THEJOURNALOFBIOLOGICAL C H E M I S T R Y Vol. 256, No. 7, Issue of April 10, pp. 3245-3252, 1981 Prinled in U.S.A.

Receptor-mediated Endocytosis of Transferrin inDevelopmentally Totipotent Mouse Teratocarcinoma StemCells* (Received for publication, September 22, 1980)

Michael Karin and Beatrice Mintz From The Institute for Cancer Research, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111

especially critical role in erythroiddifferentiation and function in the embryo as well as in later life (5, 6). The fist step in the delivery of iron to cells by transferrin involves its binding to a specific cell-surface receptor ( 7 ) . Among the kinds of cells examined, transferrin receptors are most abundant on placental cells (5) and reticulocytes (€9, tissues which are most active in iron uptake; but they have also been demonstrated on embryonic lung cells, splenic lymphocytes, choriocarcinoma cells, and nasopharyngeal carcinoma cell lines (8),as well as on B- and T-lymphoblastoid cell lines (9), mammary tumor cells (lo), andfibroblasts (11). The transferrin receptors from the rabbit reticulocyte membrane (12) and the human placental brush border membrane (13) have been solubilized, and the latter even purified to homogeneity (14). Although iron delivery to cells seems clearly to begin with the binding of transferrin to its receptor, the next step has remained in doubt. A possibility suggested by some investigators is that iron is removed from transferrin while the complex is still bound to thecell-surface receptor, whereupon the iron combines with a different acceptor found in the stroma (of reticulocytes); it is then perhaps mobilized from the stroma by an intracellular binding protein (15, 16). Other investigators have presented morphological evidence for the intracellular localization of iron-labeled transferrin, and have postulated instead that iron-laden transferrin enters the cell by endocytosis and delivers the iron to intracellular binding sites (17, 18). Whatever the route of transferrin-mediated iron incorporation into cells, a defect in any one of the steps would in all likelihood lead to an anemia, and possibly to some other metabolic disorders as well. In the mouse there are at least two genetic diseases which might result from lesions in this pathway: hereditary microcytic anemia (19) and sex-linked anemia (20); still-unrecognized candidate diseases may exist in the humanpopulation. The cultured malignant stem cell of mouse teratocarcinomas present unique experimental possibilities for genetic disTransferrin is the chief iron-transport protein in mamma- section and developmental analysis of this andother receptorlian blood, its principal function being the movement of iron mediated pathways. These cells can be stably channeled into from absorption sites (in the intestine) and storage sites (in normal differentiation if they are microinjected into normal liver, spleen, and bone marrow) to sites of iron utilization (for earlyembryos at the blastocyst stage (21). Tumor-lineage reviews, see Refs. 1-3). Judging from observations on cells in cells then participate, along with embryo-derived cells, in the culture (4),it is likely that virtually all types of cells require complete development of tumor-free mice in which the reiron for which transferrin as aniron carrier is largely respon- spective strains arerecognizable by means of cellular genetic sible. Transferrin would nevertheless be expected to play an markers. The tumor straincan contribute to formation of all somatic tissues and, in the best cases, to germ-line progeny * This research was supported by United States Public Health carrying the genes of that strain. The capacity of the teratoService Research Grants HD-01646, CA-06927, and RR-05539, and by carcinoma cells to undergo full development in vivo has served an appropriation from the Commonwealth of Pennsylvania. The costs of publication of this articlewere defrayed in part by the payment of as a basis for the proposal that they be used to produce mice Page charges. This article must therefore be hereby marked ‘‘duer- with preselected mutational changes, some of which would provide models of human genetic diseases (22).Thus, thestem tkement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cells could be mutagenized while in culture,selected for clones

The transferrin-mediated pathway of iron uptakehas been characterized in mouse teratocarcinoma stem cells in culture. These cells were chosen for the characterization because of their unique capacity t o be converted to normalcy and to undergo complete and orderly differentiation into all tissues after microinjection into early mouse embryos. Therefore, the cells provide the possibility of experimentally analyzing in vitro and in vivo the progressive developmental expression of genes, both normal and mutant, controlling the transferrin pathway. In a new developmentally totipotent euploid cell line, the teratocarcinoma cells were found to have 5.7 lo3high affinity (Kd = 6.7 n ~ specific ) cell-surface receptors for transferrin. Surface binding of transferrin to the receptors was documented at 4 “C, when the cells do not take in significant amounts of iron. At 37 “C, both surface binding and internalization occur, yielding large amounts of intracellular iron. Receptor-bound transferrin isinternalized rapidly, at the rate of one full complement of cell-surface receptors every 6 min, and is an energy-requiring process. The lysosomotropic agents W C l and chloroquine inhibit iron uptake into thecells without inhibiting internalization of transferrin, a resultsuggesting that thelysosome is an important intermediate in the iron-uptake pathway. The low pH in the lysosome can account for dissociation of iron from the apoprotein and of the transferrin from its receptor. The apotransferrin molecule escapes lysosomal degradation and is released intact into the medium by the cells, thereby becoming available for another cycle of iron transport. Itis now possible to design selection strategies for the isolation of mutant teratocarcinoma stem cells carrying different genetic lesions affecting the transferrin-mediated pathway as described here. The genetic variants would enable construction ofmice carrying specific iron transport defects.

