A discrete thyroxine-rich iodopeptide of 20,000 daltons from rabbit thyroglobulin.

Rabbit thyroids, labeled in vivo with 12'1, were quickly removed and processed under conditions designed to minimize postmortem proteolysis. These included gentle extraction at 2°C for 20 min, the use of pepstatin and phenylmethanesulfonyl fluoride, and rapid homogenization followed by boiling. The supernatants of boiled thyroid homogenates after reduction with mercaptoethanol showed several fast iodinated bands on gel electrophoresis in sodium dodecyl sulfate; of these the most prominent were 20,000 daltons (15.2% of total "'1 of homogenate supernatant), -31,000 daltons (6.8%), and -15,000 daltons (2.4%). These values for molecular mass and lz5I content were consistent among the thyroids of six rabbits processed individually within the same experiment. The 20,000and -15,000-dalton species had more than one-half of their "'I as thyroxine. The same three iodinated bands were seen with reduced extracts of thyroid, but not in unreduced samples. On basic gel electrophoresis of unreduced samples, no ''1 was found other than in 19 S thyroglobulin or its polymers. For bulk preparations, we pooled thyroids from several '2SI-injected rabbits with 250 glands obtained commercially, and isolated thyroglobulin by gel chromatography. The 20,000-dalton species was purified from reduced and alkylated thyroglobulin by gel filtration, DEAE-cellulose chromatography, and preparative gel electrophoresis. It had the following features: 1) a single band on analytical gel electrophoresis; 2) a single NH2-terminal residue, aspartic acid; 3) a ratio of "'I to I2'I which was over twice that of the parent thyroglobulin; 4) a carbohydrate content of approximately 10%; 5) 63% of its I2'I was in thyroxine, compared with a value of 19% for thyroglobulin; 6) it contained 39% of the '2SI-labeled thyroxine of thyroglobulin and 18% of the '271-labeled thyroxine (the high thyroxine content was confirmed by radioimmunoassay); and 7) its molar ratio to 660,000-dalton thyroglobulin was close to one. We conclude that rabbit thyroglobulin contains a discrete iodopeptide of 20,000 daltons which is not the product of postmortem proteolysis. It appears to be held to the larger part of the thyroglobulin molecule by disulfide bonds. Its high thyroxine content indicates an important role for it in the production and/or release of thyroid hormones.

mass and lz5I content were consistent among the thyroids of six rabbits processed individually within the same experiment. The 20,000-and -15,000-dalton species had more than one-half of their "'I as thyroxine. The same three iodinated bands were seen with reduced extracts of thyroid, but not in unreduced samples. On basic gel electrophoresis of unreduced samples, no ' ' ' 1 was found other than in 19 S thyroglobulin or its polymers.
For bulk preparations, we pooled thyroids from several '2SI-injected rabbits with 250 glands obtained commercially, and isolated thyroglobulin by gel chromatography. The 20,000-dalton species was purified from reduced and alkylated thyroglobulin by gel filtration, DEAE-cellulose chromatography, and preparative gel electrophoresis. It had the following features: 1) a single band on analytical gel electrophoresis; 2) a single NH2-terminal residue, aspartic acid; 3) a ratio of "'I to I2'I which was over twice that of the parent thyroglobulin; 4) a carbohydrate content of approximately 10%; 5) 63% of its I2'I was in thyroxine, compared with a value of 19% for thyroglobulin; 6) it contained 39% of the '2SI-labeled thyroxine of thyroglobulin and 18% of the '271-labeled thyroxine (the high thyroxine content was confirmed by radioimmunoassay); and 7) its molar ratio to 660,000-dalton thyroglobulin was close to one.
We conclude that rabbit thyroglobulin contains a discrete iodopeptide of 20,000 daltons which is not the product of postmortem proteolysis. It appears to be held to the larger part of the thyroglobulin molecule by disulfide bonds. Its high thyroxine content indicates an important role for it in the production and/or release of thyroid hormones.
