Bafilomycin A 1 Treatment Retards Transferrin Receptor Recycling More than Bulk Membrane Recycling*

Treatment of Chinese hamster ovary cells with the vacuolar proton pump inhibitor bafilomycin A 1 causes a 2-fold retardation in the rate of recycling of transfected human transferrin receptors back to the cell surface as measured using biochemical assays (Johnson, L. S., Dunn, K. W., Pytowski, B., and McGraw, T. E. (1993) Mol. Biol. Cell 4, 1251–1266). We have used quantitative fluo- rescence microscopy to determine which step(s) in the endocytic recycling pathway are affected. We show that removal of transferrin from sorting endosomes and accumulation in the peri-centriolar endocytic recycling compartment takes place normally in bafilomycin A 1 treated cells. However, the rate constant for exit of transferrin receptors from recycling endosomes ( k e ) is reduced from 0.063 min (cid:50) 1 in untreated cells to 0.034 min (cid:50) 1 in the presence of bafilomycin A 1 . This retardation appears to be dependent on the presence of inter- nalization motifs in the cytoplasmic domain since mod-ified receptors lacking these oligopeptide motifs do not show as large a decrease in recycling rate in the presence of bafilomycin A 1 . Bulk membrane recycling (meas- ured by efflux of an internalized fluorescent lipid analog, 6-[ N -[7-nitrobenzo-2-oxa-1,3-diazol-4-yl]-amino]- hexoyl-sphingosylphosphorylcholine)


Treatment of Chinese hamster ovary cells with the vacuolar proton pump inhibitor bafilomycin
Biol. Cell 4, 1251-1266). We have used quantitative fluorescence microscopy to determine which step(s) in the endocytic recycling pathway are affected. We show that removal of transferrin from sorting endosomes and accumulation in the peri-centriolar endocytic recycling compartment takes place normally in bafilomycin A 1treated cells. However, the rate constant for exit of transferrin receptors from recycling endosomes (k e ) is reduced from 0.063 min ؊1 in untreated cells to 0.034 min ؊1 in the presence of bafilomycin A 1 . This retardation appears to be dependent on the presence of internalization motifs in the cytoplasmic domain since modified receptors lacking these oligopeptide motifs do not show as large a decrease in recycling rate in the presence of bafilomycin A 1 . Bulk membrane recycling (measured by efflux of an internalized fluorescent lipid analog, 6-[N-[7-nitrobenzo-2-oxa-1,3-diazol-4-yl]-amino]hexoyl-sphingosylphosphorylcholine) is slowed from an exit rate constant of 0.060 min ؊1 without drug to 0.046 min ؊1 in the presence of bafilomycin A 1 . We conclude that bafilomycin A 1 slows bulk membrane flow, but it causes additional inhibition of receptor recycling in a manner that is dependent on a peptide motif on the cytoplasmic domain.
After internalization via clathrin-coated pits, endocytosed material rapidly enters sorting endosomes, organelles that sort recycled molecules such as transferrin (Tf) 1 bound to the transferrin receptor (Tf-R) from lysosomally destined molecules such as low density lipoprotein (1)(2)(3). We have provided evidence for a sorting mechanism in which components not carrying special retention or targeting signals are removed from the sorting endosome by a default process based on the geometry of the sorting endosome and an iterative fractionation mechanism (1,3). Most of the fluid brought into the tubulo-vesicular sorting endosome fills a spherical lumen, whereas most of the surface area is on tubular extensions that are involved in the removal of membrane components from the sorting endosome for export along the recycling pathway (1)(2)(3). Tf remains attached to its receptor throughout the recycling pathway, but it is converted to apo-Tf by the low pH of endosomes (4,5). Occupied Tf-R are transported from sorting endosomes to the endocytic recycling compartment, which in CHO cells is a collection of tubules concentrated near the centriole (6). The Tf-R are returned to the cell surface from the endocytic recycling compartment, and apo-Tf dissociates from the receptor upon return to the surface (5). It has been shown that internalized lipid analogs such as 6-[N-[7-nitrobenzo-2-oxa-1,3-diazol-4-yl]-amino]hexoyl-sphingosylphosphorylcholine (C 6 -NBD-SM) follow the same endocytic recycling pathway as Tf, indicating that in the absence of specialized signals membrane molecules will recycle efficiently from endosomes to the cell surface (3). Endocytosed fluid and ligands released from their receptors by low endosomal pH are retained in the lumen of the sorting endosome and accumulate. The vesicular portions of the sorting endosomes become the vesicles that fuse with late endosomes (7)(8)(9).
Acidification of endosomes by ATP-dependent proton pumps is important for several aspects of endocytic trafficking. Two well established functions of the low pH of endosomes are to dissociate lysosomally destined ligands from membrane-associated receptors and to strip Fe 3ϩ from transferrin (4,5,10,11). Two additional roles for endosome acidification have been proposed based on studies of drug-treated cells or mutant cell lines, but these are not well understood at present. Inhibition of endosome acidification has been associated with slowed endocytic recycling and intracellular retention of recycling receptors (12)(13)(14). In addition, the delivery of endocytosed molecules to late endosomes or lysosomes is affected by treatments that impair acidification (15,16). Unfortunately, both of these effects appear to vary among cell lines.
