De novo triiodothyronine formation from thyrocytes activated by thyroid-stimulating hormone

The thyroid gland secretes primarily tetraiodothyronine (T4), and some triiodothyronine (T3). Under normal physiological circumstances, only one-fifth of circulating T3 is directly released by the thyroid, but in states of hyperactivation of thyroid-stimulating hormone receptors (TSHRs), patients develop a syndrome of relative T3 toxicosis. Thyroidal T4 production results from iodination of thyroglobulin (TG) at residues Tyr5 and Tyr130, whereas thyroidal T3 production may originate in several different ways. In this study, the data demonstrate that within the carboxyl-terminal portion of mouse TG, T3 is formed de novo independently of deiodination from T4. We found that upon iodination in vitro, de novo T3 formation in TG was decreased in mice lacking TSHRs. Conversely, de novo T3 that can be formed upon iodination of TG secreted from PCCL3 (rat thyrocyte) cells was augmented from cells previously exposed to increased TSH, a TSHR agonist, a cAMP analog, or a TSHR-stimulating antibody. We present data suggesting that TSH-stimulated TG phosphorylation contributes to enhanced de novo T3 formation. These effects were reversed within a few days after removal of the hyperstimulating conditions. Indeed, direct exposure of PCCL3 cells to human serum from two patients with Graves' disease, but not control sera, led to secretion of TG with an increased intrinsic ability to form T3 upon in vitro iodination. Furthermore, TG secreted from human thyrocyte cultures hyperstimulated with TSH also showed an increased intrinsic ability to form T3. Our data support the hypothesis that TG processing in the secretory pathway of TSHR-hyperstimulated thyrocytes alters the structure of the iodination substrate in a way that enhances de novo T3 formation, contributing to the relative T3 toxicosis of Graves' disease.

In the body of vertebrate animals, thyroglobulin (TG) 2 is the primary (if not exclusive) original source of thyroid hormones (1) that regulate central nervous system development and function, oxidative metabolism, thermogenesis, and body weight regulation, heart rate, cardiac output, LDL cholesterol levels, and other phenotypes (2,3). The thyroid gland produces virtually 100% of the supply of L-thyroxine (T 4 ) from the body. However, other than nongenomic actions (4), the main physiological effects of thyroid hormones are brought about by gene expression changes as a consequence of 3,3Ј,5-triiodo-L-thyronine (T 3 ) interaction with nuclear thyroid hormone receptors (5).
Depending upon the species and conditions, there are somewhat differing views about the main sources of circulating T 3 . In otherwise normal thyroidectomized rats that are fully replaced with exogenous levothyroxine (i.e. normal serum T 4 ), circulating T 3 is decreased ϳ55% (6) indicating a significant thyroidal contribution to circulating T 3 . In normal humans, classic studies have estimated that only ϳ21% of daily T 3 production is derived from thyroidal secretion (the rest coming from deiodination of T 4 to T 3 by deiodinases D1 and D2) (7). However, in patients with untreated Graves' disease (a disease of thyroidal hyperstimulation by TSH receptor-stimulating antibodies (8)), thyroid tissue is markedly enriched in T 3 concurrent with increased T 3 in the circulation (9,10). Although some increased thyroidal T 3 production in Graves' disease might be derived from intrathyroidal deiodination of T 4 to T 3 (11), the aforementioned study of untreated Graves' patients reported increased thyroid tissue T 3 only after Pronase digestion (7). Moreover, mice with whole body D1/D2-double knock-out (DKO) nevertheless maintain normal circulating T 3 levels (12). Taken This work was supported, in whole or in part, by National Institutes of Health Grant R01 DK40344 (to P. A.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 1 To whom correspondence should be addressed: Brehm Tower, Rm. 5112, 1000 Wall St., Ann Arbor, MI 48105. E-mail: parvan@umich.edu. 2 The abbreviations used are: TG, thyroglobulin; PTU, propylthiouracil; T 3 , triiodothyronine; T 4 , thyroxine; TSH, thyroid stimulating hormone; TSHR, TSH receptor; D1, type 1 deiodinase; D2 type 2 deiodinase; DKO, D1/D2 double knockout; DIT, diiodotyrosine; MIT, mono-iodotyrosine; BisTris, 2-[bis(2hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol; qPCR, quantitative PCR.
cro ARTICLE together, these findings strongly imply that the thyroid gland has the capability to contribute importantly to circulating T 3 via a mechanism involving de novo T 3 formation, and this may be particularly important in Graves' disease.
The role of TG (a large homodimeric glycoprotein with a monomer molecular mass of 330 kDa and containing Ͼ2745 residues) in thyroid hormone synthesis is initiated upon its iodination (13,14). Iodination is catalyzed by thyroid peroxidase, which provides the necessary oxidation to form diiodotyrosine (DIT) and monoiodotyrosine (MIT) within TG. Favored by these same oxidizing conditions, a coupling reaction involving a DIT acceptor residue and a corresponding DIT donor residue allows for the formation of T 4 within the TG polypeptide; similarly, coupling of an MIT donor with a DIT acceptor allows for de novo T 3 formation (15,16). Classic studies report that thyroid peroxidase shows no marked specificity in its ability to catalyze TG iodination and coupling over that of lactoperoxidase or myeloperoxidase (16), whereas efficient T 4 and T 3 formation requires the TG substrate in its native conformation (17). Furthermore, despite ϳ70 Tyr residues distributed broadly along the length of the protein, T 4 and T 3 formation are restricted to relatively few sites in TG, including an evolutionarily preferred DIT-DIT coupling of Tyr 130 -Tyr 5 to yield T 4 at position 5 (1) and a preferred T 3 formation site at position 2746 of human TG (2744 of mouse TG, although the MIT coupling partner in either species remains unclear) (18,19).
During its complex trafficking through the intracellular transport pathway of thyrocytes, TG undergoes considerable post-translational processing prior to its secretion and iodination (20). Many of these post-translational modifications are regulated indirectly by TSH-induced changes in the gene expression and activity of TG processing enzymes (21). Herein, we have examined de novo T 3 formation within TG analyzed both from in vivo samples and after iodination in vitro. We directly demonstrate de novo T 3 formation in TG and establish that this ability is directly related to the degree to which thyrocytes have been exposed to prior TSHR stimulation.

