Apotransferrin and Holotransferrin Undergo Different Endocytic Cycles in Intestinal Epithelia (Caco-2) Cells*

Previous studies have demonstrated that diferric transferrin and apotransferrin compete for the binding to basolateral transferrin receptors and that transferrin-mediated iron uptake by Caco-2 cells is inhibited by apotransferrin to a larger extent than that predicted solely by receptor competition. This inhibition can have important implications in determining the net exchange of iron between intestinal cells and the basolateral milieu. Accordingly, we further characterized the endocytic cycles of apotransferrin and diferric transferrin in Caco-2 cells. We found that after internalization both apotransferrin and diferric transferrin recycled to the cell exterior, but that apotransferrin had a protracted endocytic cycle. Confocal microscopy studies revealed a different cellular distribution of apotransferrin and diferric transferrin; both were found in a compartment close to the basal membrane, but apotransferrin reached as well regions closer to the apical membrane. Moreover, the intracellular distribution of transferrin receptors was influenced by the iron load of transferrin; cells incubated with apotransferrin presented a more apical distribution of transferrin receptors than cells incubated with diferric transferrin. These results indicate for the first time that the endocytic cycle of transferrin receptors in intestinal epithelial cells is determined by the iron content of transferrin. They explain also the marked inhibitory effect of apotransferrin on transferrin-mediated iron uptake by Caco-2 cells, since incubation of cells with apotransferrin resulted in the actual sequestration of the receptor in the cell interior.

Iron is an essential nutrient involved in cellular functions related to the binding and transport of oxygen, redox reactions, detoxification, and nucleotide synthesis (1). Excess iron is toxic, and healthy individuals maintain a balance between obligatory body iron losses and iron absorption (reviewed in Ref. 2). The mechanisms that establish body iron homeostasis are, therefore, of great importance. Although the mechanisms involved in the regulation of iron absorption are not well known, a model describing the sequential passage of iron from mucin to integrin and to mobilferrin gives important clues on the molecular components involved in iron absorption (3). Using Caco-2 cells grown in bicameral inserts as a model of intestinal epithelia, it was shown that iron uptake through the basolateral endocytosis of iron-containing transferrin (Tf) 1 contributes importantly to the overall content of intracellular iron (4) and that the extent of apical iron uptake is inversely related to this content (5,6).
The endocytic cycle of Tf has been thoroughly characterized in a variety of cells (reviewed in Ref. 7). In many cell types internalized Tf first reaches the sorting endosome (pH 6.2) located in the peripheral cytoplasm, then moves to the recycling endosome (pH 6.4) and from there back to the cell surface (8). To our knowledge, the endocytic cycle of apoTf has not been characterized. Using apoTf as ligand, an apical endocytic compartment, where basolateral internalized apoTf colocalized with an apical fluid face marker, was described (9). These findings were not complemented with studies using holoTf, to determine if both ligands have a different intracellular distribution, and the rationale for the use of apoTf was not given (9).
There is a relationship between apoTf and iron flux in intestinal epithelia cells, since basolateral apoTf increases the apical to basolateral iron flux in iron-deficient Caco-2 cells (10). Furthermore, the binding of iron-containing Tfs to Tf receptors was competitively inhibited by apoTf, and apoTf inhibited Tfmediated iron uptake in Caco-2 cells (4). This inhibition was larger than expected solely by receptor competition, suggesting that apoTf affects other step(s) in the iron uptake cycle. From the above, it follows that the ratio apoTf/holoTf in the basal medium should determine the amount of cellular iron in intestinal epithelia cells, which in turn could determine the extent of body iron absorption. In this study we decided to further investigate the behavior of both Tf and the Tf receptor when Tf was internalized in the apo or in the iron-containing form. We found that both diFeTf and apoTf recycled to the cell exterior, but apoTf had a protracted endocytic cycle. Confocal microscopy studies revealed that diFeTf and apoTf reached different intracellular compartments. Moreover, Tf receptors also carried an endocytic cycle that was dependent on whether the ligand was apoTf or diFeTf. These results indicate for the first time that the endocytic cycle of the Tf receptor in intestinal epithelia cells depends on its bound ligand.