Several studies have shown that weak bases and ionophores that collapse transmembrane pH gradients can reduce the fraction of recycling receptors that are expressed on the cell (17)(18)(19)(20). However, the degree of inhibition of receptor trafficking varies greatly from study to study. Since weak bases and ionophores have multiple effects, including osmotic effects on organelles, interpretation of these experiments has not been straightforward (21). In 12-4 cells, a CHO mutant line in the end2 complementation group that has partially defective acidification of endosomes, Tf exits the sorting endosome at a normal rate, but it exits the recycling compartment more slowly than in the parental cells (22). However, the rate of bulk membrane recycling, measured by the release of internalized C 6 -NBD-SM from the cells, is identical to the rate in the parental cells, suggesting that Tf is retained in the recycling compartment by a specific retention mechanism.
Recently, it was shown that the proton pump inhibitor bafilomycin A 1 raised the pH of sorting and recycling endosomes and slowed recycling of the Tf-R in a Chinese hamster ovary cell line transfected with the human Tf-R (TRVb-1) (14). Internalization of the Tf-R was not affected, but externalization was slowed 2-fold, leading to a net intracellular accumulation of Tf-R.
The effects of bafilomycin A 1 on delivery of molecules to late endosomes and lysosomes have also been investigated (15,16). In a study of BHK cells it was found that delivery to late endosomes was blocked by treatment with bafilomycin A 1 , and in cell homogenates formation of multivesicular body/endosomal carrier vesicles was inhibited by bafilomycin A 1 (15). These multivesicular body/endosomal carrier vesicles are the intermediates between early sorting endosomes and late endosomes (9). In contrast, in HepG2 cells, delivery to late endosomes was slowed somewhat, but delivery to lysosomes was nearly completely blocked (16). Neither the basis for these differences nor the molecular role of endosome acidification in these processes is known. These discrepancies illustrate that even the organelles where bafilomycin A 1 exerts its effects are not well understood.
In this paper, we examine the effects of bafilomycin A 1 on bulk membrane traffic and on normal and mutant Tf-R recycling in wild-type and TRVb-1 cells. Using digital fluorescence microscopy we find that in bafilomycin A 1 -treated cells, Tf-R leaves sorting endosomes with normal kinetics, but its exit from the recycling compartment is slowed similar to the observations in 12-4 cells (22). However, we also find that bafilomycin A 1 treatment causes bulk membrane recycling to be significantly slowed. The slowing of membrane recycling accounts for about one-third of the reduction in the rate of Tf-R recycling.

EXPERIMENTAL PROCEDURES
Materials-Bafilomycin A 1 , obtained from Dr. K. Altendorf (University of Osnabrü ck, Germany), was dissolved in Me 2 SO to make a stock solution that was frozen in aliquots. Anti-Tf antibody (B3/25) was obtained from Hybritech Inc. (San Diego, CA). Goat anti-mouse FITCconjugated antibodies were obtained from Pierce. Human transferrin (Sigma) was iron loaded and purified as described previously (6). Transferrin was conjugated to FITC (FITC-Tf) as described previously (6) and to Texas Red (Tx-Tf) (Molecular Probes Inc., Eugene, OR) according to the manufacturer's instructions. F-R-Tf was made as described previously (23) by reacting iron loaded Tf with succinimidyl esters of rhodamine and fluorescein (8 molecules total of dye per molecule Tf) in a 100 mM sodium borate buffer, pH 9.0, for 1 h, and then removing excess dye by extensive dialysis against phosphate-buffered saline (pH 7.4). C 6 -NBD-SM was obtained from Molecular Probes Inc. (Eugene, OR).
Cell Culture and Drug Treatments-TRVb-1 cells, a previously described CHO cell line lacking the endogenous Tf-R but expressing the human Tf-R (24), were grown in bicarbonate-buffered Ham's F-12 (Life Technologies, Inc.) at pH 7.4, supplemented with 5% fetal bovine serum (Life Technologies, Inc.), 100 units/ml penicillin, 100 g/mg streptomycin, and 200 g/ml G418 in 100-mm dishes. The 12-4 endosome acidi-fication mutant cell line derived from TRVb-1 (25) was grown similarly. For microscopy experiments, cells were plated in 35-mm coverslip bottom dishes 2 to 3 days prior to experiments and grown to 50% confluence as described previously (26). For biochemical lipid efflux experiments, cells were plated in 60-mm dishes as described previously (3,22) and grown for 3 days (until 90% confluent). Bafilomycin A 1 (0.5 or 0.25 M) with a maximum final concentration of 0.1% Me 2 SO was used for all drug treatments. Non-drug-treated controls were mock treated with the same concentration of Me 2 SO as bafilomycin A 1 -treated cells in all experiments. Cells were kept in either HF-12 (Ham's F-12 medium without bicarbonate, but with 20 mM HEPES buffer, pH 7.4) or medium 1 (150 mM NaCl, 20 mM HEPES, pH 7.4, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mg/ml glucose) in experiments carried out outside of a CO 2buffered incubator.
