Endocytosis of the transferrin receptor is altered during differentiation of murine erythroleukemic cells.

During terminal differentiation of murine erythroleukemic (MEL) cells, the number of surface transferrin binding sites per cell decreases dramatically, while steady-state ligand uptake and immunoblotting studies demonstrate that the total number of transferrin receptors per cell remains constant. Since the amount of protein per cell decreases 4-fold during this 4-day period, the amount of transferrin receptor protein, relative to total soluble cell protein, increases 4-fold during this time, suggesting continued synthesis of the receptor. Supporting this, we show that the amount of transferrin receptor transcript in equal amounts of total cell RNA also increases as differentiation proceeds. Uninduced cells maintain 52% of the total transferrin binding sites on the cell surface, whereas only 22% of the receptors are on the surface in 4-day induced cells. All ligand endocytosed by either uninduced or induced cells at 37 degrees C is rapidly and completely exocytosed from the cells, suggesting that all of the cellular receptors are cycling. These studies suggest that, during MEL cell differentiation, an increasing fraction of transferrin receptors are localized to the cell interior, but are nevertheless cycling to the cell surface. This observed redistribution is due to altered kinetic parameters of the receptor. Receptor-bound 125I-labeled transferrin ligand has been followed through a single endocytic cycle. Ligand internalization occurs much more rapidly in induced cells (t1/2 = 2.9 min) than in uninduced cells (t1/2 = 6.9 min). The rates for ligand movement back out to the cell surface and its subsequent release into the medium in both uninduced and induced cells are quite similar.

During terminal differentiation of murine erythroleukemic (MEL) cells, the number of surface transferrin binding sites per cell decreases dramatically, while steady-state ligand uptake and immunoblotting studies demonstrate that the total number of transferrin receptors per cell remains constant. Since the amount of protein per cell decreases 4-fold during this 4-day period, the amount of transferrin receptor protein, relative to total soluble cell protein, increases &fold during this time, suggesting continued synthesis of the receptor. Supporting this, we show that the amount of transferrin receptor transcript in equal amounts of total cell RNA also increases as differentiation proceeds. Uninduced cells maintain 52% of the total transferrin binding sites on the cell surface, whereas only 22% of the receptors are on the surface in 4-day induced cells. All ligand endocytosed by either uninduced or induced cells at 37 "C is rapidly and completely exocytosed from the cells, suggesting that all of the cellular receptors are cycling. These studies suggest that, during MEL cell differentiation, an increasing fraction of transferrin receptors are localized to the cell interior, but are nevertheless cycling to the cell surface. This observed redistribution is due to altered kinetic parameters of the receptor. Receptor-bound '2sI-labeled transferrin ligand has been followed through a single endocytic cycle. Ligand internalization occurs much more rapidly in induced cells (tu = 2.9 min) than in uninduced cells (tH = 6.9 min). The rates for ligand movement back out to the cell surface and its subsequent release into the medium in both uninduced and induced cells are quite similar.
The delivery of iron into the cell by transferrin constitutes the first step in the incorporation of this essential cofactor into a variety of polypeptides, most notably hemoglobin and several respiratory proteins (Vogt et al., 1969). By receptormediated endocytosis, iron-saturated transferrin is internalized into endocytic vesicles (Octave et al., 1983;reviewed in Goldstein et al., 1985). Iron is released from transferrin in these acidic compartments, while apotransferrin remains tightly bound to its receptor. Both the receptor and ligand then recycle together back to the cell surface, where, upon exposure to neutral pH, the apotransferrin dissociates from the receptor (Dautry-Varsat et al., 1983). Both apotransferrin and the receptor are reutilized. *This work was supported by Grants GM09566, HL27375, and GM35012 from the National Institutes of Health. 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. Specific membrane receptors for transferrin were initially demonstrated on immature erythroid cells, which require large amounts of iron for hemoglobin production (Jandl and Katz, 1963). However, during erythroid differentiation, the composition of the plasma membrane undergoes significant alterations (Skutelsky and Danon, 1970;Skutelsky et al., 1974;Ackerman and Clark, 1972); although the transferrin receptor is abundant in erythroblasts and reticulocytes, it is completely absent in mature erythrocytes (van Bockxmeer and Morgan, 1979;Frazier et al., 1982). The mechanism by which the transferrin receptor is cleared from the erythroid cell during maturation is unknown.
