Endocytosis of folate-protein conjugates: ultrastructural localization in kb cells

Endocytosis by mammalian cells occurs for various macromolecules via ligand binding to membrane-bound receptors and internalization in clathrin-coated pits or ‘uncoated’ caveolae. There are several fates for ligands utilizing the coated pit pathway. In some systems, the ligand is uncoupled from the receptor, the receptor is recycled to the plasma membrane, and the ligand is then degraded in lysosomes (e.g. low-density lipoprotein, asialoglycoprotein) (Goldstein et al., 1985; Green and Kelly, 1992; Gruenberg and Howell, 1989; Smythe and Warren, 1991). In other systems both receptor and ligand are degraded (e.g. in the case of epidermal growth factor) (Carpenter and Cohen, 1979). Furthermore, in the case of transferrin (TF) both the receptor and ligand are rapidly recycled (Dautry-Varsat et al., 1983), although long incubation times (>60 min) with transferrin tracers result in localization of TF in lysosomelike compartments (Hopkins, 1983). Endocytic pathways utilizing ‘uncoated’ pits or caveolae are less well characterized (Watts and Marsh, 1992). Ligands that have been identified to use this pathway include tetanus toxin (Montesano et al., 1982), cholera toxin (Montesano et al., 1982; Tran et al., 1987), and the vitamin folic acid (Rothberg et al., 1990).

Recently, it has been shown that proteins covalently coupled to folic acid may be delivered into mammalian cells via the endocytic mechanism for folic acid (Leamon and Low, 1991). Studies have demonstrated that these protein-folic acid conjugates enter HeLa or KB cells via a 65 kDa folate binding protein (Leamon and Low, 1993), that these conjugates may remain catalytically active (Leamon and Low, 1991), and that they may modify physiological processes occuring in the cytosol (Leamon and Low, 1992). Leamon and Low (1991) reported that FITC-IgG-folate internalized by KB cells could be visualized by confocal fluorescence microscopy in large vesicles (0.5 μm) within the cell cytoplasm. The purpose of the present study was to determine the subcellular location of protein-folic acid conjugates in KB cells using electron microscopy and to compare their location with that of a protein taken up via the clathrin-coated-pit pathway.

KB cells were maintained as previously described (Leamon and Low, 1992) in folate-free Dulbecco’s Modified Eagle’s Medium supplemented with 10% heat-inactivated fetal calf serum, penicillin (50 units/ml), streptomycin (50 μg/ml), and 2 mM L-gluta-mine at 37°C in a 5% CO₂/95% air humidified atmosphere. The day before each experiment, 2×10⁶ KB cells were plated in 60 mm Falcon culture dishes and incubated for 18 hours at 37°C.

Methods for the conjugation of folic acid to BSA were described previously (Leamon and Low, 1992). Briefly, folic acid or 5-methyltetrahydrofolic acid (5-MeH₄ folate) was dissolved in anhydrous dimethylsulfoxide and ‘activated’ with a 5-fold excess of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide for 1 hour at 23°C. The activated folate was then reacted with BSA in 0.1 M phosphate/boric acid buffer, pH 8.5, using an 8:1 folate to BSA ratio. Excess folate and other reactants were removed from conjugated protein using a PD-10 desalting column (BioRad). The resulting conjugates contained 1-2 folates per BSA molecule.

Fifteen nm colloidal gold (CG) sol was prepared by a citrate reduction method (Frens, 1973), and 5 nm CG sol was prepared by reduction with tannic acid (Slot and Geuze, 1985). Bovine serum albumin, F-BSA, and 5-MeH₄ folate-BSA solutions of identical protein concentration were adsorbed onto 15 nm CG at pH 6.5. Human holotransferrin (Sigma, St. Louis, MO) was adsorbed onto the 5 nm CG at pH 6.2. After stabilization of the CG sols with protein, the sols were pelleted and washed with 0.1 M phosphate buffer, pH 7.2, containing 0.2 mg/ml polyethylene glycol (M_r 20,000).

KB cells were incubated in Krebs-Ringer phosphate buffered saline containing 10 mg/ml BSA with trace amounts (∼10⁻⁸ M) of F-BSA-CG, 5-MeH₄ folate-BSA-CG or BSA-CG at either 11°C for 2 hours or 37°C for 15 minutes, and the cells washed with cold Krebs buffer to remove unbound material. Some cells were fixed immediately in 3% phosphate-buffered glutaraldehyde (pH 7.3), and others were incubated in fresh buffer at 37°C for additional time periods up to 6 hours. To compare receptor-mediated endocytosis via clathrin-coated pits to the folate uptake pathway, either 5 nm TF-CG alone or TF-CG plus 15 nm F-BSA-CG was incubated with KB cells for 15 minutes at 37°C, the cells were washed with cold Krebs buffer, and labeled cells fixed with 3% phosphate-buffered glutaraldehyde or incubated at 37°C for up to 6 hours before fixation.