Fluorescent Lipid Labeling-Labeling with small unilamellar vesicles of the fluorescent sphingomyelin analog C 6 -NBD-SM was carried out exactly as described (3,22). Small unilamellar vesicles were made using C 6 -NBD-SM and dioleylphosphatidylcholine (1:1) (3). TRVb-1 cells were labeled with 5 M C 6 -NBD-SM vesicles on ice for 30 min in HF-12 (3). Cells were then washed and warmed to 37°C for 10 min to allow internalization of fluorescent lipid that had been inserted into the plasma membrane. They were then back-exchanged at 4°C against several washes of medium 1 containing 5% bovine serum albumin for 1 h. This back-exchange typically removed 98 -99% of surface-bound fluorescent lipid (3). Monolayers were then warmed to 37°C in the presence of medium 1 containing 1% bovine serum albumin to backexchange effluxed C 6 -NBD-SM for varying times as indicated for individual experiments. In biochemical experiments, cell-associated and effluxed lipid was butanol-extracted and quantified by spectrofluorometric analysis exactly as described (3,22).
Immunofluorescence Staining of Tf-R-TRVb-1 cells on coverslip bottom dishes were fixed with 2% formaldehyde for 30 min and permeabilized with 500 g/ml saponin in phosphate-buffered saline (PBS), containing 1% BSA and 40 mM methylamine for 20 min. Dishes were then stained with the B3/25 antibody at 4 g/ml for 1 h and washed in PBS with 1% BSA for 15 min. Cells were stained with a goat anti-mouse polyclonal antibody conjugated to FITC at a 1:100 dilution for 1 h, washed for an additional 15 min, and then placed in Slow-Fade (Molecular Probes, Eugene, OR) and examined by fluorescence microscopy. Under these conditions, the antibody showed a distribution indistinguishable from that of fluorescent Tf at steady state (2.5 h) in the same cells (data not shown).
Quantitative Fluorescence Microscopy-For analysis of approach to steady state of Tf in the recycling compartment, coverslip bottom dishes were pretreated with 0.5 M bafilomycin A 1 or mock pretreated with 0.1% Me 2 SO for 15 min and then labeled with Tx-Tf for varying periods. In each experiment, 3 to 4 dishes were labeled to steady state (2.5 h) with Tx-Tf in the absence of drug. Other dishes were labeled with Tx-Tf for 0 -18 min with or without bafilomycin A 1 treatment. Labeling was terminated by acid washing the dishes on ice in mild acid buffer (containing 50 mM sodium citrate, 280 mM sucrose, and 0.01 mM deferoxamine mesylate, pH 4.6) for 2 min followed by a 2-min wash in cold medium 1 on ice (27) and fixation in 2% paraformaldehyde in PBS at room temperature (first 2 min on ice) for 30 min. Dishes were stained with an anti-Tf-R mouse monoclonal antibody (B3/25) and FITC conjugated rabbit anti mouse secondary antibody as described above.
Labeled cells were examined using a 40 ϫ, NA 1.3 objective on a Leitz Diavert microscope with a cooled charge coupled device (CCD) camera (Photometrics, Inc., Tucson, AZ) using a 660 ϫ 517 pixel array as described previously (3,22). Images were either viewed and processed on a SPARC/300 workstation (Sun Microsystems, Inc., Mountainview, CA) using software from Inovision Corp. (Durham, NC), or transferred to a micro VAX (DEC, Maynard, MA) and processed on a Gould 8500 image processor (Vicom, Fremont, CA).
Corresponding pairs of Tx-Tf and FITC images (5 to 6 pairs of images per dish, Ͼ50 cells) were obtained. The images in each pair were aligned and background corrected using a median filter, and recycling compartments were defined in the FITC (antibody) image using procedures described previously (22). The recycling compartment in the antibody image was used as a mask, so that fluorescence from the antibody image and from the background-corrected Tx-Tf image was quantified from precisely corresponding pixels.
In the dishes labeled to steady state (i.e. 2.5 h labeling), each receptor in the recycling compartment should be occupied with Tx-Tf. Therefore, there should be an equal number of Tx-Tf molecules and Tf-R. If F tx is defined as Tx-Tf fluorescence power and F a is antibody fluorescence power in recycling compartments, a correction factor (C) that relates the fluorescence from the two labels of Tf-R can be determined from fluorescence measurements at steady state as shown in Equation 1.