EXPERIMENTAL PROCEDURES
Cells-Caco-2 cells, from the American Type Culture Collection (no. HTB37, Rockville, MD), were cultured in Dulbecco's minimal essential medium (Life Technologies, Inc., catalog no. 430-2100EC) supplemented with 10% fetal bovine serum (Sigma). Culture medium was changed every 3 days. For transport and binding experiments, cells were grown on 0.33-cm 2 polycarbonate cell culture inserts, with 0.4-m * This work was supported by Grants 1940568 and 1970465 and from the Fonds de Ciencia y Tecnologia a Catedra Presidencial en Ciencias (to M. T. N.). 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.
Preparation of Tfs-Human apoTf (Sigma) was obtained by extensive dialysis of Tf against 100 mM sodium citrate, pH 4.5, as described previously (11), and it was kept in saline (50 mM Na-MOPS, 94 mM NaCl, 7.4 mM KCl, 0.74 mM MgCl 2 , 1.5 mM CaCl 2 , pH 7.4) plus 0.1 mM deferrioxamine (Sigma). DiFeTf was prepared from apoTf as described elsewhere (12). Tf was labeled with 125 I to one iodine atom per molecule by the method of McFarlane (13).
Labeling of Transferrin with Fluorescent Probes-DiFeTf was labeled with fluorescein isothiocyanate (FITC, Molecular Probes, Eugene, OR) to 5-6 mol of FITC per mol of transferrin following the instructions of the manufacturer. The Tf-FITC adduct was separated from free FITC by filtration through Sephadex G-200, and it was either used at once or stored in aliquots at Ϫ20°C until use. To obtain apoTf labeled with rhodamine isothiocyanate (RITC, Molecular Probes), diFeTf was incubated with RITC and gel-filtrated through Sephadex G-200 as described above, and the diFeTf-RITC adduct was converted to apoTf-RITC by dialysis against 100 mM sodium citrate, pH 4.5. ApoTf-RITC was then equilibrated in saline containing 0.1 mM deferrioxamine (Sigma) and either used immediately or stored at Ϫ20°C until use. To check for the possibility that the Tf-dye adducts could aggregate upon storage, we filtrated previously frozen aliquots of apoTf-RITC and diFeTf-FITC through Sephadex G-200. We did not observe the formation of aggregates under these conditions. Moreover, fractions corresponding to the beginning ("head") and the end ("tail") of the Tf peaks showed similar intracellular routings when tested by confocal microscopy. Therefore, the different routing of apoTf and diFeTf reported in this work cannot be ascribed to the aggregation of one of the Tfs.
Pulse-Chase Experiments of DiFeTf and ApoTf-125 I-Tf internalization studies were done as follows. Endogenous Tf was eliminated with three incubations of insert-grown Caco-2 cells for 10 min at 37°C with saline supplemented with 5 mM glucose and 0.1 mM deferrioxamine. Cells, in triplicate wells, were then pulsed for 30 min at 37°C in the above medium with either 0.1 M 125 I-labeled diFeTf or 0.5 M 125 Ilabeled apoTf. The cells were washed three times with saline, and the internalized 125 I label was chased for 0 -30 min at 37°C. The filters were excised with a scalpel, and the 125 I radioactivity in the cells, as well as in the basolateral medium, was determined in a Packard gamma radioactivity counter.
Effect of Hyperosmolarity on the Rate of Internalization of ApoTf and HoloTf-Insert-grown cells were incubated for various times at 37°C with either 0.1 M 125 I-labeled diFeTf or 0.5 M 125 I-labeled apoTf, in either saline or saline containing 0.43 M sucrose (14). When internalization of apoTf was determined, the incubation medium was supplemented with 0.1 mM deferrioxamine. The reaction was stopped, and cells were washed two times with cold saline. Surface-bound Tf was eliminated with an acid wash (15). The filters containing the cells were then excised from the inserts with a scalpel, and the 125 I radioactivity in the cells was determined as above.
Confocal Microscopy-Cells grown in transparent inserts (Costar) were incubated from the basolateral side for 60 min at 37°C with saline (50 mM Na-MOPS, 94 mM NaCl, 7.4 mM KCl, 0.74 mM MgCl 2 , 1.5 mM CaCl 2 , pH 7.4) containing 0.1 M diFeTf-FITC, 0.5 M apoTf-RITC, and 0.1 mM deferrioxamine. Cells were washed with ice-cold saline, and surface-bound Tf was eliminated by acid wash (15). Cells were then fixed with 3% glutaraldehyde for 20 min at room temperature. The inserts, placed with their bases on top of a glass slide, were placed on the microscope stage so the direction of the laser beam was basal to apical. The optical sections had the same basal to apical direction. Samples were protected from photo-bleaching with slow-fade TM (Molecular Probes).
