Studies of transferrin recycling reconstituted in streptolysin O permeabilized Chinese hamster ovary cells.

Efficient transferrin receptor recycling is reconstituted when donor cytosol and ATP are added to the streptolysin O permeabilized cells. The rate of reconstituted recycling is dependent on the concentration of donor cytosol. The cytosol provides a factor(s) required for the transport of transferrin from the pericentriolar recycling compartment to the plasma membrane. N-Ethylmaleimide treatment of permeabilized cells inhibits both the fusion of recycling vesicles with the plasma membrane as well as the formation of functional recycling vesicles from the pericentriolar recycling compartment. Guanosine 5′-3-O-(thio)triphosphate (GTPS) does not affect reconstituted recycling in the presence of an optimal cytosol concentration. Therefore, the rate-limiting step in recycling is not regulated by GTP-hydrolyzing proteins, and hydrolysis of GTP is not required for endocytic recycling. GTPS stimulates recycling when suboptimal concentrations of cytosol are used. This stimulatory effect is not mediated by a brefeldin A-sensitive ADP-ribosylation factor protein. Addition of wild-type donor cytosol to permeabilized END2 Chinese hamster ovary cells, which recycle transferrin at half the rate of wild-type cells, reconstitutes recycling to the reduced rate of intact END2 cells but not to the wild-type recycling rate. These results indicate that the defect responsible for the slowed transferrin recycling in END2 mutants is membrane associated or that the defective protein is too large to diffuse out of the cells through the streptolysin O pores.

Receptor-mediated endocytosis is a process used by cells for rapid and specific uptake of extracellular macromolecules (McGraw and Maxfield, 1991;Trowbridge et al., 1993). Most internalized ligands are delivered to lysosomes, and most receptors are recycled back to the cell surface to mediate further rounds of ligand internalization. Membrane proteins are recycled to the cell surface by a default, bulk flow process (Dunn et al., 1989;Mayor et al., 1993). In CHO 1 cells, recycling membrane is concentrated in the pericentriolar region of the cell ( Fig. 1). This concentration of recycling membrane is a fusioncompetent, stable compartment (Yamashiro et al., 1984;Salzman and Maxfield, 1988;Marsh et al., 1995). Although the recycling compartment is in the same general region of the cell as the trans-Golgi network and the Golgi complex, the recycling compartment is distinct from these organelles (Yamashiro et al., 1984. Transfer from the pericentriolar recycling compartment to the plasma membrane is the rate-limiting step in recycling of lipids and receptors back to the cell surface, occurring with a half-time of ϳ12 min Presley et al., 1993). A similar pericentriolar recycling compartment is found in other cell lines, indicating that the pericentriolar recycling compartment is not peculiar to CHO cells (e.g. Tooze and Huttner (1990); Apodaca et al. (1994)). The transfer of recycling membrane from the pericentriolar compartment to the plasma membrane is likely to involve the formation of recycling vesicles, although there is no direct evidence for these intermediates in recycling in CHO cells.
Many vesicle-mediated membrane-trafficking steps have been reconstituted in cell-free and semi-intact cell systems (Rothman, 1994). These studies have led to the development of a model for vesicle-mediated membrane trafficking (Rothman, 1994). In this model, GTP-binding proteins and coat proteins are required for the formation of coated buds on donor membranes (Orci et al., 1986;Serafini et al., 1991). The transport vesicles are uncoated, in a process requiring GTP hydrolysis, prior to fusion with target membranes (Serafini et al., 1991). The specificity of the fusion, in part, is dependent on proteins found on transport vesicles and target membranes (v-SNAREs and t-SNAREs, respectively) (Söllner et al., 1993a;Takizawa and Malhotra, 1993). Fusion of vesicle and target membranes require the binding of NSF protein and SNAP proteins to the vand t-SNARE complex (Söllner et al., 1993b;Sü dhof et al., 1993). The interactions of SNAREs may be further modulated by small GTP-hydrolyzing Rab proteins (Pfeffer, 1992;Novick and Brennwald, 1993). A number of diverse vesicle-mediated membrane transport steps (e.g. intra Golgi, endoplasmic reticulum to Golgi, TGN to plasma membrane, synaptic vesicle release, endosome-endosome fusion) require all or some of the above mentioned factors (Gravotta et al., 1990;Miller and Moore, 1991;Ostermann et al., 1993;Peter et al., 1993). As endocytic recycling is likely to be vesicle mediated, it is reasonable to suggest that some of the proteins required for these other membrane transport processes will also be required for endocytic recycling.
To determine the biochemical requirements for endocytic recycling in CHO cells, we have modified a streptolysin O toxin (SL-O) permeabilization technique previously used to examine transport from the TGN to the cell surface of CHO cells as well as other transport steps involving the fusion of vesicles with the plasma membrane (Miller and Moore, 1991;Galli et al., 1994). SL-O permeabilization was chosen * This work was supported in part by the American Cancer Society Research Grant CB8 and the Council for Tobacco Research Grant MR2919. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by Training Grant T32 AG00189 (Cellular and Neurobiological Aspects of Aging because it forms stable pores that are large enough to allow proteins on the order of 150 kDa to pass through the plasma membrane, and most importantly SL-O toxin can be used to selectively permeabilize the cell surface (Ahnert-Hilger et al., 1989).
