Coated Vesicles from Thyroid Cells Carry Iodinated Thyroglobulin Molecules AN INTERNALIZATION OF THE THYROID PROHORMONE VIA A MECHANISM OF RECEPTOR-MEDIATED ENDOCYTOSIS*

We have tried to identify iodinated thyroglobulin molecules in purified thyroid-coated vesicles to determined whether the internalization of the thyroid prohormone could proceed via a mechanism of receptor-mediated endocytosis. Coated vesicles isolated from pig thyroids by differential centrifugation and centrifugation on 2H2O-sucrose cushion were characterized by transmission electron microscopy and analyses of the polypeptide composition by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate and Western blot using anti-clathrin heavy chain and anti-thyroglobulin antibodies. Clathrin and thyroglobulin (Tg) appeared as the two major components of the purified thyroid coated vesicles (TCV). Purified TCV fraction was homogeneous when analyzed by isopycnic centrifugation on 30% Percoll gradient. TCV had an apparent buoyant density of 1.035 g/ml. The presence of Tg molecules inside TCV was ascertained by (a) immunogold labeling on cryosections of TCV pellet and (b) identification by gel electrophoresis and radio-immunoassay of a definite fraction of Tg (3-5% of total protein) in TCV treated by Triton X-100. The detergent-treated TCV also contained protein-bound iodine: 0.5-0.7 micrograms of iodine/mg protein. Pulse-chase experiments on in vitro reconstituted thyroid follicles have been used to further document the presence of iodinated Tg molecules in coated vesicles. TCV were isolated from reconstituted thyroid follicles previously labeled with [125I]iodide to radioiodinate Tg of the follicular lumen (the pre-endocytotic compartment) and incubated with or without thyrotropin or dibutyryl cyclic AMP to activate intraluminal 125I-Tg endocytosis. Autoradiographic analyses revealed the presence of 125I-Tg in purified TCV and Triton X-100-treated TCV. 125I-Tg present in TCV represented 1-2% of the total intracellular protein-bound radioactivity. Thyrotropin and dibutyryl cyclic AMP increased 2-3-fold the 125I-Tg content of TCV. Our results clearly show that iodinated Tg, the molecular form of the thyroid prohormone known to be internalized, is present into TCV. The data suggest that coated vesicles are involved in the uptake and transport of Tg from the follicular lumen to the lysosomal compartment and therefore, that the internalization of Tg could proceed, at least for a part, via a mechanism of receptor-mediated endocytosis.

follicular lumen to the lysosomal compartment and therefore, that the internalization of Tg could proceed, at least for a part, via a mechanism of receptor-mediated endocytosis.
Thyroid hormone synthesis implies bidirectional transport of the prohormone thyroglobulin (Tg).' After synthesis and maturation, Tg is released into the intrafollicular lumen and iodinated into this extracellular compartment.
Hormones containing Tg molecules are then taken up by the cells and conveyed to lysosomes for final processing and generation of free thyroid hormones. Two different pathways of Tg endocytosis have been reported involving the formation of either large vesicles, colloid droplets or small vesicles, pinocytotic vesicles (1). The former process also named macropinocytosis share morphological characteristics with phagocytosis; it implicates the formation of pseudopods by the apical plasma membrane. The second pathway or micropinocytosis implies the formation of vesicles by invagination of the cell membrane. The macropinocytosis process enables the cell to take up large amounts of Tg whereas micropinocytosis may offer selectivity in the Tg uptake. The relative importance of macro-and micropinocytosis for the secretion of thyroid hormones is still debated. Macropinocytosis has been well documented in rat and mouse thyroid tissue stimulated by thyrotropin (TSH) (1). The number and the size of pseudopods, as well as the number of endocytotic vesicles or colloid droplets, appear to be related to the degree of TSH stimulation.
It has been suggested (2) that micropinocytosis could be the major pathway of Tg uptake in normal gland, i.e. thyroid in a low activation state. In this situation, colloid droplets are rare or absent and the amount of Tg required for the maintenance of normal plasma levels of thyroid hormone could be internalized via micropinocytotic vesicles. A process of micropinocytosis was also postulated in hyperactive glands (after multiple TSH injections) in which the thyroid hormone release is not dependent on the appearance of pseudopods and the formation of colloid droplets (3).