The transferrin receptor has been characterized in several cell types; it is a glycoprotein covalently associated with fatty acid (Omary and Trowbridge, 1981), with an apparent molecular weight of 95,000. It exists in the membrane as a disulfidebonded dimer (Seligman et al., 1979;Wada et al., 1979;Schneider et al., 1982;Van Agthoven et al., 1984).
In this paper we examine the transferrin receptor during differentiation of a murine erythroleukemic cell line (MEL' cells). The receptor is evaluated in relation to its effect on the intracellular cycling pathway of its ligand, transferrin. We find that the number of surface binding sites per cell for transferrin decreases dramatically during differentiation. However, the total number of transferrin receptor molecules per cell remains constant and, importantly, accessible to the intracellular receptor-ligand cycling pathway throughout the 4-day program of MEL cell differentiation. In fact, since cell volume and protein content decrease continuously during differentiation, there is continued synthesis of transferrin receptor and, we show, of transferrin receptor mRNA. Finally, we examine the kinetics of ligand internalization and ligand movement back to the cell surface. We find an altered rate of endocytosis of surface transferrin-receptor complexes in induced cells which can account for the observed redistribution of the transferrin receptor.

MATERIALS AND METHODS
Cells-"EL cells were maintained in suspension culture and induced to undergo differentiation, according to published procedures (Volloch and Housman, 1982). Briefly, differentiation was initiated by incubating cells at a density of 5 X lo4 cells/ml in Dulbecco's modified Eagle's medium supplemented with 1.8% (v/v) dimethyl sulfoxide, 13% (v/v) heat-inactivated fetal bovine serum, 5% (w/v) bovine serum albumin, 2 mM L-glutamine, and 1.8 mM Imferon. The extent of cell differentiation was assessed by determining the amount of globin present in cells induced for 4 days; sodium dodecyl sulfate gel-resolved protein species were scanned with a microdensitometer. The globin protein comprised at least 60% of the total celldar protein in 4-day induced cells. As expected, no globin was present in unin- The abbreviations used are: MEL, murine erythroleukemic; Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; kb, kilobase.

5455
duced cells. MEL cells, where indicated, were grown and induced to differentiate attached to fibronectin-coated dishes according to previously published procedures (Pate1 and Lodish, 1987). In all experiments involving fibronectin-coated dishes, only the cell population attached to the fibronectin was utilized. Cultures were grown at 37 "C in a humid 5% COz incubator, and cell numbers were determined using a Coulter Counter (Coulter Electronics, Inc., Hialeah, FL).
Transferrin-Human transferrin (Behring Diagnostics) was saturated with iron and radiolabeled with lZ5I using chloramine T as described previously (Karin and Mintz, 1981). Iodinated transferrin was stored at -80 "C in the presence of 0.5 mg/ml cytochrome c.
Binding Assays-All assays were carried out at either 4 or 37 "C, as indicated, using human 1261-ferrotransfemn. To assess transferrin binding at 4 'C, MEL cells, either uninduced or induced for up to 4 days, were washed twice at 4 "C with Hanks' saline solution containing 20 mM Hepes, pH 7.4 (binding medium). Cells were resuspended in this buffer to a concentration of 4 X 10' cells/ml and 250-pl aliquots were added to each well of a 24-multiwell tissue culture dish (Falcon Labware, Oxnard, CA). Binding was carried out in duplicate wells at 4 "C for 2 h with gentle agitation. Nonspecific binding was determined, also in duplicate, by addition of 50-100-fold excess of unlabeled transferrin just prior to addition of the radiolabeled ligand. Nonspecific binding remained below 10% of the specific binding in these assays. After 2 h, cells were processed with slight modifications of the procedure of Klausner et al. (1983). A 2 0 0 4 aliquot of the cell sample was layered over 150 pl of heat-inactivated fetal bovine serum in a 400-pl Eppendorf tube and centrifuged for 10 s at 12,000 X g. The tubes were frozen in liquid nitrogen, the tips containing the cell pellets were cut off, and the radioactivity of both the tip and the remaining tube containing the unbound ligand was determined using a Packard Auto-Gamma 500 y counter.
