Sixty min in PMA was used for subsequent experiments. In order to determine whether the internal receptor pool participated in the transferrin recycling pathway, the following experiment was performed. HL60 cells were incubated at 37 0 C in the presence or absence of PMA for 30 min. At this time Tf-HRP was added to a concentration equivalent to 10% receptor occupancy and further incubated for 30 mIn. The temperature was reduced to 00 C and surface ligand was removed using alternating washes of isotonic citrate/phosphate buffer (pH 3.8) and PBS supplemented with 1 mM EDT A. Cells were homogenized, membranes isolated, and vesicles were incubated with DAB and H202. This material was solubilized with 0.2% Triton X-IOO and the extracted supernatant analyzed for the ability of the transferrin receptor to bind 105 100 0) .a-a" 0- .--------. 80 c: =uc: 60 CJ Control ....e c: 40 • PMA Ci5 -D-------CJ -------• o () (fl. 20 a a 10 20 30 40 Time Figure 1. PMA reduces surface receptor number in HL60 cells. HL60 cells were incubated at 37 0 C. in the presence of PMA (0.1 ug/ml) for the times indicated. Cells were placed at 0 0 C., washed and 125I - Tf was added to measure surface receptor binding. Values are presented as the percentage of control surface binding. 106 radioligand in a soluble receptor assay. PMA treated cells exhibited a major loss in receptor activity compared to control cells (Figure 2). This result indicates that the internal to control cells receptors are not only part of the recycling pathway, but that they are a result of an increase in the rate of internalization of unoccupied receptors. Morphological, physical, and kinetic data suggest that it is within the endosome that intracellular routing decisions are made. In general, internalized receptor/ligand complexes are either transferred to lysosomes or recycled to the cell surface. In spite of the evidence supporting this view, very little is known about the features of the endosome which affect these decisions. is known about the actual structure of the endosome. In fact, little We demonstrated that receptors internalized into endocytic compartments at different times remain temporally separated (Chapter 4). Morphological analyses of endocytic compartments have described both vesicular and tubular structures. Endocytic compartments existing as tubules or a series of discreet vesicles could account for our data and can be simplified by the following models. Model 1 (Figure 3) indicates that the endosome is a tube- like structure in which receptors have restricted lateral mobility. Movement of receptors along the tube would occur by the directed flow of the membrane. Alternatively, internalized receptors could be internalized into discreet membrane vesicles. Temporal segregation would be maintained by restricting the fusion of vesicles internalized at different times (Model 2). 107 · 80 - 100 ...... s; ~ '';::::; u <C (5 '..... c: 0 () #. -,- -,- ....,- T · 40 60 · 20 · 0 -DAB +DAB +DAB -DAB -PMA +PMA Treatment Figure 2. PMA Increases the rate of internalization of unoccupied Tf receptors. Cells were incubated for 30 min at 37 0 C. in the presence or absence of 0.1 ug/ml PMA. Tf-HRP (2 x 10-9 M) was added and cells further incubated for 30 min. Surface ligand was removed at 00 C., cells homogenized, and endosomes isolated. Vesicles were incubated with and without DAB, solubilized with 0.2% Triton X-lOO, and the extracts analyzed for transferrin receptor activity in a soluble receptor assay. The data are presented as the percentage of control receptor activity. 108 Model 1 Model 2 , Transferrin « Apotransferrin 109 The experiments demonstrating temporal segregation cannot differentiate between the two endosomal models. Our approach to distinguish between these possibilities was to compare the interaction of internalized fluid phase and receptor-mediated markers. As previously described, the HRP catalyzed polymerization of DAB within membrane vesicles crosslinks the lumen of the vesicles, and renders their contents insoluble in detergent. We utilized this observation to test the models by analyzing endocytic compartments containing either fluid phase HRP or 125 1-Tf. If endosomes are tubes, the fluid phase marker HRP would be expected to diffuse freely throughout the lumen and intermingle with a receptor-mediated marker administered at a different time. If, however, endosomes exist as discreet vesicles, the markers would not be expected to mix except by vesicle fusion. Cells were given a 5 min pulse of 1251-Tf (5 x 10- 9 M). The temperature was reduced to 0 0 C and surface ligand was removed using alternating washes of isotonic citric acid/phosphate buffer (pH 3.8) and PBS supplemented with 1 mM EDTA. The temperature was raised for various times and cells were pulsed for 5 min with HRP (20 mg/ml). Control cultures were either pulsed with 1251-Tf alone or received a simultaneous pulse of 1251-Tf and HRP. Cells were washed extensively at 0 0 C, homogenized, and the 800 x g supernatant applied to 12% Percoll gradients. Endosomal fractions were isolated, treated with DAB and extracted with 0.2% SDS. The detergent extracts were centrifuged over cushions of 35% sucrose (104,000 x g, 60 min) and the radioactivity in the supernatant and 110 the pellet determined. If the pulses were separated by as little as 30 sec at either 37 0 C, or 27 0 , the endocytic markers remained separated. This was demonstrated by the ability to detergent extract essentially all of the 1251-Tf following DAB treatment (Figure 4A and 