De novo formation of T 3 within TG
We developed a simple assay to detect the presence of T 3 formation within thyroidal protein of euthyroid mice by immunoblotting using a mAb that recognizes T 3 when contained within the TG protein backbone, in parallel with immunoblotting with a polyclonal antibody against TG. The addition of free T 3 (half-maximal concentration ϳ75 ng/ml) eliminated the immunoblotted mouse TG band with mAb anti-T 3 , whereas the addition of free T 4 had little effect (Fig. 1A, blots at left; quantitation at right). Thyroidal immunofluorescence from euthyroid mice with mAb anti-T 3 was distributed primarily in the follicle lumen where extracellular TG resides in "colloid"; this immunofluorescent signal was fully blocked by the addition of free T 3 at 500 ng/ml but was only slightly diminished by free T 4 even at 2000 ng/ml (Fig. 1B). Routinely, we added 500 ng/ml of free T 4 in all immunoblotting experiments designed to detect T 3 in TG, to ensure specificity. The data in Fig. 1 indicate that a basal level of T 3 in TG is present under euthyroid (i.e. not TSHhyperstimulated) conditions.
The primary sequence of mouse TG encodes a single predicted site for Factor Xa cleavage that excises the N-terminal one-third of the protein from the C-terminal two-thirds. As Dunn et al. (18) reported that the majority of T 3 synthesized within TG is located in its C-terminal portion, we extracted mouse thyroid tissue and found by immunoblotting that a single band of intact Tg was cleaved to two fragments of the expected molecular mass after incubation with Factor Xa (Fig.  1C, left). Of these, the larger fragment corresponding to the C-terminal portion of TG was selectively enriched in T 3 (Fig.  1C, right).
There are strong cell biological arguments to suggest that T 3 contained within the TG protein does not result as a consequence of deiodination of T 4 contained within TG. Specifically, almost all of the hormone-containing-TG is localized to the extracellular thyroid follicular lumen (Fig. 1B) (22), whereas the catalytic activities of the two enzymes responsible for T 4 to T 3 conversion (deiodinases D1 and D2) are topologically facing the cytosol (23). Indeed, we examined the T 3 content of TG in the thyroids of animals devoid of D1 and D2 (12) and found that D1/D2-DKO mice had as much or more T 3 contained within TG than in that from a wild-type reference animal (Fig. 2, left). All of the detected signal was derived from bona fide proteinbound T 3 as it was completely blocked by the addition of free T 3 added to the immunoblotting conditions (Fig. 2, right). As a negative control, TG obtained from PTU-treated mice contained no detectable T 3 (Fig. 2 left). Together, these data provide strong support that de novo T 3 formation within TG is independent of T 4 deiodination.

