Isolation by fluorescence-activated cell sorting of Chinese hamster ovary cell lines with pleiotropic, temperature-conditional defects in receptor recycling.

We have isolated several Chinese hamster ovary cell lines with temperature-sensitive defects in the recycling of receptors after endocytosis. These cell lines were selected using fluorescence-activated cell sorting for retention of a pulse of labeled transferrin after a chase in the presence of unlabeled transferrin. One of these cell lines, TfT1.11, was selected for further characterization. In TfT1.11 the trapping of transferrin within the cells is paralleled by a loss of cell surface transferrin receptors. Within 4 h after the shift from 33 to 41 degrees C the surface binding of transferrin is reduced to 18% of parental cells at 41 degrees C. The trapping of transferrin and the loss of transferrin receptor from the cell surface are caused by a temperature-conditional 5.5-fold decrease in the initial rate of transferrin recycling. TfT1.11 cells also rapidly lose 89% of their ability to take up alpha 2-macroglobulin after the temperature shift to 41 degrees C. These data indicate that the TfT1.11 cell line has a pleiotropic defect in receptor recycling.

We have isolated several Chinese hamster ovary cell lines with temperature-sensitive defects in the recycling of receptors after endocytosis. These cell lines were selected using fluorescence-activated cell sorting for retention of a pulse of labeled transferrin after a chase in the presence of unlabeled transferrin. One of these cell lines, TfT 1.1 1, was selected for further characterization. In TfT1.ll the trapping of transferrin within the cells is paralleled by a loss of cell surface transferrin receptors. Within 4 h after the shift from 33 to 41 "C the surface binding of transferrin is reduced to 18% of parental cells at 41 "C. The trapping of transferrin and the loss of transferrin receptor from the cell surface are caused by a temperature-condi-tiona15.5-fold decrease in the initial rate of transferrin recycling. TfT 1.11 cells also rapidly lose 89% of their ability to take up a2-macroglobulin after the temperature shift to 41 "C. These data indicate that the TfT 1.1 1 cell line has a pleiotropic defect in receptor recycling.
Several early observations indicated that there is extensive recycling of internalized membrane and protein to the cell surface after endocytosis (reviewed in Refs. 1 and 2). During extended periods of endocytic activity cells internalize a large percentage of their plasma membrane and a large volume of fluid although the dimensions of the cells do not change. In addition, for many molecules the total amount of ligand internalized greatly exceeds the number of cell surface receptors. Cells are able to sustain continuous uptake of ligand for long periods of time even in the absence of protein synthesis (3-5). The kinetics of ligand accumulation indicate that receptor recycling is very efficient and that a single receptor can be internalized hundreds of times.
To accumulate ligand and allow for receptor recycling cells must have a mechanism to segregate material to be recycled from that to be retained.
For many receptors affinity for ligand is low at acidic pH (6)(7)(8)(9), and dissociation of ligand * This work was supported in part by National Institutes of Health Grant GM32508 and National Science Foundation Grant DCB-8903657 (to R. F. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by an American predoctoral fellowship of the American Association of University Women Educational Foundation. from receptor is coupled to endocytic acidification (for review, see 10). Segregation of most receptor-ligand complexes occurs in the endosome, an early, prelysosomal endocytic compartment with a low buoyant density (11)(12)(13)(14)(15). The pH of early endosomes is regulated in many cell types to approximately pH 6 (16-18), a pH sufficient to drive the dissociation of many ligands and receptors. This allows recycling to occur before exposure to the degradative environment of the lysosomes. Once ligands have dissociated from their receptors, segregation largely becomes a problem of the relative retention of fluid during the recycling of membrane and membranebound material (19).
Although the mechanism for dissociation of many receptors and ligands has been determined, very little is known about the mechanism for segregation and recycling. It has been suggested that the morphology of the endosome may be responsible for the segregation of receptors and ligands (20). Although the relative surface area to volume ratios of the recycling vesicles and endosomes may account for the segregation of free ligand from receptor during recycling (21), there are several membrane-bound proteins, including epidermal growth factor receptor, IgG Fc domain receptor, and mannose 6-phosphate receptor, which are retained and carried further in the pathway (11,15,22,23). This suggests that there is a specific mechanism to sort membrane-bound components from each other in addition to a means of separating membrane from fluid. Sorting for selective internalization is accomplished at the plasma membrane through cytoplasmic determinants on membrane proteins that direct their binding to adaptor proteins (adaptins) in clathrin-coated regions of the plasma membrane (24). Although a similar mechanism could play a role in receptor recycling, it is currently not known if recycling is a directed process (a recycling signal is needed) or a default pathway (a retention signal is needed). It is clear, however, that the rate of receptor recycling can be regulated in both a growth state-and cell-type-dependent manner (25-30).
