Internalization and Subcellular Localization of Transferrin and Transferrin Receptors in HeLa Cells*

The subcellular location of radiolabeled transferrin ("'I-Tf), internalized during cellular iron uptake, and the cellular distribution of transferrin (Tf) receptors were studied in cultured HeLa cells. Cells were 37 '251-Tf(Fe)z. Forty per cent of the labeled ligand was associated with cell surface receptors. The remaining 60% was internalized as shown by the inability to dissociate '"I-Tf from cells by competition with excess Tf(Fe)z or treatment of cells with 0.2 M acetic acid containing 0.5 M NaC1. Subcellular fractionation studies using sucrose density gradients indicated that internalized Tf was localized in a membranous vesicle distinct from lysosomes, Golgi apparatus, endoplasmic reticulum, or plasma membranes. The subcellular distribution of Tf receptors was studied using an assay for detergent solubilized receptors. Even without preincubation with ligand, the majority of cellular Tf receptors were lo- calized intracellularly in a vesicle with the same buoyant density as the vesicle containing internalized '''I-Tf. Using an assay for occupied receptors, we demon- strated that the same vesicle contained both internal receptors and internalized ligand. A portion (20%) of the intracellular receptor pool was insensitive to trypsin treatment of whole cells at 37 "C suggesting that during the experimental time period (20-30 min) this portion did not recycle to the cell surface. We propose that during cellular iron uptake, Tf re- ceptor-ligand

Internalization and Subcellular Localization of Transferrin and Transferrin Receptors in HeLa Cells* (Received for publication, January 5, 1983) Jamie E. Lamb The subcellular location of radiolabeled transferrin ("'I-Tf), internalized during cellular iron uptake, and the cellular distribution of transferrin (Tf) receptors were studied in cultured HeLa cells.
Cells were incubated at 37 OC with '251-Tf(Fe)z. Forty per cent of the labeled ligand was associated with cell surface receptors. The remaining 60% was internalized as shown by the inability to dissociate ' "I-Tf from cells by competition with excess Tf(Fe)z or treatment of cells with 0.2 M acetic acid containing 0.5 M NaC1. Subcellular fractionation studies using sucrose density gradients indicated that internalized Tf was localized in a membranous vesicle distinct from lysosomes, Golgi apparatus, endoplasmic reticulum, or plasma membranes. The subcellular distribution of Tf receptors was studied using an assay for detergent solubilized receptors. Even without preincubation with ligand, the majority of cellular Tf receptors were localized intracellularly in a vesicle with the same buoyant density as the vesicle containing internalized '''I-Tf. Using an assay for occupied receptors, we demonstrated that the same vesicle contained both internal receptors and internalized ligand. A portion (20%) of the intracellular receptor pool was insensitive to trypsin treatment of whole cells at 37 "C suggesting that during the experimental time period (20-30 min) this portion did not recycle to the cell surface.
We propose that during cellular iron uptake, Tf receptor-ligand complexes are internalized and directed to a nonlysosomal compartment where iron is released, followed by recycling to the cell surface of an intact Tf receptor-apo-Tf complex.
The initial event in the transfer of iron from Tf(Fe); to cells is the binding of Tf(Fe)* to specific, high affinity surface receptors. Following the formation of the receptor-ligand complex, cells internalize and accumulate iron. The precise nature * This work was supported by National Institutes of Health Grant HL 23376. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Supported by National Institutes of Health Postdoctoral Training Grant T32AM07115.
li Under tenure of a clinician-scientist award from the American Heart Association with funds contributed in part by the Utah Heart Association. of the steps involved in the internalization and accumulation of iron remain to be defined.
Most published data suggest that Tf(Fe)z is internalized, the iron is removed, and apo-Tf then is released intact from the cells (1-4). There are conflicting views, however, regarding the precise intracellular location of internalized Tf(FeIz (5-7). Agents which inhibit endocytosis (5) or raise the pH of intracellular compartments prevent cellular iron accumulation (6,8). The effect of such agents has led to the suggestion that internalized Tf(Fe)* is located in either a lysosomal compartment (6) or an acidic nonlysosomal compartment (7).