3245

3246

Transferrin Receptor-mediated Endocytosis of

with the mutation of interest, and converted by blastocyst injection into the full range of specialized tissues in mice. Preliminary experiments aimed at production of mice with a specific receptor deficiency have been described for the low density lipoprotein receptor system (23). Mice with heritable defects in the transferrin pathway would enable us to learn how specific lesions affect the development and metabolism of the whole animal and not just onekind of cell. In order to be able t o generate such disease models, it is first necessary to characterize the pathway in the culturedteratocarcinoma cells. We report here that mouse teratocarcinoma stem cells do in fact express high affinity transferrin receptors in culture. Moreover, our biochemical evidence strongly suggests that receptor-mediated endocytosis of iron-laden transferrin is the route by whichiron enters the cells. MATERIALS AND METHODS

Cells-All experiments were carried out with a new euploid line of mouse teratocarcinoma stem cells that was recently established in this laboratory from a newly induced (embryo-derived) tumor of the 129/sv inbred strain. The line, whose characteristics will bedescribed in greater detail in other publications, has a high frequency (over 90%) of karyotypically normal cells' and is developmentally totipotent after injection into embryos.' The cells ordinarily grow without a feeder layer, in medium containing fetal calf serum, but simple test media free of serum (as willbe described) wereusedfor binding assays in the present experiments. In all cases, 300,000-500,OOO cells/ dish were plated. Iron Saturationof Transferrin-10 mg of human transferrin (ironfree, from Sigma) were dissolved in 1 ml of 0.1 mg/ml ofFeNHd citrate in phosphate-buffered saline, incubated at room temperature for 3 h, and dialyzed overnight, with 2 changes, against phosphatebuffered saline at 4 "C. Saturated transferrin concentration was calculated by measuring absorption at 465 nm (E?c, = 0.57). Iodination of Transferrin-Two variants of iodinated transferrin were prepared, one with radioactive iodine and nonradioactive iron, the other with non-radioactive iodine and radioactive iron. In the first case, 50 pg of iron-saturated transferrin were iodinated with 2 mCi of lZ6I(as sodium iodide, carrier-free, obtained from New England Nuclear) by a modification of the chloramine-T method (24). The changes which were introduced were: only 5 pl of 5 mg/ml of chloramine-T in 0.05 M sodium phosphate, pH 7.5 buffer, were used; and the iodination itself lasted only 1 minbefore termination of the reaction. The iodinated transferrin was separated from the free iodide on a 10-ml Sephadex G-50 column equilibrated with phosphatebuffered saline containing 1 mg/ml of gelatin. The iodinated transferrin was stored in this buffer at -80 "C and was used within 3 weeks after preparation. Specific activity was usually about 4.10' cpm/nmol of protein. We have iodinated by the same procedure, but with nonradioactive sodium iodide, ['Fe]transferrin (prepared as described). When its biological activity, measured as iron uptake by cells, was compared with that of noniodinated PFeItransferrin, the iodination was found to have no effect on the rateof iron uptake. Iodinated transferrin thus apparently behaves identically with native transferrin. Labeling of Transferrin with "Fe-Iron-free human transferrin was saturated with 59Fe(as ferric chloride, 30 Ci/g of iron, obtained from New England Nuclear) using the nitrilotriacetate method (25). C"YFe]transferrin was separated from ["qFeInitrilotriacetate by chromatography on a 10-ml Sephadex G-50 column equilibrated with phosphate-buffered saline. No "Fe dissociated from transferrin during overnight dialysis against phosphate-buffered saline. 126 I-transferrin BindingAssay-Mouse teratocarcinoma stem cells growing on 60-mm tissue culture plates were incubated with various amounts of '"I-transferrin in Hanks' buffered saline plus 10 m~ 4-(2hydroxyethy1)-I-piperazineethanesulfonicacid and 1mg/ml of bovine serum albumin, at pH 7.25 (binding buffer), in a total volume of 1 ml. At the end of the incubation period, the radioactive medium was removed by aspiration and each plate was washed 5 times with icecold binding buffer. Whenever the cells were not treatedwith pronase subsequent to thewashing, they were solubilized in1 ml of 1 N NaOH

B. Mintz and C. Cronmiller, unpublished data.

' T. A. Stewart and B. Mintz, unpublished data.

and removed for counting in an IntertechniqueModel CG4000 gamma counter. Unless otherwise indicated, all assays were done in duplicate. Nonspecific binding was determined as theamount of "'I-transferrin bound in the presence of12.5 ~ L Mof unlabeled transferrin; this was always deducted from the amount of total binding to determine specific binding. Nonspecific binding ranged from 6-15% of total binding. Pronase Treatment of Cells-After washing off excess unbound transferrin, each plate was incubated with 1 ml of 0.25% pronase (B grade, from Calbiochem) in binding buffer for 60 min at 4 "C. At the end of the incubation period, 100 pl of fetal calf serum were added and the cells were completely detached by repeated pipetting with a Pasteur pipette. The cells were then transferred to a microfuge tube and pelleted by centrifugation for 1 min in an Eppendorf microfuge. The supernatant was separated from the cellular pellet and each was counted for level of radioactivity. RESULTS

Time Course of Transferrin-mediuted 59FeUptake-In order to determine the time course and the temperature dependence of transferrin-mediated iron uptake,teratocarcinoma cells (3. lo6 cells/plate) were incubated with 22 nM of [59Fe]transferrinfor different time periods at 37 "C and also at 4 "C. The cells were then washed to remove unbound [59Fe]transferrinand solubilized in 1 N NaOH. Nonspecific binding determined in the presence of 12.5 IJM of unlabeled transferrin did not exceed 5% of total binding. A t 37 "C, there was a time-dependent increase in the amount of 59Feassociated with the cells. The uptake of "Fa was linear in t h e fist 60 min, then slowed down slightly and reached a steady state after 2 h (Fig. 1). At 4 " C ,the uptake of "Fe by the cells was substantially lessand the steady statewas reached bythe first 15 min (Fig. 1). A t 2 h of incubation, there were approximately 36 +. 2. lo4 iron atoms taken up/cell at 37 "C and only about 1.4 k 0.2. lo4 iron atoms/cell at 4 Time Course of '251-transferrinAssociation with Terutocarcinoma Cells-Since the iron atoms are not covalently bound to the transferrin molecule, the kineticsof iron uptake might be different from the uptake kinetics of the protein moiety of the molecule. Therefore, we next determined t h e

"c.