TG secreted from thyrocytes after stimulation of TSHR is intrinsically more competent for de novo T 3 formation
Iodination of TG is enhanced by TSH stimulation of the thyroid gland (21). However, independent of effects on sodium iodide symport activity, dual-function oxidase activity, or thyroid peroxidase activity, it has been hypothesized that the intrinsic ability of TG to form T 3 may be modulated through prior thyroidal stimulation by TSH on its receptor (thyroidstimulating hormone receptor (TSHR) (24)). To test this, we subjected the TG obtained from thyroids of TSHR-KO mice, and from PTU-treated WT mice (that develop hypothyroidism and a compensatory increase in TSH leading to stimulated TSHRs) to iodination under fixed conditions in vitro. In addition to eliminating any effects of TSHR stimulation on thyrocyte enzymes that promote TG iodination in vivo, the in vitro iodination method also normalizes for TG protein, and total protein, within each iodination reaction. Prior to the in vitro iodination reaction, neither thyroids from TSHR-KO nor PTUtreated mice exhibited detectable T 3 within TG, as judged both by mAb anti-T 3 immunofluorescence and immunoblotting (Fig. 3, A and B). This is precisely why, for this experiment, we did not examine TG from control mouse thyroid glands. After iodination in vitro, TG from TSHR-KO consistently exhibited less de novo T 3 formation within TG (Fig. 3B, quantified in C).
To directly test the effect of thyrocyte stimulation by TSH on the ability of TG to facilitate de novo T 3 formation, we collected Thyroglobulin in T 3 toxicosis secretion from the rat thyrocyte cell line, PCCL3. Ordinarily, these cells are grown in the presence of TSH at a concentration of 1-10 milliunits/ml (25,26). Under standard cell culture conditions, PCCL3 cells cannot iodinate their own secreted proteins; however, upon collection of the secretion followed by iodination in vitro, de novo T 3 formation was readily detectable within TG (Fig. 4A). We cultured PCCL3 cells for several days in two concentrations of TSH that differed by an order of mag-nitude (10 and 100 milliunits/liter). TG secreted from PCCL3 cells after exposure to the higher TSH concentration consistently exhibited greater potential for T 3 formation upon iodination in vitro (Fig. 4B, quantified in graph below the blot). Recently, a specific small molecule agonist of the TSHR, MS437, was described to activate G␣ s , which mimics TSH in turning on downstream target genes (27). Upon growth of PCCL3 cells in the presence of MS437, in vitro iodination of were loaded in every other lane, resolved by reducing SDS-PAGE and electrotransfer to nitrocellulose, and the membrane cut into strips. Each membrane strip was incubated with mAb anti-T 3 (1:1000) plus an increasing concentration of either free T 3 or free T 4 . The strips were then incubated simultaneously with identical HRP-conjugated goat anti-mouse antibody followed by enhanced chemiluminescence substrate, and then examined together in a single 20-s digital image exposure. Quantitation of the T 3 signal within the TG band (gel shown at left; graph shown at right) represents the mean Ϯ S.D. from 3 independent experiments. B, immunofluorescence with the same mAb anti-T 3 (1:200) localizes T 3 -containing protein to the thyroid follicle lumen, and the immunofluorescence signal is specifically blocked upon addition of 500 ng/ml of free T 3 . All images are the same magnification; the yellow bar in each panel ϭ 50 m; nuclei are counterstained with DAPI. C, there is a single Factor Xa cleavage site in mouse TG that forms two predicted fragments. TG from normal mouse thyroid lysates Ϯ digestion with Factor Xa was resolved by SDS-PAGE (4 g of protein per lane), electrotransfer, and immunoblotting with anti-TG or anti-T 3 antibodies. Two cleavage products of the expected size were detected with anti-TG; a polypeptide consistent with the C-terminal fragment (asterisk) was enriched in T 3 . Three independent experiments yielded results identical to those shown here. The position of a pre-stained molecular weight marker is indicated.

Figure 2. Evidence for de novo T 3 formation within TG, independent of T 4 deiodination.
Gels at left: thyroid tissue homogenates from five different D1/D2-DKO mice, or wild-type C57BL/6 treated with PTU (WT PTU) to inhibit iodination in vivo, were analyzed by SDS-PAGE (4 g of protein/lane), electrotransfer, and immunoblotting with anti-T 3 (above) or anti-TG (below). Gels at right: the same samples immunoblotted in the presence of 500 ng/ml of free T 3 . Wild-type (WT) TG was run in lane 9 as a 330-kDa molecular mass marker.