* This work was supported by a grant (AM1 1043) from the National Institutes of Health. Part of this work was presented at the annual meeting of the Endocrine Society, June, 1979, Anaheim, CA, and published in its abstracts. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Thyroglobulin, a thyroidal glycoprotein of approximately 660,000 daltons, is the matrix on which the thyroid hormones thyroxine and triiodothyronine are formed. Their synthesis occurs after iodination of tyrosyl residues to form the precursors 3-iodotyrosine and 3,5-diiodotyrosine, and this process appears to take place wholly within the thyroglobulin molecule (1). Several lines of evidence suggest that the molecular structure of thyroglobulin is important to hormone formation. These include thyroglobulin's greater ability, relative to that of other proteins, to form thyroxine from iodotyrosines on in vitro iodination (I), and the finding of a limited amino acid environment surrounding thyroxine in iodopeptide fragments produced by protease digestion in vitro (2).
An understanding of thyroglobulin's role in hormone synthesis demands a detailed description of its structure, particularly that of its subunits. This has been difficult to obtain because of the protein's large size, the existence of established differences among animal species, and the occasional presence of contaminating proteolytic activity during isolation procedures. In the present paper, using rabbit thyroids, we have selected preparative methods which minimize postmortem proteolysis. We have found three iodopeptide fractions of small molecular size, two of which have most of their iodine in the iodothyronines. One of these has been isolated as a discrete glycopeptide of 20,000 daltons, and some of its properties are described.

EXPERIMENTAL PROCEDURES
Gel Electrophoresis-Analytical gels were run in a pH 8.9 buffer or in 0.1% SDS' a t porosities ranging from 4% to 14% as previously described (3,4). Additional gels were run in 0.1% SDS and 6 M urea (5). A number of molecular weight markers were used, most commonly myoglobin, human growth hormone, ovalbumin, bovine serum albumin, and its cross-linked polymer (Sigma Chemical Co.). Gels were stained with Coomassie blue R-250 for SDS gels or G-250 for disc gels (3), and RF values were calculated relative to a bromphenol blue tracker dye band. For most of the work with '"I-labeled samples, the gels were not stained but instead were immediately sliced into uniform segments of 2.5 mm or 5.0 mm and counted for radioactivity. Faster bands with high counts could be eluted from the gels with water and used for further analyses. For reduction of samples for SDS gels, we placed them in a solution which was 1% mercaptoethanol and 1% SDS, either for 16 h a t room temperature or for 3 to 5 min in boiling water. Preparative electrophoresis was done on gel slabs (12 X 16 X 0.9 cm) of 8% polyacrylamide in a chamber cooled with tap water (E-C Co.). Samples of 15 mg or less were dissolved in 1% SDS containing glycerol and bromphenol blue, then placed in four large wells, up to 0.2 ml Der well. and electrophoresed at 250 mA in an electrode buffer of 0.1% SDS.
Beckman well type counter. For determination of Iz7I, we first digested samples with 28% chloric acid at an initial temperature of 115°C (6), then diluted them with 0.1 N arsenious acid in 1 N H&04 and analyzed them in a Technicon AutoAnalyzer system by catalysis of ceric sulfate (7). Standards of potassium iodide were carried through the entire procedure including digestion, and the iodine content of samples was calculated from them. For the distribution of I2'I among the iodoamino acids, we digested protein or peptide fractions with pronase (Calbiochem), separated their iodoamino acids by descending paper chromatography in butanolethanoll N NH,OH (5:1:2), and counted for radioactivity, as previously described (2). Occasionally, additional chromatograms were run in a butanol:2 N acetic acid system (2). In calculating the distribution of I2'I among the iodoamino acids, we excluded the small amounts found at the origin and as iodide, attributing these, respectively, to undigested iodopeptides and deiodinated iodoamino acids. Together these were usually less than 10% of the 1251 on the chromatogram.