A

2,000

-370

2 1 1,500

@P

I

TIME ( M i n )

FIG. 1. Time courseof transferrin-mediated "Fe uptake into mouse teratocarcinoma stemcells. Cells on 60-mm tissue culture plates were incubated with 22nM ["'Feltransferrin in binding buffer. To measure nonspecific binding, cells were incubated with 22nM [5YFe]transferrinin the presence of 12.5 p~ unlabeled transferrin. The At the end incubation temperature was either 37 "C (0)or 4 "C (0). of the incubation period, the cells were rinsed four times with ice-cold binding buffer, solubilized in 1 ml of 1 N NaOH, and the "Fe content measured in a gamma counter. All data shown are averages of duplicate determinations (which were usually within 5% of each other) and are corrected for nonspecific binding (which did not exceed 5%).

Receptor-mediated Endocytosis binding kinetics of iodinated transferrin at three different temperatures (Fig. 2 A ) . At 4 "C, '251-transferrinrapidly binds to the cells, reaching a steady state after about 10 min of associated incubation. At 37 "C,the amountof 1251-transferrin with the cells is about 3-fold higher and begins to approach a steady state only after 30 min of incubation. At 24 "C, the rate of association of transferrin with the cells is slower than at 37 "C, but the amount associated with the cells seems to approach the amount associated at 37 "C. The incubation of cells with 1251-transferrinwas continued for up to 3 I/i h at 37 "C (Fig. 2 B ) . After the steady state was reached at about 50 min, there was little change in the amount of transsferrin associated with the cells. Therefore, no down regulation (27) of transferrin binding was apparent. Determination of Number of Receptors and Binding Affinity-The number of transferrin receptors on teratocarcinoma cells and their affinity for lZ5I-transferrinwas determined by

of Transferrin

3247

analyzing our binding data by the method of Scatchard (26). The incubation was carried out at 4 "C, a temperature at which only receptor binding occurs. There were about 5.7. io3 receptors/cell (average of three experiments, range from 5420 to 6460) and the apparent dissociation constant was K d = 6.7 & 0.3 nM (Fig. 3). Next we investigated whether there is any difference in affinity for the receptor and for the uptake system between iron-free transferrin and iron-saturated transferrin. Cells were incubated with 0.1 nM lZ5I-transferrin(iron-saturated)for 90 min at 37 "C in the presence of varying concentrations of competing unlabeled iron-free and iron-saturated transferrin. As indicatedin Fig. 4, there is no apparent difference in

1.4

fMOL TF BOUND /

io6 CELLS FIG. 3. Scatchard plot analysis of '251-tran~ferrin binding to mouse teratocarcinoma stemcells. The cells were incubated with 0.2 nM '251-transferrinand various concentrations of unlabeled transfor 90 min at 4 "C. Complete competition was observed at ferrin (TF) 1 PM unlabeled transferrin. The residual amount of '"I-transferrin observed at this concentration of unlabeled transferrin (14%of maximum) was taken to represent nonspecific binding and was subtracted from all other values. All data points are averages of triplicate determinations (which were usually within 107r of each other).

0

I 0 I

rn

100

-

80-

-1

a

5 X

60-

U

40LL

0

TIME (Hr)

FIG. 2. Time course of '251-transferrinbinding to mouse teratocarcinoma stem cells. A, the cells were incubated with 11.7 nM of 1251-transferrin(I2'Z-TF)at either 4 "C (O),24 "C (W), or 37 "C (A).At the inmcated time points, the cells were washed and processed as described under "Materials and Methods." B, long-term time course of '251-transferrinbinding. Cells were incubated with 11 nM of I25 I-transferrin a t 37 "C. At the end of the incubation period, the cells were rinsed and processed as described under "Materials and Methods." All data points are averages of duplicate determinations (which were usually within 5% of each other) and have been corrected for nonspecific binding (assayed in the presence of12.5 p~ unlabeled transferrin) which did not exceed 15%.

.\" 2 0 -

i

00.1

10

100

nM [ T F ]

FIG.4. Competition of Iz5I-transferrinbinding to mouse ter-

atocarcinoma stem cells by apotransferrin and holotransferrin. The cells were incubated for 90 min at 37 "C with 0.1 nM '"Itransferrin (TF) in the presence of various concentrations of either holotransferrin (e) or apotransferrin (0).Washing and processing of the cells was as described under "Materials and Methods." Data points are averages of triplicate determinations with nonspecific binding subtracted.