Tg molecules stored in the follicular lumen present a large heterogeneity in terms of iodine or hormone content (4). It has been reported that Tg molecules with a high iodine content are taken up more actively and degraded more rapidly than low iodinated newly synthesized Tg (5,6). The micropinocytotic process seems to be compatible with the existence of such a selective uptake of Tg. The mechanism of selection of Tg molecules would require the presence of specific binding sites at the site of endocytosis, i.e. the apical plasma membrane. Attempts have been made to identify such binding sites on thyroid slices (7), purified membrane preparation (8), or open thyroid follicles.* Data obtained with intact cell systems are consistent with plasma membrane binding activity of Tg with an apparent K. of l-5 gM. Purified thyroid membrane preparations are capable of specific recognition of Tg, and the interaction process depends on the iodination level of the ligand but also on the presence or the absence of sialic acid (8,9). It is not known, however, whether these Tg-binding sites of asialoglycoprotein receptor type (10) are expressed on the cell surface or located on intracellular membranes. Additional work is needed to ascribe a role of these binding sites in the process of Tg internalization.
A mechanism of micropinocytosis characterized by the capacity of sorting, internalization, and addressing of Tg molecules to the lysosome compartment (11,12) could follow to a large extent the general pathway of receptor-mediated endocytosis (13,14). Receptorligand complexes form on specialized domains of the plasma membrane (coated pits) and enter the cell via coated vesicles. Soon after entering the cell, clathrin coats are shed from coated vesicles. The resulting vesicles deliver receptor-ligand complexes to endosomes. Depending on the ligand, receptorligand complexes dissociate or not, and receptor and/or ligand are transported into different cell compartments (plasma membrane, lysosomes). The first type of organelles involved in the selective ligand endocytosis coated vesicles (15) have been identified in thyroid cells by ultrastructural approaches (16)(17)(18)(19)(20). It is not known whether Tg is present in thyroidcoated vesicles. The only study (21) devoted to the analysis of the protein component of purified thyroid-coated vesicles has yielded negative results; the small amount of Tg found in the coated vesicle preparations was outside the vesicles associated to the clathrin coat. We have decided to reexamine this question, taking advantage of a method of tissue disruption (22) which permits elimination of more than 98% of the enormous amount of soluble Tg coming from intrafollicular stores and thus greatly decreases possible artifactual association of Tg with subcellular organelles. This procedure successfully used to purify lysosomes (22) and analyze intralysosomal Tg and intermediate degradation products (23)

TCV-
The combination of differential centrifugation steps and a centrifugation on a 'H20-sucrose cushion yielded highly purifled coated vesicle preparations.
One to 3 mg of TCV protein (mean = 1.7 + 0.4, n = 6) were obtained from 100 g wet weight pig thyroid tissue. The protein content of the starting cell homogenate (depleted in intrafollicular Tg stores) was 1010 f 50 mg/IOO g of thyroid tissue (n = 4). The light vesicle fraction (pellet P5) obtained at the last differential centrifugation step accounted for only 1% of total protein of the homogenate. About 20% of the particulate material of the P5 pellet were recovered as the purified CV fraction sedimenting through the sucrose-*Hz0 cushion. The purity and the structure of TCV were examined by transmission electron microscopy on ultrathin sections after resin embedding (Fig. 1). Ninety % or more of the vesicles could be identified as CV. At high magnification (X 150,000), CV appeared as electron dense vesicles exhibiting the bristle border characteristic of the clathrin coat. The TCV population was rather homogeneous in size, the average diameter being about 75 nm.