To assay for endocytic recycling in SL-O toxin-permeabilized cells, the release of previously internalized transferrin (Tf) from cells is monitored. Tf is an ideal tool for monitoring endocytic recycling because it remains associated with the TR until it returns to the plasma membrane (Dautry-Varsat et al., 1983;Klausner et al., 1984). Tf is internalized through clathrincoated pits into acidic endosomal components, where iron is released from Tf. Unlike other endocytosed ligands, apo-Tf (iron-free) is not released from its receptor in the acidic environment of endosomes. Apo-Tf remains bound to the TR, and this complex is recycled back to the cell surface. At the neutral pH of the extracellular environment, apo-Tf is released from the TR. Thus, the release of previously internalized Tf into the medium can be used to monitor the recycling of the TR back to the cell surface. The kinetics of Tf trafficking and the compartments through which it passes have been extensively documented in CHO cells (e.g. Dunn et al. (1989); McGraw and Maxfield (1990); Wilson et al. (1993)).
In this report, we demonstrate that the ability to reconstitute efficient Tf recycling in SL-O permeabilized cells is dependent on the addition of donor cytosol and ATP. We have conducted studies using the alkylating agent NEM and the nonhydrolyzable GTP analog GTP␥S to further explore the molecular requirements for recycling. Furthermore, we establish the utility of this assay in biochemical complementation studies using a CHO END2 mutant cell line.

EXPERIMENTAL PROCEDURES
Cell Culture-Cells were grown in either Ham's F12 medium or McCoy's 5a medium containing 5% fetal bovine serum, 2% penicillinstreptomycin (Life Technologies, Inc. or Sigma), and 220 mM sodium bicarbonate. TRVb-1 cells were grown at 37°C in a humidified atmosphere of 5% carbon dioxide in air. TRVb-1 is a CHO cell line that does not express endogenous hamster TR and has been transfected with wild-type human TR (McGraw et al., 1987). 12-4 cells are a recessive non-conditional END2 mutant CHO cell line derived from TRVb-1 cells. The isolation and characterization of 12-4 CHO cells have been reported previously (Johnson et al., 1994). The 12-4 cells were grown at 34°C.
Ligands-Human Tf was obtained from Sigma and further purified by Sephacryl S-300 gel filtration. Fe 2 Tf and diferric 125 I-Tf were prepared as described previously (Yamashiro et al., 1984). Tf was labeled with fluorescein or rhodamine (Molecular Probes) according to the manufacturer's directions.
Materials-Streptolysin O was obtained from Murex Diagnostics (Atlanta, GA). Rabbit anti-human Tf was obtained from Boehringer Mannheim. Brefeldin A was obtained from Life Technologies, Inc. All chemicals were obtained from Sigma unless otherwise stated, and all chemicals were used at reagent grade.
Tf Efflux/Permeabilized Cell Assays-The kinetic efflux assay used in this paper is a modification of an assay that has been described in detail elsewhere (McGraw and Maxfield, 1990). For each assay, cells were grown in 24-well plates to approximately 70 -80% confluence at 2 days post-plating. The buffer used to load cells to steady-state TR occupancy with 125 I-Tf is McCoy's growth media without fetal bovine serum but supplemented with 20 mM Hepes, pH 7.4 (MB buffer).
A 24-well plate of TRVb-1 CHO cells was incubated at 37°C, 5% CO 2 in MB buffer (supplemented with 3 g of 125 I-Tf/ml of MB buffer) for 1 to 2 h. The incubation medium was removed, and cells were washed once with medium 1 (150 mM NaCl, 20 mM Hepes, 1 mM CaCl 2 , 5 mM KCl, 1 mM MgCl 2 , pH 7.4). Cells were incubated for 2 min at room temperature in a mild acid buffer (0.5 M NaCl, 50 mM MES, pH 5.0), washed three times with ice-cold medium 1, and placed on ice. Incubation in mild acid buffer followed by neutral washes released surfacebound Tf (Dautry-Varsat et al., 1983). The cells were washed twice with phosphate-buffered saline (136.9 mM NaCl, 2.7 mM KCl, 8.1 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , pH 6.9) at 4°C and incubated with 200 l/well of 0.8 units SL-O/ml for 4 min on ice. Mock-treated cells were incubated in ice-cold medium 1. In a typical experiment, cells in the first row of 6 wells were left intact (mock-treated), and the other three rows were treated with SL-O. The SL-O was removed and replaced with transport buffer (78 mM KCl, 4 mM MgCl 2 , 50 mM Hepes, pH 7.2, with KOH, 2 mM DTT). The plates were transferred to a 37°C water bath for 4 min to allow pore formation. SL-O will insert into membranes at 4°C but will only form pores at 37°C, so preloading the plasma membrane with SL-O toxin at 4°C followed by removal of unbound SL-O prior to incubation at 37°C ensures that the plasma membrane is selectively permeabilized (Ahnert-Hilger et al., 1989). The cells were checked by light microscopy for morphology characteristic of SL-O permeabilization (e.g. Miller and Moore (1991)). In initial experiments, uptake of the membrane-impermeant nuclear dye, propidium iodide, was used to confirm permeabilization. Estimations using this method to monitor permeabilization showed that 95% (or greater) of the cells were permeabilized by the SL-O treatment. Following the incubation at 37°C, cells were placed on ice and incubated in fresh ice-cold transport buffer for 10 -30 min. The transport buffer was removed, and 300 l of the regeneration solutions were added to the respective wells: donor cytosol and ATP regeneration solution (1 mM ATP, 8 mM creatine phosphate, and 40 units/ml creatine phosphokinase) in transport buffer, an ATP regeneration solution in transport buffer, or transport buffer alone. Intact cells were incubated in medium 1. All regeneration solutions contained approximately 20 g/ml of unlabeled Fe 2 Tf. For most experiments, mouse liver cytosol at 3 mg/ml was used. Cells were incubated in the regenerating solutions for 5 min at 4°C, and the plate was then transferred to a 37°C water bath. The medium (effluxed Tf) was removed and collected at various time points (one well per time point, six time points per condition), and the cells were washed once in phosphate-buffered saline. This wash was pooled with the efflux medium. The cells were solubilized in 1% Triton X-100, 0.1 N NaOH. The efflux and cell-associated cpm were determined, and the rate of efflux was calculated as the slope of the line of a plot of the natural log of the fraction Tf cell-associated versus time .