This correction factor can then be used in time course experiments to determine the fraction of Tf-R occupied (f o ) and the fraction of unoccupied Tf-R (f u ) by Equation 2 since the antibody will label Tf-R in the recycling compartment at all times.
pH Measurements-The methods for measuring pH of sorting and recycling endosomes were as described in Ref. 22. Briefly, TRVb-1 or 12-4 cells were plated in coverslip bottom dishes 2 days before the experiment and grown until cells were well spread but not yet confluent. Tf-R expression levels were matched by growing 12-4 cells in Ham's F12 medium, which contains iron salts, and TRVb-1 cells in McCoy's 5A medium, which lacks iron salts. Tf-R are up-regulated in TRVb-1 cells grown in McCoy's 5A medium. Dishes were preincubated in medium 1 containing 0.5 M bafilomycin A 1 for 25 min and then incubated in the same medium with 10 g/ml Tf labeled with both rhodamine and fluorescein (F-R-Tf). Fluorescence images of sorting endosomes were collected between 5 and 15 min of labeling with a Zeiss 63 ϫ planapochromat NA 1.4 oil immersion objective, and images of recycling compartments were collected between 10 and 20 min of labeling with a Zeiss 40 ϫ plan-Neofluor NA 0.9 water immersion objective. Images were collected with a Zeiss Axiovert microscope using a Bio-Rad MRC-600 confocal attachment, with a 515-545 nm fluorescein and 575 nm rhodamine emission filter sets, and 488 nm excitation. For calibration, cells were fixed and maintained in a range of pH calibration buffers with 10 M nigericin. The average value of the ratio of rhodamine to fluorescein fluorescence power from endosomes was graphed against pH, and the resulting pH calibration curves were used to assign pH to ratios determined in the experimental dishes.
Image processing was as described previously (22). Sorting endosomes were identified as small punctate structures, and recycling compartments appear as single large structures. Fluorescence from sorting endosomes was quantified after applying a size criterion that elimi-nated the large recycling compartments. Recycling compartments were identified by visual inspection of processed images. Ratios of rhodamine to fluorescein fluorescence were calculated for individual compartments, and histograms were plotted.
Measurement of Surface Transferrin Receptor-To determine the effect of bafilomycin A 1 on surface expression of receptors, cells grown in 6-well clusters (Becton Dickinson, Lincoln Park, NJ) for 2 days before use were incubated in McCoy's binding buffer (McCoy's 5A medium, without NaHCO 3 , 100 units/ml penicillin, 100 mg/ml streptomycin, 20 mM HEPES, pH 7.4) for 45 min at 37°C. The cells were then washed and incubated in 0.25 M bafilomycin A 1 at 37°C for the indicated times. The dishes were placed on an ice slurry, washed 3 times with ice-cold medium 1, and incubated with 3 g/ml 125 I-Tf in medium 1 for at least 2 h. The cells were then washed 6 times with ice-cold medium 1 and solubilized. The radioactivity was determined by gamma counting (Wallac Inc., Gaithersburg, MD). One 6-well cluster was preincubated for 45 min in McCoy's binding buffer as described above and then placed on ice with 3 g/ml 125 I-Tf in medium 1 for at least 2 h, and surface Tf binding was determined. The surface Tf binding measured for this plate served as the untreated control value. For each plate the nonspecific Tf binding was determined by including a 200-fold excess of unlabeled Tf in two of the wells during binding on ice. The average surface Tf binding per plate was calculated by computing first an average nonspecific binding from the two competition wells and subtracting this value from the value of each of the 4 experimental wells. Typically, the nonspecific counts were less than 10% of the total. The 4 background-corrected experimental wells were then averaged to give a value for the plate, which could be compared with the similarly determined control value. The data presented are the average values of at least 5 separate experiments Ϯ S.E.
Effect of Bafilomycin A 1 on Surface Tf-R-We assumed that the Tf-R cycle between two pools, a surface and an internal pool, and that exchange between the two pools occurs with simple first order kinetics. Previous studies with CHO cells indicate that internalization and recycling rate measurements are consistent with this assumption within experimental error (1, 3).
Given these assumptions, at steady state surface Tf-R (Tf s ) and internal Tf-R (Tf i ) are related by Equations 3-5. where k e is the rate constant for externalization, k i is the rate constant for internalization, Tf i is the number of internal Tf receptors per cell, Tf s is the number of surface Tf receptors per cell, and Tf tot is total Tf receptors per cell. As a rapid assay for the effects of bafilomycin A 1 , the surface binding of Tf was measured before and after addition of bafilomycin. It has been shown previously that in TRVb-1 cells the addition of bafilomycin A 1 changes k e but does not affect k i (14).

Effects of Bafilomycin A 1 on pH of Recycling Endosomes and
Sorting Endosomes in Wild-type and 12-4 Cells-It was shown previously that an end2 CHO cell line, 12-4 cells, has alkaline sorting endosomes relative to wild-type cells (0.5 pH units higher than TRVb-1) and more alkaline recycling endosomes (Ϸ0.2-0.4 pH units higher) (22). To compare the effects of bafilomycin A 1 on endosome pH in 12-4 cells and TRVb-1 cells, pH measurements were made ( Fig. 1 and Table I). In TRVb-1 cells, both sorting and recycling endosomes were alkalinized by bafilomycin A 1 treatment as described previously (14). Both of these endocytic compartments were also alkalinized by bafilo-mycin A 1 treatment of 12-4 cells. These results confirm that the mutation in 12-4 cells only partially affects the H ϩ ATPase-dependent acidification of endosomes.