To begin to identify molecules involved in this process we selected cell lines defective in receptor recycling using labeled transferrin (Tf)' as a marker for its receptor. Tf, a major iron transport protein, binds with high affinity to surface receptors and is internalized during endocytosis. Unlike most other ligands, however, Tf escapes lysosomal degradation by recycling back to the cell surface bound to its receptor (31, 32). A single cycle of internalization and recycling is completed ' T h e abbreviations used are: Tf, transferrin; TfR, transferrin receptor; a2M, a,-macroglobulin; a-MEM, minimal essential medium, a-modification; Cy5, cyanine5.18-OSu: FITC, fluorescein isothiocyanate; MES, 2-(A"morpho1ino)ethanesulfonic acid CHO, Chinese hamster ovary. within 10-15 min (4, 5, 16). To isolate recycling defective mutants cells were selected for retention of labeled Tf after a chase with excess unlabeled Tf. In this paper we describe the isolation of.these cell lines and the initial characterization of the clone TfI'l.ll, which shows a temperature-dependent, pleiotropic defect in receptor recycling.

EXPERIMENTAL PROCEDURES
Tissue culture supplies were purchased from GIBCO, and other materials were purchased from Sigma, unless otherwise noted.
A 7-26% polyethylene glycol fraction (33) was dialyzed against distilled water, centrifuged (40,000 X g), and dialyzed against buffer A (0.15 M NaCl, 0.02 M NaPO,, pH 6.0). The solution was then fractionated by zinc affinity column chromatography (34). The a,M bound to the affinity column (2.5 X 27 cm) was eluted as described (35) except that 50 mM acetate, pH 4.5, was used for elution. The eluted protein was then dialyzed against buffer A and fractionated by gel permeation on a Sephacryl S-300 column (2.5 X 90 cm). The purified asM was then activated with methylamine as described (33).
Synthesis of Fluorescent Conjugates-FITC-dextran (70,000 daltons) was prepared by the dibutyltin-dilaurate method with the modifications described previously to reduce free dye contamination (36).
Cy5.18-OSu has properties similar to those of Cy5.12-OSu (37) and was a generous gift from Dr. Alan Waggoner, Carnegie Mellon University. The resulting dye to protein ratios were 2.75 for the conjugate used in mutant isolation experiments and 5.0 for the conjugate used in the characterization experiments. aeM was labeled with Cy5, with a resulting dye to protein ratio of 7.1. All of the Cy5 protein conjugates had binding specificities of greater than 90%. Typical ratios of Cy5 fluorescence to background fluorescence from unlabeled cells were 130:l for Tf (10 pg/ml) and 45:l for n2M (10 pg/ml).
Flow Cytometry-A dual laser FACS 440 flow cytometer (Becton Dickinson Immunocytometry Systems) equipped with argon and krypton lasers was used for all analyses. FITC fluorescence (488 nm excitation, 400 milliwatts) was collected using a 530 nm band pass filter (30 nm band width), and Cy5 fluorescence (647 nm excitation, 200 milliwatts) was collected using a 670 nm band pass filter (13.5 nm band width). A FACStar''."' equipped with argon and heliumneon lasers was used for sorting.