It is not practical to correct each value for nonspecific radioactivity, as is commonly done for studies of intact cells. The data presented in this paper are derived from total radioactivity per well. However, in FIG. 1. Schematic of the endocytic recycling pathway in CHO cells. Apo-Tf, iron-free transferrin; MTOC, microtubule organizing center. In CHO cells, recycling membranes (receptors and lipids) are concentrated in a fusion-competent compartment in the pericentriolar region of the cell. Transport from this compartment to the cell surface is the rate-limiting step in recycling. Recycling vesicles mediating transport between this compartment and the cell surface are shown; however, these recycling vesicles have not been identified in CHO cells.
other experiments of intact cells, using the same cell line and the same batches of iodinated Tf as used for the studies presented in this report, nonspecific radioactivity (that is, not competed by incubation with a 200-fold excess of unlabeled Tf) did not account for more than 10% of the total cell-associated radioactivity. Initial experiments revealed that nonspecific counts in the SL-O permeabilized cell assay fell within 10% of total counts.
Cytosol-Mouse liver cytosol was prepared from fresh mouse liver or livers that had been stored intact in liquid nitrogen. Livers were washed in transport buffer, minced, and washed again. The buffer was decanted and the liver weighed. Fresh transport buffer containing a protease inhibitor mixture (10 M Leupeptin, 1 mM 1,10-O-phenanthroline, 0.5 mM benzamidine, 2 g/ml soybean trypsin inhibitor) was added. The volume of buffer added was equal to the weight of the liver. The livers were then homogenized in a Dounce homogenizer. The homogenate was centrifuged at low speed (20,000 ϫ g) for 15 min, and this supernatant centrifuged at 100,000 ϫ g for 1 h. The 100,000 ϫ g supernatant was assayed for protein concentration using the BCA method with bovine serum albumin as a standard (Pierce). The cytosol was frozen in a dry ice/ethanol bath and stored in liquid nitrogen. The average concentration of liver cytosol was approximately 30 mg/ml. Sephadex G-25 M (Pharmacia Biotech Inc.) columns were used for gel filtration of cytosol. In additional experiments, the cytosol was dialyzed to remove small molecules. 10 ml of 10 mg/ml mouse liver cytosol was dialyzed in dialysis tubing with a molecular mass cut off of 12-14 kDa (Spectra/Por 2) for 24 h at 4°C against 1 liter of transport buffer. The buffer was changed once during dialysis.
Immunoprecipitation-100 l of the efflux medium was taken from the 20-min time point of mock-treated, complete, and buffer-treated permeabilized cells. Anti-transferrin antibody was added, and the aliquots were incubated for 1 h at 4°C, with end-over-end mixing. 100 l of Sepharose A beads were added and incubated with the efflux-antibody complex overnight at 4°C. The solution was spun at 15,000 rpm for 1 min to pellet the beads. The supernatant and the pellet were separated and counted. Free Tf in the efflux was determined by calculating the percentage of counts in the pellet. Approximately 90% of counts (Tf) was pelletable in both intact and permeabilized cells.
Fluorescence Microscopy-TRVb-1 cells were plated in coverslip dishes 2 days prior to use. Coverslip dishes are constructed by punching a hole in a 35-mm dish and attaching a glass coverslip to the bottom using paraffin wax. The cells were incubated in 10 g/ml fluorescein-Tf (or Cy3-Tf), 2.0 mg/ml ovalbumin in MB buffer for 1-2 h. SL-O permeabilization and reconstitution of recycling were performed exactly as described above. The cells were allowed to efflux for 20 min at 37°C. The cells are washed six times in cold medium 1 and fixed in 3.7% formaldehyde for 20 min.

Reconstitution of Tf Recycling in SL-O Permeabilized Cells-
CHO cells were incubated with iodinated Tf to achieve steadystate TR occupancy. Unbound and surface-bound Tf was removed, and the plasma membrane was permeabilized with SL-O (see "Experimental Procedures"). Recycling was monitored by following the loss of Tf from cells and the concomitant appearance of Tf in the medium. The results of a representative experiment are shown in Fig. 2. Addition of donor cytosol is required for efficient reconstitution of recycling. In intact cells and in permeabilized cells incubated with donor cytosol, the amount of cell-associated Tf decays, and the amount of Tf in the medium increases as a function of time at 37°C. Incubation of permeabilized cells with 3 mg/ml donor cytosol, ATP, and an ATP-regenerating system reconstitutes Tf recycling to near the level observed in intact cells (compare Fig. 2, A and B). Tf recycling from cells incubated in transport buffer containing ATP and no donor cytosol was only slightly more efficient than incubation in transport buffer alone (Fig. 2,C and D).