In this paper, we report that in cultured HeLa cells internalized Tf is localized in a unique intracellular compartment which is distinct from lysosomes, plasma membrane, or Golgi apparatus. The majority of cellular Tf receptors are localized intracellularly and are in the same compartment. Tf may reach this compartment in association with internalized Tf receptors.
Binding Studies-Binding of ligand to whole cells was performed in suspension in 1.5-ml Eppendorf tubes (unless otherwise indicated, all operations were performed at 0-4 "C). Cells were washed three times in PBS (10 mM phosphate, pH 7.2, 3 mM KCl, 140 mM NaCl) and incubated (37 or 0 "C) in 1 ml of MEM containing 4 mg. ml" of bovine serum albumin and 10 nM '*'I-Tf(Fe)z. Nonspecific binding was determined by incubating cells under the same conditions but in the presence of 2-5 p~ Tf(Fe),. After incubation, cells were washed three times with PBS and solubilized with 0.1% sodium dodecyl 0.5 ml of 10% trichloroacetic acid and the precipitate removed by centrifugation. After centrifugation, 1.1 ml of "color reagent" was mixed with 0.5 ml of supernatant and incubated at 37 "C for 60 min.
The absorbance at 720 nm was recorded. The color-reagent was prepared daily by mixing stock solutions of ammonium molybdate and ascorbic acid. The reagent contained 28 mM ammonium molybdate and 40 mM ascorbic acid. One unit of enzyme activity represents 1 nmol of glucose 6-phosphate hydrolyzed per min. a modification of the procedure of Avruch and Wallach (11). Samples The plasma membrane marker, 5'-nucleotidase, was assayed using were incubated with 45 mM Tris-HC1, pH 9.0, 10 mM MgC12, 1.5 mM 5'-AMP, 3.85 nM [3H]AMP (0.035 pCi/assay, specific activity 15 Ci. mmol") in a final reaction mixture of 600 pl. After 30 min at 37 "C, 0.2 ml of 0.25 M ZnS04 and 0.2 ml of 0.25 M Ba(OH)2 were added sequentially at room temperature and the samples mixed after the addition of each reagent. The precipitate was removed by centrifugation at 1000 X g for 30 min. An aliquot (0.25 ml) of the supernatant was mixed with 5 ml of scintillation fluid (Aquasol 2, New England Nuclear) and counted in a Beckman LS-233 scintillation counter. A unit of enzyme activity is equivalent to 1 pmol of AMP hydrolyzed per min.
Galactosyl transferase, a marker for the Golgi apparatus, was assayed by a modification of the method of Brew et al. (12) as described by Rome et ai. (13). A unit of enzyme activity is equivalent to 1 pmol of galactose transferred per min.
The method of Horvat et al. (14) was employed to measure the lysosomal enzyme marker hexoseaminidase. p-Nitrophenyl-N-acetyl-(3-D-ghcosaminide (3.15 mM) was incubated with 100 pl of a sample containing 0.1% Triton X-100. The final reaction volume was 1.0 ml.
After 30 min at 37 "C, the reaction was quenched with 1.0 ml of 500 mM glycine, pH 10.5. The liberated p-nitrophenolate was measured spectrophotometrically at 420 nm. One unit of enzyme activity represents 1 @mol of substrate converted per min.
Leucyl-8-naphthylamidase was assayed according to the procedure of Peters et ai. (15). One unit of enzyme activity is equivalent to 1 nmol of substrate hydrolyzed per min.
Protein Determination-Protein was measured by the method of Lowry et al. (16) using bovine serum albumin as the standard.
Preparation of Cells for Sucrose Gradient Analysis-HeLa cells, grown either on monolayers or in spinner culture, were harvested and washed three times in PBS. Monolayer cells were harvested by the use of a rubber policeman. The cells were resuspended in 10 mM Tris-HCI, 1 mM EDTA, pH 7.2, containing 250 mM sucrose and homogenized in a tight-fitting Dounce homogenizer. The percentage of disrupted cells (generally greater than 80%) was monitored by phasecontrast microscopy. Intact cells, nuclei, and large cellular debris were removed by centrifuged (200 X g, 10 min, 0-4 "C).