Thyroglobulin in T 3 toxicosis
secreted TG led to a marked increase in de novo T 3 formation (Fig. 4C, quantified in graph above the blot). Stimulatory G proteins in thyrocytes activate adenylyl cyclase, which catalyzes cAMP synthesis (28). Upon incubation of PCCL3 cells in the presence of dibutyryl cAMP, in vitro iodination of secreted TG again demonstrated increased de novo T 3 formation (Fig. 4D, quantified in the graph above the blot). These data strongly suggest that TSH, working through TSHR, stimulatory G proteins, and cAMP production, alters the intrinsic ability of TG to form T 3 upon TG iodination.
There are several significant post-translational modifications that may have structural consequences on TG synthesized in TSH-stimulated thyrocytes, which may impact its hormonogenic potential, including de novo T 3 formation. These include the following: increased N-linked glycosylation (29) promoted in part by stimulation of oligosaccharyl transferase activity (30) and up-regulation of N-acetylglucosaminyltransferase 1 (31); increased complex sugars added in the thyrocyte Golgi complex including galactose (32); a marked decrease in the level of ␣2,6-bound sialic acid (33); down-regulation of tyrosine sulfation (34); increased formation of dityrosine cross-bridges (35) that can form non-disulfide-linked covalent TG dimers (36) and also increased TG phosphorylation (37), which can occur within TG carbohydrate, phosphotyrosine, and phosphoserine residues.
Recently, Fam20C was identified as a secretory pathway kinase that phosphorylates hundreds of secreted proteins, with a marked preference for Ser residues within the consensus sequence Ser-X-Glu/phospho-Ser (38). We observed a TSH dose-dependent 3-fold stimulation of Fam20C mRNA levels in PCCL3 cells (Fig. 5A). To examine the potential contribution of increased Fam20C mRNA on de novo T 3 formation in TG, we eliminated this increase using siRNA knockdown of Fam20C in TSH-stimulated PCCL3 cells (Fig. 5B). Upon in vitro iodination of secreted TG, TSH-stimulated cells with knockdown of Fam20C showed significantly less de novo T 3 formation (Fig.  5C, quantified in D). Moreover, treatment with calf intestinal phosphatase to dephosphorylate TG secreted from TSH-stimulated PCCL3 cells decreased de novo T 3 formation in TG by 26% (data not shown). Taken together, these data suggest that TG phosphorylation (37) is one of the TSH-stimulated posttranslational modifications that contribute to altering the structure of TG to increase de novo T 3 formation within TG.
Graves' disease involves thyroidal overactivity leading to thyrotoxicosis, with a particular predilection to increased circulating T 3 , i.e. relative T 3 toxicosis (9). To determine whether a TSHR-stimulating immunoglobulin could also promote enhanced de novo T 3 formation, PCCL3 cells were cultured for 5 days in the presence or absence of KSAb1, a strong TSHRstimulating immunoglobulin (39). After the culture period, serum-free secretion from PCCL3 cells was collected for the ensuing 24 h and the secreted proteins were iodinated in vitro. From the results shown in Fig. 6A, left (quantified at right), it was clear that TG secreted from PCCL3 cells that had been previously incubated with KSAb1 showed significantly increased de novo T 3 formation.
Thyrotoxicosis and complications of Graves' disease have been found to improve when the elevated concentrations of  Fig. 1B) from WT mice (Control), or mice treated for 6 weeks with PTU in vivo to block thyroidal iodination and induce hypothyroidism with high TSH, or mice bearing genetic deletion of TSHR (TSHR-KO). Both of the latter sets of animals lacked immunofluorescently detectable T 3 , which is present in the thyroid gland of control mice. B, thyroid tissue extracts from TSHR-KO mice or mice rendered hypothyroid with PTU, both before and after iodination in vitro, were resolved by SDS-PAGE (3 g of protein per lane), electrotransfer to nitrocellulose, and immunoblotting with mAb anti-T 3 and anti-TG. Wild-type TG was run in an adjacent lane as a 330-kDa molecular mass marker. C, quantitation of the relative T 3 /TG band intensity ratio from five independent experiments; mean Ϯ S.D.; *, p Ͻ 0.05 comparing thyroids from TSHR-KO to PTU-treated animals.