To determine the cellular localization of the Tf receptor, cells were incubated with either 0.5 M apoTf or 0.1 M diFeTf as above. The cells were then fixed with 3% paraformaldehyde, treated with 0.5% Triton X-100, and reacted with OKT9 monoclonal anti-Tf receptor antibody (the kind gift of Dr. A. Orellana). Second antibody was goat anti-mouse IgG 1 (Sigma) labeled with RITC. Fluorescence was determined in the confocal microscope as described above.
Quantification of Light Intensity-Fluorescence intensity due to Tf or Tf receptor distribution along the apical to basal axis was quantified from composites of basal to apical fluorescence given by the computer of the confocal microscope (for an example, see Fig. 2C). These composites Insert-grown Caco-2 cells were pulsed for 15 min at 37°C with 125 Ilabeled apoTf or 125 I-labeled diFeTf, followed by a chase with unlabeled apoTf or diFeTf. At the times shown, the 125 I radioactivity remaining in the cell was determined. Shown is one of three similar experiments.

FIG. 2. Colocalization of apoTf and diFeTf in Caco-2 cells.
Caco-2 cells grown in transparent inserts were simultaneously incubated from the basolateral side for 60 min at 37°C with diFeTf-FITC and apoTf-RITC. Cells were then fixed with glutaraldehyde and examined with a confocal microscope. Optical sections were from the basal end to the apical end, so the sectioning defined the basal to apical axis (the z axis), while the optical cuts defined cell sections perpendicular to this axis. Shown is one of four independent experiments. A, 1.5-m optical cuts at (clockwise) 1.5, 9, 13.5, and 19.5 m from the basal side. ApoTf fluorescence was pseudo-colored with red and diFeTf fluorescence was pseudo-colored with green. Colocalization of both labels is displayed in yellow. B, gallery of 2-m optical sections starting from the basal toward the apical end. C, integration of fluorescence intensity in the basal to apical axis of a group of cells from those shown in B.
were divided into 10 equal sections following the basal to apical axis, and the relative intensity of the bands was determined using the SigmaScan program (Jandel Scientific, San Rafael, CA).
Data Analysis-Curve fitting was done using the GraphPad Prism program (GraphPad Software Inc., San Diego, CA). The experiments shown were repeated two to four times. Variability between experiments was Ͻ10%.

Endocytosis of ApoTf and DiFeTf, Pulse-Chase Kinetics-We
recently reported that iron uptake from 0.1 M diFeTf was 65% inhibited by 0.5 M apoTf, while the expected inhibition, based on receptor competition, was only 25% (4). One possibility for this extra inhibition was different cell handling of these Tfs. Hence, we determined the kinetics of the endocytic cycle of apoTf and diFeTf in Caco-2 cells (Fig. 1). Internalized diFeTf will be named "diFeTf," although it should lose most of its iron during endocytosis (4). DiFeTf presented a characteristic endocytic cycle kinetics with a t1 ⁄2 of about 25 min. ApoTf, instead, exhibited a protracted cycle with a t1 ⁄2 of about 60 min (Fig. 1).
Simultaneous Subcellular Localization of Internalized ApoTf and DiFeTf-We next studied the intracellular localization of internalized apoTf and diFeTf by confocal microscopy (Fig. 2). A high degree of colocalization (yellow spots) were observed in a basal region, while in more apical regions a predominance of the pseudo-red apoTf fluorescence was observed ( Fig. 2A). A gallery of optical cuts (Fig. 2B) and integration of the signal in the z axis (Fig. 2C) showed a distinctive pattern of distribution for both labels, with a preferential basal distribution for diFeTf and a more apical distribution for apoTf. Densitometric analysis indicated that diFeTf distributed mainly in the first half of the basal to apical axis, while apoTf distributed also into the second third of the axis (Fig. 3).