3248

Receptor-mediated Endocytosis

competition of '251-transferrinbinding by either iron-free or saturated transferrins. At 2.5 rn unlabeled transfenin, 50% inhibition of '251-transferrinbinding is observed. Internalization of Bound Transferrin-The large differences between transferrin binding at 4 and 37 "C and the complete inhibition of transferrin-mediated iron uptake at 4 "C might be due to endocytosis of bound transferrin, which is likely to be a temperature-sensitive process. In order to determine whether or not bound transferrin is internalized, we have made use of two established procedures previously used to discriminate between membrane-bound and internalized epidermal growth factor (27, 28). Teratocarcinoma cellswere incubated for different time periods with '251-transferrinat 4, 24, and 37 "C, the excess unbound lZ5I-transferrinwas washed off, and the cells were treated with 1 ml of binding buffer containing 0.25% pronase for 60 min at 4 "C. The cellswere then transferred to a microfuge tube and pelleted by centrifugation. Only surfacebound transferrin is accessible to proteolytic attack by pronase and is released by itintothe medium, while internalized transferrin is protected from proteolysis and sediments in the cellular pellet. In the experiment illustrated in Fig. 5, 97% of the Iz5I-transferrinthat was bound to cells at 4 "C was liberated by pronase, while after 45 min of incubation at either 24 or 37 "C, only 30% of the bound transferrin was liberated by pronase (Fig. 5C). When the time course of the pronase-sensitive fraction of 125 I-transferrin binding is examined (Fig. 5A), there is no appreciable difference in the rateor extent of binding at 4,24, or 37 "C. But when the time course of the pronase-resistant fraction of '251-transferrinis plotted (Fig. 5B), it is quite clear that whileinsignificant amounts of '251-transferrin became pronase-resistant at 4 "C, large amounts of '251-transferrin became resistant to proteolysis when the incubation was carried out at either 24 or 37 "C. At 37 "C, the "internalization" process was very rapid and started to plateau after about 20 min of incubation. The internalization showed a 10-min lag until receptor occupancy reached a steady state; thereafter, linear uptake was observed. Whencellswere incubated with [59Fe]transferrin, about 70%of the cell-associated radioactivity was liberated by pronase treatment if the initial incubation was carried out at 4 "C, and only 3 4 % was liberated if the initial incubation was at 37 "C (Table I). These results indicate that the 59Feis located inside the cell at 37 "C, while most of it is surfacebound at 4 "C. The second procedure used to discriminate between surfacebound and internalized transferrin was acetic acid removal of surface-bound material (28). Teratocarcinoma cells were incubated with '251-transferrinfor 60 min at either 4 or 37 "C. The unbound transferrin was removed by washing and the cell monolayers were then treated with acetic acid for 6 min at 4 "C. If the initial incubation was at 4 "C, 88.1% of the bound '251-transferrinwas released into the medium. However, if the initial binding was at 37 "C, only 19.8% of the bound material was released by the acetic acid treatment (Table 11, Experiment 1). When the cells wereincubated with lZ5I-transferrin for 60 min at 37 "C, washed, incubated an additional 10 min at 37 "C, washed again, and then treated with acetic acid, only 11.0% of the cell-associated radioactivity was released into the medium (Table 11, Experiment 2). Thus the internalization of transferrin at 37 "C has been demonstrated by two different methods. If protection from proteolytic attack truly represents internalization of surface-bound transferrin, then it should be possible to use this method to follow the fate of surface-bound material in a "chase" type of experiment. Therefore, cells were

of Transferrin

r

I-

60-

$

50-

z v)

W

40-

. O

4" ?

,

;

20

10

,

30

I

-

40

TIME ( M m )

FIG. 5. Time course of membrane binding and internalization of '261-transferrin in mouse teratocarcinoma stem cells. The cells were incubated with 11.7 nM L251-transferrin (TF)at either 4 "C (O),24 "C (B), or 37 'C (A).At the end of the incubation period, the cells were washed and treated with pronase as described under "Materials and Methods." Pronase-sensitive radioactivity is the radioactivity liberatedto the supernatant (A). Pronase-resistant radioactivity is the radioactivity in the cellular pellets ( B ) .C, percentage of the total Iz5I-transferrin bound of pronase-resistant '251-transferrin to cells. Data have been corrected for nonspecific binding.

TABLEI Effect of pronase treatment on the release of teratocarcinoma cellassociated 59Fe Cells were incubated with 22 nM [s9Fe]transferrin at either 37 or 4 "C. At the end of the indicated incubation periods, the cells were then pronase-treated washed 4 times with ice-cold binding buffer and as described under "Materialsand Methods." Radioactivity Binding temperature

Length of incubation

"C

min

37

60 120 60 120

4

Cell pellet

Sur:z-

cpm/plate

110

2216 3896 47 41

Released B

76 172

104

3.3 3.8 70.1 71.7

first incubated with '251-transferrinat 4 "C, a temperature at which only surface binding could occur and internalization would be inhibited. Then the cells were washed freeof excess unbound '251-transferrin,and 1 mlof binding buffercontaining

Receptor-mediated Endocytosis of Transferrin 1.25 p~ of unlabeled transferrin at either 4, 24, or 37 "C was added to each culture plate. The release of radioactivity into the medium and the distribution of cell-associated radioactivity were followed. In the case of cells left at 4 "C, most of the '251-transferrin that remained boundwas pronase-sensitive (89%)and no changes were detected in the extent of pronase resistance during the chase (Fig. 6). At 37 "C, there was a very rapid increase in the amount of material that became pronaseresistant. At the 0 time point, only 12%of the material was pronase-resistant. After 6 min, 86% of the material that remained cell-associated was pronase-resistant. After 6 min, there was no further increase in the extent of pronase resistance. Carrying out the chase at 24 "C slowed downthe process,

L

0

TABLE I1 Release of teratocarcinoma cell-associated 'Z51-transferrinby acetic acid treatment Cells were incubated with 12 nM '251-transferrinfor 60 min at either 4 or 37 "C. The cells were washed 5 times with ice-cold binding buffer and were then incubated with 1 ml of 0.2 M acetic acid, 0.5 M NaCl for 6 min a t 4 "C and rinsed with an additional 1 ml of that solution (4 and 37 "C, Experiment 1) or were incubated an additional 10 min with 1 ml of binding buffer at 37 "C (37 "C, Experiment 2). washed once, and then treated with the acetic acid solution. The acetic acid solution was removed into counting tubes and the cells solubilized in 1 ml of 1 N NaOH and both fractions were counted. AU data have been corrected for nonsDeciiic binding.