Glutaraldehyde-fixed cells were washed in 0.15 M Millonig’s phosphate buffer, pH 7.3, postfixed in 1% osmium tetroxide/1.5% potassium ferrocyanide in phosphate buffer, dehydrated in a graded ethanol series, and embedded in Poly/Bed 812 (Polysciences, Warrington, PA). Some preparations were embedded as a monolayer so that the cells could be sectioned parallel to their growth plane to examine large areas of the plasma membrane. Thin sections (60-80 nm) were stained with uranyl acetate and lead citrate, and examined in a JEOL JEM-100CX at 80 kV. To determine the percentage of the CG conjugates present in different endocytic compartments at various time points, the thin sections were systematically examined and the location and number of CG particles noted as follows: (1) plasma membrane, (2) cave-olae, (3) small 50-90 nm tubulovesicular structures, (4) multi-vesicular bodies, (5) dense endosomes, (6) dense endosomes with membranous components typical of secondary lysosomes, (7) morphologically undefined irregular endosomes (refer to Figs 1G and 5). Counting was done directly from the microscope screen at a magnification of ×6,600. The data for the F-BSA-CG and BSA-CG conjugates are the result of 4 separate labeling experiments.

The distribution of the F-BSA-CG particles at three time points is contained in Table 1. Cells pulse labeled for 15 minutes with F-BSA-CG at 37°C had F-BSA-CG clusters on the plasma membrane in or near 70-90 nm caveolae up to 6 hours following pulse labeling (Fig. 1A,B,C). At 15 minutes, F-BSA-CG particles could be found in dense tubular endosomes (Fig. 1D), in MVBs (Fig. 1E), and on the rim of clear MVBs similar to Fig. 1F. After 1 hour, F-BSA-CG localized in the above mentioned compartments, in irregular endosomes that appeared to be formed by the fusion of three or more vesicles (Fig. 1G), and in various MVBs associated with the Golgi complex (Fig. 2). By 6 hours F-BSA-CG particles were still found in the above endosomal compartments, but were also found in dense endosomes, some of which contained membranous debris characteristic of secondary lysosomes. These were located in the juxtanuclear region of the cell in proximity to the Golgi complex (Fig. 3). Control cells incubated with BSA-CG had a few CG particles located in small clear vesicles at 15 minutes and large dense MVBs and secondary lysosome-like structures by 6 hours (Fig. 4). There was a considerable difference in the amount of BSA-CG internalized compared to F-BSA-CG. Only an occasional CG particle could be found in less than 10% of thin-sectioned KB cells incubated with BSA-CG. In contrast, 50% or greater of thin-sectioned KB cells contained clusters of F-BSA-CG particles. A rare finding at 6 hours was the presence of large irregular endosomes surrounded by a fine granular to filamentous area of cytoplasm devoid of organelles (Fig. 5A). There appeared to be some release of F-BSA-CG particles into the cytosol (Fig. 5B). Incubation of KB cells with 5-MeH₄ folate-BSA-CG resulted in internalization via caveo-lae followed by a high degree of localization in MVBs (Fig. 6) as occurred with cells incubated with F-BSA-CG.

To determine if the F-BSA-CG particles were binding specifically to the cell surface compared to BSA-CG conjugates, incubation studies were carried out at 11°C. At this temperature non-specific endocytosis is halted, but binding may still occur. Bovine serum albumin-CG conjugates were not detected microscopically on or in KB cells after the initial 2 hour binding at 11°C and incubation for 5 and 30 minutes at 37°C. At 60 minutes and 6 hours, occasional BSA-CG particles were found in large dense endosomes in a few cells. In contrast, F-BSA-CG conjugates were found on the plasma membrane near uncoated pits 5 minutes after cell washing and in MVBs and tubular endosomes at 30 and 60 minutes. At 1 and 6 hours, F-BSA-CG conjugates localized in dense tubular endosomes and MVBs, which were often found in proximity to the Golgi complex. The initial binding at 11°C did not result in BSA-CG or F-BSA-CG conjugates residing in any different endocytic compartments compared to cells that were pulse labeled at 37°C. However, there was an overall decrease in the number of CG particles found within the cells (data not shown).

At 15 minutes the TF-CG was localized in clathrin-coated pits (Fig. 7A) and coated vesicles (Fig. 7B), various tubulovesicular endosomes (Fig. 7C), and on the rim of clear MVBs (Fig. 7D). At 1 hour TF-CG was localized in clear vesicles (Fig. 7E), and MVBs (Fig. 7F), and MVBs in the juxtanuclear region of the cell (Fig. 7G). By 6 hours, TF-CG was primarily located in large dense endosomes (Fig. 7H). The location of TF-CG at the early and late time points is consistent with what is known about TF and the TF receptor.