In this procedure, the SL-O toxin was bound to the cells at 4°C, and excess SL-O was removed before warming the cells to 37°C. The technique ensures that the plasma membrane is selectively permeabilized and that intracellular organelles are left intact (Ahnert-Hilger, et al., 1989). To confirm the integrity of internal membranes, SL-O permeabilized cells incubated with cytosol, ATP, and an ATP regenerating system were stained using wheat germ agglutanin (WGA) (Fig. 3). WGA binds with high affinity to high molecular weight acidic glycoproteins found primarily on the plasma membrane, in the lumen of endosomal compartments and Golgi, and to some extent in the nuclear envelope (Fitzgerald et al., 1981;Raub et al., 1990). To determine accessibility of internal membrane compartments to rhodamine-WGA (Rh-WGA), two different experiments were performed. In one, following a 15-min, 37°C incubation with cytosol and ATP, SL-O permeabilized cells were incubated with 1.0 g/ml Rh-WGA on ice for 30 min (Fig.  3B). In the other, 1.0 g/ml Rh-WGA was added to the cytosol during the 15-min incubation at 37°C (Fig. 3C). In a control sample, SL-O permeabilized cells following a 15-min incubation at 37°C with cytosol and ATP were fixed and permeabilized with detergent (saponin, 100 g/ml) and then incubated with Rh-WGA. In detergent-permeabilized cells, WGA stained the plasma membrane and intracellular membrane compartments (Fig. 3A). In SL-O permeabilized cells not treated with detergent, WGA stained the plasma membrane and the nuclear envelope; however, it does not stain intracellular membrane compartments (Fig. 3, B and C). This result demonstrates that the SL-O permeabilization procedure does not permeabilize internal membrane compartments. Reconstituted Tf recycling is also blocked by incubation at 4°C, regardless of the addition of cytosol or ATP (not shown). This result is consistent with the 4°C block of Tf recycling in intact cells. To demonstrate that the Tf released in the medium is not attached to membrane sheets, the medium was collected following a 15-min efflux and was spun for 1 h at 100,000 ϫ g. Approximately 90% of the counts released from intact cells were in the supernatant, and approximately 87% of the counts released from SL-O permeabilized cells, incubated in cytosol and ATP, were in the supernatant. Furthermore, the Tf released from SL-O permeabilized cells can be precipitated with anti-Tf antibody (not shown). In aggregate, these results demonstrated that the radioactivity in the medium was Tf released from cells by fusion of recycling vesicles with the plasma membrane, since if intact vesicles containing Tf were released by lysis of cells, the Tf in the lumen of the vesicles, would not be accessible to the antibody. In panels B through D, the data shown are from SL-O permeabilized cells incubated in 3 mg/ml mouse liver cytosol and ATP-regenerating system in transport buffer (B), an ATP-regenerating system in transport buffer (C), or transport buffer alone (D). The total cpm per time point (efflux plus cell associated) did not vary more than 18% within a given condition.
The loss of cell-associated Tf from CHO cells fits a single exponential decay. The recycling rate constant is the slope of a plot of the natural log of the fraction of Tf cell-associated versus time. In all permeabilized cell conditions, the fraction of Tf that is cell-associated at time zero is similar to that found in intact cells, demonstrating that SL-O permeabilization does not result in a significant aberrant loss of Tf from cells, as would be expected if the permeabilization was destroying the integrity of the recycling compartment (Fig. 4A). The data of Fig. 2 plotted to calculate the recycling rate constant are shown in Fig. 4A. In this experiment, the rate of Tf efflux in the presence of cytosol is 70% of the intact rate, and that of incubation with ATP or transport buffer alone is 34 or 21% of the intact rate, respectively. A summary of a number of experiments is shown in Fig.  4B. SL-O permeabilized cells recycle Tf at ϳ80% of the intact rate when incubated with 3 mg/ml donor cytosol and ATP, ϳ35% when incubated with ATP, and ϳ25% when incubated with transport buffer alone. Both the initial rate of Tf recycling and the extent of Tf released from the cells is dependent on addition of donor cytosol and ATP to the permeabilized cells (Fig. 4C). After 20 min of incubation, little further Tf is released from either permeabilized or intact cells. In both intact cells and permeabilized cells incubated with donor cytosol, about 20% of the total Tf remains internal after a 60-min incubation at 37°C. An explanation for this plateau at 20% is that the data are not corrected for nonspecific Tf binding, which is typically 10% of the total cpm bound (see "Experimental Procedures"). Regardless of the reason for the plateau at 20%,

FIG. 3. Integrity of internal membranes is maintained in SL-O permeabilized cells.
In panels A through C, cells were permeabilized with SL-O and incubated in cytosol, ATP, and an ATP regenerating system followed by 15 min at 37°C to allow recycling. In panel A, the cells were fixed and treated with Rh-WGA in the presence of detergent (100 g/ml saponin) for 30 min. In panel B, the cells were returned to 4°C, incubated in Rh-WGA in transport buffer for 30 min, washed for 10 min in transport buffer to remove excess label, and fixed. In panel C, Rh-WGA was added to the cytosol solution during the 37°C incubation. At the end of the assay, the cells were returned to 4°C, washed for 10 min in transport buffer, and fixed. Panel D is a schematic of the optical section photographed in these experiments. The plasma membrane appears as a ring at the optical section (solid arrows). The edge of the cell at its bottom was only visible if planes below the optical section Tf recycling from permeabilized cells incubated with ATP and donor cytosol is similar to recycling in intact cells.