The polypeptide composition of purified TCV analyzed by SDS-polyacrylamide gel electrophoresis is reported in Fig. 2A The relative abundance of Tg, as compared with clathrin in purified TCV, was somewhat different from experiment to experiment. The ratio of Tg monomer to clathrin heavy chain estimated from densitometric tracings of Coomassie-stained gels varied from 1 to 3 to almost 1 to 1. These values must be considered as rough estimates since the ability of Tg and clathrin to bind the dye could be different. The electrophoretic analysis of the purified TCV revealed that numerous other minor bands copurify with clathrin; two groups of polypeptides with apparent molecular mass of 100-110 and 45-55 kDa could correspond to clathrin assembly factors (33). The polypeptides of about 35 kDa are probably clathrin light chains (34).
Peroxidase, a membrane-bound microsomal enzyme, was assayed to estimate the contamination of the purified TCV fraction by endoplasmic reticulum and Golgi vesicles. Peroxidase activity was measured after sonication of the particulate samples. The specific activity of the enzyme in TCV fraction was equal or lower than 0.1 unit/mg, a value 15-20 times lower than that measured in the crude vesicle fraction (Pl). Considering the specific activity values of peroxidase in TCV and Pl, one can estimate the contamination of TCV by microsomes to be below 5%.
We have used the isopycnic centrifugation on iso-osmotic Percoll gradients (Fig. 3) to test the homogeneity of the TCV preparations.
Purified TCV sedimented as a rather symmetrical narrow peak of low buoyant density (about 1.035 g/ml) in autogenerated 30% Percoll gradients. Purified CV and CV present in the crude thyroid particulate fraction (sedimenting from 8 X lo3 g X min to 9 X lo6 g X min) exhibited the same apparent density; CV, present in the crude thyroid vesicle fraction, identified by detection of clathrin by Western blotting were mainly recovered in fraction I of the Percoll gradient (Fig. 3) and mitochondria, sedimented as a broad peak with density ranging from 1.04 to 1.06 g/ml; lysosomes present in fraction VI had a density ranging from 1.08 to 1.14 g/ml (22).

Localization of Tg in Purified TCV Fractions-The
possibility of an artifactual association of Tg to TCV along the purification steps appeared very unlikely. Indeed, purified TCV did not exhibit any binding activity for exogenous labeled Tg, whatever the Tg concentration, from 80 pg/ml to 4 mg/ml (Table I).
Immunogold labeling with anti-Tg antibodies and chemical treatments have been used to precise the location of Tg in purified TCV preparations. The electron microscope observations of cryosections of isolated TCV further documented the structure of the vesicles. Using this sample preparation procedure, the membrane of the vesicles is clearly visible (Fig.  4). The average internal diameter of the vesicle was close to 50 nm whereas the mean diameter of the vesicle (with its coat) was about 85 nm. The interior of the vesicle appeared electron dense. Immunogold labeling with anti-Tg antibodies was used to visualize the presence of Tg into TCV. Anti-Tg antibody-protein A-gold complexes were observed either inside the vesicles or over the coat (Fig. 4). The specificity of the labeling was demonstrated by the absence of gold particles Purified TCV (about 1 mg of protein) or a crude thyroid vesicle fraction sedimenting between 8 x 10" g x min to 9 X 10" g X min (10 mg of protein) were mixed with 30% Percoll and centrifuged at 60,000 x g for 30 min. Fractions of 1 ml were collected from the top to the bottom and the vesicle distribution was determined by A:M measurements.
The fractions of the gradient containing crude thyroid vesicles were pooled as indicated on the figure @ool I-VI) and centrifuged at 150,000 x g for 60 min. The pellets were resuspended in TS buffer and assayed for protein; aliquots of 100 fig of protein were fractionated on 6% acrylamide gels and transferred onto nitrocellulose.