The thyroxine content of the 20,000-dalton iodopeptide was also assessed by radioimmunoassay after pronase digestion, using a double antibody system similar to that of Chopra (8), described in detail elsewhere (9).
Amino Acid Analyses-Samples were hydrolyzed in constant boiling HC1 at 105°C for 36 and 72 h in sealed, evacuated tubes, and measured on an automatic amino acid analyzer (Dionex, Sunnyvale, CA) connected to an integrator (Supergrator 1, Columbia Scientific Co.). The recommended Dionex program was slightly altered to identify glucosamine and tryptophan. For tryptophan we hydrolyzed in 3 N mercaptoethanesulfonic acid (10) instead of HCI.
NHt-Terminal Analyses-Samples were reacted with dansyl chloride, followed by hydrolysis in constant boiling HCl a t 105OC in sealed, evacuated tubes. We identified the dansyl amino acids by two-and three-dimensional thin layer chromatography, using the method of Percy and Buchwald ( l l ) , a s previously described (3).
Other Analyses-The protein content of samples was measured by the method of Lowry et al. (12) using bovine serum albumin as standard. We estimated the total carbohydrate content by the alkaline ferricyanide method of Krystal and Graham (13). using their correction factors for hydrolytic destruction. Cathepsin D activity was measured by the release from hemoglobin of peptides soluble in trichloroacetic acid (14).
Disulfide Bond Cleavage-Samples were reduced with mercaptoethanol, 50 mol per mol of disulfide bond, at 25°C for 4 h, and alkylated with acrylonitrile (15), 2 mol per mol of mercaptoethanol, for 30 min a t 25"C, pH 8.0. The reaction was stopped with mercaptoethanol and excess reagents removed by dialysis against water or by gel fdtration. In the samples used for amino acid analyses, we alkylated with iodoacetic acid (16) or oxidized thyroglobulin with performic acid (17).
Animals-All isotopic experiments were conducted with young female New Zealand White rabbits, maintained on a standard diet of laboratory chow containing approximately 0.8 pg of iodine per g.
Preparation of Thyroid Extracts-Three rabbits were killed 6 days after each received an intraperitoneal injection of 0.5 mCi of camer-free NaIz5I. The thyroids were removed quickly on ice and cut into three slices per lobe. The pooled slices were placed either into 0.06 M sodium phosphate buffer, pH 7.0, or into the same buffer plus the proteolytic enzyme inhibitors pepstatin M) and phenylmethanesulfonyl fluoride M), and extracted at 2°C for 20 min, followed by centrifugation a t 122,000 X g for 35 min at 4°C. Portions of the extract were placed immediately on polyacrylamide gels for electrophoresis. Other aliquots were fwst made 1% SDS, with or without mercaptoethanol, and run on gels in SDS.
Preparation of Boiled Thyroid Homogenates-Six rabbits were killed 3 days after each received an intraperitoneal dose of 0.5 mCi of Na'*'I. Each thyroid was removed quickly, cut into three slices per lobe, added to 1 ml of a buffer which was 0.05 M sodium phosphate, pH 7.0, 10"' M phenylmethanesulfonyl fluoride, and lo-' M pepstatin, homogenized with three strokes of a loose fitting glass homogenizer on ice, then placed for 5 min in boiling water, followed by centrifugation at 122,000 X g for 35 min. Portions of each supernatant were analyzed by polyacrylamide gel electrophoresis in SDS with mercaptoethanol.