Reconstituted Tf Recycling Is Dependent on the Concentration of Donor Cytosol-To further define the requirement for cytosol in the reconstitution of Tf recycling in SL-O permeabilized cells, a range of concentrations of donor cytosol was examined. The rate of Tf recycling was dependent on the concentration of donor cytosol added to SL-O permeabilized cells (Fig.  5). The data in Fig. 5 were collected using mouse liver donor cytosol. A similar dependence on the cytosol concentration was observed when CHO cytosol was used (data not shown). We chose to use mouse liver cytosol because larger quantities of cytosol can be made in a single preparation. The complementing activity supplied by the cytosol is heat labile, as boiling the cytosol destroys the activity. The required activity cannot be mimicked by addition of bovine serum albumin, demonstrating that a specific activity is provided by the cytosol. Furthermore, removing molecules of less than 5000 Da by gel filtration of the cytosol does not remove the complementing activity, demonstrating that cytosolic factor(s) other than additional free nucleotides are required for efficient recycling.
Cytosol Provides an Activity Required for Transport of Tf from the Pericentriolar Recycling Compartment-Monitoring reconstituted Tf recycling by following the release of iodinated Tf into the medium is ultimately dependent on the fusion of recycling vesicles with the plasma membrane. Thus, the biochemical assay cannot be used to study the individual recycling steps that precede fusion of recycling vesicles with the plasma membrane. However, by monitoring recycling using fluorescent microscopy, transport of Tf from the pericentriolar recycling compartment to the plasma membrane can be distinguished from fusion of recycling vesicles with the plasma membrane.
The efflux of fluorescent Tf during the basic incubation conditions is shown in Fig. 6. After a 20-min incubation, most of the Tf has been released from intact cells, with the residual fluorescence visible in the pericentriolar region (Fig. 6, A and  B). SL-O permeabilization does not alter the morphology of the Tf-containing pericentriolar recycling compartment (Fig. 6C). Consistent with the biochemical studies, most of the internal Tf is released from SL-O permeabilized cells during a 20-min incubation with donor cytosol and ATP (Fig. 6, C and D). Unlike intact cells, the residual Tf fluorescence in SL-O permeabilized cells incubated with cytosol and ATP is not solely concentrated in the pericentriolar region but is also found in a punctate pattern distributed throughout the cell (Fig. 6, B and D). In SL-O permeabilized cells incubated for 20 min with ATP alone or in transport buffer alone, fluorescent Tf remains in the pericentriolar region (Fig. 6, E and F). These results demonstrate that a factor provided by the donor cytosol is required for transport of Tf from the pericentriolar recycling compartment to the plasma membrane.
Mouse Liver Donor Cytosol Does Not Restore Wild-type Tf Recycling to END2 Cells-The SL-O permeabilized cell recycling assay can be used to study cell lines defective in Tf recycling. Mutant CHO cells of the END2 complementation group recycle Tf at half the rate of wild-type cells Johnson et al., 1994). Tf recycling was reconstituted in a SL-O permeabilized recessive END2 mutant cell line. Addition of 3.0 or 6.0 mg/ml donor mouse liver cytosol (Fig. 7) or wildtype CHO cytosol (not shown) reconstituted Tf recycling to the level of intact END2 cells but not to the wild-type recycling rate. Thus, wild-type donor cytosol was unable to complement the recycling defect in END2 cells. Confirming that the defective protein in END2 cells responsible for slowed Tf recycling is not exchangeable through the SL-O pores is the finding that cytosol prepared from END2 cells supports Tf recycling in wild-type cells to the same extent as does wild-type cytosol (data not shown). These results demonstrate that the protein in END2 cells responsible for the slowed recycling is either membrane-associated or is too large to diffuse through the SL-O FIG. 5. Reconstituted Tf recycling is dependent on the concentration of donor cytosol. The average recycling rate constants as a function of donor cytosol are presented. The data presented represent experiments performed with mouse liver cytosol. Plus (ϩ) ATP denotes incubation with 1 mM ATP and an ATP-regenerating system (see "Experimental Procedures"). The data for the bar labeled "boiled" were collected using cytosol that was boiled for 5 min prior to use in the assay. The data for the bar labeled "gel filtered" were collected using cytosol that had been passed over a gel filtration column (PD-10, Pharmacia) to exclude molecules less than 5000 Da. Bovine serum albumin was added and used at a concentration 3 mg/ml in place of donor cytosol. pores. The observation that the slowed Tf recycling phenotype of intact END2 cells is observed in the permeabilized cells provides additional evidence that the recycling kinetics measured in the permeabilized cells reflects recycling kinetics in intact cells.
NEM Treatment of Permeabilized Cells Inhibits Tf Recycling-Previous studies of various membrane-trafficking steps reconstituted in cell-free and permeabilized cell systems have used mild NEM treatment to establish a requirement for the cytosolic fusion factor, NSF (Block et al., 1988;Beckers and Balch 1989;Diaz et al., 1989;Sü dhof et al., 1993). Since the release of Tf from cells requires the fusion of recycling vesicles with the plasma membrane, it is likely that recycling also requires the activity of NSF. Pretreatment of donor cytosol with NEM does not significantly inhibit cytosol-dependent recycling (Fig. 8A). Furthermore, preincubation of cytosol at 37°C for 30 min in the absence of ATP, a treatment known to inactivate NSF (Block et al., 1988;Goda and Pfeffer, 1991), does not inactivate the required activity provided by the donor cytosol (Fig. 8A). These results indicate that the activity provided by donor cytosol is not inactivated by mild NEM treatment and that the donor cytosol does not provide NSF.