3249

I

I

I

1

I

I

6

12

I8

24

30

36

I

TIME ( M i n )

FIG.7. Release of surface-bound '2611-transferrin.The experimental details pertaining to incubation in '2sI-transferrin, washing, and asecond incubation with unlabeled transferrin were as described in Fig. 6 . A t the indicated time points, the binding buffer was removed and both the radioactivity still associated with the cells and radioactivity released were determined. The amount of '251-transferrinthat was cell-associated was converted to per cent of 'z51-transferrinbound to thecells at the0 time point. 0, the second incubation was at 4 "C; 0,the second incubation was at 37 "C.

Radioactivity Binding temperature

"G 4

37, Experiment 1 37, Experiment 2

Left on plate

R$Ezp

cpm/plate

684

20,529 12,031

Released %

5,059 5,060

88.1 19.8

1.496

11.0

2ol u

0

20

40

60

80

100

TIME(Min)

FIG.6. Internalization of surface-bound 1z61-transferrin.Teratocarcinoma stem cells were incubated with 11.8 n~ '*'I-transferrin at 4 "C for 45 min. The cells were washed 5 times with ice-cold binding buffer and then incubated with 1 ml of binding buffer containing 1.25 p~unlabeled transferrin at either 4 "C (=), 24 "C (A),or 37 "C (0).The binding buffer was already a t the indicated temperatures to minimize warm-up time. At the indicated time points, the binding buffer was removed to a counting tube and the cells left on the plate were treated with pronase as described under "Materials and Methods." The per cent of cell-bound 1251-transferrinthat was pronase-resistant (remaining in the cellular pellet) is plotted against the second incubation time. Data have been corrected for nonspecific binding.

FIG.8. Release of cell-bound '"1-transferrin at 4 and 37 "C. Teratocarcinoma stem cells were incubated with 11.8 nM "'I-transferrin for 60 min a t 37 "C. Then thecells were washed five times with binding buffer and incubated with 1 ml of binding buffer containing 1.25 PM unlabeled transferrin at either 37 "C (0)or 4 "C (0). At the indicated time points, the medium was removed into counting tubes and cells left on the plate were solubilized in 1 ml of 1 N NaOH. The radioactivity left on the cells was converted to per cent of radioactivity bound to the cells at the 0 time point. All data were corrected for nonspecific binding. but results were qualitatively the same as for the 37 "C chase. There was a time-dependent increase in the extent of cellassociated transferrin that became pronase-resistant. These results demonstrate that progressive internalization of surface-bound transferrin occurs at 24 or 37 "C. Fig. 7 illustrates the kinetics of '251-transferrindissociation from the cells. The experimental scheme was as described above, except that this time the release of '251-transfenininto the medium was monitored. A t 4 "C, '251-transferrindissoci-

3250

Receptor-mediated Endocytosisof Transferrin

TABLEI11 Effects of inhibitors on "'I-transferrin and ("Fe]transferrin uptake into teratocarcinoma cells Teratocarcinoma ceUs were preincubated for 30 min at 37 "C in binding buffer without inhibitors or in the presence of the indicated inhibitors. Then, '251-transferrin(final concentration 12 nM),[5gFe]transferrin (final concentration 22 nM),or unlabeled transferrin (final concentration 12.5 PM) was added to the appropriate plates which were incubated for an additional 90 min. The plates were washed 5 times and 1 ml of 1 N NaOH was added to those plates incubated with ["Fe]transferrin to solubilize the cells. The plates incubated with transferrin were incubated for 6 min at 4 "C, with 1 ml of 0.2 M acetic acid, 0.5 M NaCl, and rinsed with an additional 1 ml of this solution. The acetic acid solution was removed into counting tubes and the cells solubilized in 1 N NaOH and counted. Surface-bound '251-transferrinis represented by radioactivity released by incubation with acetic acid. Internalized '251-transferrinis represented by radioactivity left in cells after incubation with acetic acid. See text for further details. Surface-bound Condition

12sII-transferrin

fmol/plate

2.6 210.0

119.4 100.0 41.5 Control 42.2 100.5 10 mM NaCN 1087.8 107.3 128.2 89.9 10 mM NaF 107.4 109.8 45.6 10 mM NaCn + 10 mM NaF38.0 45.4 84.6 35.1 10 mM413.1 NH4C1 43.3 38.2 139.6 166.7 68.8 1 mM chloroauine 82.1 58.1 24.1