Sucrose Density Gradients-Sucrose density gradients were prepared using a Buchler gradient maker and a peristaltic pump. All sucrose solutions (w/w) were prepared in 10 mM Tris-HCI, and 1 mM EDTA, pH 7.2. One ml of 60% sucrose was placed in the bottom of Beckman cellulose nitrate tubes (5/8 X 4 inches) and a linear (50-15%) sucrose gradient layered over the top. Approximately 1-2 ml of cell homogenate, prepared as described above, were carefully loaded onto the gradients. After centrifuging overnight at 100,000 X g in a Beckman SW 28 rotor at 15 "C, the gradients were fractionated manually. The sucrose concentration was measured at 25 "C using a Bausch and Lomb refractometer and converted to sucrose density (8. ml-').
Soluble Ti Receptor Assay-HeLa cells were washed three times with PBS and then twice with 10 mM Tris-HC1, 150 mM NaCl, pH 8.0, and finally resuspended in the same buffer containing 0.1% Triton X-100. The cells were homogenized using a Brinkman Polytron (Westbury, NY) using 2 30-s bursts at setting 6-7. The supernatant obtained after centrifugation of the homogenate (20,000 X g, 30 min, 4 "C) was used as a source of solubilized Tf receptors. The supernatant (cell extract) was diluted to a protein concentration of 2 mg . m1-I.
When assaying Tf receptor activity across a sucrose density gradient, aliquots were simply made 0.1% in Triton X-100 and mixed thoroughly to solubilize the Tf receptor and disrupt membrane-bound organelles.
Iodination of Plasma Membrane-HeLa cells were incubated for 1 h at 37 "C in the presence of 1 p~ Tf(Fe)2. After three washes in PBS at 0 "C, approximately 23 X lo6 cells were iodinated according to a modification of the method of Hubbard and Cohn (17). The final 1.0ml reaction mixture contained 10 milliunits of glucose oxidase, 20 pmol of glucose, 10 pg of lactoperoxidase, 1 nmol of NaI, and 10 pl of 1251 (1-3 mCi). The glucose oxidase was always added last. After 40 min at 0 "C, the reaction was quenched with 2 ml of Eagle's MEM. The cells were washed several times with 5.5 mM EDTA and finally with PBS. Iodinated cells were combined with unlabeled control cells which had been incubated with 1 ~L M Tf(Fe)2 at 37 "C for 1 h, and subjected to Dounce homogenization as described above. Proteinbound IZ5I was determined by measuring the trichloroacetic acidprecipitable radioactivity in the presence of 50 mM KI.
Remouai of Cell Surface '"I-Tf at Low pH-HeLa cells growing on 35-mm diameter tissue culture dishes were incubated with '261-Tf(Fe)2 (3 nM) at 37 "C for 60 min then shifted to 0 "C. After four washes with 2 ml of PBS (4 "C), 1.5 ml of acid-stripping solution (0.5 M NaCl in 0.2 M acetic acid) was added to each plate. After 8 min at 0-4 "C, the acid-stripping solution was removed and placed in counting vials. Each plate was again washed with 1 ml of acid-stripping solution. This wash was added to the initial 1.5 ml and the "acid-strippable" radioactivity in the final 2.5 ml was determined. The cells were then solubilized with 1 ml of 1% sodium dodecyl sulfate and the radioactivity, which represented "acid-resistant'' Tf receptors, was determined. During acid-stripping, both cells and solutions were maintained at 4 "C.