SL-O permeabilized cells were treated with NEM, and untreated donor cytosol was added to determine if an NEMsensitive factor(s) required for Tf recycling remained associated with the permeabilized cells. NEM treatment of permeabilized cells significantly inhibited cytosol-dependent recycling, reducing the rate by ϳ70% (Fig. 8B). This demonstrates a requirement for a NEM-sensitive factor that remains associated with cells following SL-O permeabilization. Complete inhibition of cytosol-supported Tf recycling is achieved by pretreating the SL-O permeabilized cells for 4 min at 37°C with 2 mM NEM, whereas identical treatment of donor cytosol only partially inhibits its ability to support recycling (Fig. 8B). These results demonstrate that Tf recycling requires a cell-associated NEMsensitive factor, and although they do not demonstrate a role for NSF, they are consistent with a requirement for NSF in Tf recycling.
In NEM-treated Cells, Tf Is Trapped in Vesicles That Accumulate Near the Plasma Membrane-Fluorescence microscopy was used to determine which steps of Tf recycling are inhibited by NEM treatment. In NEM-treated permeabilized cells, following a 20-min incubation at 37°C with donor cytosol and ATP, fluorescent Tf was found distributed between a punctate pattern near the periphery of the cells and the pericentriolar recycling compartment (Fig. 9). This distribution was indistinguishable from the distribution of Tf in intact cells treated with NEM (Fig. 9). The accumulation of fluorescent Tf in punctate structures near the plasma membrane suggests a NEM block in the fusion of recycling vesicles with the plasma membrane and thus is consistent with a requirement for NSF in Tf recycling. The persistence of Tf in the pericentriolar recycling compartment following NEM treatment indicates that exit of Tf from this compartment is also inhibited by NEM treatment.
Reconstituted Tf Recycling Is Stimulated by GTP␥S-Small GTP-hydrolyzing proteins and heterotrimeric G proteins are involved in regulating many membrane-trafficking steps (for reviews, see Bomsel and Mostov (1992); Pfeffer (1992); Zerial and Stenmark (1993); Novick and Brennwald (1993)). The ef- The data were then analyzed to determine the percent inhibition. Percent inhibition of Tf recycling supported by cytosol and ATP was calculated as described previously (Beckers et al., 1989). In this analysis, 100% inhibition represents the rate supported by addition of ATP alone, and 0% inhibition represents the rate supported by the addition of 3 mg/ml donor cytosol and ATP (i.e. complete conditions). Thus, percent inhibition represents inhibition of Tf recycling promoted by addition of donor cytosol. In panel A, the results of NEM treatment of donor cytosol are presented. The results presented in the bar labeled pretreatment at 37°C with no ATP were collected from reactions in which the cytosol was preincubated at 37°C in the absence of ATP for 30 min at 37°C. This treatment inactivates NSF (Block et al., 1988). This cytosol was supplemented with ATP and an ATP regenerating system when added to the permeabilized cells. Panel B represents the results of experiments where the permeabilized cells were treated with NEM for 15 min at 4°C. The NEM was quenched by incubation for 15 min at 4°C in transport buffer supplemented with DTT (at 2ϫ the concentration of NEM). The percent inhibition of recycling is shown for cells treated with 1 or 2 mM NEM and cells pretreated with NEM at 37°C for 4 min followed by 15 min at 4°C in transport buffer containing DTT. fect of GTP␥S on reconstituted recycling was examined to establish a requirement for GTP-hydrolyzing proteins in Tf recycling. Addition of GTP␥S to permeabilized cells incubated with cytosol and ATP had no effect on the rate of Tf recycling (Fig.  10A). Thus, hydrolysis of GTP is not required for Tf recycling. For these experiments, molecules less than 5000 Da were removed from the donor cytosol by gel filtration; therefore the failure of GTP␥S to inhibit recycling was not due to competition from GTP in the cytosol preparations. To confirm that free nucleotides in the donor cytosol were not competing with the added GTP␥S and thereby blocking effects of GTP␥S on recycling, experiments were conducted with mouse liver cytosol that was extensively dialyzed to remove molecules less than 12,000 Da. As is the case with the gel-filtered cytosol, no stimulatory or inhibitory effect was observed when either 50 M (Fig. 10A) or 200 M (not shown) GTP␥S was added to reconstituted recycling reactions containing dialyzed cytosol.
GTP␥S stimulated Tf recycling in SL-O permeabilized cells when added to reactions containing suboptimal concentrations of donor cytosol (Fig. 10A). Tf recycling was stimulated by ϳ60% when 50 M GTP␥S was added to permeabilized cells in the absence of donor cytosol. The stimulatory effect of GTP␥S decreased with increasing amounts of cytosol. Stimulation of recycling was observed over a range of GTP␥S concentrations, from 50 to 200 M GTP␥S (not shown).
The stimulation of Tf recycling by GTP␥S when added to the reconstitution system in the absence of donor cytosol is blocked by addition of an excess of GTP. A 5-fold excess of GTP only partially reduces the stimulation observed in the absence of added GTP, whereas a 10-fold excess of GTP is required to significantly diminish the GTP␥S stimulation of recycling (Fig. 10B).
GTP␥S stimulates endosome-endosome fusion in suboptimal cytosol concentrations . It is proposed that GTP␥S promotes the association of a member of the ADPribosylation factor family (ARF) with endosome membranes and that the membrane-bound ARF recruits other soluble proteins required for fusion (Lenhard et al., 1992;Colombo et al., 1992). GTP␥S stimulation of recycling could occur by a similar mechanism. The fungal metabolite, brefeldin A, blocks exchange of GTP for GDP on ARF and consequently inhibits association of ARF to membranes (Donaldson et al., 1992;Helms and Rothman, 1992;Dascher and Balch, 1994). If the effect of GTP␥S on recycling in suboptimal donor cytosol concentrations is mediated by an ARF protein, then preincubation of permeabilized cells with brefeldin A would block GTP␥S stimulation of recycling. Cells were preincubated with radioactive Tf for 1.5 h with 5 g/ml brefeldin A included during the last 15 min of this incubation and in all subsequent incubations. The cells were permeabilized and the effect of GTP␥S on recycling determined. Brefeldin A did not affect GTP␥S stimulation of recycling (Fig. 10A), although it did cause tubulation of the Tf-containing recycling compartment in CHO cells . These results suggest that GTP␥S stimulation of Tf recycling in suboptimal cytosol concentrations is not mediated by a brefeldin A-sensitive member of the ARF family.