Thyroglobulin in T 3 toxicosis
circulating TSHR-stimulating immunoglobulins are reversed, for example, by plasmapheresis (40). To examine reversibility of enhanced de novo T 3 formation in TG, we washed out the KSAb1 and collected TG secreted over the next 24 h, or the subsequent 48 h thereafter. Upon iodination in vitro, it was apparent that beyond 24 h after washout of KSAb1, the subsequent secretion contained TG that had reverted back to a decreased ability for de novo T 3 formation (Fig. 6B, left, quantified at right).
One clinical assay to screen for TSHR-stimulating immunoglobulins involves measuring responses to antibodies contained within the sera of human Graves' patients, in a cultured rat thyrocyte cell line (41). With this in mind, we cultured PCCL3 cells for 5 days in the presence of control sera from 5 individuals without Graves' disease or from 2 patients with Graves' disease. After exposure to the human sera, serum-free secretion from PCCL3 cells was collected for the ensuing 24 h and the secreted proteins were iodinated in vitro. From the results shown in Fig. 7, culture with each of the 5 control sera resulted in subsequent TG secretion with approximately the same potential for de novo T 3 formation. By contrast, culture with unpurified sera from two Graves' disease patients both suggested an increase in the ability of the subsequently secreted TG to form T 3 upon iodination.
Finally, we collected secretion from cultured primary human thyrocytes that were obtained from normal thyroid tissue of patients undergoing total thyroidectomy for localized thyroid cancer. In monolayer culture in TSH-containing medium to which additional iodide was not added, there was no detectable T 3 found within TG secreted into the culture medium, regardless of the TSH dose (Fig. 8A). However, when the TSH concentration in the culture medium was increased by an order of magnitude, upon iodination in vitro, T 3 formation in the subsequently secreted TG was clearly apparent (Fig. 8B). Taken together, these data indicate that both human TSHR-stimulating immunoglobulins, and human thyrocytes responding to stimulation of TSHR, increase de novo T 3 formation in TG.

Discussion
Pioneering work of Dunn and others (15) helped lead to the identification of a few selected Tyr residues of the ϳ70 on the

Thyroglobulin in T 3 toxicosis
TG protein that are favored for T 4 and T 3 formation. Remarkably, within the huge TG polypeptide, the most frequent site of T 4 formation resides just 5 residues from the N terminus (42), whereas more than half of all T 3 in TG is formed just 3 residues from the C terminus (24). It is known that DIT located at Tyr 130 is the DIT "donor" to form T 4 at Tyr 5 , but the precise mechanism of T 3 formation at the C terminus has not been established (19), although it has been postulated to involve one of several potential upstream MIT donor residues in the TG polypeptide (43) including residue 2520 of human TG (44).
In this study, we have utilized a simple immunoblotting procedure to specifically identify T 3 within TG (Fig. 1A). Essentially all of the immunodetectable T 3 in TG from mouse thyroid tissue (Fig. 1B) resides within the C-terminal portion of the

Figure 6. Effects of TSHR-stimulating immunoglobulin (KSAb1) on de novo T 3 formation in TG.
A, left: PCCL3 cells were preincubated Ϯ mouse mAb KSAb1 TSHR-stimulating immunoglobulin (3 g/ml). Secreted TG was iodinated in vitro followed by SDS-PAGE (3 g of protein/lane), electrotransfer, and immunoblotting with mAb anti-T 3 and anti-TG, as indicated. Wild-type TG was run in an adjacent lane as a 330-kDa molecular mass marker. Right, quantitation of the relative T 3 /TG band intensity ratio from 6 independent experiments is shown; mean Ϯ S.D.; *, p Ͻ 0.05 compared with control lacking KSAb1 pretreatment. B, reversion of the stimulating effect of KSAb1. Left, the bathing media containing secreted TG was collected 1 day after removing KSAb1, and again for another 2 days after the first media collection. Secreted TG was iodinated in vitro followed by SDS-PAGE (2 g of protein/lane), electrotransfer, and immunoblotting with mAb anti-T 3 and anti-TG, as indicated. Right, quantitation of the relative T 3 /TG band intensity ratio from TG collected on days 2 ϩ 3, compared with day 1 (normalized to 1.0) in 6 independent experiments is shown; mean Ϯ S.D.; *, p Ͻ 0.05.