4B). These results are consistent with the second model which suggests that endocytic compartments are discreet vesicles. Similar results were obtained if the pulses were reversed (Figure 5). As a control, various dilutions of HRP were used in the second pulse to determine whether the level at which DAB crosslinking could be detected. When given a simultaneous pulse, even concentrations 1/10 that of normal, the inability of detergent to extract 125 1-Tf following DAB treatment was detectable (Figure 6). Thus, even low levels of mixing or dilution would have been detectable. Morphological studies have described ligand as being initially internalized into vesicular compartments and tubular structures at later times (4). The time frames used for these studies may not have been long enough to detect a more "mature" tubular endosome structure. One way to determine whether endocytic vesicles eventually form tubes is to test whether the vesicles mIX at later times after internalization. This possibility was tested by giving a pulse of Tf-HRP followed by a chase of various times in the presence of unlabeled Tf. If endosomes containing Tf-HRP mixed with compartments containing intracellular receptors, with time, internal receptors could be inactivated with DAB treatment. Cells were incubated with Tf-HRP (1 x 10- 8 M) for 90 min at 0° C. 111 A 80 - - 100 ::E c.. () Q) - ~ . .0 CO U 60 ~ ..... - -I" >< UJ 40 E ::J E 'x CO 20 ~ 0 0 - 1'"""- - ::E Control 0 .5 23456 B Control 0 .5 1 2 3 4 5 Time Between Pulses Figure 4. Temporally segregated receptor-mediated and fluid phase markers do not mix within the endocytic pathway. HeLa cells were pulsed with 125 1-Tf followed by a pulse of HRP at either 37 0 C. (A) or 270 C. (B) and the time between pulses varied from 0 to 6 min. Surface ligand was removed at 0 0 C., cells homogenized, and membranes isolated. Vesicles were treated with DAB and the amount of detergent extractable radioactivity measured. Data were normalized to the maximum extractable radioactivity. 112 Control 0 .5 1 2 3 Time Between Pulses Figure 5. Temporally segregated fluid phase and receptor-mediated markers do not mix within the endocytic pathway. HeLa cells were pulsed with HRP followed by a pulse of 125I-Tf at 27 C. and the time between pulses varied from 0-3 min. Surfacebound ligand was removed at 00 C., cells were homogenized, and membranes isolated. Vesicles were treated with DAB and the amount of detergent radioactivity measured. Data were normalized to the maximum extractable radioactivity. 0 113 100 - 80 · - ~ a.. () · Q) :0 - c:tS ~"- 60 X W 40 - c: 20 - ~ 0 0 · (5 "- 0 () Control 0.2 2 20 Concentration H RP (mg/ml) Figure 6. Low levels of HRP are sufficient to catalyze DAB-vesicle crosslinking. HeLa cells were given a simultaneous pulse of 125I-Tf and HRP. The concentration of HRP was varied from 0 to 20 mg/ml. At the end of the pulse period, surface ligand was removed at 0° C., cells were homogenized, membranes isolated, and incubated with DAB and H202. The incubation mixture was extracted with 0.2% Triton x100, and the solubilized radioactivity measured. Data were normalized to control values. 114 Unbound ligand was washed away, and the temperature raised to 37 0 C for 5 min. The temperature was reduced to 0 0 C and surface bound ligand removed as described previously. Cultures were returned to 37 0 C in the presence of unlabeled Tf (1.25 x 10-6 M) for various times, cells homogenized, and membrane isolated. This material was treated with DAB, extracted with 0.2% Triton X-lOO, and the extract analyzed for the ability of solubilized Tf receptors to bind ligand. Recovery of intracellular receptor activity was seen within 5 min and continued to increase to near control values by 30 min (Figure 7). These results suggest that the pulse of Tf-HRP was recycled, and within that time, had not intermingled with other compartments containing Tf receptors. 115 100 ·u . - 60 - T .s: . - c 0 0 40 - ~ -0 20 - e T T T 80 >. <C - . 0 o 5 . 10 30 Time (min) Figure 7. Endocytic vesicles do not fuse with compartments containing unoccupied receptors. HeLa cells were incubated with Tf-HRP (1 x 10- 8 M) for 90 min at 00 C. Cells were washed to remove unbound ligand, and the temperature raised to 37 0 C for 5 min. Cells were returned to 00 C., surface-bound ligand was removed, and the temperature was raised to 37 0 C. for the indicated times. Cells were homogenized, membrane isolated, and vesicles were treated with DAB. This material was extracted with 0.2% Triton X-IOO, and the extract analyzed for the ability of solubilized Tf receptors to bind ligand. Data are presented as a percentage of control receptor activity. 116 References 1. Lamb, J.E., Ray, F., Ward, J.H., Kushner, J.P., and J. Kaplan. 1983. Internalization and subcellular localization of transferrin and transferrin receptors in HeLa cells. J. BioI. Chern. 258: 8751 2. May, W.S., Jacobs, S., and P. Cuatrecasas. 1984. Association of phorbol ester-induced hyperphosphorylation and reversible regulation of transferrin membrane receptors in HL60 cells. Proc. Nat. Acad. Sci. USA 81: 2016 3. Klausner, R.D., Harford, J., and J. van Renswoude. 1984. Rapid internalization of the transferrin receptor in K562 cells is triggered by ligand binding or treatment with a phorbol ester. Proc. Nat. Acad. Sci. USA 81: 3005 4. Geuze, H.J., Slot, J.W., and G.J.A.M. Strouse 1983. Intracellular site of asialoglycoprotein receptor-ligand uncoupling: Doublelabel immunoelectron microscopy during receptor-mediated endocytosis. Cell 32: 277