Steady state labeling of the transferrin receptor with '"I-transferrin was assayed at 37 "C according to this same procedure with the following changes. Cells were washed as above, then resuspended in buffer containing lo-' M ferric ammonium citrate to saturate exocytosed apotransferrin during the assay (Ciechanover et al., 1983a). Incubation with the ligand was carried out for 45 min at 37 "C, and cell samples were subsequently processed as described. Control wells containing no radiolabeled ligand were included in both assays (4 and 37 "C) and used for cell counting and protein assays (Lowry et al., 1951) upon completion of the binding incubations.
Internalization of 'Z6Z-Transferrin-The procedure detailing '%Itransferrin binding to its surface receptor and measurement of its subsequent internalization and recycling has been described (Ciechanover et al., 1983a;Karin and Mintz, 1981). Briefly, a saturating concentration of '=I-transferrin (250 nM) was allowed to bind to the surface of uninduced or 4-day induced MEL cells at 4 "C. Cells were washed several times to remove unbound ligand and incubated at 37 'C with prewarmed binding medium (1 X lo6 cells/ml) supplemented with unlabeled transferrin (500 nM). At the indicated times, the cells were chilled by addition of ice-cold phosphate-buffered saline, centrifuged for 30 s in an Eppendorf microfuge, and resuspended in 1 ml of 0.25% Pronase in binding medium. Following a 1h incubation at 4 "C, the cells were again centrifuged (1 rnin), and the radioactivity in the cell pellet, Pronase-released supernatant, and binding medium was determined by y counting.
Kinetics of Receptor Recycling-The transferrin receptor can be followed by tracing its ligand through the successive steps of one complete cycle of internalization and subsequent reappearance of iron-free transferrin in the medium. A kinetic model depicting the parameters of the cycling transferrin has been described in detail (Ciechanover et al., 1983a). Here, a simplified version of this model was utilized where k,, is the first order rate constant for dissociation of ferrotransferrin from the cell surface receptor, k, is the first order rate constant for internalization of the receptor-ligand complex, and k2 is the rate constant for movement of the receptor-ligand complex back to the cell surface. As apotransferrin dissociates from its receptor into the medium within 15 s (Dautry-Varsat et al., 1983), kz can be approxi-mated from the rate of appearance of apotransferrin from the cells into the medium. We used the following equations to describe these events: where S equals the fraction of total ligand on the cell surface, C represents the fraction of ligand inside the cell, and M is the fraction of ligand released from the receptor into the medium, either directly by dissociation from surface receptors (h) or upon completion of one cycle of endocytosis (4). The half-times, 4 , for the above equations are equal to In 2lk.
We calculated the sum (h + kl) by measuring the first order rate of disappearance of labeled ligand from the cell surface. We then solved for the slope kl in a graph plotting C(t) versus l/(k,, for the first 4 min, during which time exocytosis of internalized ligand is minimal (Ciechanover et al., 1983a). We assumed the rate of ligand dissociation from the cell surface, h, to be the same in both uninduced and induced cells and confirmed this assumption by calculating k,,, from both induced and uninduced cells, by subtracting kl from (h + kl). Finally, we computed kz, the rate of ligand movement back to the cell surface and its subsequent dissociation from the cell, by following the disappearance of ligand from the cell interior into the medium.
Using these constants, we calculated a curve predicting a single cycle of endocytosis of the receptor by solving the differential equation e-bl. We subtracted background values of M and C at 0 min, then normalized the data so that at t = 0, S = 1, C = 0, and M = 0. The fraction of surface ligand, intracellular ligand, and ligand released to the medium is plotted, normalized to the value of 1.0. We then compared our calculated curve to the observed experimental data.