Bulk Preparation of 1251-labeled Thyroglobulin-Seven rabbits were each given 1.1 mCi of NaI2'I intraperitoneally and killed 3 days later. Their thyroids were removed, pooled, homogenized in 0.05 M sodium phosphate, pH 7.0, and centrifuged. The supernatant was placed on a column (1.5 X 90 cm) of Bio-Gel A-5m a t 4°C in the same phosphate buffer made 0.02% with sodium azide. Thyroglobulin eluted as a single peak of I2'I. Fractions containing it were pooled, dialyzed, and lyophilized. We prepared 12'I-thyroglobulin from animals of the same sex and breed (Pel-Freez, Rogers, AR) by slicing and homogenizing 250 frozen thyroids and separating thyroglobulin on a similar column of A-5m (4.4 X 90 cm) followed by dialysis, lyophilization, and pooling with the I2'II-labeled thyroglobulin. The A-5m gave a wide separarion of the thyroglobulin peak from that of cathepsin D activity, and the isolated thyroglobulin showed no detectable cathepsin D activity.
In addition, we made two other preparations of '"I-labeled thyroglobulin by a similar procedure, but with the labeled and unlabeled thyroids pooled with each other prior to homogenization.
Isolation of 20K"Thyroglobulin labeled with Iz5I was reduced and alkylated, and its components partially purified by gel filtration on a column of Bio-Gel A-5m in SDS (Fig. 1   The peak of lZ5I at tube 95 in Fig. 1 contained most of the 20K as well as smaller amounts of the -15,000-dalton iodopeptide and unidentified heavy components. The fractions comprising this peak (the shaded area in Fig. 1) were pooled, dialyzed against water for 24 h, lyophilized, dissolved in NH4HC0:3 in 6 M urea, 3.0 mmho, pH 8.8, and further fractionated on a column of DEAE-cellulose (Whatman DE52) (Fig. 2). From analytical gels, the material eluting at 8.8 mmho had most of its stain for protein in the 20,000-dalton zone, with traces of both heavier and lighter components. Fractions comprising this peak were pooled, dialyzed, and lyophilized. An aliquot was then dissolved in 2% SDS/ 1% mercaptoethanol and further fractionated by preparative gel electrophoresis. A center strip was cut from this preparative gel, and 5mm segments were counted for Iz5I (Fig. 3), showing a major band in the 20,000-dalton area (the shaded area in Fig. 3). The remainder of this band was then homogenized in water, centrifuged, and the supernatant dialyzed and lyophilized.
We used this preparation of 20K, referred to as Preparation I, for the analytical studies recorded below, unless otherwise specified. Preparation I1 was isolated in a similar manner from the same thyroglobulin sample alkylated with iodoacetic acid instead of acrylonitrile. It was used particularly for analyses of half-cystine and glutamic acid, since S-cyanoethyl cysteine elutes with glutamic acid on the amino acid analyzer. Preparations I11 and IV were made from S-cyanoethylated thyroglobulin by methods similar to those of Preparation I, but with labeled and unlabeled thyroid homogenates pooled before thyroglobulin isolation. Preparation V was oxidized with performic acid prior to isolation of 20K.

Thyroid Extracts-On
basic gels of 4% polyacrylamide, thyroid extracts with or without inhibitors showed over 85% of the '''1 in a slow band (Fig. 4, left panel). This band was (mm)

FIG. 3.
Purification of 20K by preparative gel electrophoresis. The partially purified material from Fig. 2 (4.1 mg) was dissolved in a solution which was 2% SDS and 1% mercaptoethanol, and electrophoresed at 250 mA on a gel slab (16 X 12 X 0.9 cm) of 8% polyacrylamide in 0.1% SDS. A 1-cm longitudinal strip was cut from the center and 5-mm horizontal segments counted for Ig5I. The ordinate scale is radioactivity per gel segment. The shaded area was identified as 20, OOO daltons by its mobility. T , I I identical in position with that of 19 S thyroglobulin, run on parallel gels and identified by analytical ultracentrifugation in a previous study (2). The only other "'I-containing band corresponded to 27 S thyroglobulin. There was no iodination of albumin or other small molecules.