Isolation of Recycling Defective Cell Lines-CHO WTB cells (obtained from Dr. Brian Storrie, Virginia Polytechnic Institute and State University) were maintained a t 33 "C in a-MEM supplemented with 10% calf serum (HyClone Laboratories, Logan, UT), 100 units/ ml penicillin, 100 pg/ml streptomycin, 0.29 g/liter L-glutamine, and 4 g/liter proline (c-a-MEM). Cells were mutagenized by incubation in c-n-MEM with 250 gg/ml ethylmethane sulfonate for 24 h and allowed to recover for 2 weeks prior to the first sort. Cells were incubated a t 41 "C for 0.5, 1, 1.5, or 2 h and labeled for 30 min with 2.5 pg/ml Cy5-Tf in a-MEM. The cells were then incubated for 1 h in c-a-MEM with 1 mg/ml unlabeled human diferric Tf and 2 mg/ ml FITC-dextran a t 4 "C (to determine the init,ial amount of Tf internalized) or a t 41 "C (for samples to be sorted). After labeling, the cells were washed six times with ice-cold a-MEM salts (116 mM NaCI, 5.4 mM KC1, 0.2 mM CaCI,, 0.8 mM MgSO,, 10 mM NaHaP04, pH 7.4) and suspended by scraping into a-MEM. Samples were maintained on ice during analysis and sorting.
The 0.1% of the cells which showed the highest Cy5-Tf and FITCdextran labeling were sorted from samples from each incubation time (1,000 cells sorted). Parallel sorts using less stringent criteria (1% of the cells) were unsuccessful. Cells were sorted into 4 ml of ice-cold c-WMEM supplemented with additional antibiotics (0.1% gentamicin, 1% fungizone, and 0.1% nystatin) and plated at 33 "C. After 2 days the sorted cells were maintained in c-a-MEM without additional antibiotics. Each sort was maintained as a separate population (four incubation times), and each population was expanded to obtain sufficient cells for a second sort (19 days). Each population was sorted a second time (same expression time and sort criteria), expanded, and analyzed. After expansion of the second sort, a distinct doublepositive population was present in the 2-h population and was sorted.
The Tff cell lines were cloned from this third sort by limiting dilution.
Transferrin Binding-Cells were plated in 12-well tissue culture plates (Corning Glassworks, Corning, NY) 2 days in advance for approximately 80% confluence at the time of use. The plates were shifted to 41 "C for various times and washed with a-MEM salts. The cells were then labeled for a t least 1 h a t 4 "C in a-MEM containing Cy5-Tf at the indicated concentrations. The cells were washed six times with a-MEM salts, suspended by scraping into a-MEM salts, and analyzed immediately (within 30-60 s after removal of TO. For the determination of TfR affinity, cells were incubated in a-MEM for 30 min at 4 "C, washed two times with a-MEM salts, and labeled for 1 h at 4 "C in a-MEM containing 1 mg/ml bovine serum albumin and various concentrations of Cy5-Tf (2-200 pg/ml, three samples/concentration). The samples were then washed and analyzed as above. Mean fluorescence values were corrected for nonspecific binding, determined a t each concentration by competition using 5 mg/ml unlabeled Tf. For Scatchard analysis the concentration of Tf added was used as an estimate of free Tf (since the fraction of input Tf bound to cells was negligible under the conditions used).
Transferrin Ezternalization-Cells were plated as described for Tf binding experiments.
Samples of T f f l . l l a n d W T B were shifted to 41 "C for 3 h or maintained a t 33 "C, washed with a-MEM, and incubated for 30 min at the appropriate temperature in 0.5 ml of N -MEM containing 1 mg/ml bovine serum albumin and 10 pg/ml Cy5-Tf. The cells were then chilled quickly to 4 "C by washing twice with ice-cold a-MEM salts. To remove surface-bound Tf cells were incubated at 4 "C in 0.5 ml of stripping buffer (150 mM NaCI, 100 p M desferrioxamine mesylate (Ciba-Geigy), 50 mM MES (Research Organics, Cleveland, OH), pH 4.5, for 5 min followed by two successive 5-min incubations in n-MEM salts containing 100 p~ desferrioxamine mesylate. The cells were then warmed quickly to the appropriate temperature by the addition of 2 ml of warm c-a-MEM with 0.5 mg/ ml unlabeled Tf. After various times the cells were again chilled quickly to 4 "C with two washes of ice-cold a-MEM salts. The cells were maintained a t 4 "C until all samples had been collected, washed five times with a-MEM salts, and suspended by scraping immediately prior to analysis by flow cytometry. The data were fit by one or two exponentials using a nonlinear least squares fitting program (38). Nonlinear fits to all data were made using four models: a single exponential (two free parameters), a single exponential with a nonrecycled component (three free parameters), two exponentials (four free parameters), and two exponentials with a nonrecycled component (five free parameters). The root mean square error (normalized to a percent error by division by the mean Y value for each data set) was used to judge goodness of fit. For the 33 "C data the first model yielded significantly higher error values (11.6 and 16.8% for WTB and TfTl.11, respectively) than the other three. Since the error values for these three models were within 0.03% of each other, the second model (single exponential with a nonrecycled component) was chosen (error values were 6.5 and 6.8% for WTB and Tffl.ll, respectively). For the 41 "C data, the first two models gave higher error values (21.2 and 14.5% for WTB, 12.9 and 10.2% for TfTl.ll) than the last two (9.6-9.7% for both WTB and TfTl.ll). The decrease in error of less than 0.1% between the third and fourth models was not considered to justify the addition of a free parameter, so the third model was chosen. Results are also shown for the second model for TfTI.ll at 41 "C since the difference in error between the second and third models was only 0.6%.