Northern Gel Analysis-Uninduced and induced cells were grown both in suspension cultures and on fibronectin-coated dishes, as indicated. Total RNA was isolated by guanidine thiocyanate extraction followed by centrifugation through a cesium chloride cushion (Chirgwin et al., 1979). Equal amounts of RNA (20 pgllane) were electrophoresed on 1.0% agarose, 6% formaldehyde gels, and transferred to a nylon membrane (Biotrans, ICN). The membrane was hybridized overnight at 42 "C in 35% formamide, 5 X SSC to the nick-translated cDNA probe, a 4.9-kb BamHI fragment of pCD-TR1 encoding the human transferrin receptor (a generous gift from Dr. F. Ruddle, Yale University, New Haven, CT) (Kuhn et al., 1984). The transfer and hybridization conditions were as recommended in the ICN Biotrans instruction booklet. After hybridization, the membrane was washed in 0.1 X SSC, 0.1% sodium dodecyl sulfate at 50 "C, then subjected to autoradiography.
Gel Electrophoresis and Zmmunoblotting Procedure-Total cell protein was dissolved in sodium dodecyl sulfate, subjected to polyacrylamide gel electrophoresis in sodium dodecyl sulfate according to the method of Laemmli (1970), using a 10% (w/v) acrylamide separating gel. It was then transferred electrophoretically to sheets of nitrocellulose (Towbin et al., 1979;Braell and Lodish, 1982). The nitrocellulose sheet was reacted sequentially with goat anti-human transferrin receptor antibody, rabbit anti-goat IgG, and lZ6I-protein A. The nitrocellulose sheet was thoroughly washed after each incubation, dried, and exposed to Kodak XAR film. Quantitation of the transferrin receptor at each stage of cell induction was accomplished by excising each band from the nitrocellulose and determining the associated radioactivity with a y counter.
Reagents-Iron-free human transferrin was purchased from Behring Diagnostics. Na1261 and affinity-purified 'Z61-protein A labeled with Bolton and Hunter reagent (40 mCi/mg) were purchased from Amersham Corp. Dulbecco's modified Eagle's medium and fetal bovine serum were obtained from GIBCO. All other reagents used in these studies were purchased from standard commercial sources. transferrin receptor antibodies. This antibody is able to crossreact with the murine transferrin receptor; it specifically immunoprecipitated the transferrin receptor from both 35Slabeled human hepatoma (HepG2) cells and MEL cells yielding a single labeled polypeptide band of the same molecular weight (95,000) (data not shown). Visualization of the receptor band in the Western blot was enhanced by incubating with rabbit anti-goat IgG prior to addition of lZ51-protein A. Upon autoradiography (Fig. l), both the uninduced and the induced cells exhibited a single band of M, -95,000 present in equal amounts regardless of the extent of cell differentiation, indicating that the total number of transferrin receptors per cell remained constant throughout the differentiation period. This conclusion was confirmed by quantitating the amount of radioactivity in each band excised from the nitrocellulose. The transferrin receptor band from uninduced cells contained 450 cpm of lZ51; 2-day induced cells, 490 cpm; 3-day, 452 cpm; 4-day, 445 cpm; &day, 464 cpm.

Immunological
However, as differentiation proceeds, the cells divide and the cell volume decreases significantly (Volloch and Housman, 1982). The amount of total soluble protein per cell decreases approximately 3.5-fold from 0-to &day induced cells? Although the amount of transferrin receptor per cell remains constant throughout differentiation (Fig. l), the amount of receptor protein relative to total cellular protein actually increases during this time (Fig. 2). To show this, equal amounts of total soluble cell protein (200 pg) were examined by immunoblotting techniques and the amount of transferrin receptor in each band increased as differentiation progressed. The radioactivity in each excised band was as follows: uninduced cells, 370 cpm of lZ5I; 3-day induced cells, 674 cpm; 4-day induced cells, 914 cpm; 5-day induced cells, 1337 cpm. There is an overall increase of 3.6-fold in the amount of transferrin receptor from 0-5 days suggesting that the rate of synthesis of the transferrin receptor may increase throughout differentiation.