On 4% gels in SDS, most of the 1251 was at the origin or in a slow band of -290,000 daltons (Fig. 4, center panel). When the sample was reduced with mercaptoethanol, several new bands appeared, including one of -20,000 daltons (8.2% of the total "'1 on the gel), -15,000 daltons (2%), and -37,000 daltons (1.6%) (Fig. 4, right panel). A similar pattern was seen for the extract without enzyme inhibitors, in which the 20,000-dalton band contained 8.1% of the IZ5I, the 15,000-dalton peak 1.1%, and the -37,000-dalton peak 3.2%.
Boiled Homogenates of Thyroid-The supernatants from homogenates of each of the six rabbit thyroids showed five discrete peaks of '"I on 4% gels in SDS after mercaptoethanol (Fig. 5 ) . Gels of 8% in SDS, both with and without 8 M urea, were also run to provide sharper peaks for the low molecular weight components. Table I shows that the six preparations were similar to one another in distribution of lZ5I among these peaks. The molecular sizes of the individual '"I-containing bands were estimated from standards of known molecular weight run on parallel gels, and were quite reproducible among the six homogenate supernatants. For Band B, taken from the 4% gels, the mean estimated molecular weight (k S.E.) was 233,000 k 6,000. Bands C, D, and E, from the 8% gels, were, respectively, 32,000, 21,000, and 15,000, all with S.E. values less than 1,000.
The distribution of Iz5I among the iodoamino acids of Bands A to E from an individual rabbit is shown in Table 11. The smaller bands contained much more of their as thyroxine than the larger ones, a finding confirmed in other experiments.
The thyroxine values shown for Band E are probably artifactually low. In another experiment with more of the -15,000dalton material available, chromatography in a butano1:acetic acid system separated an iodopeptide which had co-migrated with 3-iodotyrosine in the butano1:ethanol:ammonia system. The I2'I distribution of the -15,000-dalton iodopeptide in this latter experiment was 21% as diiodotyrosine, 8% as 3-iodotyrosine, 67% as thyroxine, and 3% as triiodothyronine. This distribution is representative of most other determinations of the -15,000-dalton species. We do not have an explanation for the high triiodothyronine content obtained in Band E of Table 11. Isolation of20K" Fig. 6 shows gels at various steps in the purification of 20K for Preparation I. The A-5m column after reduction and alkylation removed excess reagents and most of the heavy components of thyroglobulin. The DEAE-cellulose separated 20K from a noniodinated material of similar

Distribution of among five bands (A to E) from electrophoresis of reduced supernatants of boiled thyroid homogenates on gels of
I%polyacrylamide in 0.1% SDS Data from six rabbits are presented individually. Table entries   molecular size and from the -15,000-dalton iodopeptide. Preparative gel electrophoresis also separated 20K from the -15,000-dalton and -31,000-dalton species. In the DEAEcellulose step of Preparation 111, we eluted 20K a t a conductivity of 7.0 mmho, and on analytical gel electrophoresis found almost all of the stain in one major band a t 20K with several faint bands visible on overloaded gels. For many purposes, isolation with the A-5m and DEAE-columns is satisfactory and avoids the limitations and losses of the preparative gel step. Properties of 20K"Analytical electrophoresis of the purified 20K of Preparation I showed a single heavy band of 20K (Fig. 6d). Very faint traces of several slower bands were occasionally present in this and the other three preparations.
In all preparations there was only one band of l2.'I, corresponding exactly to the stained 20K band.
Dansylated 20K showed a single NH2-terminal residue, aspartic acid. This result was confirmed in Preparation I11 as well. Measurements of the ' "1 peak of 20K on SDS gels of 6%, 8%, lo%, 12%, and 14% polyacrylamide gave molecular weight estimates of 19,800, 21,300, 19,200, 20,000, and 21,200, respectively, when compared with protein markers on parallel gels. Preparation IV on a similar series of gels gave a mean value of 20,000. Gels of 4% polyacrylamide gave somewhat higher values, in keeping with the glycopeptide nature of 20K (see below) (18).