RESULTS
Evidence for Internalization of '251-Tf(Fe)2-Inc~bation of HeLa cells at 37 "C for 60 min resulted in a steady state binding of '251-Tf(Fe)2, confirming our earlier observations (9). When high concentrations of unlabeled Tf(Fe)2 were added to cells at 37 "C a time-dependent loss of cell-associated radioactivity occurred (Fig. 1, Group C). Greater than 90% of cell-associated radioactivity was lost after 60 min with appar- Group B cells were shifted to 37 "C after 240 min (arrow) and the amount of cell-associated radioactivity determined over the next 60 min. The inset shows a semilogarithmic plot of the dissociation of cell-associated radioactivity for Groups B (after temperature shift) and C at 37 "C and Group A at 0 "C. ent first order kinetics, yielding a dissociation rate constant (37 "C) of 9.24 x 10" min". When unlabeled Tf(FeIz was added to cells incubated with 1251-Tf(Fe)2 at 0 "c ( fig. 1,   Group A), the loss of radioactivity also exhibited first order kinetics. The dissociation rate constant (0 "C) was calculated to be 1.11 X IO-* min". These values are similar to those we have reported previously (9). However, when cells were incubated at 37 "C with '251-Tf(Fe)2 and then shifted to 0 "C in the presence of nonradioactive Tf(FeI2, only 40% of the cellassociated radioactivity was lost (Fig. 1, Group 23). This 40% dissociated with the same kinetics as the radioactivity displaced from cells which had always been maintained at 0 "C. The retained radioactivity remained cell-associated even after 4 h at 0 "C. Upon shifting these cells to 37 "C ( Fig. 1, Group  B ) the remaining cell-associated radioactivity was lost with kinetics which were identical to those calculated for cells maintained continuously at 37 "C. These results suggest that, following incubation of cells with 1251-Tf(Fe)2 at 37 "C, a component of cell-bound Tf is internalized. Shifting cells to 0 "C prevents dissociation of this component.
When HeLa cells were incubated with '251-Tf(Fe)2 at 0 "C for 105 min, 92% of the cell-associated radioactivity appeared to be bound at the cell surface, since it could be removed by incubation with 0.2 M acetic acid containing 0.5 M NaCl. If binding was allowed to proceed a t 37 "C for 75 min, only 26% of the cell-associated radioactivity could be removed by acid treatment. The remainder of the cell-associated radioactivity (74%) was acid-resistant and presumably located within the cell. Intracellular, cell-associated, lZ5I-Tf determined in these experiments (as a percentage of total cell-associated 1251-Tf) is in close agreement with the data obtained by the temperature shift experiments discussed previously and is similar to that reported by Karin and Mintz (5) and Bleil and Bretscher (18).
Intracellular Localization of '251-Tf-The experiments described above indicate that in the steady state approximately 60-74% of cell-associated '2sI-Tf(Fe)2 is internalized at 37 "C.
To determine the subcellular localization of this internalized Tf, we employed linear sucrose density gradients. Cells were incubated with Iz5I-Tf(Fe), at 37 "C for 1 h and then placed a t 0 "C in the presence of unlabeled Tf(Feh (10 p M ) for 3-4 h. The cells were washed, homogenized, and the homogenate applied to a sucrose density gradient (15-50%). Subcellular fractions were identified using as enzyme markers glucose-6phosphatase (endoplasmic reticulum), galactosyl-transferase (Golgi apparatus), hexoseaminidase (lysosomes), and 5'-nucleotidase (plasma membranes). Representative sucrose density gradients are shown in Fig. 2. Two peaks of 1251 radioactivity were always observed. One was a major peak of radiolabeled material which entered the gradient and the other a quantitatively minor component of apparent lower buoyant density. The lower buoyant density peak was precipitable by trichloroacetic acid but was nonsedimentable (100,000 x g, 6 h). '2'I-Tf(Fe)2 added to cellular homogenates and then applied to a sucrose gradient exhibited a sedimentation pattern of similar low buoyant density. These observations suggest that the minor component of apparent low buoyant density represents free Iz5I-Tf.
The major peak of lZ5I radioactivity, with a buoyant density of 1.123 g.ml", was sedimentable, and the radioactivity in this component could be released by detergent treatment. Thus, the major peak of radioactivity represents membranebound lZ5I-Tf. The free l2'1-Tf was likely generated as a result of mechanical damage to cell membranes during homogenization.