Northern Analysis of Transferrin Receptor mRNA-We carried out Northern analyses on RNA from both uninduced and induced MEL cells to examine the possibility that the transferrin receptor continues to be synthesized during cell differentiation. Probing Northern blots of equal amounts of total cellular RNA with human transferrin receptor cDNA revealed the 28 and 15 S rRNA bands as well as a 4.9-kb putative transferrin receptor transcript. Upon washing these blots under high stringency conditions, the rRNA bands disappeared while only the 4.9-kb transcript was retained (Fig. 3). The amount of transferrin receptor mRNA, isolated from cells differentiated in suspension, increased as differentiation progressed. Attachment of MEL cells to a fibronectin monolayer allows differentiation to proceed longer than in suspension; after 6 days, most of the cells enucleate and form reticulocytes (Pate1 and Lodish, 1987). Fig. 3 also shows an increase in the transferrin receptor mRNA up to 6 days of differentiation on fibronectin. These observations support the proposal that the transferrin receptor continues to be synthesized during cell development, perhaps at an increased rate.
Binding of Transferrin to the Surface of MEL Cells-We next examined receptor function and distribution in MEL cells throughout the differentiation period. The number of functional transferrin binding sites on the surface of MEL cells was determined by binding lZ51-ferrotransferrin for 2 h at 4 "C to prevent endocytosis. Data in Fig. 4 and Table I show that the number of transferrin receptors on the cell surface decreased significantly as cell differentiation pro- Cells were incubated with various concentrations of lZ61-transferrin at 4 'C for 2 h, as detailed under "Materials and Methods." The cells were subsequently separated from unbound ligand by centrifugation through a layer of fetal bovine serum. The data shown each represent the mean of duplicate determinations of cell-associated radioactivity corrected for nonspecific binding. Nonspecific binding was also determined in duplicate samples in the presence of 50-100-fold excess of unlabeled ligand and never comprised more than 10% of the specific binding. Data are expressed as molecules of transferrin per cell uersus ligand concentration. 0, binding to uninduced cells; A, 3-day induced cells; 0, 4-day induced cells.

Number of transferrin binding sites
The numbers of functional binding sites, both surface receptors and total cellular receptors, were determined either through binding of lZ61-transferrin at 4 "C or ligand uptake studies carried out at 37 "C (see "Materials and Methods"). The binding data were subjected to Scatchard analysis and fitted using linear regression analysis to obtain the Kd value. The total numbers of receptors per cell were obtained from the saturation points of the data from the 37 "C ligand uptake studies.

Cells receptors/cell
No. of (X 106) K d

(X1O4 M)
Surface (4 'C) Uninduced 3.2 11.6 3-day induced 1.4 6.1 4-day induced 1.1 6.9 days. Further, the receptors remaining on the surface of induced cells were of the same, or perhaps slightly higher, affinity than the receptors of uninduced cells (Table I). These assays were also performed using 1251-mouse transferrin as ligand and identical results, with respect to both receptor number and affinity, were obtained. However, the assays using '251-h~man transferrin exhibited slightly lower nonspecific binding, so this ligand was used throughout these studies. Other investigators have also examined the specific binding of human transferrin to mouse teratocarcinoma stem cells (Karin and Mintz, 1981) and to MEL cells (Wikzynska and Schulman, 1980) and have reached the same conclusion.

Total Number of Functional Transferrin Receptors in MEL
Cells-The decrease in the number of transferrin receptors on the surface of induced cells prompted examination of the total number of functional receptorsjcell. Cells were incubated with 1251-ferrotransferrin for 45 min at 37 "C, to allow steady-state accumulation of the receptor-ligand complex. When steady state is reached, the number of cell-associated ligand molecules should equal the total number of cycling transferrin binding sites. The data depicted in Table I demonstrate that, after a 45-min incubation with the ligand, each uninduced cell contained 6.2 x lo5 transferrin molecules. Since these cells have 3.2 X lo5 surface receptors, 48% of the receptors for transferrin were localized intracellularly. However, in 4-day induced cells, 78% of the receptors were intracellular. Further, the total number of cellular transferrin molecules, and thus functional transferrin receptors, remained constant throughout a 4-day program of cell differentiation (Table I). This result, combined with the finding that induced cells exhibited decreased numbers of cell surface receptors, but a constant amount of immunoreactive transferrin receptors (Fig. l), suggested that cell differentiation was accompanied by selective internalization of receptor molecules, rather than actual loss of receptors from the cell.
