Identification of a Heparin-binding Region of Rat Thyroglobulin Involved in Megalin Binding*

We recently showed that thyroglobulin (Tg) is a heparin-binding protein and that heparin inhibits binding of Tg to its endocytic receptor megalin (gp330). Here we have identified a heparin-binding region in the carboxyl-terminal portion of rat Tg and have studied its involvement in megalin binding. Rat thyroid extracts, obtained by ammonium sulfate precipitation, were separated by column fractionation into four Tg polypeptides, with apparent masses of 660, 330, 210, and 50 kDa. As assessed by enzyme-linked immunoadsorbent assays and ligand blot binding assays, megalin bound to intact Tg (660 and 330 kDa) and, to a even greater extent, to the 210-kDa Tg polypeptide. Furthermore, the 210-kDa Tg polypeptide inhibited megalin binding to intact Tg by ∼70%. Solid phase assays showed binding of biotin-labeled heparin to intact Tg and to the 210-kDa Tg polypeptide. We characterized the 210-kDa Tg polypeptide by matrix-assisted laser desorption/ionization mass spectrometry analysis and found that it corresponds to the carboxyl-terminal portion of rat Tg. We developed a synthetic peptide corresponding to a 15-amino acid sequence in the carboxyl-terminal portion of rat Tg (Arg689–Lys703), containing a heparin-binding consensus sequence (SRRLKRP) and demonstrated heparin binding to this peptide. A rabbit antibody raised against the peptide recognized intact Tg in its native conformation and under denaturing conditions. This antibody markedly reduced heparin-binding to intact Tg, indicating that the region of native Tg corresponding to the peptide is involved in heparin binding. Furthermore, the anti-Tg peptide antibody almost completely inhibited binding of megalin to Tg, suggesting that the Tg region containing the peptide sequence is required for megalin binding. Physiologically, Tg binding to megalin on thyroid cells may be facilitated by Tg interaction with heparin-like molecules (heparan sulfate proteoglycans) via adjacent binding sites.

We recently showed that thyroglobulin (Tg) is a heparin-binding protein and that heparin inhibits binding of Tg to its endocytic receptor megalin (gp330). Here we have identified a heparin-binding region in the carboxyl-terminal portion of rat Tg and have studied its involvement in megalin binding. Rat thyroid extracts, obtained by ammonium sulfate precipitation, were separated by column fractionation into four Tg polypeptides, with apparent masses of 660, 330, 210, and 50 kDa. As assessed by enzyme-linked immunoadsorbent assays and ligand blot binding assays, megalin bound to intact Tg (660 and 330 kDa) and, to a even greater extent, to the 210-kDa Tg polypeptide. Furthermore, the 210-kDa Tg polypeptide inhibited megalin binding to intact Tg by ϳ70%. Solid phase assays showed binding of biotin-labeled heparin to intact Tg and to the 210-kDa Tg polypeptide. We characterized the 210-kDa Tg polypeptide by matrix-assisted laser desorption/ionization mass spectrometry analysis and found that it corresponds to the carboxyl-terminal portion of rat Tg. We developed a synthetic peptide corresponding to a 15-amino acid sequence in the carboxyl-terminal portion of rat Tg (Arg 689 -Lys 703 ), containing a heparin-binding consensus sequence (SRRLKRP) and demonstrated heparin binding to this peptide. A rabbit antibody raised against the peptide recognized intact Tg in its native conformation and under denaturing conditions. This antibody markedly reduced heparin-binding to intact Tg, indicating that the region of native Tg corresponding to the peptide is involved in heparin binding. Furthermore, the anti-Tg peptide antibody almost completely inhibited binding of megalin to Tg, suggesting that the Tg region containing the peptide sequence is required for megalin binding. Physiologically, Tg binding to megalin on thyroid cells may be facilitated by Tg interaction with heparin-like molecules (heparan sulfate proteoglycans) via adjacent binding sites.
Megalin (gp330) is a member of the low density lipoprotein receptor family (1, 2) expressed on the apical surface of certain absorptive epithelial cells, including thyroid cells (3,4). Based on the assumption that physiologically megalin binds to ligands to which it is exposed in different organs (5), we postulated that megalin on thyroid cells is a receptor for thyroglob-ulin (Tg). 1 Tg is synthesized in thyrocytes and released into the follicle lumen, where it is stored as the major component of colloid (6,7). Hormone secretion requires uptake of Tg by thyrocytes, with transport to lysosomes, where proteolytic cleavage leads to release of hormones from mature Tg molecules (6). Internalization of Tg may result from pseudopod ingestion, but under most conditions uptake occurs by micropinocytosis (vesicular internalization), which can take place both by nonselective fluid phase uptake and receptor-mediated endocytosis (6 -17). In previous studies (18,19) we showed that megalin is a high affinity receptor for Tg and that it can mediate Tg endocytosis by cultured thyroid cells. We also demonstrated that heparin almost completely inhibits binding of megalin to Tg (18,19), and, in studies designed to investigate the mechanism, we showed that Tg is a heparin-binding protein, as are certain other megalin ligands (19). In the present study we show that a 15-amino acid sequence in the carboxylterminal portion of rat Tg, rich in positively charged residues, is involved in binding of Tg to heparin and to megalin.
A previously described mouse anti-megalin monoclonal antibody, designated 1H2, has been shown to react with ectodomain epitopes in the second cluster of ligand-binding repeats (21). A rabbit antibody against human Tg, cross-reactive with Tg from other species, was purchased from Axle (Westbury, NY). Alkaline phosphatase (ALP)conjugated goat anti-rabbit IgG and horseradish peroxidase-conjugated goat anti-rabbit IgG were obtained from Bio-Rad (Hercules, CA). ALPand horseradish peroxidase-conjugated goat anti-mouse IgG were obtained from Sigma.
Purification and Fractionation of Rat Tg-Tg was prepared from frozen rat thyroids by ammonium sulfate precipitation and column fractionation, as described previously (22). Fifty rat thyroids (Pel-Freeze Biologicals, Rogers, AR) were homogenized in phosphate-buffered saline (PBS), pH 7.4, with an electric homogenizer. The thyroid homogenate was centrifuged for 10 min at 10,000 rpm, and the supernatant (the thyroid extract) was recovered. The extract was incubated in 45% ammonium sulfate for 6 h at 4°C, followed by centrifugation at 3000 g for 30 min. The pellet was resuspended in PBS and dialyzed overnight at 4°C in 2 liters of PBS. The dialyzed material was then fractionated by elution with PBS, using a 90 ϫ 2.6 cm Sephadex G-200 column (Sigma). 3-ml fractions were collected. The protein concentration of each fraction was measured using a commercial kit (Bio-Rad).