Marker enzyme activities associated with the major radiolabeled peak (presumably "'I-Tf in a vesicle) did not reveal any concordance with the peaks of activities of hexosaminidase, 5'-nucleotidase, galactosyl-transferase, or glucose-6phosphatase. There was, however, a consistent coincidence between the peaks of activity for leucyl-P-naphthylamidase and 1251-radioactivity. This enzyme has been reported to be associated with plasma membranes (13,19) but in our studies there was a clear separation between leucyl-6-naphthylamidase activity and 5'-nucleotidase activity. To determine which of these activities was associated with plasma membranes, an independent plasma membrane marker was used. Cells were iodinated at 0 "C using the combined lactoperoxidase-glucose oxidase procedure of Hubbard and Cohn (17). The cells were then homogenized and the homogenate applied to a sucrose Cellular Localization of Transferrin and Transferrin Receptors gradient. As demonstrated in Fig. 2, the dense (1.153 g. ml-') peak of iodinated material exhibited the same buoyant density as 5'-nucleotidase and was well separated from leucyl-@naphthylamidase activity. These results demonstrate that in HeLa cells 5'-nucleotidase, but not leucyl-0-naphthylamidase, is a good marker for plasma membranes. Leucyl-0naphthylamidase, however, was always coincident with the  CELL EXTRACT (pll peak of internalized labeled lZ5I-Tf and therefore served as a useful marker for this compartment. These results also demonstrate that internalized lZ5I-Tf(Fe), was not associated with plasma membrane vesicles but resided in a distinct vesicle.

Zdentification of "Znternal" Tf Receptors-Karin and Mintz
( 5 ) and Ward et al. (9) have suggested the existence of a pool of intracellular Tf receptors. We developed an assay for Triton X-100 solubilized Tf receptors in order to investigate the relationship between surface Tf receptors and internal receptors, and to determine whether Tf receptors were present in the vesicle in which we detected internalized lZ5I-Tf. The assay capitalized on the observation that receptor-bound lZ5I-Tf could be specifically precipitated with (NH4)2S04. The data in Fig. 3 and Fig. 4 define some of the properties of the assay. In the presence of a Triton X-100 extract of cells, lz5I-Tf could be precipitated at a concentration of 45% (NH.&S04, while the precipitation of "T-Tf(Fe), alone required a higher concentration (Fig. 3). This differential precipitation of lZ5I-Tf(Fe), could be completely abrogated by the inclusion of a high concentration of nonradioactive Tf(FeI2. At a ligand concentration of 2 nM, the assay was linear with respect to added extract and at least 60% of the added lZ5I-Tf(Fe), was specifically precipitated (Fig. 4a). The data in Fig. 4b demonstrate that the formation of a precipitable ligand-receptor complex was time dependent, saturable, and reversible. Scatchard analysis (20) of binding of '251-Tf(Fe)2 to Triton X-100 solubilized receptors at 0 "C revealed a straight line (Fig. 4c), indicating a homogeneous population of noninteracting receptors. A KD of 18.8 nM was determined by linear regression analysis according to the method of Scatchard (20). This value is in good agreement with that obtained for cell-associated Tf receptors (9). Thus, Triton X-100 solubilized Tf receptors appear to possess binding characteristics similar to those of cell-associated receptors described previously (9).