To determine how similar the pathways in TfT 1.11 and WTB were at 41 "C, an additional fit was made using the third model with the rate constants fixed to those obtained with WTB (the model thus has only two free parameters). The error obtained (14.2%) was significantly larger than the others for this data set (including the first model, which also had only two free parameters); the results are shown in Table IV for comparison only. For all of the samples there was a significant but variable fraction of the Tf which was released very rapidly from the cells (tIr2 < 30 s).
This may represent the "fast" external compartment described by McKinley and Wiley (39) for adherent cells, which is caused by trapping of ligand between the cells and the plates at low temperatures. No attempt at fitting this component was made since the halftime of release was shorter than the first point (30 s). To estimate the amount of cell-associated Tf excluding this external compartment, the nonlinear fits were extrapolated to 0 min. The raw data were then converted to a percentage of this value. a2-Macroglobulin Uptake-Cells were plated as described for the Tf binding experiments, shifted to 41 "C for various times, and labeled for 6 min at 41 "C in a-MEM with 10 pg/ml Cy5-asM. To clear receptors of unlabeled a2M from the serum, the cells were incubated without serum at either 33 or 41 "C for 30 min prior to labeling. After labeling the cells were chilled quickly to 4 "C, washed twice with a-MEM salts, suspended by scraping into a-MEM salts, and analyzed by flow cytometry. Short term uptake was measured instead of surface binding because very little binding could be detected at 4 "C even after a 1-h incubation. The temperature dependence of a2M binding has been noted for other cell types as well (40). Measurements of cell-associated asM reflect both surface bound and internalized ligand because the off rate, like the on rate, is quite slow. Cellular ATP Leuels-Cells were plated as for the Tf binding experiments. The cells were shifted to 41 "C for 4 h or maintained at 33 "C prior to harvesting by scraping at 4 "C. The amount of ATP/ sample was determined using a luciferin/luciferase ATP determination kit for somatic cells (Sigma) and a Thorn EM1 PMT (800 V) coupled to a Thorn EM1 photon counter (Thorn EM1 Gencom Inc., Fairfield, NJ). The average number of cells/sample was determined from parallel samples.
Protein Synthesis-Cells were plated as for the Tf binding experiments. The plates were shifted to 41 "C for 4 h or maintained at 33 "C, washed with a-MEM salts, and incubated at the appropriate temperature for 15 min in assay medium (deficient Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 100 units/ ml penicillin, 100 pg/ml streptomycin, 0.292 g/liter L-glutamine, 0.105 g/liter L-leucine, 0.146 g/liter L-lysine, 4 g/liter proline, and 0.3 mg/ liter L-methionine). 8 pCi of Tran35S-label (ICN; Irvine, CA) was added to half of the wells, and the incubation continued for 45 min. The cells were then washed three times with a-MEM salts. The cells that did not receive radioactivity were suspended by trypsin treatment and counted with a Coulter Counter (Coulter Corp; Hialeah, FL). The cells receiving radioactivity were lysed in 85 pl of lysing solution containing 0.1% sodium dodecyl sulfate, 0.1 mg/ml deoxyribonuclease I, 1 mM CaCI2, and 1 mM MgC12. The lysate from each well was transferred to 2.5-cm squares of Whatman 3MM, soaked for 20 min in 5% trichloroacetic acid containing 0.5 mg/ml L-methionine, and washed once in 95% ethanol. The filter papers were dried, and the radioactivity was measured in a liquid scintillation counter using Ecolume (ICN) as the scintillant.