Thyroglobulin in T 3 toxicosis
molecule (Fig. 1C) and this immunoreactivity is independent of T 4 to T 3 conversion (Fig. 2). However, from TG secreted both in mouse thyroid glands (Fig. 3) and a (rat-derived) thyrocyte cell line (Fig. 4), the degree of prior stimulation of TSHRs exerts a strong influence over the efficiency of de novo T 3 formation upon TG iodination. Similar effects are observed either with high TSH itself (Fig. 4B), a TSHR agonist (Fig. 4C), a cAMP analog (Fig. 4D), or a TSHR-stimulating immunoglobulin (Fig.  6A). Moreover, these stimulating effects on de novo T 3 formation are entirely reversible within a day after removing the TSHR-stimulating immunoglobulin (Fig. 6B). Our data are consistent with the work of Fassler et al. (24), who proposed that TSH alters the utilization of hormonogenic sites on TG through changes in TG structure, which is presumed to be a consequence of altered TG post-translational processing.
Of the many potential TSH-stimulated changes in gene expression of TG post-translational processing enzymes, we have examined Fam20C, a novel secretory pathway kinase often referred to as "casein kinase" (38). Of the potential sites that are both predicted casein kinase (45) and canonical Fam20C targets, multiple sites are specifically conserved between rat, mouse, and human TG; three of these fall within the ChEL domain; and one of these (at position 2721 of human TG) has been directly established to be a phospho-Ser residue by mass spectrometry of human TG (46) and is close to a primary site of T 3 formation (24). Our current evidence suggests that TSHR stimulation can increase Fam20C mRNA levels within 48 h, and this increase contributes (along with other changes) to enhanced de novo T 3 formation within TG (Fig. 5).
The observation that TSHR stimulation up-regulates the efficiency of de novo T 3 formation upon TG iodination fits plausibly with the notion of a direct increase in de novo T 3 formation within TG in Graves' disease, thereby contributing to increased intrathyroidal and secreted T 3 in this condition (7,11), which contributes to a state of relative T 3 toxicosis (47,48). Indeed, when collecting the secretion from PCCL3 cells incubated with serum from two Graves' disease patients, the secreted TG showed a clearly increased predisposition to form T 3 upon iodination, in comparison to various control sera (Fig. 7). Moreover, TSHR activation also promotes enhancement of T 3 formation in TG from human thyrocytes, as directly demonstrated in cultures of normal human thyroid tissue hyperstimulated with TSH (Fig. 8).
Interestingly, from repeated experiments, we were unable to detect any demonstrable evidence that TSHR stimulation alters TG in a way that enhances its efficiency in de novo T 4 formation (data not shown), and this selectivity for de novo T 3 formation is consistent with previous reports (49). Of course, in Graves' disease, there is also a general increase of T 4 secretion that may be attributed to many TSHR-stimulated activities including increased iodide uptake, DUOX function, TPO function, TG synthesis, and endocytosis of colloid, to name but a few (50). More work is needed at both the TG structural level, and at the thyroid cell biological level, to understand how and why T 3 is formed preferentially at residue 2746 of human TG (2744 of mouse TG), and how TSHR stimulation selectively increases de novo T 3 formation. However, our data support the hypothesis that the TSHR-stimulated effects reflect alterations in TG posttranslational processing that impact structurally on the carboxyl-terminal region of TG molecules to enhance de novo T 3 formation.

Figure 7. Stimulation of de novo T 3 formation by serum from two human Graves' disease patients.
A, PCCL3 cells were incubated for 5 days in media containing 40% serum from two patients with Graves' disease (#6 and #7), or 5 euthyroid controls (#1-#5). After washing, TG secreted thereafter into serum-free media bathing the cells was iodinated in vitro and analyzed by SDS-PAGE (2 g of protein/lane), electrotransfer, and immunoblotting with anti-T 3 and anti-TG, as indicated. Wild-type TG was run in an adjacent lane as a 330-kDa molecular mass marker. B, quantitation of the relative T 3 /TG band intensity with the ratio set to a value of 1.0 for control serum #1. For all sera tested, secreted TG was analyzed in at least three independent cultures and iodination in vitro; mean Ϯ S.D.; *, p Ͻ 0.05 compared with controls. Figure 8. De novo T 3 formation in TG secreted from primary culture of human thyrocytes. Human thyrocytes were grown at two different TSH concentrations: 100 or 1000 milliunits/ml, as described under "Experimental procedures." A, TG secreted into serum-free media that was not iodinated in vitro served as a negative control; analyzed by SDS-PAGE (2 g of protein/lane), electrotransfer, and immunoblotting with mAb anti-T 3 and anti-TG, as indicated. B, TG secreted from human thyrocytes grown as in panel A was iodinated in vitro and analyzed as in panel A.

Animal thyroid tissues used in this study
Mice were housed and fed as per an approved institutional protocol. C57BL/6 (WT) mice were 8 -11 months old; D1/D2 double knock-out (DKO) mice (12) were 10 months old; TSHR-KO mice (52) were 3-5 months old. Where indicated, WT mice were fed low-iodide chow containing 0.15% PTU (Envigo) for 6 -15 weeks. Thyroid glands were lysed by sonication in RIPA buffer Ϯ SDS and containing a protease inhibitor mixture. Lysates were cleared at 12,000 ϫ g for 10 min at 4°C and total protein was determined by BCA or Bramhall assay (53).