Reduction of Surface Tf-R by Treatment with Bafilomycin A 1 -Tf-R recycles between the cell surface and internal endosomal compartments. Consequently, the amount of Tf-R on the cell surface is determined by the rates of internalization and recycling (Equations 3-5 under "Experimental Procedures"), and modification of either rate by bafilomycin A 1 should alter the amount of surface Tf-R in a time-dependent fashion. Surface Tf-R should plateau at a level that reflects the new steadystate rates of recycling and internalization in the presence of bafilomycin A 1 . A time course for surface Tf binding following bafilomycin A 1 treatment is presented in Fig. 2. Following bafilomycin A 1 treatment, the amount of surface Tf binding decreases in cells expressing the wild-type Tf-R to a value that is ϳ54% of the pretreatment value. The new steady-state value is reached within 30 min. In cells expressing the Tf-R with a deletion of 58 of the 61 amino acids of the cytoplasmic domain (⌬3-59), there is only a very small reduction in the surface expression of Tf-R following bafilomycin A 1 treatment (Fig. 2). These results are consistent with previous results showing that bafilomycin A 1 decreases the k e of Tf-R from TRVb-1 cells without affecting k i in a manner that depends on the cytoplasmic domain (14).
12-4 cells have a decreased rate of Tf-R recycling that may be related to their defect in endosome acidification. Since bafilomycin A 1 further alkalinizes endosomes in 12-4 cells, we used the surface binding assay to determine whether bafilomycin A 1 can further slow receptor recycling in 12-4 cells. As shown in Fig. 2, surface Tf-R was reduced significantly by bafilomycin A 1 treatment of 12-4 cells. This is consistent with a further reduction in the k e of the Tf-R, perhaps as a consequence of more complete endosome neutralization induced by bafilomycin A 1 in 12-4 cells (Fig. 1).

Recycling Compartment Is Not Significantly Affected in Bafilomycin A 1 -treated Cells-While both sorting and recycling en-
dosomes are alkalinized in bafilomycin A 1 -treated cells, the point at which trafficking of Tf through the cell is slowed, resulting in net intracellular retention, is unclear. A step-bystep analysis of transit of the Tf-R through the recycling pathway was undertaken in which the sorting of Tf-R from lysosomally directed markers, trafficking of Tf-R to a peri-centriolar recycling compartment, and its exit from the cell were sequentially studied. Sorting endosomes in TRVb-1 cells (i.e. endosomes that contain both lysosomally directed and recycling markers) are observed as punctate structures in the periphery of the cell (1). After short pulses (e.g. 2 min), most Tf is colocalized in the sorting endosomes with lysosomally destined markers such as low density lipoprotein (1,3,8,22). Tf and other membraneassociated markers rapidly leave the sorting endosomes and move to the peri-centriolar endocytic recycling compartment with a half-time of ϳ2 min (3). Thus, if retention in sorting endosomes is contributing in a rate-limiting manner to retention of Tf-R in bafilomycin A 1 -treated cells, this retention should be readily detected in a pulse-chase experiment.
TRVb-1 cells were treated or mock treated with bafilomycin A 1 , pulsed for 2 min with Tx-Tf, acid washed to remove surfacebound Tf, and then chased for 4 or 8 min with FITC-Tf included in the last 2 min of the chase to identify sorting endosomes. In both treated and mock treated cells, Tx-Tf had begun to move to the recycling compartment after 4 min of chase (not shown). After 8 min chase, the initial Tx-Tf pulse was almost entirely redistributed into the peri-centriolar recycling compartment in both treated and mock treated cells (Fig. 3). This indicates that bafilomycin A 1 does not have a large effect on removal of Tf from sorting endosomes and delivery to the recycling compartment.
Tf Receptors Are Retained in the Endocytic Recycling Compartment-Since iron loaded Tf remains tightly bound to its receptor at neutral pH, re-internalization of Tf is a complication of experiments to measure efflux in bafilomycin A 1 -treated cells. The rate of approach to steady-state labeling of Tf-R provides a method to measure efflux kinetics that is not affected by futile recycling (14). Here, we adopted this assay to use fluorescence microscopy to measure efflux from the endocytic recycling compartment based on the ratio of Tx-Tf to total Tf-R (detected by antibody labeling). To calibrate the measurements, we measured the ratio at steady-state labeling with a saturating concentration of Tx-Tf. TRVb-1 cells were labeled for 2.5 h with Tx-Tf to ensure full occupancy of all intracellular Tf-R. We stained the same cells for the Tf-R using an antibody against the Tf-R (B3/25) that is not blocked by receptor occupancy and an FITC-labeled secondary antibody. A quantitative fluorescence measurement of both Tx-Tf and the Tf-R was thereby obtained under conditions in which they should be present in all endocytic compartments at the same levels. There was complete colocalization of Tf and Tf-R in these double labeled cells. The ratio of the brightness of the two stains (FITC/Tx) gives a calibration factor (C) that when multiplied by the Tx brightness (F tx ) in cells pulsed for a short period gives the fraction of occupied Tf-R in the recycling compartment (f o ; see "Experimental Procedures").