RESULTS
Cell Sorting-To isolate cells with temperature-conditional defects in receptor recycling we explored the feasibility of using fluorescence-activated cell sorting t o select cells that were unable t o recycle labeled Tf administered after a shift to a higher temperature. A continuous incubation in the presence of labeled Tf during the temperature shift could not be used t o label the cells since normal WTB cells were observed t o retain some labeled transferrin after chase. The accumulation of labeled Tf occurred at a rate of approximately 5% of the total internal Tf for each h of labeling and was inhibited by the addition of excess unlabeled Tf during the pulse (data not shown). This phenomenon was observed for both Cy5-and FITC-Tf and was also observed in CHO K1 and Swiss 3T3 cells (data not shown). The retention of a small fraction of the internal Tf has been described in other cell lines as well (27,41). To avoid this problem cells were incubated at 41 "C for various times to allow expression of heat-sensitive lesions, pulse labeled with Cy5-Tf for 30 min, and chased for 1 h with excess unlabeled Tf. Cy5 was used as the fluorescent marker because it gave very high signal to noise ratios (typically 1 3 0 1 for the initial labeling: see "Experimental Procedures"). T o ensure that the sorted cells were capable of internalization, FITC-dextran was included in the chase media as a marker of fluid phase endocytosis. The desired mutants were selected t o be positive for Cy5-Tf (caused by failure t o recycle it during the chase) and also positive for FITC-dextran (normal for internalization). Examples of the sort conditions are shown in Fig. 1. Double-positive cells were sorted from the samples chased at 41 "C (Fig. IC, quadrant 11).
There was no detectable enrichment for double positives after the first sort (data not shown). After the second sort  (Fig. 2). The trapping reaches a maximum at 3.5 h and then decreases. After 5 h at 41 "C, the uptake of Tf during the 30-min pulse is only 35% of the uptake at 33 "C. Although there appears to be an optimal time at which to select for T f trapping mutants, the expression time may be different for each recycling mutant isolated.
Initial Characterization of Clones-To verify that the clones had temperature-conditional defects in receptor recycling clones were screened for temperature-dependent loss of surface TfR. The results are shown in Table I. Since these clones were derived from a single mutagenesis and carried as a bulk population for several generations after sorting it is possible that they may be derived from a single mutation. However,  Log FlTC Fluorescence one of the clones, TfT1.2, appears to be different from the others, as it shows a 30% reduction in Tf binding (relative to parental) even at the permissive temperature. The clone TfT1.11 was selected because it had normal levels of TfR at the permissive temperature and showed a significant loss of TfR at the nonpermissive temperature.
After a 2-h incubation at 41 "C, TfTl.11 shows a small   Fig. 3, A and B), and the majority of the cells are clearly positive for both dextran accumulation and retention of Cy5-Tf (Fig. 30). However, TfT1.11 shows some decrease in the extent of dextran labeling (Fig. 30). This defect in dextran accumulation is not caused by a difference in the initial rate of dextran internalization but by a difference in the efflux of fluid after endocytosis.' The retention of Cy5-Tf in TfT1.ll is temperature dependent, as there is a significant amount of Tf retained after chase at 41 "C ( Fig. 4B) but little trapping at 33 "C (Fig. 4C). TfI'l.ll is similar in size to WTB (Fig. a), in contrast to the majority of the cells after the second sort (see above). The phenotype of TfT1.11 is not caused by a gross alteration in cellular metabolism. There is no significant decrease in the levels of ATP in TfI'l.ll cells and no significant difference in the relative rates of protein synthesis after temperature shift ( Table 11).
Loss of Surface Transferrin Receptors- Fig. 5 shows the kinetics of loss of apparent cell surface Tf binding sites from TfT 1.11 cells during incubation at 41 "C. Within 4 h after the shift to 41 "C the binding decreased to less than 30% of the initial binding and less than 25% of the increased binding to WTB at the elevated temperature. The loss of receptors occurred with a t1r2 of approximately 2 h and began almost immediately after shifting the temperature. Scatchard analysis (Table 111) indicates that the loss of binding is caused by a decrease in the number of surface receptors rather than a change in receptor affinity. After a 4-h incubation at 41 "C there was no difference in the TfR affinity of WTB and TfT 1.   At 33 "C the kinetics of loss of labeled Tf were similar for WTB and TfTl.ll although the fraction of Tf which is not recycled (at least over 90 min) is higher in Tffl.ll (Fig. 6A).