Analysis of the Thyroid Fractions-Each 3-ml thyroid fraction was subjected to SDS-polyacrylamide gel electrophoresis under nonreducing conditions followed by Coomassie staining. To study the immunoreactivity of the thyroid fractions to a rabbit anti-Tg antibody, Western blotting and enzyme-linked immunoadsorbent assays (ELISA) were performed.
For Western blotting, 10 g of each thyroid fraction was subjected to SDS-polyacrylamide gel electrophoresis under nonreducing conditions and blotted onto nitrocellulose membranes, which were incubated with the rabbit anti-Tg antibody (1:500 in TBS, 0.05% skimmed milk) followed by horseradish peroxidase-conjugated goat anti-rabbit IgG (1: 2500) and autofluorography using a chemiluminescent substrate kit (Kirkegard & Perry Laboratory, Gaithersburg, MD).
For ELISAs, 96-well microtiter plates were coated overnight at 4°C with each thyroid fraction (100 l/well), adjusted to a concentration of 100 g/ml in PBS, or, as control, with OVA at the same concentration. Wells were then blocked with bovine serum albumin (Sigma), washed, and incubated with the rabbit anti-human Tg antibody (1:500), followed by ALP-conjugated goat anti-rabbit IgG (1:3000). After incubation with p-nitrophenyl phosphate (Sigma), absorbance at 405 nM was determined with a EL-311 ELISA microplate reader. Absorbance obtained in bovine serum albumin-coated wells was considered as the blank.
Binding of Megalin to Tg Fractions by ELISA-To determine the megalin binding ability of each thyroid fraction, solid phase assays were performed as described (18). Briefly, ELISA plates were coated overnight at 4°C with each thyroid fraction or, as a control, with OVA. For coating, thyroid fractions or OVA were adjusted to a concentration of 100 g/ml in PBS and added in a volume of 100 l to the wells. The amounts of coated proteins were calculated by subtracting the amount of protein recovered after coating from the initial amount of protein added to the wells, assessed with a commercial kit (Bio-Rad). The mean amount of coated protein was 1.5 g/well for fractions corresponding to 660-and 330-kDa Tg, 1.7 g/well for fractions corresponding to the 210-kDa Tg polypeptide, 1.6 g/well for fractions corresponding to the 50-kDa Tg polypeptide, and 1.5 g/well for OVA. Following coating, wells were blocked with bovine serum albumin, washed with TBS containing 0.05% Tween-20, and incubated with purified rat megalin (5 g/ml) for 1 h at room temperature, in binding buffer: TBS, 5 mM CaCl 2 , 0.5 mM MgCl 2 , 0.5% bovine serum albumin, 0.05% Tween-20. To detect bound megalin, wells were washed with TBS, 0.05% Tween-20 and incubated with the mouse monoclonal anti-megalin antibody (1H2, 20 g/ml), followed by ALP-conjugated goat anti-mouse IgG secondary antibody (1:3000). After incubation with p-nitrophenyl phosphate, absorbance was determined at 405 nM. The amount of bound megalin was calculated using a standard obtained by coating the wells with 5 g/ml of purified megalin and was normalized for the amount of coated proteins. To study the ability of the 210-kDa or of the 50-kDa Tg polypeptide to inhibit binding of megalin to intact Tg, experiments were performed using ELISA wells coated with preparations of Tg containing 660-and 330-kDa Tg. Wells were incubated with megalin alone or with megalin preincubated overnight at 4°C with the fractions corresponding to pure 210-or 50-kDa Tg polypeptides at a concentration of 100 g/ml. Because some overlapping between different fractions was seen by Coomassie staining in Ϯ 2 fractions (Ϯ6 ml) at the border of the elution between regions containing the three Tg components found (660-and 330-kDa Tg, 210-kDa polypeptide, and 50-kDa polypeptide), to obtain thyroid fractions that were devoid of contaminating Tg polypeptides, we selected thyroid fractions that were separated from the border by at least 10 ml of elution. Only such pure fractions were used in this study, unless otherwise specified.
Binding of Megalin to Tg Polypeptides by Ligand Blot Assay-Thyroid fractions containing pure Tg polypeptides (5 g/fraction) or, as a control, OVA (5 g), were subjected to SDS-polyacrylamide gel electrophoresis under nonreducing conditions and transferred onto nitrocellulose membranes, which were incubated for 3 h at room temperature with purified megalin (10 g/ml in TBS, 5 mM CaC1 2 , 0.5 mM MgC1 2 , 2% skimmed milk) followed by 1H2 (20 g/ml) and horseradish peroxidase-conjugated goat anti-mouse IgG (1:10,000). The blots were analyzed using a chemiluminescent substrate kit.
Binding of Tg Polypeptides to Heparin by Solid Phase Assays-To investigate whether heparin can bind to the Tg polypeptides found in the thyroid extract, solid phase binding assays were performed as described (19). Briefly, 96-well microtiter plates were coated overnight at 4°C with thyroid fractions containing pure Tg polypeptides or, as a control, with OVA at a concentration of 100 g/ml in PBS, as described above. After blocking with bovine serum albumin, plates were incubated for 3 h at room temperature with a biotin-labeled heparin-albumin complex (Sigma) (0.1 g/ml) or, as a control, with biotin-labeled albumin (Sigma) (0.1 g/ml) in PBS, 0.05% Tween-20, 0.5% bovine serum albumin, followed by ALP-conjugated streptavidin (Vector, Burlingame, CA; 1:3000) and p-nitrophenyl phosphate. Absorbance was determined at 405 nM. The amount of bound biotin-labeled ligands was calculated using standard results obtained by coating the wells with 1 g/ml of biotin-labeled heparin or of biotin-labeled albumin and was normalized for the amount of coated proteins. For inhibition experiments biotin-labeled heparin was added to the wells alone or together with unlabeled heparin (500 units/ml), with thyroid fractions containing pure 210-or 50-kDa Tg polypeptides (100 g/ml), or, as a control, with OVA (100 g/ml).