To determine whether the GTP␥S stimulation of recycling in suboptimal cytosol concentrations is dependent on the microtubule cytoskeleton, microtubules were depolymerized with nocodazole, and recycling in permeabilized cells was measured. Depolymerization of microtubules did not affect the cytosol-dependent reconstituted recycling (not shown) nor did it affect the GTP␥S stimulation of recycling in suboptimal cytosol concentrations (Fig. 10A). These results are in agreement with studies of intact CHO cells, where it has been shown that recycling is not dependent on an intact microtubule cytoskeleton . DISCUSSION In this report, we demonstrate that efficient Tf recycling from SL-O permeabilized CHO cells is reconstituted when the cells are incubated with donor cytosol and ATP. The rate of reconstituted recycling is dependent on the concentration of cytosol added to the permeabilized cells. Fluorescent microscopy analysis of reconstituted recycling demonstrates that donor cytosol provides an activity required for efficient transport of Tf from the pericentriolar recycling compartment. Addition of ATP to permeabilized cells in the absence of donor cytosol promotes only a modest increase in recycling over that achieved by incubation in transport buffer alone. The results of this condition (ATP only) serve as an internal control for the degree and extent of SL-O permeabilization in this system. The reconstitution of Tf recycling in SL-O permeabilized CHO cells supplemented with only ATP was previously reported (Galli et al., 1994). In that study, the requirement for cytosol in reconstitution of recycling was not investigated. Our observation that efficient recycling is dependent on donor cytosol is in agreement with results of a study of Tf recycling to the basolateral membrane in semi-intact Madin-Darby kidney cells (Pod- FIG. 9. CHO cells were labeled to steady state with rhodamine-Tf. SL-O permeabilized and intact (panels A and B, respectively) cells were treated with 1 mM NEM for 15 min at 4°C followed by 15 min in 2 mM DTT at 4°C, or SL-O permeabilized and intact cells were incubated in transport buffer for 30 min at 4°C (panels C and D, respectively). Intact cells (B and D) were then incubated in transport buffer with Fe 2 Tf and SL-O permeabilized cells (A and C) were incubated in 3 mg/ml cytosol and ATP for 5 min at 4°C. The cells were warmed to 37°C 20 min as in the normal recycling assay. The small arrows in panels A and B indicate Tf containing vesicles near the plasma membrane, and the large arrow indicates Tf in the pericentriolar recycling compartment. bilewicz and Mellman, 1990).
The slowed Tf recycling characteristic of CHO mutants of the END2 complementation group can be recapitulated in permeabilized END2 cells. This result provides evidence that the system is faithfully reconstituting Tf recycling. As in wild-type cells, cytosol stimulates reconstituted recycling in END2 cells above the rate supported by ATP alone. Donor cytosol from mouse liver or wild-type CHO cells restores Tf recycling to the slow rate characteristic of intact END2 cells. These "wild-type" donor cytosols, however, are unable to restore wild-type rates of recycling to permeabilized END2 cells. These results indicate that the factor responsible for the reduced rate of Tf recycling in mutants of the END2 complementation group is membrane associated or is too large to diffuse through the SL-O pores.
Tf recycling is inhibited by mild NEM treatment of permeabilized cells. Finding Tf in a punctate distribution near the plasma membrane following mild NEM treatment of permeabilized cells is consistent with a block in fusion of recycling vesicles with the membrane and consequently is consistent with NSF being at least one of the factors inactivated by the mild treatment of cells with NEM. Tf recycling is partially inhibited when the v-SNARE, cellubrevin, is cleaved by tetanus toxin, which is consistent with a role for NSF in Tf recycling (Galli et al., 1994). The failure of untreated donor cytosol to rescue the inhibition in recycling caused by NEM treatment of permeabilized cells may be because the NSF trimer is too large to efficiently diffuse through the SL-O pores, either in or out of the permeabilized cells, or that the NSF required for recycling may be associated with the membrane. In this regard, it has been suggested that NSF is required for vesicle formation (Wattenberg et al., 1992). Although our findings are consistent with a requirement for NSF in recycling, we cannot rule out the possibility that NSF is not required and the mild NEM treatment is inactivating other proteins required for recycling. For example, endosome-TGN transport involves a protein other than NSF that is sensitive to NEM (Goda and Pfeffer, 1991). In this regard, it is interesting that mild NEM treatment inhibits transport of Tf from the pericentriolar recycling compartment. This block could be due to NEM inactivation of a protein other than NSF. Rigorous demonstration that NSF is necessary for Tf recycling requires the use of reagents that specifically inactivate NSF (e.g. antibodies). The recent finding that apical transport in Madin-Darby kidney cells does not require NSF provides precedent for NSF-independent fusion with the plasma membrane (Ikonen, et al., 1995).