Cells were treated or not treated with bafilomycin A 1 and labeled with Tf, followed by immunofluorescence staining of Tf-R. Digital fluorescence images of Tx-Tf and FITC-labeled Tf-R were obtained (Fig. 4). The recycling compartments were identified in the pairs of images using digital image processing (22). Using the calibration factors determined at steady state, the fraction of occupied receptors in the recycling compartment after various labeling periods (0 -18 min) was determined. The fraction of unoccupied receptors decayed with time as shown in Fig. 5. This decay could be fit to a single exponential, and rate constants for exit (k e ) were determined. The rate constant for the exit of unoccupied Tf-R from the recycling compartment of mock treated cells was 0.063 Ϯ 0.019 min Ϫ1 , which agrees well with the rate of 0.057 Ϯ 0.005 min Ϫ1 obtained using conventional 125 I-Tf efflux experiments (28). In bafilomycin A 1 -treated cells, the rate of efflux from the endocytic recycling compartment was reduced to 0.034 Ϯ 0.039 min Ϫ1 (Fig. 5 and Table II). This reduction in rate is in close agreement with the overall reduction in the rate of efflux from cells treated with bafilomycin A 1 (14) (see Table II). This indicates that the reduced overall rate of exit of Tf-R from the cell is due to slowed exit from the recycling compartment.
Recycling of the Bulk Membrane Marker C 6 -NBD-SM Is Slowed in Bafilomycin A 1 -treated Cells-In acidification defective 12-4 cells, Tf-R exit from the recycling compartment is slowed, but exit of the bulk membrane marker C 6 -NBD-SM is unaffected (22). We wanted to see if the same was true in bafilomycin A 1 -treated cells. In drug-treated cells C 6 -NBD-SM colocalized with Tf and had a similar distribution to C 6 -NBD-SM in non-drug-treated cells (Fig. 6). Using a previously described assay we measured efflux of C 6 -NBD-SM from cells (22) with or without bafilomycin A 1 treatment (Fig. 7).
As described previously for TRVb-1 and 12-4 (3,22), efflux kinetics are well fit by a double exponential composed of a major (Ϸ75%) slow component and a minor (Ϸ25%) fast component (t1 ⁄2 2-3 min). The nature of the fast component is unknown but may represent direct traffic from sorting endosomes to the plasma membrane. The slow component corresponds to the rate of exit of C 6 -NBD-SM from the recycling compartment (3,22). Double exponential least square fits showed no difference in amplitude or rate of the fast component of lipid exit between bafilomycin A 1 -treated and mock treated cells (data not shown). In cells treated with 0.5 M bafilomycin A 1 , the rate of the slow component of lipid exit was 0.042 Ϯ 0.003 min Ϫ1 . In mock treated cells, the rate of the slow component of lipid exit was 0.059 Ϯ 0.006 min Ϫ1 , a rate that agrees closely with previously determined values (3,22). When a lower dose (0.25 M) of bafilomycin A 1 was used, similar rates were obtained (bafilomycin A 1 , 0.060 Ϯ 0.007 min Ϫ1 ; mock, 0.046 Ϯ 0.004 min Ϫ1 ; Table II). Thus, bafilomycin A 1 significantly reduces the recycling of bulk membrane in TRVb-1 cells.
Effects of Altering the Internalization Motif of the Tf-R on Its Retention in a Bafilomycin A 1 -treated Cell-To explore the role of the cytoplasmic domain of the Tf-R in the bafilomycin A 1triggered retention, we examined various mutations of the internalization sequence. Since slowed recycling will result in the depletion of surface Tf-R, we examined the effect of bafilomycin A 1 on surface expression of unaltered Tf-R or Tf-R in which the internalization domain on the cytoplasmic tail was altered. Table III shows surface Tf-R on bafilomycin A 1 -treated cells as a percentage of the control (mock treated) value after 30 min of drug treatment for various Tf-R.
The surface expression of ⌬3-59 Tf-R is only slightly affected by bafilomycin A 1 treatment ( Fig. 2 and Table III). The surface expression of a Tf-R containing a mutation in the internalization motif (Y20C Tf-R) is affected to an intermediate extent by bafilomycin A 1 treatment. We examined the effect of bafilomy-

FIG. 5. Recycling of Tf-R in bafilomycin A 1 -treated and control cells.