Composition of 2OK-Using the conditions and correction factors of Krystal and Graham (13) for hydrolytic destruction, we found the total carbohydrate content to be 10.0% by weight. This value is close to that of ihtact thyroglobulin (2). This method does not measure sialic acid, which represents 1.7% of the residue weight of rabbit thyroglobulin (2), and
Entries are the contribution, by percentage, of each iodoamino acid to the total 1251 of the band.

Thyroxine-rich Iodopeptide from Thyroglobulin
does not distinguish the relative contributions of individual monosaccharides.   Table IV were taken from the sum of its amino acids, iodine, and monosaccharides. We also eluted 20K from 8% gels and ran it in the Lowry reaction, correcting for the considerable background eluted from blank gels run in parallel. By this procedure the 20K gave 45% of the color value of bovine serum albumin or thyroglobulin, in keeping with its low content of tyrosine and tryptophan.
Using the data of Table IV    acid to the total '"1 of 20K or of thyroglobulin.
(' Entries are the contribution, by percentage, of each iodoamino ronine. Actual values for thyroxine may have been higher, since its ratio of ' 9 to lZ7I is usually lower than that of the iodotyrosines for a number of days after injection of Na'"1. From analytical gels, we found that 20K contained 11.8% of the total '"I of thyroglobulin in this preparation. Thus, we can derive from Table IV that it contained 39% of the ''51labeled thyroxine of thyroglobulin and 18% of the unlabeled thyroxine. From these data we also calculate that 20K contributed 3.1% of the weight of thyroglobulin. For Preparation 111, 20K represented 2.1% of thyroglobulin's weight, and for Preparation IV, 3.6%.
In Preparation IV we confirmed the chromatographic identification of thyroxine by radioimmunoassay. This gave a value of 20.2 nmol of thyroxine per mg of 20K compared with 33.3 calculated from data similar to those of Table IV for Preparation IV. For comparison, the parent thyroglobulin for this preparation contained 4.6 nmol of thyroxine per mg of protein by radioimmunoassay, and 5.2 nmol per mg calculated from the specific activity data.

DISCUSSION
Our study demonstrates that reduced rabbit thyroglobulin contains several iodinated components of small molecular weight and that they are not the products of postmortem proteolysis. Quantitatively, the most important of these components is the 20,000-dalton species. Its high thyroxine content and its high specific activity of iodine imply a preferred site for thyroxine formation, and although it comprises only about 3% of the protein's weight, it contains over one-third of the newly formed hormone. We have found a similar thyroxineenriched peptide of 20,000 to 25,000 daltons in other species including rat and man.' The other two small iodinated bands of thyroglobulin, -31, C!OO daltons and -15,000 daltons, have not yet been extensively characterized. The very high iodothyronine content of the -15,000-dalton component suggests that it is physiologically important.
It is not clear how these iodopeptides, particularly 20K, relate to the parent thyroglobulin. Thyroglobulin is isolated from the thyroid principally as a 19 S form of 660,000 daltons, which can dissociate into 12 S half-molecules. There is controversy over whether the two 330,000-dalton half-molecules are identical (19,20). The 3% of thyroglobulin's weight found in 20K would represent approximately 1 mol of 20K per mol of 660,000-dalton thyroglobulin. We can consider several possibilities for interpretation of this relationship. ( a ) There may be 1 mol of 20K in one of two nonidentical 330,000-dalton half-molecules. ( b ) 20K may be a separate component of the 660,000-dalton protein in addition to, rather than as part of, the two half-molecules; its attachment to the remainder of thyroglobulin could occur by disulfide formation during iodination (21), a post-translational event. ( c ) There may be variability in the synthesis of thyroglobulin, with only some of the molecules containing 20K; thyroglobulin has already been found heterogeneous in its composition of amino acids in two other species (4, 22) and even in samples isolated from different parts of the same thyroid (23). (cl) Finally, 20K may represent an early step in the physiological, ie. antemortem, proteolysis of thyroglobulin. Further work is necessary to permit a choice among these or other possibilities.