%fmoi/plate control

Internalized "'1-transferrin

taken

=Fe

up

76fmol/piate control

% control

100.0 101.7 245.6

22.7

104.3

ated from the cells with simple fiist order kinetics, the halflife being 39 min. The dissociation of '=I-transferrin at 37 "C is somewhat .more complex. There was a 6-7 min lag before first order dissociation proceeded, with a half-life of 16 min. In the experiment illustrated in Fig. 8, the cells were incubated with '251-transferrinat 37 "C, a temperature at which both surface binding and endocytosis occur, and the dissociation of all associated transferrin was followed, again at both 4 and 37 "C. This time, at 4 "C, only 50% of the cell-associated radioactivity was released, with a half-life of 33 min. At 37 "C, there was no delayin the release of transferrin intothe medium, and the dissociation half-life was 14 min. At 37 "C, 80%of the original cell-associated 1251-transferrin was released after 1 h of incubation. A t all the time points, the '251-transferrin released from the cells at 37 "C was 100%trichloroacetic acid-insoluble, thereby indicating that no degradation of internalized transferrin had occurred. Effects of Inhibitors on Uptake of Transferrin and IronIf transferrin internalization and iron uptake are in fact due to a classical endocytosis mechanism, then the two processes should be energy-dependent (29).Moreover, inmost internalization systems thus far studied, the internalized material is finally found in the lysosomes,where it is susceptible to proteolysis (30). In order to establish the energy requirement and the lysosomal involvement in transferrin internalization and iron uptake, we examined the effects of various inhibitors on the two processes. Teratocarcinoma cells werepreincubated for 30 min, either in control binding buffer or in the presence of the inhibitors listed in Table 111. Either 1251-transferrin or [59Fe]transferrin was then added as indicated and the incubation was continued; after another90 min, bindingand uptake of '251-transferrinor "Fe were measured. A t least two independent experiments, each carried out in duplicate, were performed for each inhibitor; the results between duplicates agreed within 10%.The data in Table 111 are fromone representative experiment; each value in the table is the average of a duplicate determination. The energy inhibitors, CN- and F-, had no effect on binding of I2'I-transferrin to cell-surfacereceptors. While CN- didnot affect internalization of Iz5I-transferrin,F- did inhibit it to some extent, and the combination of these two inhibitors led to a stronger inhibition of transferrin uptake. The inhibition of 5gFeuptake was even stronger than the inhibition of transferrin uptake. Clearly, energy is required both for internalization of transferrin and for transferrin-mediated iron uptake. The lysosomotropic ion NH4+did not affect either binding or internalization of 1251-transferrinbut did cause a marked inhibition of 59Fe uptake. The other lysosomotropic agent,

chloroquine, caused some inhibition of '251-transferrinbinding, but did not lead to any further inhibition of '251-transferrin uptake. The inhibition of 59Feuptake was much stronger and could not be accounted for solely by diminished transferrin binding and internalization. Chloroquine, being a lipid-soluble agent, might interact with teratocarcinoma cell membranes in such a way as to perturbtransferrin binding to those cells. DISCUSSION

Using '251-labeledhuman transferrin as a probe, we have demonstrated the presence of specifk high affinity receptors for transferrin on mouse teratocarcinoma stem cells. The number of receptors/cell is 5.7. lo3, a figure much lower than reported for choriocarcinoma, embryonic lung, and splenic lymphocytes, which ranged from 2.5. lo5 to 3.7 lo5 sites/cell (8).The apparent dissociation constant at 4 "C was 6.7 nM, as determined by Scatchard analysis (26)of '251-transferrinbinding to cells. Using the data presented in Fig. 2A (for binding at 4 " C ) , the association rate constant of transferrin with the receptor can be measured (assuming a bimolecular reaction) and is k 1 = 0.886. lo7M" rnin". From the data shown in Figs.7 and 8, the half-life for transferrin dissociation is 36 3 min. The dissociation rate constant is therefore k l = 1.39. lo-' rnin", and the calculated dissociation constant is K d = k - l / k l = 1.9310-*/0.886. IO7 = 2.2. lo-' M. This is in good agreement with the apparent dissociation K d = 6.7. lo-' M. These values are also in good agreement with the dissociation constants reported for the solubilized receptor from human placental cell membranes orfrom culturkd human choriocarcinoma cell membranes, whichwere 2.4 f 1.2 mi%f and 6.7 f 0.17nM, respectively (8, 13). '251-transferrinbinding (Fig. 2) is much more rapid than transferrin-mediated 59Feuptake, which reaches a steady state only after 2 h at 37 "C (Fig. 1).Another difference between transferrin-mediated 59Feuptake and '251-transferrin binding is the temperature effect on the twoprocesses.While the amount of 59Fe uptake at 37 "C is 25-fold greater than the amount of uptake at 4 "C, the level of '251-transferrin associated with the cells at 37 "C is only about 2-3-fold higher than at 4 "C, and binding to surface receptors is identical at the two temperatures. The amount of iron taken up is in the range of500-600 fmol/106 cell42 h in confluent cells,while the amount of '251-transferrinassociated with the cells at 37 "C is only 16 fmol/106 cells. All these differences strongly suggest that at 37 "C,iron uptake is uncoupled from the uptake of the protein at some stage. At 4 'C, the degree of iron uptake corresponds well to thatof transferrin binding. The two phenomena follow a similar rapid time course, and the amount of e