Characterization of the 210-kDa Tg Polypeptide-To characterize the 210-kDa polypeptide found in the thyroid extracts, we used site-specific proteolysis combined with matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS), as described previously (23)(24)(25)(26). Following 6% nonreducing SDS-polyacrylamide gel electrophoresis of thyroid fractions containing pure 210-kDa Tg polypeptide, the 210-kDa band was cut out with a razor blade and subjected to in-gel proteolytic digestion with the protease Lys-C (Roche Molecular Biochemicals). The proteolytic digest was analyzed at the Protein Chemistry Facility at Columbia University (New York, NY) on a PerSeptive Voyager DE-RP mass spectrometer in the linear mode. The observed molecular masses of the peptides generated by proteolysis were compared with the masses of the theoretical Lys-C digests of the amino-terminal 213 amino acids and of the carboxyl-terminal 967 amino acids of rat Tg, the only published portions of the rat Tg cDNA sequence (27,28), which were generated by computer. Peptides containing cysteine residues were excluded from analysis because of alterations in molecular mass caused by covalent cross-linking with acrylamide in the SDS-polyacrylamide gel electrophoresis.
Development of Synthetic Tg Peptides-Seven synthetic Tg peptides were used in this study. The sequences of these peptides and their position within the amino acid sequence of rat Tg are shown in Table I. A 15-amino acid peptide corresponding to a sequence (Arg 689 -Lys 703 ) in the carboxyl-terminal portion of rat Tg containing a Cardin and Weintraub heparin-binding consensus sequence (29 -32) (SRRLKRP) was synthesized by Genemed Biotechnologies (South San Francisco, CA) and by the Peptide-Protein Core of Massachusetts General Hospital (Charlestown, MA). The peptide was named Tg peptide 1. No appreciable differences in the extent of heparin binding between the two preparations of Tg peptide 1 were seen (not shown). Unless otherwise specified, in the experiments performed with Tg peptide 1 the preparation from Genemed Biotechnologies was used. Another 15-amino acid peptide in which the 4 inner arginine and lysine residues of Tg peptide 1 were substituted with glycine was synthesized by Genemed Biotechnologies and was named "control peptide." To study the effect of modifications of the amino acid sequence on the heparin binding ability of Tg peptide 1 several "mutant" peptides were synthesized by the Peptide-Protein Core of Massachusetts General Hospital. A peptide in which the 15 amino acid residues of the synthetic Tg peptide 1 were randomly shuffled was named "scrambled peptide." Four synthetic peptides in which one of the four inner lysine or arginine residues in the sequence of Tg peptide 1 was substituted with glycine were named mutants 1 to 4.
A second synthetic 15-amino acid Tg peptide (Tg peptide 2) corresponding to another 15-amino acid sequence (Met 796 -Ile 810 ) containing another heparin-binding consensus sequence (ARRTRG) within the 967-amino acid carboxyl-terminal sequence of rat Tg was synthesized by the Peptide-Protein Core of Massachusetts General Hospital.
Experiments with Synthetic Tg Peptides-The ability of megalin and heparin to bind to the synthetic Tg peptides was determined in solid phase assays as described above, by coating ELISA microtiter wells with 100 g/ml of the peptides. The amount of coated peptides, calculated as described above, ranged from 1.6 to 1.8 g/well.

Met796-Ile810
To compare the heparin binding affinity of Tg peptide 1 with that of intact Tg, heparin-binding assays were performed using microtiter wells coated with 100 g/ml of Tg peptide 1 or of intact Tg (660-and 330-kDa Tg), which were incubated with biotin-labeled heparin (0.1 g/ml), alone or in the presence of various concentrations of Tg peptide 1 or of intact Tg, respectively, after overnight preincubation. The ability of Tg peptide 1 to inhibit binding of intact Tg to megalin or to heparin was assessed by using wells coated with intact Tg, which were incubated with megalin or biotin-labeled heparin alone, or with megalin or biotin-labeled heparin preincubated overnight at 4°C with Tg peptide 1 or with the control peptide at various concentrations. Similar experiments were also performed to test the ability of several heparin-binding peptides/protein (lactoferrin, lipoprotein lipase, protamine, and polylysine, all at a concentration of 10 M) to inhibit binding of biotin-labeled heparin to intact Tg. Heparin binding experiments were also performed to compare the ability of Tg peptide 1 with that of intact Tg to inhibit binding of biotin-labeled heparin to lactoferrin, lipoprotein lipase, protamine, and polylysine. For this purpose, microtiter wells coated with 100 g/ml of heparin-binding peptides/proteins (mean amounts of coated proteins: lactoferrin, 1.6 g/well; lipoprotein lipase, 1.6 g/well; polylysine, 2 g/well; protamine, 1.9 g/well) were incubated with biotin-labeled heparin, alone or in the presence of intact Tg (100 nM), of Tg peptide 1 (47 M), or, as controls, of the control peptide (47 M) or of OVA (10 M).
The heparin binding abilities of Tg peptide 1 and of intact Tg were also tested by affinity chromatography, as described previously (33). Intact Tg or Tg peptide 1 were applied to heparin-agarose (Sigma) columns (0.5 ϫ 2.5 cm) equilibrated with 50 mM NaCl, 10 mM NaH 2 PO 4 , pH 7.4. Columns were then extensively washed with equilibration buffer, followed by elution with a 20-ml linear NaCl gradient (50 mM to 1.2 M). Fractions of 1 ml were collected, and the protein absorbance was measured with a spectrophotometer (at 280 nM for intact Tg and at 225 nM for Tg peptide 1). The amount of protein was calculated using standard curves obtained by serial dilutions of a known concentration of intact Tg or of Tg peptide 1. The molarity of the fractions was analyzed by measurement of their conductivity, based on a standard curve obtained by serial dilution of a buffer containing 1.2 M NaCl, 10 mM NaH 2 PO 4 .
Development and Use of a Rabbit Antibody against Tg Peptide 1-A rabbit antiserum against Tg peptide 1 was raised by Cocalico (Reamston, PA), and the anti-Tg peptide 1 antibody was immunoaffinity purified using agarose beads (Pierce) coupled with Tg peptide 1. The ability of the anti-Tg peptide 1 antibody to recognize intact Tg and the 210-and 50-kDa Tg polypeptides or Tg peptide 1 was tested by ELISA. For this purpose, ELISA well microtiter plates were coated with Tg peptide 1, with the control peptide, with intact Tg, or with Tg fractions containing pure 210-kDa Tg polypeptide or pure 50-kDa polypeptide at various concentrations followed the anti-Tg peptide 1 antibody at various dilutions, ALP-conjugated goat anti-rabbit IgG secondary antibody (1:3000), and p-nitrophenyl phosphate. The ability of the anti-Tg peptide 1 antibody to recognize intact Tg was also tested by Western blotting. For this purpose 10 g of intact Tg were subjected to polyacrylamide gel electrophoresis after boiling in Laemmli buffer and blotted onto nitrocellulose membranes, which were incubated with the anti-Tg peptide 1 antibody (1:200) followed by horseradish peroxidaseconjugated goat anti-rabbit IgG.