Immunoreactive clathrin was detected as in panel B of on TCV when anti-Tg antibodies were preabsorbed with purified Tg (result not shown). The location of gold particles over the coat does not necessarily indicate that Tg molecules are outside the vesicles. Indeed, due to the length of the immunoglobulin cross-bridge, a protein A-gold particle bound to a Tg-Anti-Tg antibody complex located on the internal side of the CV membrane can be found either in the center of the vesicle or on the other side of the membrane, on the coat. Biochemical approaches were used to further ascertain that TCV actually contain Tg molecules. TCV were submitted to three different treatments: (i) high ionic strength treatment (1.5 M NaCl) to eliminate peripheral membrane proteins associated to particles on the basis on charge-charge interactions; (ii) a treatment by 0.5 M Tris, pH 7.0, or 10 mM Tris, pH 8.5, to induce the dissociation of the clathrin coat and therefore the removal of protein artifactually attached or entrapped into the coat structure; and (iii) a treatment by 0.05% Triton X-100 to solubilize contaminating membrane or vesicles which could contain Tg. Coated vesicles are supposed to be not or only slightly affected by such a detergent treatment (35). As shown in panel A of Fig. 5, the polypeptide composition of purified TCV was not altered by the 1.5 M NaCl treatment. As expected, vesicles which sedimented after 0.5 M Tris treatment, no longer contained clathrin. Tg was still present in the resulting uncoated vesicle fraction. In contrast with previous treatments, the exposure of TCV to MMEP buffer containing 0.05% Triton X-100 yielded a sedimentable vesicle fraction with the same amount of clathrin, but with a reduced Tg content (panel B of Fig. 5). Intact Tg was recovered in the detergent supernatant together with some other components but no clathrin. The Triton X-100 treatment induced the release of soluble Tg and/or the solubilization of membrane-bound Tg present in contaminating non-coated vesicles. Indeed, under the conditions used, Triton X-100 was capable of solubilizing Tg from uncoated vesicles obtained by treatment of TCV by 10 mM Tris, pH 8.5, or 0.5 M Tris, pH 7.0 (Fig. 6).
Attempts have been made to quantitate the amount of Tg retained in Triton X-loo-treated TCV using a Tg radioimmunoassay (Fig. 7). Results of panel A of Fig. 7  which were roughly parallel to the standard curve generated with pure Tg. TCV resuspended in MMEP buffer or 0.5 M Tris and pelleted by centrifugation at 140,000 x g had almost the same Tg content as untreated TCV; less than 10% was recovered in the supernatant.
The Triton X-100 treatment induced the release of 50-60% of immunoreactive Tg, a value in keeping with the observations made on acrylamide gels. It must be noted that the total amount of immunoreactive Tg was increased about 1.5-fold in detergent-treated TCV as compared with that of TCV which were only freezed-thawed (column D versus column A of Fig. 7B). Assuming that the detergent treatment plus the freezing-thawing step allowed detection of all the Tg present, we estimated that Tg accounts for about 10% of total protein in purified TCV and 3-4% of total protein in Triton X-loo-treated TCV.
Iodine Content of TCV-Purified TCV supernatant and pellet of Triton X-loo-treated TCV were assayed for the presence of organic iodine (Table II). TCV preparations had a rather high iodine content: 1.2 f 0.2 (n = 3) pg/mg of protein. The treatment of TCV by 0.05% Triton X-100 (as compared with buffer alone) reduced the iodine content of the sedimentable vesicles. However, 40-60% of protein-bound iodine remained associated with Triton X-loo-treated TCV. The proportion of Tg and iodine recovered into the detergenttreated TCV are therefore very similar. As observed for Tg, uncoating of TCV by 0.5 M Tris treatment did not cause a significant release of iodine (result not shown). These data are in keeping with the presence of iodinated Tg molecules in TCV. Identification of Iodinated Tg in TCV: Pulse-Chase Experiments on in Vitro RTF-Thyroid cells cultured in the presence of TSH in plastic Petri dishes allowing cell attachment reorganized in histiotypic structures-thyroid follicles with morphological characteristics and metabolic activities corresponding to those of the intact tissue. The properties of in vitro RTF to iodinate Tg stored in the neoformed follicular lumina and to internalize iodinated Tg from that compartment, will be described in a separate report. For this study, we have taken advantage of this in vitro system to label Tg molecules which are subjected to the endocytotic process. Tg was metabolically labeled by incubating RTF with ['*"I]iodide for 1 h. After addition of methimazole, a peroxidase inhibitor, to block further incorporation of radioiodide, RTF were incubated for a short period of time (15 min) without or with TSH or dibutyryl cyclic AMP to activate '*'I-Tg internaliza- Procedures" for the procedure used to separate cells from the intrafollicular material). The CV isolation procedure used up to now on intact thyroid tissue have been applied to labeled RTF-derived cells to answer three questions. (a) Can we find internalized iz51-Tg molecules in CV? (b) How much of the internalized labeled prohormone is present in purified TCV and Triton X-lOOtreated TCV? (c) Does TSH and/or dibutyryl cyclic AMP increase the amount TCV-associated ""I-Tg? The electrophoretic analysis reported in panel A of Fig. 8 indicates that the CV fraction extracted from RTF-derived cells exhibited a purity comparable to that of CV extracted from the thyroid tissue. The fraction could be identified by the presence of clathrin.