Reconstituted Tf recycling is not inhibited by GTP␥S, and therefore, hydrolysis of GTP is not required for Tf recycling. The failure of GTP␥S to inhibit recycling distinguishes the mechanism for recycling from the mechanism for transport from the TGN to the cell surface because the latter is inhibited by GTP␥S (Miller and Moore, 1991). In light of the findings that G proteins of the Rab and ARF families and heterotrimeric G proteins are involved in intracellular trafficking of membrane proteins, it is surprising that GTP␥S does not affect reconstituted recycling supported by donor cytosol (van der Slujis et al., 1992;D'Souza-Schorey et al., 1995;Bomsel and Mostov, 1992). The recycling assay described in this report measures transport from the pericentriolar recycling compartment to the plasma membrane (Fig. 1). It may be that GTP hydrolysis is required in a earlier step in the endocytic pathway, such as transport from sorting endosomes to the recycling compartment. GTP␥S inhibition of trafficking steps preceding transport from the pericentriolar recycling compartment would not be detected in this assay because at steady-state there is approximately five times as much Tf in the pericentriolar recycling compartment as in sorting endosomes . Additionally, this assay may not be sensitive to GTP␥S effects on components required to support multiple rounds of recycling vesicle formation and consumption. Regardless, it is interesting that the overall rate-limiting step in the constitutive recycling of membrane back to the cell surface does not require GTP hydrolysis. Future studies using the reconstituted recycling assay will provide insight into the molecular mechanism of regulating recycling.
Stimulation of fusion with the plasma membrane is commonly observed for regulated secretion events (Gomperts, 1990). GTP␥S stimulates Tf recycling in suboptimal cytosol concentrations. GTP␥S may promote the association of soluble . The values compared percent stimulation calculated from the rates of matched experiments. Percent stimulation is calculated as the rate of efflux in the presence of GTP␥S divided by the rate of efflux of the same condition (concentration of cytosol) in the absence of GTP␥S. Additionally displayed in this panel is the pretreatment with BFA. 5 g/ml BFA was added to the 125 I-Tf incubation media (see "Experimental Procedures") 15 min prior to the start of the assay. BFA was maintained at 5 g/ml throughout the remainder of the assay to prevent recovery from any effects of BFA. BFA treatment did not exceed 2 h total (time at 37 and 4°C) and did not exceed 1 h at 37°C. The last bar of the graph represents experiments conducted after first incubating the cells in 10 g/ml nocodazole. The nocodazole was added to the 125 I-Tf incubation media for 1 h and was maintained throughout the assay. For all experiments where GTP␥S was added to cytosol, the cytosol was first passed over a gel filtration column to remove excess free nucleotides. B, competition curve of GTP to 50 M GTP␥S. Increasing concentrations of GTP was added to an ATP-regenerating system, and 50 M GTP␥S, which was then added to cytosol-depleted SL-O permeabilized cells. The rate of efflux in experimental conditions is compared to cells that were treated with only an ATP regenerating system. Percent stimulation was calculated as in A. Values represent the average of n Ն 2 Ϯ S.E. proteins required for recycling to the recycling compartment, and hence, in limiting amounts of added donor cytosol, GTP␥S would increase recycling. Although the permeabilized cells are preincubated to allow soluble proteins to diffuse from cells, some fraction of the cytosolic proteins will remain inside the cells. Finding GTP␥S has its greatest effect on recycling in the absence of donor cytosol is consistent with the interpretation that GTP␥S is promoting the association of soluble proteins with the recycling compartment membrane. Our results are similar to those described for endosome-endosome fusion . GTP␥S stimulates endosome-endosome fusion at suboptimal cytosol concentrations in a cell-free assay . It is proposed that GTP␥S promotes the recruitment of ARF and other soluble proteins to endosome membranes (Lenhard et al., 1992;Colombo et al., 1992). ARF proteins are required for the binding of coat proteins to membranes (Donaldson et al., 1992;Helms and Rothman, 1992;Stamnes and Rothman, 1993;Robinson and Kreis, 1992) and have been implicated in a number of membrane transport processes (Balch et al., 1992;Lenhard et al., 1992). Brefeldin A blocks exchange of GTP for GDP on ARF and blocks ARF association with membranes (Donaldson et al., 1991;Helms and Rothman, 1992). GTP␥S stimulation of recycling in suboptimal concentrations of donor cytosol is not affected by brefeldin A, suggesting that the stimulatory effect of GTP␥S is not due to an increase in an ARF association with recycling membrane. In future studies, the reconstitution system will be used to explore the role of G proteins in regulating endocytic recycling.
A trivial explanation for the failure of GTP␥S to effect recycling in the presence of cytosol is that free nucleotides added with the donor cytosol block the effect of GTP␥S. This is extremely unlikely. First, in other studies the effects of GTP␥S have been seen using similar gel filtration techniques for removing free nucleotides from the cytosol (Melancon et al., 1987;Mayorga et al., 1989;Carter et al., 1993) Second, we have demonstrated that a 10-fold excess of GTP is required to compete with the GTP␥S. Thus, the gel-filtered or dialyzed cytosol would have to provide approximately 2 mM GTP to effectively compete 200 M GTP␥S. It is unlikely that the concentrations of free nucleotides are this high in the gel-filtered or dialyzed cytosol.
Reconstituted Tf recycling shares some of the characteristics of other membrane-trafficking processes: inhibition by NEM, dependence on ATP and donor cytosol, and partial inhibition by tetanus toxin cleavage of cellubrevin (Galli et al., 1994). Unlike many of these other processes, cytosol-dependent Tf recycling is not inhibited by GTP␥S. This finding may point to important mechanistic differences among various membrane transport steps. In this regard, it is of interest to note that the recent findings that demonstrate apical and basolateral transport in Madin-Darby kidney cells have different molecular requirements (Ikonen et al., 1995)