Effect of Hyperosmolarity on the Internalization of ApoTf and DiFeTf-The observed differences in endocytic cycle kinetics and intracellular localization of apoTf and diFeTf could be due to different mechanisms of internalization, e.g. clathrin-medi- ated or fluid phase endocytosis. Since only clathrin-mediated endocytosis is inhibited by hyperosmolarity (14), we tested its effect on the internalization of diFeTf or apoTf (Fig. 4). Internalization rates (molecules of Tf internalized ϫ h Ϫ1 ϫ insert Ϫ1 ϫ 10 Ϫ4 ) under isotonic and hypertonic conditions were 18.2 and 8.1 for apoTf and 45.5 and 18.8 for diFeTf, respectively (Fig. 4). Hence, hyperosmolarity inhibited to the same extent the internalization of apoTf and diFeTf, although the internalization rate of diFeTf was 2.5-fold larger than the rate of apoTf internalization. These results indicate that in Caco-2 cells both apoTf and diFeTf undergo clathrin-mediated internalization, and hence are internalized bound to the Tf receptor.
Intracellular Localization of Tf Receptors when Cells Were Incubated with ApoTf or DiFeTf-A different intracellular distribution of apoTf and diFeTf would result if one of the ligands, e.g. apoTf, dissociated from its receptor after internalization. Hence, we probed the location of Tf receptors after basolateral incubation of insert-grown Caco-2 cells with either apoTf or diFeTf. We found that, while incubation with diFeTf resulted in a receptor distribution mostly in the first half of the basal to apical axis, incubation with apoTf produced a more even distribution of receptors in this axis (Fig. 5). Thus, the distribution of Tf receptors followed closely in each case that of apoTf or diFeTf, an indication that these Tfs carried on their endocytic cycles bound to Tf receptors. DISCUSSION In its simplest form, iron absorption can be described as the regulated passage of lumenal iron through the intestinal epithelia. This passage is regulated by the intracellular iron levels of intestinal epithelial cells (6), and its adequate functioning maintains body iron homeostasis. It is therefore important to advance knowledge of the mechanisms responsible for the regulation of intracellular iron concentration in intestinal epithelial cells. With this in mind, we characterized the endocytic cycle of apoTf and diFeTf in polarized Caco-2 cells.
Pulse-chase experiments indicated that internalized apoTf took twice as long as diFeTf to externalize back to the basal medium. Therefore, apoTf not only competes with diFeTf for binding to the Tf receptor (4) but it actually sequesters the Tf receptor in intracellular compartments, rendering it unavailable for iron uptake. Blood plasma has a relation of apoTf to FeTf that varies from about 2 in individuals in iron balance to 10 or more in individuals with anemia secondary to iron deficiency (16). Because of the high concentration of Tf in plasma (about 35 M), even in conditions of iron deficiency, there should be enough FeTf in plasma to provide for a full supplement of iron to intestinal cells (4). The simple mechanism of apoTf-mediated receptor competition/sequestration described here should effectively down-regulate Tf-mediated iron uptake in these cells. Thus, enterocytes of iron-deficient individuals would have lower intracellular iron levels and hence enhanced iron absorption.
Confocal microscopy revealed differences in the distribution of internalized apo and diFeTf and of Tf receptors. It is puzzling why apoTf and diFeTf carry out different endocytic cycles since diFeTf delivers its iron in the process (4). The similar intracellular distribution found for Tfs and Tf receptors indicate that both apoTf and diFeTf carry on their endocytic cycles bound to their receptors. Two alternatives come to mind to explain these different distributions. It may be that after reaching a common basal compartment, e.g. the early endosome (7), the apoTf-Tf receptor complex could be targeted to an apical compartment, and from there back to the basolateral membrane. DiFeTf, on the other hand, could recycle back to the basolateral membrane directly from the basal compartment. Another possibility is that both apoTf and diFeTf pass through the apical compartment but that apoTf is retained longer than diFeTf in this compartment. Although the molecular bases that underlie this process are unknown, its functioning should help to regulate intracellular iron levels and iron absorption by intestinal cells.
In summary, we determined that apoTf and diFeTf undergo different endocytic cycles in polarized Caco-2 cells. The protracted cycle of apoTf sequesters the Tf receptor inside the cell, making it less available for iron uptake. We propose that this mechanism regulates the extent of iron uptake by making it a function of plasma iron levels, and that it is part of the mechanism that determines the net flow of iron between the intestinal cell and blood plasma.