TCV prepared from RTF-derived cells had a low Tg content; no Tg band was visible after Coomassie Blue staining. The autoradiograms shown on panel B of Fig. 8 demonstrate the presence of intact lz51-Tg molecules in TCV and Triton X-loo-treated TCV extracted from either untreated or TSH-or dibutyryl cyclic AMP-treated RTF. The amount of ""I-Tg in purified TCV was clearly augmented by both TSH and the cyclic AMP derivative. The quantitative analysis of lZ51-Tg distribution in cell subfractions and isolated CV is reported in Table III. Purified TCV collected as pellets after differential centrifugation and centrifugation on the *H20-sucrose cushion contained 1.6-3.0% of the cell-associated l*'I-Tg and accounted for 20-30% of '"'I-Tg present in the vesicle fraction which sedimented at 140,000 X g. The treatment of TCV by 0.05% Triton X-100 led to a partial release of radioactivity; however, 60-70% of '451-Tg remained bound to sedimentable vesicles after the detergent treatment. The radioactivity released by Triton X-100 mainly corresponded to labeled components with a molecular weight lower than that of Tg (panel B of Fig. 8). The amount of lZ51-Tg present in Triton X-loo-treated TCV expressed in percentage of total cell radioactivity was 2-fold higher in stimulated than in untreated cells (Table III). DISCUSSION The method of Nandi et al. (24) initially developed for the purification of brain CV gave satisfying results when applied to the thyroid epithelial tissue both in terms of yield and purity. The overall yield for TCV (1.7 mg of protein/100 g wet weight of tissue) was 2-3-fold lower than the yield for brain CV or CV from a number of sources (25). The difference is likely related to the presence of a very large non-cellular compartment, follicular lumina, in the thyroid. Indeed, intrafollicular soluble Tg stores represent 6080% of total proteins in normal glands. The analyses of (a) the structure of purified vesicles by transmission electron microscopy, (b) the polypeptide composition of the total vesicle fraction by polyacrylamide gel electrophoresis, and (c) the dispersity of the vesicle population by sedimentation on Percoll density gradients, give evidence for the purity of TCV preparations.
Considering the 180-kDa clathrin heavy chain as a CV marker and assuming that clathrin represents 0.15% of cellular proteins (37), one can estimate from the proportion of clathrin among purified TCV proteins (about 30-40%) that TCV were purified 200-300-fold as compared with the homogenate. Purified TCV preparations reproducibly exhibited a rather high Tg content. The fact that Tg present in TCV preparations could not be released by high salt treatment seems to indicate that Tg was packed inside vesicles. Immunogold labeling experiments definitely show the presence of Tg inside morphologically defined CV; however, one cannot exclude that part of TCV-associated Tg could be present in contaminating vesicles. Thus, to eliminate this potential problem and to try to better estimate the actual Tg content of TCV, we have used a final purification step based on the observations of Pearse (35) who reported that Triton X-100 up to 1% added at the beginning of CV purification or to partially purified CV was able to solubilize contaminating membranes or vesicles without affecting the protein content of the coated vesicles. The Triton X-100 treatment of purified TCV did not change the amount of vesicle-associated clathrin but decreased by 40-60% (depending on the preparatioan) the Tg content of TCV. From these data, we are inclined to think that purified TCV preparations contained a small proportion of non-coated vesicles with a high Tg content. These vesicles could correspond to the scarce structures with very electron dense content observed on ultrathin sections prepared from resin-embedded samples (panel A of Fig. 1). Such vesicles might represent Tg-rich exocytotic or endocytotic structures.