Immunofluorescence with mAb anti-T 3
Fresh thyroid tissues from wild-type, PTU-treated, or TSHR-KO mice were immersion fixed with 10% formaldehyde and paraffin-embedded. Six-m tissue sections were de-paraffinized with Citrosolv, followed by antigen retrieval with Retrieve-All, and permeabilization with Triton X-100 (0.2%). First antibody incubation with mAb anti-T 3 (1:200) in TBS plus 0.2% Tween 20 and 3% BSA Ϯ free T 3 or free T 4 as described, was followed by a secondary antibody (noted above) diluted 1:5000 in the same buffer without hormone competitor. Images were captured using a ϫ40 oil objective in an Olympus Flu-oView 500 laser scanning confocal microscope.

Site-specific protease digestion of TG
Factor Xa protease has only one predicted cleavage site (Peptide Cutter, Expasy) in WT mouse TG, at amino acid 1036. Thyroid homogenate protein from C57BL/6 WT mice (140 g) in RIPA buffer was adjusted to 1% SDS, 5 mM DTT and boiled for 1 min. The sample was then diluted to 0.01% SDS, 0.05 mM DTT in a total volume of 100 l and digested with 8 l of Factor Xa (New England Biolabs) for 15 min at 25°C. Digestion was terminated by boiling in denaturing gel sample buffer containing 100 mM DTT for 5 min before SDS-PAGE.

PCCL3 cell culture and treatments
PCCL3 cells were cultured in DMEM/F-12 supplemented with 5% FBS plus penicillin/streptomycin and a four-hormone mixture containing 1 g/ml of insulin, 1 nM hydrocortisone, 5 g/ml of apo-transferrin, and 1 milliunit/ml of TSH.
For experiments with varying TSH concentrations, PCCL3 cells were seeded at 50,000 cells/well in 24-well plates. After 24 h, the cells were grown for 5 days in complete medium containing TSH at either 10 or 100 milliunits/liter. The cells were then washed in PBS and re-fed at the same TSH concentrations in serum-free media and cultured for 1 (100 milliunits/liter of TSH) or 3 days (10 milliunits/liter of TSH), leading to comparable amounts of secreted TG and total protein in the bathing media. For experiments studying the TSHR agonist MS437 (27), the PCCL3 cells were seeded as above and grown for 3 days in complete medium with 10 milliunits/liter of TSH plus 10 M MS437 or vehicle (DMSO, 0.1%). The cells were then re-fed under the same conditions in serum-free bathing media, which were finally collected either at 2 (MS437) or 3 days (vehicle). For experiments studying the effects of dibutyryl-cAMP, PCCL3 cells were seeded as above. After 24 h, the cells were grown in complete medium with 10 milliunits/liter of TSH Ϯ 0.1 mM dibutyryl cAMP. The media were changed on days 2 and 4, and the cells then re-fed under the same conditions in serum-free media, which were finally collected either at 2 (dibutyryl cAMP) or 3 days (negative control). For experiments studying the mouse mAb KSAb1 TSHR-stimulating immunoglobulin (39), the PCCL3 cells were seeded at 25,000 cells/well in 48-well plates. After 24 h, the cells were grown for 5 days in complete medium containing 10 milliunits/liter of TSH Ϯ purified KSAb1. The cells were then re-fed under the same conditions in serum-free media, which were finally collected either at 1 day (KSAb1-stimulated) plus further days in fresh medium (this is referred to as "KSAb1 washout" for cells that had previously been KSAb1-stimulated) or 3 days (negative control).
Human sera were collected from patients with Graves' disease, or controls, with their consent under an approved Institutional Review Board (IRB) research protocol. For treatment of PCCL3 cells with human sera, the cells were initially plated as in the KSAb1 experiments described above. After 24 h, the cells were grown for 5 days in complete medium containing 10 milliunits/liter of TSH ϩ 40% human serum. The cells were then re-fed with serum-free media containing 10 milliunits/liter of TSH and cultured for 24 h before collecting the media for TG analysis and iodination in vitro.