In order to verify the specificity of the assay for soluble receptors, we examined a Triton X-100 extract of mature red

Identification of a Nonrecycling Internal Pool of Receptors-
Trypsinization of HeLa cells at 37 "C for 20-30 min leads to almost complete loss of Tf receptor binding activity (5% of control value) (9). However, when Triton X-100 extracts of trypsinized cells were assayed for solubilized receptor binding activity, 20-40% of total control activity was detected (Fig.  5 ) . When Triton X-100 extracts from control and trypsinized cells were mixed, the total Tf binding activity of the mixture equaled the sum of the components. This finding demonstrates that trypsinization of whole cells has no inhibitory effect on the soluble receptor assay. These results suggest that some portion of the Triton X-100 soluble Tf receptor activity was refractory to the effects of trypsinization of whole cells. Subcellular Localization of Tf Receptors-In order to determine the location of intracellular Tf receptors, cells were incubated in Tf-free media for 60 min at 37 "C, homogenized, applied to a sucrose gradient, and each fraction assayed for soluble Tf receptor activity. As demonstrated in Fig. 6 , the peak of detectable receptor activity was coincident with leucyl-p-naphthylamidase activity with a broad shoulder in more dense fractions. When Triton X-100 was omitted from the assay mixture, only 15% of the Tf receptor binding activity could be detected (Fig. 6). This observation indicates that a large proportion of cellular Tf receptors are membrane bound and are located in a nonlysosomal, intracellular vesicle. Some of the receptors assayed probably included those on plasma membranes. To discriminate between surface receptors and internal receptors, cells were trypsinized at 0 "C, homogenized, and the homogenate applied to a sucrose gradient. Gradient fractions were assayed for soluble Tf receptor activity. Eighty-five per cent of control activity was detected, the bulk of which was coincident with leucyl-B-naphthylamidase activity (Fig. 7). A broad shoulder of Tf receptor activity was detected in higher density fractions.
To define the location of trypsin-resistant receptors, cells were trypsinized at 37 "C for 20 min, homogenized, and then applied to a sucrose gradient. Under these conditions, only 30% of control Tf receptor activity was detected using the soluble Tf receptor assay. The bulk of receptor activity was coincident with leucyl-P-naphthylamidase activity (Fig. 7). Again, a broad shoulder of Tf receptor activity was detected in higher density fractions. Thus, the trypsin-resistant recep-   37 and 0 "C. HeLa cells were incubated for 20 min (37 or 0 "C) with either PBS or PBS containing trypsin (0.25%).

FIG. 7. Distribution of Tf receptors from control cells and cells trypsinized at
Cells were then washed with media containing 10% fetal calf serum to inactivate tryptic activity. The cells were homogenized and samples of the homogenate applied to a linear sucrose density gradient. After fractionation of the gradient, Triton X-100-solubilized Tf receptors, hexosaminidase, and leucyl-0-naphthylamidase were assayed. The peak fractions of activity of hexosaminidase (*) and leucvl-8-naDhthvlamidase (**) are shown. Specific receptor .binding in-control ceils extracts (0-200 pl) and binding was determined. Specific binding to In the assay mixture, the identical amount of IZ5I-Tf(Fe), was added as was present in the samples used in A. 1251-Tf-occupied receptors were precipitated either immediately (0 min) or after a 60-min incubation at 0 "C. tor pool does not appear to be restricted to a specific subcellular site, at least on the basis of buoyant density.

Evidence That Internal Tf Receptors and Internalized T f Are in the Same
Vesicle-To determine whether Tf receptors and internalized "'I-Tf were present in the same vesicle, we designed an assay for occupied internal receptors. Cells were incubated a t 37 "C with lZ5I-Tf for 60 min and then shifted to 0 "C for 4 h in the presence of excess unlabeled Tf(Fe),. The cells were then harvested and homogenized. The homogenate was applied to a sucrose gradient and the leucyl-o-naphthylamidase associated peak of radioactivity isolated. This vesicular fraction was solubilized with Triton X-100 followed by the immediate addition of 45% (NH4),S04 and IgG. After centrifugation, radioactivity in the pellet was determined. AS shown in Fig. 8A, under these conditions, an lZ51-Tf-receptor complex was precipitated by (NH4),S04. The addition of unlabeled Tf(Fe), to the solubilized vesicular fraction did not reduce the amount of radioactivity detected in the (NH2),S04 precipitate.