Tg molecules which remained associated to Triton X-lOOextracted TCV and accounted for 3-4% of the detergentinsoluble protein fraction meet the criteria for a CV-transported protein. They remain into the vesicles resulting from Tris buffer-mediated TCV uncoating and are extracted from the uncoated vesicles by a Triton X-100 treatment. The amount of Tg transported into CV could be somewhat underestimated if one considers that all the Tg extracted by the detergent exclusively corresponds to Tg present in contaminating vesicles. Indeed, some authors (38) think that the resistance of CV to Triton X-100 treatment could vary from tissue to tissue and that in some cases, the detergent could remove some receptors and proteins present inside CV.
Results of organic iodine assay in TCV and detergenttreated TCV are in keeping with the presence of iodinated Tg molecules inside TCV. If we assume that all the TCV-associated iodine is on Tg molecules, one can calculate that the iodination level of Tg would range from 60 to 80 iodine atoms/ Tg molecule. This value is compatible with a process of endocytosis which would involve a CV-mediated sorting of Tg molecules with a high iodine content. It must be noticed that the average iodination level of Tg stored in the follicular lumina of the pig thyroid glands used in these studies ranged from 32 to 42 iodine atoms/Tg molecule.

CV. In conclusion, our observations
show for the first time that the endocytosis of the thyroid prohormone Tg coulc proceed via a mechanism involving coated vesicles. It is there fore reasonable to think that Tg internalization, at least ix part or in certain situations, brings into play a process o receptor-mediated endocytosis. Efforts will now be made tc try to characterize the Tg receptor(s) which should median the sorting and the selective uptake of Tg molecules from the follicular lumen.
The identification of radioiodinated Tg in TCV prepared from prelabeled RTF reinforce the data obtained on TCV isolated from intact thyroid tissue. The Tg content of TCV preparations obtained from RTF was almost insensitive to Triton X-100. Triton X-100 mainly caused the release of labeled components with molecular weights lower than that of Tg. These molecular species could likely correspond to Tg degradation products generated in post-endocytotic contaminating organelles. The difference of effect of Triton X-100 on TCV isolated from intact tissue and TCV isolated from RTF was not unexpected since, potential contamination by Tgcontaining vesicles, seems to be higher in the former than in the latter situation. In the same way, the low Tg content of CV extracted from RTF-derived cells as compared with that of CV purified from intact tissue could be related to differences in the concentration of Tg in the pool from which Tg is internalized, i.e. the follicular lumina. Indeed, although Tg accumulates in the follicular lumina of RTF, the concentration of Tg which is achieved after 4 days of culture is much lower than that reached in follicules in intact tissue.
The pulse-chase type of experiments conducted on RTF not only demonstrate the transfer of iodinated Tg molecules into CV, but clearly show that the process is hormonally regulated. TSH and dibutyryl cyclic AMP are shown (a) to be activators of the endocytosis of prelabeled lz51-Tg in the RTF system (the endocytosis being evidenced by the decrease of luminal ?-Tg and the concomitant increase of cell-associated lZ51-Tg) and (b) to increase the iz51-Tg content of CV. Interestingly, TSH and dibutyryl cyclic AMP had a more marked effect on the labeling of CV than on either the labeling of the bulk of vesicles or on the overall cell labeling. The relative low amount of ""I-Tg found inside TCV (l-2% of intracellular radioactivity) seems to be compatible with the short half-life of CV or the short residence time of the ligand in CV (39,40). It must be noticed that experiments on RTF were conducted at 20 "C to decrease post-endocytotic events and to try to increase CV accumulation (41). Since CV are supposed to carry only very few molecules (l-4) of internalized ligand (35,42), it is likely that the TSH-induced increase of the labeling of CV corresponds to an increase in the number of CV rather than an increase of the lz51-Tg content of the