*

Transferrin Receptor-mediated of Endocytosis iron taken upat 4 "C is 14.lo3 atoms/cell. Since there aretwo iron-binding sites/molecule of transferrin, this corresponds to 7. io3transferrin molecules/cell, a number which is very close to that of 5.7. lo3 molecules/cell based on amount of "'1transferrin binding at 4 "C. We conclude that at 4 "C all the iron associated with the cells is attached totransferrin molecules bound to surface receptors. In contrast to theearlier report by Jandl and Katz (7), and in agreement with Hamilton et a2. (8), both apo- and hobtransferrin bind equally well to the transferrin receptor of teratocarcinoma cells (Fig. 4), and arealso internalized to the same extent (data not shown). All of the datadescribed in the present report areconsistent with the hypothesis of receptor-mediated endocytosis of transferrin. While earlier experiments carriedout mainly by Hemmaplardh and Morgan (17) and Morgan and Appleton (31) suggested that either "'I-transferrin or a ferritin-transferrin conjugate do enter rabbit reticulocytes, thoseexperiments lacked the proper controls to correct for nonspecific binding and to demonstrate that the endocytosis is indeed receptormediated. Our experiments have presentedthe first biochemical demonstration of transferrin internalization. When incubated at 4 "C, most (97%) of the '251-transferrin bound to teratocarcinoma cells is released by pronase treatment. When cells are incubated with '251-transferrinat 37 "C, only 30% of the cell-associated material is released by pronase (Fig. 5C). Similarly, when cells are incubated with [59Fe]transferrinat 4 "C, 70% of the cell-bound iron is released by pronase. When incubated at 37 "C, only 3-455 of cell-bound iron is released by pronase (Table I). Internalization of bound transferrin was also demonstrated by a different method: 88% of the labeled transferrin bound at 4 "C is released by acetic acid treatment of the cells; only 20% of the bound protein is released if the cells are incubated with transferrin at 37 "C (Table 11).The most likely interpretation of these results is a temperaturedependent endocytosis of transferrin that occurs at 37 "C but not at 4 "C. When teratocarcinoma cells are incubated with "'I-transferrin at 4 "C, a temperature at which only surface-receptor binding can occur, and then shifted to either 24 or 37 "C, the surface-bound transferrin that was releasable by pronase is internalized and becomes resistant to pronase. At 37 "C, it takes 6 min for all the surface-bound transferrin to be internalized (Fig. 6). When the release of surface-bound transferrin from the cells is followed (Fig. 7), we find that at4 "C release occurs immediately, while at 37 "C there is a 6-7-min lag period before release of transferrin is observable. This is the same length of time required for all of the surface-bound transferrin to be internalized. The interpretation is that surface-bound transferrin is internalized before simple dissociation from the plasma membrane receptorcan take place. After it hasbeen internalized, the transferrin is released by the cells into the medium by still unknown mechanisms, with a halflife of14 min. Similar observations have been madefor internalization of lowdensity lipoproteinby human fibroblasts (32). In contrast to the situation just discussed, when the cells are incubated with '251-transferrinat 37 "C, a temperature at which both surface binding and internalization of bound transferrin occur, and the release of bound transferrin is followed, release can be detected immediately, even at 37 "C (Fig. 8). Since the initial binding was at 37 "C, there is already a pool of transferrin inside the cell so that release can be detected without delay. At 4 "C, only 50% of the cell-associated transferrin that is due to prior incubation at 37 "C is actually released by the cells, while a t 37 "C most of the cell-associated transferrin can

3251 be released. Since about %-?4 of the steady-state amount of cell-associated transferrin is intracellular transferrin, only that fraction (%-%) which represents surface-bound material can be released at 4 "C, while the intracellular material cannot leave the cell. At 37 "C, all the transferrin associated with the cell can dissociate, In contrast to the epidermal growth factor and low density lipoprotein systems (27,32), theradioactivity released to the medium does not seem to be in the form of low molecular weight degradation productsof the internalized protein, since all of it is trichloroacetic acid-insoluble. This is not surprising, since in vivo the half-life of transferrin is much longer than the half-life for plasma iron, so that transferrin must be recycled and not consumed during iron delivery (33). We believe that, like epidermal growth factor (34) and low density lipoprotein (35), internalized transferrin reaches the lysosome, where the iron that is tightly bound to transferrin can be released because of the low pH. The stability constant of the iron-transferrin complex a t physiological pH appears to be the of the order of loz9 (36), and only at low pH can iron dissociate from the apoprotein. The low pH in the lysosome is possibly responsible not only for the dissociation of iron from the transferrin molecule but also for the dissociation of transferrin from its receptor. The inhibition of iron uptake by the lysosomotropic agents chloroquine and ammonia (Table I11 and Ref. 11) suggests that the uptake of iron by the cells does involve the lysosome at some step. Ammonia has no effect on '251-transferrin binding and internalization, while chloroquine causes a slight inhibition of transferrin binding but no further inhibition of internalization. By elevating the pH in the lysosome (37), those agents can prevent the release of iron from the transferrin molecule and transferrin from the receptor. Transferrinis not the only example of a protein that may enter the lysosome and there escape degradation. The receptor for low density lipoprotein, in contrast to low density lipoprotein itself, also escapes degradation in the lysosome (32). Like endocytosis in general (29),the endocytosis of transferrin is also energy-dependent, and incubation of teratocarcinoma cells with a combination of inhibitors of both glycolysis and oxidative phosphorylation (F- and CN-) inhibits both internalization of "'I-transferrin and 59Feuptake, but does not inhibit receptor binding. The final question is: can the observed endocytosis of transferrin actually acount for all the iron uptake in the cells? The maximal rate of endocytosis is 1 fmol/106 cells/min at a transferrin concentration that saturates 77% of the available receptors (Fig. 5B).At 100%saturation (atwhich iron uptake was measured), the endocytosis rate would be 1.29 fmol/106 cells/min. Since every transferrin molecule carries two iron atoms, this will correspond to 2.6 fmol of iron/106 cells/min. Another estimate is based on the finding that it takes about 6 min to internalize the whole complement of cell-surface receptors for transferrin at 37 "C. This is equivalent to internalizing lo3 transferrin molecules/min/cell. Since each transferrin molecule carries two iron atoms, this equals 2.10' Fe atoms/cell/min, or 3.3 fmol of iron/106 cells/min. These values compare favorably with rates of 4-5 fmol of iron/1O6 cells/ min, that were measured using "Fe-labeled transferrin. We suggest that there are at least four major steps in the transferrin pathway: (i) binding of iron-laden transferrin to specific high affinity cell-surface receptors; (ii) internalization of the iron-transferrin (and probably iron-transferrin-receptor) complex via an endocytotic, energy-requiring mechanism; (iii) fusion of endocytotic vesicles containing iron-transferrin complexes with lysosomes, and subsequent liberation of iron from transferrin and transferrin from the receptor due to the low pH within the lysosome; and (iv) the ultimate release of iron-free transferrin into the medium. The iron-free transfer-