First-order rate constants and maximum fraction lost were estimated by nonlinear least squares fitting of a single exponential (Table IV). The products of these values (the initial rate) were similar for WTB and Tffl.ll. At 41 "C the loss of Tf is accelerated in WTB whereas it is slowed in TfTl.11 (Fig. 6B). When cells are shifted to 41 "C the initial rate of recycling increases by a factor of 3 in the WTB cells, but a significant fraction of the Tf is cleared from the cells through a slow pathway (tllz = 27 min). In TfTl.11 cells recycling through the fast pathway is either absent or altered dramatically. When the data are fit using the same model as for WTB  3 h (B). Cells were chilled to 4 "C and surface Tf removed by incubation at low pH (pH 4.5). The cells were warmed rapidly in the presence of excess unlabeled Tf to allow recycling and were chilled to 4 "C at the indicated times to halt externalization. Data shown are mean fluorescence values and standard deviations for triplicate samples from three experiments ( n = 9) normalized to the calculated internal Tf at t = 0 (see "Experimental Procedures"). Curues are single exponentials fit to the individual data points for the 33 "C samples and two exponentials fit to the 41 "C data (see Table IV).

TABLE IV Kinetic constants from exponential fits of the Tf recychg data
The data in Fig. 6 were analyzed as described under "Experimental Procedures." For 33 "C samples, first-order rate constants were estimated from nonlinear least squares single-exponential fits of the data (model 2), allowing for a component that is not recycled from the cell (t,,> > 90 min). For the 41 "C samples, the slow component was fit by a second exponential (assumes that all of the Tf is recycled, model 3). Results of two alternative fits for TfT 1.11 at 41 "C are also shown. Calculated as the product of the rate constant (ln(2)/t,) and the fraction of Tf which is recycled through the fast (or only) component.
Model 3 but with the rate constants fixed to those obtained for the WTB 41 "C data.
only a single component is present a t 41 "C in TfT1.ll with a rate more than seven times slower than WTB. Regardless of which model is used, the analysis leads to the conclusion that the TfTl.ll mutant retains Tf and TfR at the nonpermissive temperature because of an alteration in the rate of receptor externalization.
Reduced az-Macroglobulin Uptake-To determine whether the Tf recycling defect in TfTl.11 is caused by a general defect in the recycling pathway, aZM accumulation was measured for TfTl.ll and WTB cells after the temperature shift. Cells were incubated for various times at 41 "C and labeled with Cy5-cupM for 6 min at 41 "C. Differences in azM uptake in these short incubations are presumed to reflect differences in the number of surface receptors, as there is no difference between the two cell lines in the amount of dextran internalized in a 6-min pulse a t either 33 or 41 "C.' Interestingly, T f T l . l l cells have 1.4 times as many a2M receptors as WTB cells at 33 "C (Fig. 7). a2M uptake is lost rapidly after the temperature shift (t1,2 of approximately 1.5 h) and declines by greater than 90% by 4 h. The loss of surface azM receptor indicates that the defect in TfT 1.11 is not specific for the TfR but is a pleiotropic defect in receptor recycling.

DISCUSSION
Most endocytosis-defective mutants described previously have been isolated by selection against some aspect of the endocytic pathway (16, [43][44][45][46]. In all of these selection schemes mutant cells were isolated by killing normal cells. We have isolated cell lines with temperature-conditional defects in receptor recycling during endocytosis using the trapping of labeled Tf as a marker selectable by fluorescenceactivated cell sorting. There are several advantages of using cell sorting for selection of mutants. Positive selection by sorting allows the isolation of new classes of mutants which would be difficult to isolate using negative selection techniques. Mutants are isolated under conditions that are not lethal, and the selection does not require the perturbation of the endocytic pathway in normal cells. Flow sorting has been used recently to select CHO cell lines with temperatureconditional defects in the expression of membrane glycoproteins on the cell surface (47).