To investigate whether the anti-Tg peptide 1 antibody could recognize intact Tg in its native conformation, we performed immunoprecipitation experiments. The anti-Tg peptide 1 antibody (100 g) was incubated overnight at 4°C with 25 l of protein A-agarose beads (Amersham Pharmacia Biotech). Beads were repeatedly washed and then added to intact Tg (125 g in 500 l), followed by incubation overnight at 4°C. The beads were extensively washed, resuspended in nonreducing Laemmli buffer, boiled, and centrifuged at 14,000 rpm. The supernatant was subjected to SDS-polyacrylamide gel electrophoresis and Western blotting, which was performed using a horseradish peroxidase-conjugated mouse anti-Tg antibody from Dako Corporation (Carpinteria, CA), diluted 1:200. As a negative control we used protein A beads coupled with normal rabbit IgG. To calculate the amount of intact Tg precipitated by the anti-Tg peptide antibody, the pixel density of the bands obtained by Western blotting was measured in scanned images using a personal computer software (NIH Imager 2.1). As a reference value we used the pixel density of the bands produced by loading 10 g of intact 660-and 330-kDa Tg.
To test the ability of the anti-Tg peptide 1 antibody to inhibit binding of biotin-labeled heparin to intact Tg, ELISA solid phase binding assays were performed as described above. In inhibition experiments, wells coated with intact Tg were co-incubated with biotin-labeled heparin in the presence of the anti-Tg peptide 1 antibody (100 g/ml) or, as a control, of normal rabbit IgG (100 g/ml). To test the ability of the anti-Tg peptide 1 antibody to inhibit binding of megalin to intact Tg, ELISA solid phase binding assays were performed as described above.
In inhibition experiments, wells coated with intact Tg were preincubated with the anti-Tg peptide 1 antibody (100 g/ml) or, as a control, with normal rabbit IgG (100 g/ml), followed by incubation with purified megalin (5 g/ml) in the presence of the anti-Tg peptide 1 antibody (100 g/ml) or, as a control, of normal rabbit IgG (100 g/ml).

RESULTS
Composition of the Rat Thyroid Extract-Rat thyroids were homogenized, and the thyroid extract, obtained by ammonium sulfate precipitation, was subjected to column size fractionation. As shown in Fig. 1A, Coomassie staining revealed the presence of four polypeptides, at the following estimated molecular masses: 660, 330, 210, and 50 kDa. The first two polypeptides were eluted together in fractions from 130 to 168 ml, and corresponded to intact Tg. The 210-kDa polypeptide was eluted from 169 to 231 ml, and the 50-kDa polypeptide was eluted from 232 to 288 ml.
The immunoreactivity of the four polypeptides with a rabbit anti-human Tg antibody was evaluated by ELISA and Western blotting. Fractions corresponding to all four polypeptides were immunoreactive by ELISA, although the reactivity of the fractions corresponding to the 50-kDa polypeptide was relatively weak (not shown). As shown in Fig. 1 (B and C), the 660-, 330-, and 210-kDa polypeptides were immunoreactive by Western blotting, whereas the 50-kDa polypeptide was not.
Binding of the Tg Polypeptides to Megalin-To investigate the megalin binding capacity of the four polypeptides, solid phase binding assays were performed as described previously (18). As assessed by ELISA, megalin bound to the fractions corresponding to the 210-kDa polypeptide to an even greater extent than to those corresponding to intact Tg ( Fig. 2A). Binding of megalin to fractions corresponding to the 50-kDa polypeptide was also observed, although ligand blot binding assays revealed binding of megalin only to the three larger polypeptides (Fig. 2B). No binding to OVA coated wells or to blotted OVA was observed.
To investigate whether the 210-kDa polypeptide competes with intact Tg for binding to megalin, microtiter wells coated with fractions corresponding to intact Tg (660 and 330 kDa) were incubated with megalin alone or in the presence of fractions containing the 210-kDa polypeptide or the 50-kDa polypeptide. As shown in Fig. 2C, the 210-kDa polypeptide reduced megalin binding by ϳ70%, whereas no inhibition was produced by the 50-kDa polypeptide. No inhibition was produced by OVA, which was used as a control (not shown).
Binding of Heparin to the 210-kDa Polypeptide-To investigate whether the 210-kDa Tg polypeptide contains heparinbinding sites, solid phase binding experiments were performed. ELISA wells coated with intact Tg with fractions containing the 210-kDa polypeptide or the 50-kDa polypeptide were incubated with biotin-labeled heparin or, as control, with biotin-  labeled albumin. Biotin-labeled heparin (Fig. 3), but not biotinlabeled albumin (not shown), bound both to intact Tg and to the 210-kDa Tg polypeptide to similar extents, whereas no binding to the 50-kDa polypeptide or to OVA was observed. When wells coated with intact Tg were incubated with biotin-labeled heparin plus fractions containing the 210-kDa polypeptide, binding was reduced by ϳ60%, whereas no appreciable inhibition was produced by the 50-kDa polypeptide or by OVA. Furthermore, an excess of unlabeled heparin almost completely inhibited binding of biotin-labeled heparin to the 210-kDa Tg polypeptide, whereas no inhibition was produced by OVA.
The 210-kDa Polypeptide Corresponds to the Carboxyl-terminal Portion of Rat Tg-To identify the portion of Tg to which the 210-kDa polypeptide corresponds, we characterized the polypeptide by combining site-specific proteolysis with MALDI-MS (23)(24)(25)(26). Following 6% nonreducing SDS-polyacrylamide gel electrophoresis of a fraction containing the 210-kDa polypeptide, the 210-kDa band was cut out and subjected to in-gel proteolytic digestion with the protease Lys-C. The observed molecular masses of the peptides generated by proteolysis were compared with the masses of computer generated theoretical Lys-C digests of the known sequences of the aminoterminal 213 amino acids and of the carboxyl-terminal 967 amino acids of rat Tg, the only portions of rat Tg cDNA whose sequences have been reported (Ref. 27, GenBank TM accession number 1729963, and Ref. 28, GenBank TM accession number 1729964).
The theoretical Lys-C digest of the amino-terminal portion of rat Tg produced 6 peptides greater than 1000 Da. Because cysteine residues are known to cross-link to acrylamide moieties during electrophoresis and proteolysis, peptides containing cysteine are altered in their mass. Therefore, of the 6 peptides in the theoretical Lys-C digest of the amino-terminal sequence of rat Tg, 5 were eliminated from the analysis because they contain cysteine. The mass of the remaining peptide did not correspond to any of the observed masses of the Lys-C digest of the 210-kDa Tg polypeptide.