In the ligand uptake study described above, it was important to determine whether all cell-associated transferrin was cycling, rather than being sequestered in an intracellular compartment inaccessible to the cycling pool. Thus uninduced cells and cells induced for either 2 or 4 days were incubated with 1Z51-transferrin for 2 h at 37 "C; unbound ligand was removed and the samples were reincubated with the same concentration of unlabeled transferrin for 30 min. Cell-associated radioactivity was reduced by over 95% of its initial value, demonstrating that all internalized transferrin, presumably together with their bound receptors, were able to return to the cell surface; in the process, all of the lZ5I-labeled ligand was released to the medium and the transferrin receptors presumably rebound unlabeled transferrin.
Kinetics of Receptor-bound 1251-Transferrin-T~ elucidate the mechanism by which the observed redistribution of the transferrin receptor during differentiation occurs, the receptor was followed through a single cycle of endocytosis by tracing its labeled ligand, '251-transferrin. A saturating amount of 'Ttransferrin was bound to the surface of uninduced or 4-day induced MEL cells at 4 "C. Unbound ligand was removed, and the cells were incubated for various times at 37 "C in binding medium supplemented with unlabeled transferrin. After quick chilling, the cells were recovered by centrifugation and incubated with Pronase for 1 h at 4 "C. Samples were subsequently centrifuged and the radioactivity associated with the cell pellet after Pronase digestion (internal, C), the Pronasereleased supernatant (cell surface, S), and the binding medium ( M ) was determined. The Pronase-sensitive fraction contained surface ligand, accessible to proteolysis, whereas the intracellular ligand was protected from Pronase digestion and was recovered in the cell pellet. The radioactivity in the medium consisted of ligand (presumably apotransferrin) released from the cell at the completion of the single cycle of endocytosis, as well as surface-bound ligand (holotransferrin) that dissociated directly into the medium without undergoing endocytosis.
In uninduced cells, approximately 20% of the surface-bound I-transferrin was internalized within 5 min and was subsequently exocytosed into the medium (Fig. 5 ) . The rate constant, kl, describing the internalization event was calculated from the data to be 0.10 min". The rate constant for movement of the internalized ligand back to the cell surface, kz, was found to be 0.31 min". Finally, the value of &, the rate FIG. 5. Calculated and observed recycling kinetics of receptor-bound 1a61-transferrin in uninduced MEL cells. 1251-transferrin was bound to uninduced cells at 4 'C as described under "Materials and Methods." Unbound ligand was removed and cells were incubated at 37 "C in binding medium supplemented with unlabeled transferrin. At the indicated times, the cells were chilled in phosphate-buffered saline and treated with Pronase. The radioactivity in the Pronase-resistant fraction (0, intracellular), Pronase-sensitive fraction (0, surface), and the medium (A) was determined. At t = 0 , 4 % of the cell-associated radioactivity could not be removed by Pronase. This background value was subtracted from all data points of intracellular radioactivity. The data points for the amount of cell surface radioactivity were also normalized to a value of 1.0 at t = 0 min. The calculated curves (shown in solid lines) were obtained from the integrated form of the differential equation describing our model of endocytosis and recycling (see "Materials and Methods") and were solved for a single endocytic cycle using the normalization so that at t = 0, surface = 1.0, intracellular = 0, and medium = 0. The rate constants obtained are k,, = 0.050 rnin", kl = 0.10 min", and k2 = 0.31 rnin". of dissociation of ligand from the cell surface into the medium, was 0.050 rnin".
These rate constants, together with the integrated form of the differential equation describing our simple model of recycling (see "Materials and Methods") generated calculated curves, depicted in Fig. 5, as solid lines, superimposed on our observed data points. The close agreement of our data with the predicted fit confirms that our model is applicable, and that the calculated rate constants are valid.