TRVb-1 cells in coverslip bottom dishes were pretreated with bafilomycin A 1 or mock pretreated with 0.1% Me 2 SO for 15 min and then labeled for the indicated lengths of time with Tx-Tf in HF-12. Tf-R in these cells was stained with mouse monoclonal B3/25 and an FITCconjugated goat anti-mouse secondary antibody, and corresponding Tx-Tf and FITC-Tf-R receptor images were taken of each field. 5-6 fields were taken at random from each dish and were selected using phase contrast. Recycling compartments were isolated in the doublelabeled images, and total Tx-Tf and FITC-Tf-R fluorescence in each field was quantified. The ratio of Tx-Tf fluorescence to FITC-Tf-R fluorescence was then normalized as described under "Experimental Procedures" to give mole fraction of occupied Tf-R. The corresponding mole fraction of unoccupied Tf-R was plotted versus time and fitted to a single exponential. Thus, this experiment measures the exit of unoccupied receptors. Each curve is an average of 3-4 experiments on different days Ϯ S.D.

FIG. 6.
Colocalization of C 6 -NBD-SM and Tf in bafilomycin A 1treated cells. TRVb-1 cells in coverslip bottom dishes were pretreated with bafilomycin A 1 for 15 min, then chilled, and labeled at 4°C with C 6 -NBD-SM vesicles (5 M total lipid) and 5 g/ml Tx-Tf, and incubated for 10 min at 37°C to allow internalization of surface-bound lipid. Cells were then chilled on ice, and remaining surface-bound lipid was removed by back-exchange. Cells were warmed up and chased an additional 6 min to clear sorting endosomes. Cells were visualized using a 40 ϫ objective. A, Tx-Tf; B, C 6 -NBD-SM. Micron bar, 10 M. cin A 1 on the surface expression of a Tf-R mutant that is more rapidly internalized than the wild-type Tf-R. This mutant Tf-R, G31Y, contains two functional internalization motifs: the native internalization motif, YTRF, at positions 20 to 23, and a second internalization motif at positions 31 to 34 which was created by mutating glycine 31 to a tyrosine (29). The G31Y shows a net loss from the surface similar to that of the wildtype Tf-R which only contains a single internalization motif (29,30). The effect of bafilomycin A 1 on the recycling of a Tf-R mutant, Y20C, G31Y Tf-R, containing the artificial G31Y internalization motif but with an alteration in the native motif was also examined. This double mutant is internalized at the rate of the wild-type Tf-R (29). Surface Tf binding in cells expressing the Y20C, G31Y Tf-R is reduced to the same extent as the wild-type Tf-R, demonstrating that a Tf-R containing the G31Y internalization motif is slowly recycled in the presence of bafilomycin A 1 (Table III).
For wild-type Tf-R, the net reduction of surface expression upon addition of bafilomycin A 1 is due to changes in the k e of the Tf-R with no change in k i (14). Based on Equation 5 (see "Experimental Procedures"), the percentage reduction in sur-face Tf-R is dependent on the k e prior to drug treatment, the new k e after drug treatment, and k i . The small redistribution of ⌬3-59 from the surface to internal compartments can be accounted for by its slow rate of internalization and small effect of bafilomycin A 1 on k e for this receptor lacking most of the cytoplasmic domain (14). Interestingly, the Y20C, S34Y Tf-R, and the Y20C, G31Y Tf-R redistribute similarly to the wildtype Tf-R. Since all three receptors have similar internalization rates, these data indicate that they also have similar bafilomycin A 1 -induced retention. This indicates that retention is based on cytoplasmic sequences similar to the requirements for rapid internalization.

DISCUSSION
Intracellular Routing of the Tf-R on Treatment with Bafilomycin A 1 -We find that bafilomycin A 1 does not significantly alter the exit of the Tf-R from early sorting endosomes nor its subsequent accumulation in the peri-centriolar endocytic recycling compartment. Since sorting endosomes are essentially completely neutralized in bafilomycin A 1 -treated cells, these data show that proper sorting of recycling membrane components does not require the sorting endosome lumen to be acidic.
The half-time for efflux of internal Tf-R to the cell surface in bafilomycin A 1 -treated cells is ϳ20 min as compared with ϳ10 min for control cells (14). Exit of Tf-R from recycling compartment was slowed approximately 2-fold relative to nontreated cells. This rate constant (0.034 min Ϫ1 ) was similar to the rate constant for exit of 125 I-Tf from drug-treated TRVb-1 cells (0.035 min Ϫ1 ; 14). The rate constant for exit of Tf-R from recycling compartments in non-drug-treated cells was 0.063 min Ϫ1 (Table II). Thus, exit from the recycling compartment is rate-limiting in bafilomycin A 1 -treated cells, and the slowing at this stage of the Tf-R itinerary is responsible for the overall slowdown in Tf-R recycling kinetics.