The theoretical Lys-C digest of the carboxyl-terminal portion of rat Tg produced 18 peptides greater than 1000 Da. Of these 18 peptides, 5 were eliminated from the analysis because they contain cysteine. As shown in Table II, following MALDI-MS analysis of the 210-kDa Tg polypeptide, the masses of 8 of the remaining 13 peptides in the theoretical digest of the known sequence of the carboxyl-terminal portion of rat Tg corre-sponded to the observed masses of the Lys-C digest of the 210-kDa Tg polypeptide. Furthermore, 2 of the 8 matching peptides were derived from the extreme carboxyl-terminal portion (Ala 878 -Lys 898 and Glu 899 -Lys 911 ) of rat Tg. Based on its molecular mass, we estimated that the 210-kDa Tg polypeptide consists of approximately 1800 amino acid residues. Thus, the results obtained by MALDI-MS analysis indicate that the 210-kDa Tg polypeptide represents the carboxyl-terminal twothirds of monomeric Tg.
A 15-Amino Acid Sequence in the Carboxyl-terminal Portion of Tg Is Involved in Heparin Binding-Heparin binding to certain ligands of other members of the low density lipoprotein receptor family is dependent on the presence in their sequence of clusters of positively charged amino acids (arginine and lysine) (29 -32). We identified a 15-amino acid sequence in the carboxyl-terminal portion of rat Tg with 6 arginine and lysine residues, corresponding to Arg 689 -Lys 703 (RELPSRRLKRP-LPVK), containing a putative Cardin and Weintraub (29) heparin-binding consensus sequence: SRRLKRP. To investigate this further, a synthetic peptide (Tg peptide 1) corresponding to the 15-amino acid sequence was developed, as well as a control peptide in which the four inner lysine and arginine residues were substituted with glycine (Table I). Solid phase assays were performed to investigate the heparin binding ability of the synthetic Tg peptide 1. As shown in Fig. 4A, biotin-labeled heparin was found to bind to Tg peptide 1, whereas there was no binding of biotin-labeled heparin to the control peptide. There was no binding of biotin-labeled albumin (not shown), which was used as a control. Binding of biotin-labeled heparin to Tg peptide 1 was almost completely inhibited by an excess of unlabeled heparin (not shown). Binding of biotin-labeled heparin to Tg peptide 1 was inhibited when biotin-labeled heparin was added to wells coated with Tg peptide 1 in the presence of various concentrations of Tg peptide 1 itself, with a mean calculated constant of dissociation (K d ) of ϳ12.5 M (Fig. 4B). The affinity of Tg peptide 1 was roughly 250-fold lower than that obtained for intact Tg (mean K d , ϳ47.3 nM) (Fig. 4B). We also tested the strength of binding of intact Tg and of Tg peptide 1 to heparin-agarose columns. Following loading to the column, bound proteins were eluted with a linear NaCl gradient. As shown in Fig. 4C, both intact Tg and Tg peptide 1 were retained on the column and eluted by NaCl, indicating that they had been bound to heparin-agarose. Similar amounts of Tg (81.7% of the amount applied to the column) and of the synthetic Tg peptide (80.5% of the amount applied to the column) were eluted by NaCl. However, in confirmation of the results obtained in solid phase binding assays, the molar amount of NaCl necessary to release ligands from immobilized heparin was greater for intact Tg (ϳ502 mM at protein peak) than for Tg peptide 1 (ϳ305 mM at protein peak), indicating a greater avidity of intact Tg for heparin binding.
To further investigate the strength of heparin binding of Tg peptide 1, we compared its ability to inhibit heparin binding to intact Tg with that of several other heparin-binding peptides/ proteins. As shown in Fig. 5A, Tg peptide 1 inhibited binding of biotin-labeled heparin to intact Tg to a lower extent than lactoferrin, lipoprotein lipase, polylysine, and protamine. No inhibition of binding of biotin-labeled heparin to intact Tg was produced by the control peptide or by OVA, which were used as controls. As shown in Fig. 5B, binding of biotin-labeled heparin to intact Tg was completely inhibited by Tg peptide 1 when it was added at a concentration of 250 M. The calculated mean constant of inhibition (K i ) was 13.6 M, which is consistent with the above reported K d value of heparin-binding to Tg peptide 1, calculated in direct heparin binding assays (Fig. 4B).
We then investigated whether Tg peptide 1 would inhibit

thirds of rat Tg by combining site-specific proteolysis with matrixassisted laser desorption/ionization mass spectrometry (MALDI-MS)
A thyroid extract was subjected to column size fractionation and a fraction corresponding to the 210-kDa Tg polypeptide (number 62, eluate from 184 to 186 ml) was subjected to in-gel proteolytic digestion with the protease Lys-C. The observed molecular masses of the peptides generated by proteolysis were compared to the masses of a theoretical, computer generated, Lys-C digest of the known sequence of the carboxyl-terminal 967 amino acids of rat Tg. Peptides containing cysteine were excluded from the analysis. The masses in Daltons of eight matching peptides are shown.  4. A, biotin-labeled heparin binding to Tg peptide 1. A peptide matching a 15-amino acid sequence in the carboxyl-terminal portion of rat Tg, containing six positively charged residues (arginine and lysine) was synthesized (Tg peptide 1), as well as a control peptide in which the four inner lysine and arginine residues were substituted with glycine. 96-well microtiter plates were coated with intact Tg, with Tg peptide 1, or with the control peptide at a concentration of 100 g/ml and incubated with biotin-labeled heparin (0.1 g/ml) followed by ALP-conjugated streptavidin. Results are expressed as the means Ϯ S.E. ng of heparin bound obtained in three experiments, normalized for the mean amount of coated protein (intact Tg, 1.5 g/well; Tg peptide and control peptide, 1.7 g/well). B, inhibition of biotin-labeled heparin binding to intact Tg and to Tg peptide 1. Microtiter plates were coated with intact Tg or with Tg peptide 1 at a concentration of 100 g/ml and incubated with biotin-labeled heparin (0.  5 g). B, inhibition of biotin-labeled heparin binding to intact Tg by Tg peptide 1. Experiments were performed as in A, using Tg peptide 1 or the control peptide at various concentrations. Results are expressed as ng of heparin bound, normalized for the mean amount of coated Tg. The figure is representative of one of three experiments. C, inhibition of biotin-labeled heparin binding to lipoprotein lipase, lactoferrin, polylysine, or protamine by intact Tg but not by Tg peptide 1. Microtiter wells were coated with 100 g/ml of lipoprotein lipase, lactoferrin, polylysine, or protamine and incubated with biotin-labeled heparin (0.1 g/ml), alone or in the presence of intact Tg (100 nM) or of Tg peptide 1 (47 M). Results are expressed as mean Ϯ S.E. ng of heparin bound obtained in three experiments, normalized for the mean amount of coated proteins (lipoprotein lipase, 1.6 g/well; lactoferrin, 1.6 g/well; polylysine, 2 g/well; protamine, 1.9 g/well).