There was no colocalization of TF-CG and F-BSA-CG at 15 minutes. At 1 hour, TF-CG and F-BSA-CG colocalized in irregular vesicles, similar to those in Fig. 1G, that appeared to result from the fusion of multiple vesicles (Fig. 8A), and in various MVBs (Fig. 8B,C). By 6 hours TF-CG and F-BSA-CG frequently colocalized in large dense endosomes (Fig. 8D).

Folic acid and 5-MeH₄ folate are taken up by animal cells via a folate receptor (FBP). In the model proposed for receptor-mediated uptake by the FBP of MA104 cells, 5-MeH₄ folate binds to FPB clustered around caveolae (Roth-berg et al., 1990). These caveolae are proposed to invagi-nate, forming a distinct compartment within the cytoplasm that remains physically associated with the plasma membrane. An internal proton gradient is then generated, facilitating the release of folate from the receptor and permitting the vitamin to move across the caveolae membrane by an integral membrane anion carrier (Kamen et al., 1991; Rothberg et al., 1990). This process was termed potocyto-sis to differentiate it from classical endocytosis mediated by clathrin-coated pits (Anderson et al., 1992).

There is no evidence to suggest that caveolae function as transport vesicles in an endocytic pathway (Rothberg et al., 1992). This is supported by studies that utilized CG-labeled anti-FBP antibody and demonstrated that the folate receptor remained permanently associated with the caveo-lae of MA104 cells (Rothberg et al., 1990). Also, studies on a 22 kDa protein associated with caveolae, termed ‘cave-olin’, showed that the caveolin could not be found associated with any endocytic vesicles (Rothberg et al., 1992). The MA104 cells are only capable of internalizing 5-MeH₄ folate via FBPs on the plasma membrane, whereas KB cells can internalize 5-MeH₄ folate and folic acid equally well. The KB cells also have both plasma membrane and intracellular FBPs (McHugh and Cheng, 1979). This may indicate that fundamental differences exist in the mechanisms of folate transport in these two cell types. The KB cell may transport folate similarly to rabbit proximal kidney tubule cells. Electron microscopic localization of [³H]folate in kidney proximal tubule cells demonstrated that the folate resided in various endocytic compartments and lysosomes (Hjelle et al., 1991). The study by Hjelle et al. (1991) also showed that endosomes (dissociated from the plasma membrane) were labeled with a CG-anti-FBP anti-body, suggesting that more than one endocytosis pathway for folate may exist. Another source of variation from the system in MA104 cells may be differences in glycosylation or isoforms of the FBP (Wang et al., 1992). The FBP in MA104 cells is a 38 kDa protein, whereas the KB cell FBP in this study was 65 kDa (Leamon and Low, 1993).

The results of this present study are in agreement with the potocytosis model in that the F-BSA-CG particles bound to KB cells in or near small uncoated pits, or caveolae; however, our results differ from this model in the the F-BSA-CG particles were localized within MVBs and other tubulovesicular elements characteristic of classical endocytic structures. Since our study followed the ligand and not the receptor, it is uncertain at what point, or in what endocytic compartment, the F-BSA-CG dissociated form the receptor. Micrographs in which the F-BSA-CG particles are attached to the rim of endocytic vesicles (Fig. 1F) would suggest that a high-affinity membrane binding site still exists for the conjugate in this compartment. Other micrographs in which the F-BSA-CG is free in the matrix of MVBs (Fig. 1E) presumably show the ligand dissociated from the receptor. The distribution of the F-BSA-CG particles, presented in Table 1, indicates that the MVB is the primary structure for routing these conjugates through the endocytic pathway. The increased frequency of F-BSA-CG particles in small tubules or vesicles as well as lysosomal compartments at 6 hours likely indicates that after the initial peak concentration in MVBs at 1 hour, conjugates were being shunted from the MVBs to other compartments.

In summary, we propose that proteins conjugated to folic acid enter via uncoated pits or caveolae as suggested by others. However, the conjugates then enter an endocytic pathway that shares compartments with the clathrin-coated pit pathway. The delivered proteins remain sequestered primarily in MVBs and other compartments of the tubulovesicular system, with some eventually residing in secondary lysosomes after longer incubation times. Interestingly, our results are similar to those from studies that showed that the caveolae pathway for cholera toxin and the clathrin-coated-pit pathway of α2-macroglobulin shared various endocytic compartments (Tran et al., 1987).

The manner in which folate-protein conjugates are able to modify physiological processes is still open to question, but there is some indication that certain endosomes may release their contents directly into the cytosol. It is also possible that the presence of the colloidal gold particle prevents extensive release or transport into the cytosol, which would occur if the particle were not present. The pathway for these folate-protein conjugates may also not be representative of folate transport in general, since the presence of the protein may alter transport in an unknown manner. Additional experiments that simultaneously track both the FBP and folic acid-protein conjugates and that utilize markers or antibodies for the various endocytic compartments should further clarify the mechanisms involved in transport of folic acid-protein conjugates.

The authors thank Deborah Van Horn and Phyllis Lockard for their assistance in sample preparation for electron microscopy.