3252

Receptor-mediated Endocytosis

of Transferrin

rin is then available for rebinding of iron. Mutations occurring at any one of these stepswould lead to impairment of the normal iron cycle thereby causing gross metabolic defectsand possible anemias at the organismic level. Identifying thestepsinthetransferrinpathway in developmentally versatile mouse teratocarcinoma stem cells lays thegroundwork for devising selection strategies thatwill enable isolation of variant cells carrying different mutations in this pathway.The m e of teratocarcinoma cells capable of conversion to normalcy followed by complete differentiation after injection into embryos (21, 22) should make it possible to study theeffects of lesions in the transferrin pathway not only at thecellular level in all kinds of specialized cells as they differentiate, but also in whole animals.

1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12. 13. 14.

Chern. 254,9943-9946 15. Fielding, J., and Speyer, B. E. (1974) Biochim.Biophys. Acta 363,387-396 16. Workman, E. F.,Jr., and Bates, G. W. (1974) Biochem. Biophys. Res. Commun. 58, 787-794 17. Hemmaplardh, D., and Morgan, E. H. (1977) Br. J. Haematol. 36,85-96 18. Sullivan, A. L., Grasso, J . A,, and Weintraub, L. R. (1976) Blood 47, 133en143 19. Edwards, J. A., and Hoke, J . E. (1975) Blood 46,381-388 20. Edwards, J. A,, and Hoke, J. E. (1978) J.Med. (Westbury)9,353364 21. Mintz, B., and Illmensee, K. (1975) Proc. Natl. Acad. Sci. U. S. A . 72,3585-3589 22. Mintz, B. (1979) in Models for the Study of Inborn Errors of Metabolism (Hommes, F. A,, ed) pp. 343-354, Elsevier/NorthHolland, Amsterdam 23. Goldstein, J. L., Brown, M. S., Krieger, M., Anderson, R. G. W., REFERENCES and Mintz, B. (1979) Proc. Natl. Acad. Sei. U. S. A . 76, 28432847 Putnam, F. W., ed (1975) in The PlasmaProteins, Vol. 1, pp. 26524. Greenwood, F. C., Hunter, W. M., and Glover, J. S. (1963) Bio316, Academic Press, New York chem. J.89, 114-123 Aisen, P., and Brown, E. B. (1975) Prog. Hematol. 9, 25-56 25. Bates, G. W., and Schlabach, M. R. (1973) J. Biol. Chem. 248, Lane, R. S. (1976) in Structure and Function of Plasma Proteins 3228-3232 (Allison, A. C., ed) pp. 53-75, Plenum Press, New York Hutchings, S. E., and Sato, G. H. (1978) Proc. Natl. Acad. Sci.U. 26. Scatchard, G. (1949) Ann. N . Y. Acad. Sci. 51, 660-672 27. Carpenter, G., and Cohen, S. (1976) J.Cell Biol. 71, 159-171 S. A.75,901-904 Galbraith, G. M. P., Galbraith, R. M., Temple, A,, and Faulk, W. 28. Haigler, H. T., Maxfield, F. R., Willingham, M. C., and Pastan, I. (1980) J. Biol. Chem. 255, 1239-1241 P. (1980) Blood 55, 240-242 Van Bockxmeer,F. M., and Morgan, E. H. (1979) Biochim. 29. Silverstein, S. C., Steinman, R. M., and Cohn, 2. A. (1977) Annu. Reu. Biochem. 46,669-722 Biophys. Acta584, 76-83 30. Goldstein, J. L., Anderson, R. G. W., and Brown, M. S . (1979) Jandl, J. H., and Katz, J. H. (1963) J. Clin. Inuest. 42,314-326 Nature 279,679-685 Hamilton, T. A., Wada, H. G., and Sussman, H. H. (1979) Proc. 31. Morgan, E. H., and Appleton, T.C . (1969) Nature 223,1371-1372 Natl. Acad. Sci. U. S. A . 76, 6406-6410 Larrick, J. W., and Cresswell, P. (1979) Biochim. Biophys. Acta 32. Goldstein, J. L., Basu, S . K., Brunschede, G. Y., and Brown, M. S . (1976) Cell 7, 85-95 583,483-490 Larson, S. M., Rasey, J . S., Allen, D. R., Nelson, N. J., Grunbaum, 33. Katz, J. H. (1961) J. Clin. Invest. 40, 2143-2152 Z., Harp, G. D., and Williams, D. L. (1980) J. Natl. Cancer Inst. 34. Gorden, P., Carpentier, 3. L., Cohen, S., and Orci, L. (1978) Proc. Natl. Acad. Sci. U. S. A . 75,5025-5029 64,41-53 Octave, J . N., Schneider, Y. J., Hoffman, P., Trouet, A,, and 35. Anderson, R. G. W., Brown, M. S., and Goldstein, J. L. (1977) Cell 10,351-364 Crichton, R. R. (1979) FEBS Lett. 108, 127-130 36. Korman, S. (1960) Ann. N. Y. Acad. Sci. 88,460-473 Leibman, A,, and Aisen, P. (1977) Biochemistry 16, 1268-1272 Wada, H. G., Hass, P. E., and Sussman, H. H. (1979) J. Biol. 37. Okhuma, S., and Poole, B. (1978) Proc. Natl. Acad. Sci. U. A. 75,3327-3331 Chem. 254, 12629-12635 Seligman, P. A., Schleicher, R. B., and Allen, R. H. (1979) J . Biol.

s.