1, a rabbit antibody against Tg peptide 1 was used. As determined by ELISA, this antibody recognized Tg peptide 1, intact Tg, and fractions corresponding to the 210-kDa Tg polypeptide (not shown), supporting the conclusion that the 210-kDa Tg polypeptide corresponds to the carboxyl-terminal portion of rat Tg. In contrast, the anti-Tg peptide 1 antibody did not react with the control peptide nor with the 50-kDa Tg polypeptide (not shown). Furthermore, as shown in Fig. 6A the anti-Tg peptide 1 antibody recognized 660-and 330-kDa Tg by Western blotting. This result indicates that recognition of the Tg epitope corresponding to Tg peptide 1 by the antibody is not likely to be mainly dependent on its conformation in the native protein, because Western blotting experiments were performed under denaturing conditions (boiling of the protein in SDS), whereby native conformation is lost.
To investigate whether the anti-Tg peptide 1 antibody can recognize native, intact Tg, we performed immunoprecipitation experiments. As shown in Fig. 6B, intact 660-and 330-kDa Tg were precipitated by protein A beads coupled with the anti-Tg peptide 1 antibody but not by protein A beads coupled with normal rabbit IgG, which was used as a control. The mean pixel density of the bands obtained (660-kDa Tg band ϩ 330-kDa Tg band) was 44.8 pixels/cm 2 . Because the mean pixel density obtained by loading 10 g of purified Tg was 48.2 pixels/cm 2 and the amount of Tg used in immunoprecipitation experiments was 125 g, we calculated that the mean amount of intact Tg precipitated was 7.44% of the initial amount. The results indicate that the anti-Tg peptide 1 antibody can recognize intact Tg in its native conformation and, therefore, that the sequence corresponding to Tg peptide 1 lies in the external part of rat Tg.
The anti-Tg peptide 1 antibody was also used in heparin binding inhibition experiments. As shown in Fig. 6C, when biotin-labeled heparin was added to wells coated with intact Tg in the presence of the anti-Tg peptide 1 antibody, heparin binding was markedly reduced (by ϳ70%), whereas no effect was produced by normal rabbit IgG, which was used as a negative control. The results support the conclusion that the sequence corresponding to Tg peptide 1 is a Tg heparin-binding region.
Role of Charge in Heparin Binding to Tg Peptide 1-As reported above, the heparin binding affinity of intact Tg was much greater than that of Tg peptide 1, suggesting that other heparin-binding regions may be present in the Tg molecule. By further analysis of the primary structure of the known sequence of rat Tg, we identified a second heparin-binding consensus sequence (ARRTRG), corresponding to Ala 801 -Gly 806 . Another 15-amino acid synthetic Tg peptide (Tg peptide 2) was produced, corresponding to Met 796 -Ile 810 (MASLWARRTRGN-VFI), which included this second heparin-binding consensus sequence (Table I). As shown in Fig. 7, no appreciable binding of biotin-labeled heparin to this peptide was seen as compared with Tg peptide 1.
To study to what extent charge and sequence are responsible for the heparin binding ability of Tg peptide 1 several "mutant" synthetic peptides were developed (Table I). As shown in Fig. 7, biotin-labeled heparin bound to a synthetic peptide in which the 15 amino acid residues of Tg peptide 1 were randomly shuffled (scrambled peptide). The extent of binding of biotinlabeled heparin to the scrambled peptide was ϳ80% of binding of Tg peptide 1. Four mutant peptides were also used in which one of the four inner arginine and lysine residues in the sequence of Tg peptide 1 was substituted with glycine (mutants 1-4). As shown in Fig. 7, these single amino acid substitutions resulted in a dramatic reduction of the heparin binding ability of the 15 amino acid sequence of Tg peptide 1. The extent of heparin binding to these four mutant peptides ranged from 20 to 33% of binding to Tg peptide 1.
The Tg Sequence Corresponding to Tg Peptide 1 Is Involved  FIG. 6. A, a rabbit antibody raised again Tg peptide 1 recognizes intact Tg by Western blotting. 10 g of a preparation containing 660-and 330-kDa Tg were subjected to polyacrylamide gel electrophoresis under denaturing conditions (the protein was boiled in Laemmli buffer) and transferred onto a nitrocellulose membrane that was incubated with the anti-Tg peptide 1 antibody followed by horseradish peroxidase anti-rabbit IgG. The figure is representative of one of three experiments. Arrows indicate the bands corresponding to 660-and 330-Da Tg. B, immunoprecipitation of intact Tg with the anti-Tg peptide 1 antibody. A preparation containing 660-and 330-kDa Tg was incubated with protein A beads coupled with the anti-Tg peptide antibody (lane 2) or, as a control, with normal rabbit IgG (lane 3), followed by Western blotting, which was performed using a horseradish peroxidase-conjugated mouse anti-Tg antibody. In lane 1 the Tg preparation (10 g) not subjected to immunoprecipitation was loaded. The figure is representative of one of three experiments. Arrows indicate the bands corresponding to 660-and 330-kDa Tg. C, inhibition of heparin binding to intact Tg by the anti-Tg peptide 1 antibody. Wells coated with 100 g/ml of intact Tg were incubated with biotin-labeled heparin (0.1 g/ml), alone, or in the presence of the anti-Tg peptide 1 antibody, or, as a control, of normal rabbit IgG (RIgG), followed by ALP-conjugated streptavidin. Results are expressed as the means Ϯ S.E. ng of heparin bound obtained in three experiments, normalized for the mean amount of coated intact Tg (1.5 g/well).