The above experiment was repeated but Triton X-100 treated vesicles were incubated at 0 "C for 60 min prior to the addition of (NH4),S04. Under these conditions, there was an approximate 30% reduction in precipitable radioactivity (Fig.  8 A ) . This reduction probably represents fewer occupied receptors owing to a dilution of the ligand and receptor following disruption of the vesicle. The addition of excess unlabeled Tf(Fe), to the solubilized vesicle, followed by incubation at 0 "C for 60 min reduced the radioactivity detected in the (NH4),S04 precipitate to background levels. This reduction represents the dissociation of internalized '251-Tf(Fe)2 from internal receptors and subsequent binding of the excess unlabeled ligand.
In another set of experiments, cells were incubated at 37 "C for 1 h in the absence of any Tf. The cells were then incubated at 0 "C for 4 h in the presence of unlabeled Tf(Fe), (6 p~) . The leucyl-@-naphthylamidase-associated vesicle was isolated as above and solubilized with Triton X-100. ""I-Tf(Fe), was added to the solubilized extract. (NH4)2S04 was added either immediately or after a 60-min incubation at 0 "C and the precipitable radioactivity determined. When (NH4),S04 was added immediately, either in the presence or absence of unlabeled Tf(Fe),, precipitable radioactivity did not differ from background levels. When (NH4),S04 was added after a 60min incubation at 0 "C in the presence of added unlabeled Tf(Fe),, precipitable radioactivity did not differ from background levels. When (NH4),S04 was added after a 60-min incubation a t 0 "C in the absence of added unlabeled Tf(Fe),, a significant amount of radioactivity was precipitable. The amount of 1251-Tf(Fe)2 added (Fig. 8B) to the solubilized vesicular extract was identical (cpm) to the amount present in the assay mixture when vesicles from whole cells (preincubated with T-Tf(Fe),) were used (Fig. 8A). The amount precipitated by (NH4),S04, however, was significantly lower (compare Fig. 8, A and B ) . This difference likely reflects a higher concentration of both the ligand and the receptor within the vesicle recovered from cells preincubated with Iz5I-Tf(Fe),. These results indicate that the immediate addition of (NH4),S04 to Triton X-100 solubilized vesicles, "traps" the ligand-receptor complexes.
Examination of sucrose gradient fractions (Fig. 9) using the above assay, revealed that ligand-receptor complexes had the same buoyant density as either intracellular Tf-receptors or intracellular "'I-Tf. These results indicate that internalized Tf and cellular receptors were present in the same vesicle. Internalized Iz5I-Tf, Tf receptors, and Tf receptor-ligand complexes all co-sedimented at a buoyant density co-incident with the peak fraction of leucyl-b-naphthylamidase activity.

DISCUSSION
The studies presented here support the concept that Tf is internalized in the process of cellular iron uptake. Internalization of Tf has been reported by others using morphologic techniques, (22, 23) temperature shift experiments (5), and subcellular fractionation studies (7, 24). The evidence for internalization of Tf in our studies was derived from both temperature shift experiments (Fig. 1) and subcellular fractionation studies (Fig. 2).
Morgan (8) and Octave et al. (6) suggested that internalized Tf was localized within lysosomes. Their conclusions were based largely on the finding that agents which increase intracellular pH cause a decrease in cellular iron uptake. This finding is compatible with the fact that the release of iron from Tf is facilitated by an acidic pH (25).
We found no association between the buoyant density of internalized lZ5I-Tf and hexosaminidase activity, a lysosomal marker enzyme associated with all types of lysosomes; nor did we find any association between internalized lz5I-Tf and marker enzymes for the Golgi apparatus (galactosyl transferrase), smooth endoplasmic reticulum (glucose-6-phosphatase), or plasma membrane (5'-nucleotidase). We did, however, find coincident buoyant densities for leucyl-@-naphthylamidase activity and intracellular lz5I-Tf. It was initially suggested that leucyl-@-naphthylamidase activity was a marker for plasma membranes (19). However, we observed a clear separation between the buoyant density of leucyl-Pnaphthylamidase activity and that of 5'-nucleotidase activity (Fig. 2). When HeLa cell plasma membranes were iodinated at 0 "C and fractionated on sucrose gradients, the buoyant density of the 1251-labeled membrane fraction was the same as that of 5"nucleotidase activity. There was no coincidence between the peak fractions of lZ51 radioactivity and leucyl-0naphthylamidase activity. The precise role of leucyl-@-naphthylamidase is not known but its activity was useful as it was always coincident with intracellular vesicles containing lZ5I-Tf.