This same procedure was followed for 4-day induced cells, as seen in Fig. 6. Importantly, the rate of internalization of ligand, kl, in these cells was 0.24 min-l, a dramatic increase from that observed in uninduced cells. (Graphically, this can be seen in the more rapid drop in the level of surface-bound ligand, and the more rapid and larger increase in the fraction of intracellular ligand.) Further, the rate of exocytosis, b, was 0.27 rnin", just slightly slower than that of uninduced cells (0.31 min"). Finally, the dissociation rate constant, k,,, in the 4-day induced cells was not significantly different from uninduced cells, as expected (0.040 min"). These data suggest an increased rate of receptor-bound ligand internalization in the induced cells and an equivalent or slightly slower rate of movement back to the cell surface and subsequent ligand release into the medium. This would result in a net redistribution of the transferrin receptor to the interior of the cell in 4-day induced cells of the magnitude expected from Table I.

DISCUSSION
Upon treatment with dimethyl sulfoxide, murine erythroleukemic cells are induced to differentiate, providing an ex- The experimental details are as described in Fig. 5. The observed data are depicted as: 0, intracellular; 0, surface; A, medium, while the calculated curves are superimposed on this data in solid lines. The rate constants used are = 0.040 min", kl = 0.24 min", and = 0.27 rnin". At t = 0, 3% of the cell-associated radioactivity could not be removed by Pronase, and consequently this value was subtracted from all data points of intracellular radioactivity. The data points for the amount of cell surface radioactivity were normalized to a value of 1.0 at t = 0 min. tremely useful cell system in which to examine the transferrin receptor during erythroid differentiation. MEL cells proliferate in culture as large, nucleated, hemoglobin-poor cells arrested at a developmental stage resembling proerythroblasts (Gordon and Rubin, 1982). Treatment with dimethyl sulfoxide or a variety of other agents (Marks, 1978;Nishioka and Silverstein, 1978) induces MEL cell differentiation which strongly parallels normal erythropoiesis and results in the appearance of globin mRNA (Ross et al., 1972), the accumulation of hemoglobin and other erythrocyte-specific proteins (Eisen et al., 1977, a and b), and eventually the cessation of cellular division and nuclear activity (Gaedicke et al., 1974). In suspension culture, differentiated MEL cells are nucleated and nondividing, contain large amounts of hemoglobin and developmentally resemble small orthochromatic erythroblasts (Eisen et al., 1977a;Gordon and Rubin, 1982) and reticulocytes (Volloch and Housman, 1982). The terminally differentiated cells (at 4 days) used in our experiments with suspension cultures resembled late erythroblasts. Cells that were induced to differentiate on fibronectin-coated dishes continue to differentiate to the reticulocyte stage (Pate1 and Lodish, 1987).

Induced Cells
MEL cells grown in culture in the absence of any inducer possess approximately 3.2 X 10' transferrin receptor molecules on the surface of each cell (Table I). This is similar to the number of receptors found on the surface of an uninduced human fetal erythroleukemic cell line, K562 cells (Hunt et al., 1984).
Results from 4 "C binding and 37 "C '251-transferrin uptake studies demonstrate that about 52% of the total number of functional cellular receptors in steady state are localized on the surface of uninduced cells; 6.2 X 10' total receptors/cell were detected. This result demonstrates the existence of an intracellular pool of receptors; a finding which has been noted in other cell systems as well. A human hepatoma cell line, HepG2, maintains about two-thirds of the total transferrin receptors inside the cell (Ciechanover et al., 1983b), 44% of Redistribution the total receptors in human K562 cells are localized intracellularly (Hunt et al., 1984), and the majority of transferrin receptors in cultured HeLa cells are also found inside of the cell (Lamb et al., 1983).
Following induction of MEL cells, the number of surface transferrin receptors decreases by at least %fold over a 4-day differentiation period while the total number of receptors per cell remains constant (Fig. 1, Table I). Immunofluorescence microscopy of the receptor in uninduced and induced cells also reveals a dramatic decrease in the number of receptors on the surface of induced cells relative to those on the membrane of uninduced cells (data not shown). A reduction in the number of surface receptors during hemin-induced differentiation of human K562 cells has also been demonstrated by Hunt et al. (1984). These results are in contrast to results of Hu et al. (1977), who found a doubling in the number of receptor molecules on the surface of MEL cells, upon induction, as measured by transferrin binding activity at 37 "C. However, these investigators used 60% iron-saturated transferrin in their assays, rather than 100% iron-saturated transferrin, reporting that the latter ligand gave variable results.