in Binding to Megalin-To investigate whether the heparinbinding sequence of Tg corresponding to Tg peptide 1 is also involved in megalin binding, we first tested the megalin binding ability of Tg peptide 1. As shown in Fig. 8, megalin did not bind to wells coated either with Tg peptide 1 or with the control peptide. Furthermore, no inhibition of megalin binding to Tg was produced by Tg peptide 1 or the control peptide. However, our previous findings that heparin inhibits binding of megalin to intact Tg and dissociates megalin-bound Tg (18,19) indicate that heparin and megalin-binding sites of Tg are closely related. Therefore, we investigated further the possible involvement of the Tg sequence corresponding to Tg peptide 1 in megalin binding. To this purpose we used the rabbit antibody raised against Tg peptide 1 in megalin binding inhibition experiments. As shown in Fig. 8, when wells coated with intact Tg were preincubated with the anti-Tg peptide 1 antibody followed by megalin plus the anti-Tg peptide 1 antibody, megalin binding was almost completely inhibited (by ϳ90%), whereas no effect on megalin binding to Tg was produced by normal rabbit IgG, which was used as a control. The results indicate that the Tg region corresponding to Tg peptide 1 is involved in binding to megalin. DISCUSSION In the present study we have identified a heparin-binding region in the carboxyl-terminal portion of rat Tg and have obtained evidence that it is closely related to a megalin-binding region. To characterize Tg-binding regions we used rat thyroid extracts separated by column size fractionation and, in confirmation of previous studies (6), found that they contain four major Tg polypeptides, with estimated molecular masses of 660, 330, 210, and 50 kDa. In an extension of our earlier studies (18,19), we showed that megalin bound not only to intact Tg but also to the 210-kDa Tg polypeptide, and to an even greater extent than to intact Tg. In addition, the 210-kDa Tg polypeptide inhibited binding of intact Tg to megalin by ϳ70%, indicating that it includes a major Tg-binding site for megalin. Furthermore, heparin bound to the 210-kDa Tg polypeptide, and co-incubation with the 210-kDa polypeptide substantially reduced (by ϳ60%) heparin binding to Tg, indicating that the 210-kDa Tg polypeptide contains heparin-binding sites. MALDI-MS analysis provided evidence that the 210-kDa Tg polypeptide represents the carboxyl-terminal two-thirds of rat Tg, and in support of this, a rabbit antibody raised against a synthetic peptide deduced to be in the carboxyl-terminal portion of rat Tg recognized the 210-kDa Tg polypeptide as well as intact Tg.
We postulated that even though intact Tg is anionic (pIϳ4.6), it interacts with heparin through regions containing clusters of positively charged amino acids (arginine and lysine), as previously shown for the majority of heparin-binding proteins (29 -32). By analysis of the published partial sequences of rat Tg (27,28), we identified a 7-amino acid sequence (SRRLKRP) in the carboxyl-terminal portion, corresponding to Ser 693 -Pro 699 , with characteristics of a "consensus sequence" for heparin binding, as defined by , namely XBBXBBX, where B is a basic amino acid and X is any nonbasic/nonacidic amino acid. We obtained a synthetic peptide (Tg peptide 1) corresponding to a 15-amino acid sequence in the carboxyl-terminal portion of rat Tg (Arg 689 -Lys 703 : RELPSR-RLKRPLPVK) that contains the heparin-binding consensus sequence and showed that biotin-labeled heparin binds to this peptide but not to a 15-amino acid control peptide in which the four inner arginine and lysine residues were replaced by glycine.
The heparin binding affinity of Tg peptide 1 was lower than FIG. 7. Binding of biotin-labeled heparin to Tg peptides 1 and 2 and effects of amino acid modifications in Tg peptide 1 sequence on its heparin binding ability. Tg peptide 1 was synthesized by the Massachusetts General Hospital Peptide-Protein Core. Microtiter wells were coated with 100 g/ml of Tg peptide 1, of Tg peptide 2, of a scrambled peptide in which the 15 amino acid residues of Tg peptide 1 were randomly mixed, or of four mutant peptides in which one of the four inner arginine and lysine residues in the sequence of Tg peptide 1 was substituted with glycine. The sequences of these peptides are reported in Table I. After coating, wells were incubated with 0.1 g/ml of biotin-labeled heparin followed by ALP-conjugated streptavidin. Results are expressed as the means Ϯ S.E. ng of heparin bound obtained in three experiments, normalized for the amount of coated peptides (1.6 -1.8 g/well).

FIG. 8.
Binding of megalin to intact Tg, but not to Tg peptide 1 and inhibition of megalin binding to intact Tg by the anti-Tg peptide 1 antibody. ELISA wells coated with intact Tg, with Tg peptide 1 or with the control peptide, all at a concentration of 100 g/ml, were incubated with megalin (5 g/ml) followed by 1H2, as in Fig. 2. In inhibition experiments with Tg peptide 1 megalin was added to wells coated with intact Tg in the presence of Tg peptide 1 (100 g/ml) or of the control peptide (100 g/ml). In inhibition experiments with the anti-Tg peptide 1 antibody, wells coated with intact Tg were preincubated with the anti-Tg peptide 1 antibody (100 g/ml), followed by incubation with megalin in the presence of the anti-Tg peptide 1 antibody (100 g/ml). Normal rabbit IgG (RIgG) was used as a negative control. Results are expressed as the means Ϯ S.E. ng of megalin bound obtained in three experiments, normalized for the mean amount of coated protein (intact Tg, 1.5 g/well; Tg peptide 1 and control peptide, 1.7 g/well).
that of intact Tg, as shown by calculations of the K d in solid phase assays (Tg peptide 1, 12.5 M; intact Tg, 47 nM) and by the finding that Tg peptide 1 bound to a heparin-agarose column to a lower extent than intact Tg. Furthermore, the extent of inhibition of heparin binding to Tg produced by Tg peptide 1 was lower than that produced by several known heparin-binding proteins or polypeptides (lipoprotein lipase, lactoferrin, protamine, and polylysine), and Tg peptide 1 was a weaker inhibitor than intact Tg of heparin binding to the above heparin-binding peptides/proteins. The apparently anomalous ability of Tg peptide 1 to reduce binding of heparin to intact Tg to a greater extent than the 210-kDa Tg polypeptide in our assays may be explained by the much smaller molar amounts of the peptide in the 210-kDa polypeptide. Thus, the 210-kDa polypeptide was used at a concentration of 0.476 M (100 g/ ml), with a concentration of the sequence corresponding to Tg peptide 1 of 0.0047 M, as compared with concentrations of Tg peptide 1 (used alone) up to 250 M.