A notable feature of the endosomal recycling pathway is that the Tf-R normally moves through the entire pathway at the same rate as bulk membrane markers such as C 6 -NBD-SM. This indicates that concentrative mechanisms similar to clathrin-coated pits are not required for rapid and efficient recycling of the Tf-R after internalization (3). However, bafilomycin A 1 treatment differentially affects C 6 -NBD-SM and Tf-R. Bulk membrane exit rate constants measured with C 6 -NBD-SM are reduced to ϳ75% of non-drug-treated value, and rates for exit of Tf-R are reduced to 50 -55% of the control value (Ref. 14 and Table II). Thus, in bafilomycin A 1 -treated cells, as in 12-4 cells (22), Tf-R is actively retained relative to bulk membrane.
It was previously reported that Tf-R constructs with a large deletion in the cytoplasmic domain (⌬3-59) or with point mutations in the internalization motif (Y20C or F23A) showed reductions in k e to 70 -90% of control values when treated with 0.25 M bafilomycin A 1 (14). These values are similar to the reduction in k e for bulk membrane suggesting that the same or FIG. 7. Recycling of bulk membrane in bafilomycin A 1 -treated and control cells. TRVb-1 cells were pretreated with 0.1% Me 2 SO with or without 5 M bafilomycin A 1 and were allowed to internalize C 6 -NBD-SM for 10 min. Fluorescent lipid was stripped from the cell surface using a BSA back-exchange at 0°C. Cells were warmed and allowed to efflux lipid into 4 ml of back-exchange medium (medium 1 with 1% BSA) at 37°C. 400-l aliquots of medium were removed at intervals with replacement with 400 l of prewarmed chase medium. After the termination of the experiment cell-and medium-associated lipid was butanol-extracted and spectrofluorometrically quantified as described under "Experimental Procedures." Cells were maintained in drug ϩ Me 2 SO or Me 2 SO as appropriate throughout the course of the experiment. A representative experiment is shown. For a summary of all experiments see Table II. Endosome Acidification and Membrane Trafficking similar sequences may be involved in internalization and in bafilomycin A 1 -triggered retention in the endocytic recycling compartment. Consistent with this interpretation mutants Y20C, S34Y and Y20C, G31Y have internalization rate constants that are restored to the wild-type value by creation of a new internalization motif (29), and the constructs are retained in bafilomycin A 1 -treated cells almost to the same extent as wild-type receptor (Table III). We cannot rule out that a sequence other than the YTRF could also be involved in retention of the Tf-R in bafilomycin A 1 -treated cells.
Mechanism of the Bafilomycin A 1 Effect-Some form of communication between the abnormally neutral lumen of the recycling compartment and the cytoplasmic domain of the receptor tail is required if motifs on this tail are important for slowed recycling of the receptor in bafilomycin A 1 -treated cells. One possibility is that alkaline intra-endosomal pH could trigger a change in the distribution or location of Tf-R in recycling compartments, which would allow direct recognition by a cytoplasmic protein of their cytoplasmic tails. This could occur if, for example, receptors are aggregated in regions of an abnormally neutral recycling compartment, allowing adaptins or other proteins to bind the clustered cytoplasmic tails in a nonspecific manner due to the greater avidity of multivalent binding or to other mechanisms leading to active retention. It is also possible that a pH change in the recycling compartment lumen can be transduced to the outside of the recycling compartment through other proteins and trigger an active retention of Tf-R.
We have shown previously that if Tf multimers are made by cross-linking, they are retained in the recycling compartment of TRVb-1 cells expressing either the wild-type Tf-R or ⌬3-59, indicating that retention of multimers involves either slow diffusion of the aggregate or retention of aggregates of Tf-R via their lumenal domains (31). The recycling compartments of TRVb-1 cells containing multimeric Tf are functionally normal, since a pulse of monomeric Tf can still enter and leave the recycling compartment at the normal bulk rate (31). Thus, there are at least two mechanisms by which Tf-R can be retained in the recycling compartment of TRVb-1 cells, one requiring the cytoplasmic domain and one independent of the cytoplasmic domain.
The effect of bafilomycin A 1 on bulk membrane trafficking is less than on recycling of the Tf-R. This reduction in bulk flow (to ϳ75% of pretreatment k e values) could result from slowed fusing of vesicles with the recycling compartment, slowed formation of exit vesicles, or formation of smaller exit vesicles. Currently, we have no evidence to distinguish these possibilities.
Most previous studies have used weak bases or ionophores to collapse intracellular pH gradients. These agents generally cause vacuolization of endosomal compartments and have other nonspecific effects. In our assays, the weak base primaquine causes a much larger net internalization of surface Tf-R than bafilomycin A 1 , and this large net internalization was not dependent on the cytoplasmic tail domain (data not shown).
The difference between primaquine and bafilomycin A 1 is probably due directly to the additional changes (e.g. organelle swelling) that treatment with weak bases causes in endosomal compartments.
Conclusions-Bafilomycin A 1 reduces the externalization of Tf-R from wild-type TRVb-1 cells by slowing the exit of Tf-R from the recycling compartment in a rate-limiting manner. This is partly affected by a slowdown of the rate of bulk flow from the recycling compartment and partly by active retention of the Tf-R in the recycling compartment by a mechanism dependent on the YTRF internalization motif on the cytoplasmic tail.