In an attempt to discern a mechanism by which the receptor can undergo this redistribution, we examined the kinetic parameters of the receptor. We found that 4-day induced cells internalized receptor-bound "'I-transferrin 2.4-fold faster than uninduced cells (Figs. 5 and 6). In addition, the rate of ligand movement back to the surface and its release into the medium in induced cells was just slightly slower (0.27 min-') than the rate in uninduced cells (0.31 min-'). Therefore, this would generate an alteration in the steady-state distribution of transferrin receptors in induced cells; an increased fraction of receptors inside the cell. Further, all transferrin receptors remain functional in the ligand-receptor pathway throughout differentiation. Uninduced and induced cells incorporated equal amounts of '251-ferrotransferrin at 37 "C and subsequently exocytosed the internalized '251-labeled transferrin completely from the cell.
Importantly, we showed here that the reduction in surface receptors results from an increased rate of internalization of the receptor, rather than a change in the affinity of surface receptor for its ligand (Table I) or actual loss of these molecules from the cell. The total number of receptors per cell remains constant throughout differentiation, as measured by binding assays (Table I) and immunoblotting procedures (Fig.  1).
Selective phosphorylation of the transferrin receptor may be a mechanism by which the receptor could be more rapidly endocytosed by the cell. Hunt et al. (1984) have reported differences in phosphorylation of the human K562 transferrin receptor as hemin-induced differentiation proceeded, noting a shift to the more phosphorylated forms of the receptor. This result is supported by the observation that, using phorbol esters to activate protein kinase C, phosphorylation of the transferrin receptor is accompanied by its internalization and a down-regulation of surface receptors (May et al., 1984;Klausner et al., 1984). It is probable that phorbol esterinduced receptor phosphorylation is regulating the subsequent receptor internalization and redistribution. In fact, this downregulation is completely reversible, and is accompanied by dephosphorylation of the receptor (May et al., 1984). Similar results were obtained by Hunt and Marshall-Carlson (1986), treating human erythroleukemic K562 cells with the drug trifluoperazine, which inhibits calmodulin-dependent and calcium-activated phospholipid-dependent kinases. The number of surface transferrin receptors decreased by approximately half, and this was reversible upon removal of the drug.
Perhaps increased phosphorylation of the transferrin receptor during differentiation promotes increased endocytosis of bound transferrin, causing a redistribution of the receptor population. The net result leaves less receptors on the surface at any given time during differentiation, but all receptors are still cycling.
MEL cells divide twice during the early stages of differentiation; they become smaller (Volloch and Housman, 1982) and contain 3-4-fold less protein per cell.' The number of transferrin receptors/cell remains constant during this time (Fig. 1). However, the number of transferrin receptors per cell protein mass increases (Fig. 2), as does the abundance of transferrin receptor mRNA per total cellular RNA (Fig. 3). Thus MEL cells are incorporating more transferrin ligand, normalized to cell protein, as differentiation proceeds. The requirement for hemoglobin (and thus, iron) increases during differentiation, which may provide the rationale for the increase in the number of transferrin receptors, relative to the amount of cell protein, and also the increased rate of endocytosis of the ligand-receptor complexes.
During erythropoiesis, some mechanism must exist to clear the cell membrane of transferrin receptors, since they are not present in mature erythrocytes. However, although a number of mechanisms have been proposed, the ultimate fate of the transferrin receptor molecule during cell maturation is controversial (Hunt et al., 1984;Pelicci et al., 1982;Testa et al., 1982, Harding et al., 1985. The experiments presented here describe the recycling events of the transferrin receptor during the first 4 days of differentiation. This may be too early in the differentiating program to provide any clues as to the mechanism by which the receptor is finally lost from the cell. However, the MEL cell system now established in our laboratory (Pate1 and Lodish, 1987), which allows synchronous differentiation of MEL cells through the enucleation stage, is an ideal system in which to examine the ultimate fate of the transferrin receptor.