After screening several clones for a temperature-conditional loss of surface TfR, Tffl.11 was chosen for further analysis. This cell line shows temperature-dependent trapping of labeled Tf at 41 "C ( Figs. 3 and 4) and a concomitant loss of surface Tf binding (Fig. 5). Scatchard analysis shows that this loss of binding is caused by a decrease in the number of surface receptors at the nonpermissive temperature (Table  111). Analysis of the clearance of internalized Tf showed that TfI'l.ll has a temperature-sensitive defect in the rate of receptor recycling (Fig. 6). After temperature shift, the initial rate of clearance of internalized Tf from TfTl.ll cells was reduced 5.5-fold compared with WTB cells, and 63% of the internal Tf was released through a very slow component ( tl/z = 67 min; Table IV). It is not currently known if the Tf released through the slow process was intact or degraded.
O'Keefe and Draper (44) described a mutant, AF192, isolated from mouse L cells by selecting for resistance to a diphtheria toxin-Tf conjugate, with an aberrant Tf cycle. Like WTB (0) and TfTl.ll (A) cells were incubated a t 41 "C for the indicated times and labeled with Cy5-a2M at 41 "C for 6 min. Nonspecific uptake was less than 10% for all samples. Data shown are averages of duplicate samples from two separate experiments ( n = 4) normalized to WTB at 33 "C.
TfT1.ll, this cell line shows a reduction in Tf surface binding which is caused by a shunting of 25% of the TfR to a pathway with a very slow rate of return to the cell surface ( t1/2 greater than 100 min). However, the loss of TfR from the cell surface in AF192 is not as severe as the loss in TfI'l.11 at 41 "C (a 25% reduction compared with a 82% reduction in T f T l . l l after the temperature shift). Like AF192 cells, TfTl.11 cells at 41 "C show an increase in the amount of Tf released through a slow pathway. Unlike AF192, however, there is also a 2.5-3-fold difference in the rate constants for recycling from both the fast and the slow pathways in TfI'l.11 (at 41 "C there is a fraction of Tf which recycles slowly in WTB cells as well). When the TfT1.11 data are fit using the WTB rate constants, 94% of the Tf appears to recycle through the slow pathway (Table IV). It is possible that these two cell lines have the same defect, with the difference in severity caused by differences between temperature-conditional and nonconditional alleles. However, it was not determined if the defect in AF192 was specific for Tf or affected other ligands as well.
There is considerable evidence that intrinsic properties of receptors affect recycling. A point mutation in the cytoplasmic domain of the epidermal growth factor receptor which eliminates kinase activity causes the receptor and a significant portion of the ligand to be redirected from the lysosomal pathway to the recycling pathway (23) whereas proteolytic processing of the insulin receptor at the cell surface results in redirection of the receptor from the recycling to the degradative pathway (48). Deletion of the extracellular growth factor homology region of the low density lipoprotein receptor inhibits acid-dependent ligand dissociation and receptor recycling (9). In addition, aggregation of receptors by multivalent ligands or antibodies causes a redirection of receptors to the lysosome (11,49). These data suggest that a defect in the TfR which causes aggregation or a change in conformation could result in the loss of TfR from the surface. However, the fact that at 41 "C TfTl.11 shows a loss in surface a,M-receptor as well as TfR indicates that the defect in TfT1.ll is not specific for TfR but is a more general defect in the pathway of receptor recycling.
Although there are intrinsic properties of receptors which determine whether they will be recycled, once that decision has been made there seems to be little distinction among receptors. The rates of externalization of several receptors, including Tf, a2M, and mannosylated proteins, are identical, suggesting that the receptors are moving as a single cohort (50). Treatment with growth factors and phorbol esters causes a rapid redistribution of many receptors including TfR, aZM receptor, mannose 6-phosphate receptor, and mannose receptor from an intracellular pool to the plasma membrane (26,27,51), because of an increase in the rate constant for receptor externalization (26,27). It is possible that TfTl.11 contains a defect in one of the components involved in regulating this rate. Isolation and further characterization of pleiotropic mutants defective in receptor recycling should lead to insights into the general mechanism of receptor recycling and permit identification of molecules required for this process.