From our studies, we concluded that internalized lZ5J-Tf was localized in a nonlysosomal compartment. Van Renswoude et al. (7), using erythroleukemia cells, fractionated on colloidal silica gradients, also concluded that internalized Tf was localized in a nonlysosomal compartment. We cannot exclude the possibility that internalized '251-Tf was localized in a minor lysosomal species.
The results of our studies with an assay for solubilized Tf receptors demonstrate that 70-80% of cellular receptors are intracellular. This result is independent of preincubation of cells with ligand. This figure corresponds with that reported by Bleil and Bretscher (18), who localized HeLa cell Tf receptors using an antibody technique. Intracellular pools of receptors for insulin, (26) mannose terminal glycoproteins (27), a,-macroglobulin-protease complexes,' and mannose phosphate terminal glycoproteins (28) have been demonstrated by others. A large proportion of the intracellular receptor pool is sensitive to trypsinization at 37 "C ( Fig.5) suggesting that this population of receptors communicates with the cell surface through the endocytic pathway. There was, however a pool of trypsin-insensitive receptors representing about 30% of the total soluble cellular receptor population. We detected no major differences in the buoyant densities of the trypsin-sensitive and trypsin-resistant receptor pools. Analysis of the equilibrium binding behavior of the total soluble cellular Tf receptor population also did not reveal any evidence for receptor heterogeneity (Fig. 4c) of 3T3L1 cells. They noted that 70% of cellular insulin receptors were intracellular and a significant proportion of these were protease resistant. Deutsch et al. (26) demonstrated that protease-resistant receptors under certain circumstances could repopulate the cell surface. Studies are in progress in our laboratory to extend this observation to the proteaseresistant pool of Tf receptors. Intracellular Tf receptors appeared to be located in the same vesicular compartment as internalized '"I-Tf (Fig. 8) suggesting that Tf recept~r-''~I-Tf(Fe)~ complexes were internalized as an intact unit. As discussed previously, this vesicular compartment was distinct from lysosomes. At 37 "C, a large proportion of intracellular Tf receptors can be hydrolyzed by extracellular proteases. This demonstrates that intracellular Tf receptors are susceptible to proteolysis. Since one of the hallmarks of lysosomes is high concentrations of proteases and other hydrolyases, the above observations further indicates that intracellular Tf receptors are not localized in lysosomes.
Studies in a variety of systems have suggested that internalized ligands pass through a nonlysosomal compartment prior to localization in lysosomes (29)(30)(31)(32)(33). Aside from a low luminal pH (7,34), few properties of this nonlysosomal compartment are known. We suggest that in most systems dissociation of ligand-receptor complexes occur in this compartment. This would allow for separate fates for the ligand and the receptor. The ligand being localized in the fluid phase might be directed to lysosomes while membrane-bound receptors might recycle to the plasma membrane. In the case of Tf(Fe)2-Tf receptor complexes, the acidic pH of the nonlyso-soma1 vesicle would promote dissociation of iron from lZ5I-Tf(Fe)? (25) and actually enhance binding of apo-Tf to the Tf receptor, thus preventing dissociation of the complex (7, 21,35). This would permit both "unloading" of iron and recycling of an intact apo-Tf-Tf receptor complex.
Raising the pH of acidic intracellular compartments prevents the release of iron from Tf (8) but does not appear to prevent the recycling of Tf (8) or of the Tf r e~e p t o r .~ However, for those ligands whose ultimate fate is the lysosome, increases in the luminal pH traps both unoccupied receptors and receptor-ligand complexes (28,32,(36)(37)(38). These differences in response to increased vesicular pH suggest either that the Tf receptor is directed to a different acidic compartment from other receptors or that different receptors within the same compartment respond differently.