The greater heparin binding affinity of intact Tg as compared with Tg peptide 1 suggests that other heparin-binding regions are present in the Tg molecule and/or that the native conformation of Tg is important in heparin binding. To pursue this, we further analyzed the two known partial sequences (27,28) of rat Tg and identified another putative heparin-binding consensus sequence (XBBXBX), from Ala 801 to Gly 806 (ARRTRG). We developed a second 15-amino acid synthetic Tg peptide (Tg peptide 2), corresponding to Met 796 -Ile 810 (MASLWAR-RTRGNVFI), which included this heparin-binding consensus sequence. However, we found no binding of biotin-labeled heparin to this peptide. No other heparin-binding consensus sequences could be identified in the two known partial sequences of rat Tg. To obtain information about the remaining sequences of Tg, we analyzed the complete sequence of mouse Tg (Ref. 34, GenBank TM accession number 008710), which is highly homologous to rat Tg, and identified only the same two heparinbinding consensus sequences found in rat Tg, corresponding to Tg peptide 1 and Tg peptide 2. Nevertheless, a contribution of other regions of Tg to its heparin binding affinity cannot be excluded. Thus, heparin-binding segments of proteins can be different from the Cardin and Weintraub model, and separate regions rich in positively charged amino acid residues can be brought together by the folding of the protein (32). The heparin binding affinity of intact Tg may thus depend in part on the conformation of the molecule. The optimal way to study how conformation affects the heparin binding ability of a protein is to analyze its three-dimensional structure, to locate regions rich in positively charged residues, and to determine whether they are brought together by the folding of the protein. However, this approach is not currently possible in the case of Tg, because its three-dimensional structure is not known.
Another way to study putative heparin-binding regions of a protein is by site-directed mutagenesis studies. However, this approach is unlikely to be feasible in the case of Tg. So far only Arvan and associates (35) have succeeded in transfecting a nearly full-length Tg gene in cultured cells. However, they found that point mutations in the carboxyl-terminal portion abolished secretion of the protein, because of its retention in the endoplasmic reticulum, thereby making it impossible to obtain preparations of purified mutated Tg. The above study supports the notion that certain Tg mutations described in humans, especially in the carboxyl-terminal portion, result in misfolding of the protein in the endoplasmic reticulum with impaired secretion by thyroid cells (7, 36 -38).
Our approach to the identification of a heparin-binding region, namely through the use of synthetic peptides, can be criticized because such peptides may have different conforma-tions from those of the corresponding segments in the native protein. This may be one reason why intact Tg has a greater heparin binding affinity than Tg peptide 1. Despite this reservation, evidence obtained in the present study supports the conclusion that a continuous sequence of the 15-amino acid Tg peptide 1 corresponds to an active heparin-binding region present on the surface of native, intact Tg. Thus, a rabbit antibody raised against Tg peptide 1 recognized native, intact Tg as shown by immunoprecipitation experiments, indicating that its antigenic determinants are present on the surface of the Tg molecule. Furthermore, this antibody recognized Tg by Western blotting, under denaturing conditions, where conformational epitopes are lost, suggesting that the epitopes with which the antibody reacts are probably not entirely conformational (39). In addition, the anti-Tg peptide 1 antibody markedly reduced heparin binding to Tg, indicating that the region recognized by the antibody is involved in heparin binding. However, the inhibition of heparin binding to intact Tg produced by the anti-Tg peptide 1 antibody was not complete (ϳ70%), which supports the conclusion that the Tg region corresponding to Tg peptide 1 is not entirely responsible for the heparin binding ability of intact Tg.
Our experiments performed with "mutants" of Tg peptide 1 provide evidence in favor of an important role of the charge of this sequence in determining its heparin binding ability. Thus, when the sequence of Tg peptide 1 was randomly mixed, the peptide still retained ϳ80% of its heparin binding capacity, whereas substitutions of any of the inner positively charged residues resulted in a striking reduction of heparin binding, down to ϳ20 -30% of the binding ability of the wild type sequence. The inability of Tg peptide 2 to bind heparin, even though it contains a putative heparin-binding consensus sequence, may be attributed to the lower number of positively charged residues as compared with Tg peptide 1 (three versus six positive residues). However, other factors, including conformation of the peptide and spatial relationship between positively charged residues, may also be important.
The main aim of the present study was to identify heparinbinding regions that are related to megalin-binding sites. As we have shown previously (18,19), heparin competes with megalin for Tg binding and can dissociate Tg from megalin. Therefore, we postulated that heparin-binding regions of Tg are closely related to megalin-binding sites. Although megalin did not bind to Tg peptide 1 and the peptide did not inhibit binding of megalin to intact Tg, the antibody raised against the synthetic Tg peptide 1 almost completely inhibited megalin binding to Tg, indicating that the Tg sequence corresponding to the peptide is involved in megalin binding. In this regard, previous studies dealing with proteins that bind to heparin and to cell surface receptors, including members of the low density lipoprotein receptor family, have provided evidence that heparin and receptor-binding sites lie adjacent to each other (40). For some of these proteins efficient binding and uptake by cell surface receptors requires binding to heparin-like molecules, notably heparan sulfate proteoglycans (40 -49). In these cases, binding is believed to occur via side-by-side binding sites for heparan sulfate proteoglycans and for the receptor. These considerations raise the possibility that Tg may interact with cell surface heparan sulfate proteoglycans. Indeed, heparan sulfate proteoglycans are expressed by thyroid cells in a TSH-dependent manner (50,51), and we have obtained preliminary evidence suggesting that heparan sulfate proteoglycans can facilitate binding of Tg to cultured thyroid cells. 2 Thus, the inhibition produced by the antibody against Tg peptide 1 on megalin binding to intact Tg may be explained by a steric effect. The finding that the carboxyl-terminal portion of rat Tg (210-kDa polypeptide), which bound very avidly to megalin, only partially inhibited binding of megalin to intact Tg suggests that other binding sites for megalin may be present in the amino-terminal portion of Tg. In this regard, multiple binding sites for megalin that are separate but functionally related to a heparin-binding domain have been previously demonstrated in the 39-kDa low density lipoprotein receptor-associated protein, also known as RAP, a surrogate megalin ligand (52).
The identification of a heparin-binding region of Tg that is also implicated in binding to megalin is of potential interest in the understanding of the pathogenesis of certain thyroid diseases. For example, mutations of the Tg gene resulting in amino acid substitutions are associated with certain sporadic or familial, congenital forms of goiter, and some of these mutations are located in the carboxyl-terminal portion of Tg (7,(35)(36)(37)(38). The mechanisms by which these mutations lead to a Tg accumulation in thyroid follicles and, consequently, to goiter, have not been completely clarified, although, as noted above, a defect in intracellular trafficking of Tg because of misfolding of the molecule in the endoplasmic reticulum is thought to be important (7,(35)(36)(37)(38). We postulate that modifications of the Tg structure may also result in impaired Tg endocytosis via megalin, thereby promoting excessive retention of Tg in the colloid. Clearly, additional studies are needed to further characterize the Tg-binding site for megalin and to investigate whether mutations of the Tg molecule may be responsible for altered storage of Tg as a consequence of impaired Tg-megalin interactions.