Fam20C effects in PCCL3 cells
PCCL3 cells were pretreated under basal conditions in complete media containing 10 milliunits/liter of TSH for 5 days and then plated at 100,000 cells/well in 12-well plates. For mRNA measurements, cells were then either continued in basal media or shifted to complete media containing either 100 or 1000 Thyroglobulin in T 3 toxicosis milliunits/liter of TSH for 48 h. RNA was purified, reversetranscribed, and qPCR was performed in duplicate using a StepOnePlus thermal cycler (Applied Biosystems) using SYBR Green and the following primers: Fam20C, 5Ј-gaggcacaatgcggagatag-3Ј and 5Ј-gaggcactctgcggaaatc-3Ј; and HPRT1, 5Ј-ctcatggactgattatggacaggac-3Ј and 5Ј-gcaggtcagcaaagaacttatagcc-3Ј. Data were analyzed by comparative C T (⌬⌬C T ). For knockdown of Fam20C, PCCL3 cells grown in complete media containing 1,000 milliunits/liter of TSH were plated at 100,000 cells/well in 12-well plates. After 24 h 10 nM Fam20C siRNA (ID304334, Origene) or scrambled duplex oligonucleotide (SR30004, Origene) were transfected using RNAiMAX according to the manufacturer's instructions. After 6 h, media were replaced with complete media containing 1000 milliunits/liter of TSH for 18 h. At that time, the cells were re-fed under the same conditions in serum-free bathing media, which were finally collected after 48 h. The bathing media were used for in vitro iodination of TG and the lysed cells were used for RNA purification followed by qPCR.

Primary human thyrocyte culture and treatments
Normal human thyroid tissue was obtained from a patient undergoing total thyroidectomy for thyroid cancer at the National Institutes of Health Clinical Center. The patient provided informed consent on an approved IRB research protocol and materials were received anonymously with approval of research activity through the Office of Human Subjects Research, National Institutes of Health. Primary thyrocytes were prepared and propagated as described (54).
Human thyrocytes were plated in DMEM, 10% FBS plus penicillin/streptomycin at 100,000 cells/well in 12-well plates. After 24 h, the cells were cultured for 1 day in serum-free medium plus 0.1% BSA, and after one further day, the cells were re-fed and cultured for 6 days with DMEM plus penicillin/ streptomycin and TSH at a concentration of 100 or 1000 milliunits/ml. These conditions were tested in two biological replicates, and results in the replicates were identical.

In vitro enzymatic iodination of TG
Iodination in vitro (55) included lactoperoxidase (30 ng/l), glucose (2 g/l), glucose oxidase (0.352 ng/l), 100 M NaI, and 50 -250 ng/l of thyroidal protein. Incubations were initiated with the addition of the glucose oxidase, incubated for 2 h at 37°C, and stopped by addition of gel sample buffer and boiling for 5 min. To roughly normalize the amount of TG protein in the samples being compared in each experiment, when required we diluted thyrocyte-secreted protein with a known quantity of protein from serum-free medium bathing 293T cells that do not express TG. However, in pilot experiments we found that normalizing the TG protein content may not be necessary as it did not affect the ratio of T 3 formed per unit TG.

Western blotting
Samples (2-4 g of total protein per lane) were subjected to SDS-PAGE under either reducing (Figs. 1-3) or nonreducing conditions (all other figures). Pre-stained molecular mass markers as well as WT TG (330 kDa) were run in lanes adjacent to the experimental samples. Electrotransfer to nitrocellulose was performed for 7 min at 20 V using the iBlot transfer apparatus (Invitrogen). Blocking was performed for 30 min at room temperature with 5% BSA in TBS plus 0.05% Tween 20 (TBS-T) and washed with TBS-T. Primary mouse mAb anti-T 3 was diluted at 1:1000 containing (unless otherwise indicated) 500 ng/ml of free T 4 (to eliminate any possibility of T 4 cross-reactivity) and incubated overnight at 4°C. Primary rabbit polyclonal anti-TG was diluted 1:5000 in 5% BSA/TBS-T and incubated for 1 h at room temperature. Species-specific HRPconjugated secondary antibodies (1:5000 dilutions in blocking buffer) were incubated for 30 min at room temperature. Bands were visualized using the WesternBright Sirius kit as directed by the manufacturer (Advansta). Images were captured in a Fotodyne work station with a digital camera; exposure times averaged 20 s for anti-T 3 blots and 5 s for anti-TG blots.

Quantitation of T 3 /TG band intensity ratio
Band intensities were quantified using ImageQuant 5.2 (Molecular Dynamics). The ratio of intensities of the bands corresponding to the T 3 immunoreactivity within the TG band to the direct TG immunoreactivity from polyclonal anti-TG within the same band was calculated, with the control value set to 1.0. In Fig. 5D, the mean of control values was set to 1.0.

Data analysis
Statistical analyses were done using unpaired Student's t test with two-tailed p value (Figs. 3-5B and 6) or by one-way ANOVA followed by Dunnett's test (Figs. 5, A and D, and 7). The level of significance for all statistical tests was set to p Ͻ 0.05. Statistical values were calculated with GraphPad Prism version 6. Data are presented as mean Ϯ S.D.